diff options
author | Todd Rinaldo <toddr@cpan.org> | 2011-12-23 20:17:58 +0000 |
---|---|---|
committer | Chris 'BinGOs' Williams <chris@bingosnet.co.uk> | 2011-12-23 21:30:46 +0000 |
commit | c05040198c1a12131b23c4530e9dbb3468d26f76 (patch) | |
tree | 360bac2a0eb81c5a35b5d73d2e658720cb7fbdca /dist | |
parent | 52c7aca67283e2d012cedda5e24ec7effb0dd0aa (diff) | |
download | perl-c05040198c1a12131b23c4530e9dbb3468d26f76.tar.gz |
Move Tie-File out of cpan/ and into dist/
Tie::File has not been changed on CPAN since 2003. It has meanwhile been
actively maintained in p5p.
Signed-off-by: Chris 'BinGOs' Williams <chris@bingosnet.co.uk>
Diffstat (limited to 'dist')
39 files changed, 8960 insertions, 0 deletions
diff --git a/dist/Tie-File/lib/Tie/File.pm b/dist/Tie-File/lib/Tie/File.pm new file mode 100644 index 0000000000..3862da2367 --- /dev/null +++ b/dist/Tie-File/lib/Tie/File.pm @@ -0,0 +1,2633 @@ + +package Tie::File; +require 5.005; +use Carp ':DEFAULT', 'confess'; +use POSIX 'SEEK_SET'; +use Fcntl 'O_CREAT', 'O_RDWR', 'LOCK_EX', 'LOCK_SH', 'O_WRONLY', 'O_RDONLY'; +sub O_ACCMODE () { O_RDONLY | O_RDWR | O_WRONLY } + + +$VERSION = "0.98"; +my $DEFAULT_MEMORY_SIZE = 1<<21; # 2 megabytes +my $DEFAULT_AUTODEFER_THRESHHOLD = 3; # 3 records +my $DEFAULT_AUTODEFER_FILELEN_THRESHHOLD = 65536; # 16 disk blocksful + +my %good_opt = map {$_ => 1, "-$_" => 1} + qw(memory dw_size mode recsep discipline + autodefer autochomp autodefer_threshhold concurrent); + +sub TIEARRAY { + if (@_ % 2 != 0) { + croak "usage: tie \@array, $_[0], filename, [option => value]..."; + } + my ($pack, $file, %opts) = @_; + + # transform '-foo' keys into 'foo' keys + for my $key (keys %opts) { + unless ($good_opt{$key}) { + croak("$pack: Unrecognized option '$key'\n"); + } + my $okey = $key; + if ($key =~ s/^-+//) { + $opts{$key} = delete $opts{$okey}; + } + } + + if ($opts{concurrent}) { + croak("$pack: concurrent access not supported yet\n"); + } + + unless (defined $opts{memory}) { + # default is the larger of the default cache size and the + # deferred-write buffer size (if specified) + $opts{memory} = $DEFAULT_MEMORY_SIZE; + $opts{memory} = $opts{dw_size} + if defined $opts{dw_size} && $opts{dw_size} > $DEFAULT_MEMORY_SIZE; + # Dora Winifred Read + } + $opts{dw_size} = $opts{memory} unless defined $opts{dw_size}; + if ($opts{dw_size} > $opts{memory}) { + croak("$pack: dw_size may not be larger than total memory allocation\n"); + } + # are we in deferred-write mode? + $opts{defer} = 0 unless defined $opts{defer}; + $opts{deferred} = {}; # no records are presently deferred + $opts{deferred_s} = 0; # count of total bytes in ->{deferred} + $opts{deferred_max} = -1; # empty + + # What's a good way to arrange that this class can be overridden? + $opts{cache} = Tie::File::Cache->new($opts{memory}); + + # autodeferment is enabled by default + $opts{autodefer} = 1 unless defined $opts{autodefer}; + $opts{autodeferring} = 0; # but is not initially active + $opts{ad_history} = []; + $opts{autodefer_threshhold} = $DEFAULT_AUTODEFER_THRESHHOLD + unless defined $opts{autodefer_threshhold}; + $opts{autodefer_filelen_threshhold} = $DEFAULT_AUTODEFER_FILELEN_THRESHHOLD + unless defined $opts{autodefer_filelen_threshhold}; + + $opts{offsets} = [0]; + $opts{filename} = $file; + unless (defined $opts{recsep}) { + $opts{recsep} = _default_recsep(); + } + $opts{recseplen} = length($opts{recsep}); + if ($opts{recseplen} == 0) { + croak "Empty record separator not supported by $pack"; + } + + $opts{autochomp} = 1 unless defined $opts{autochomp}; + + $opts{mode} = O_CREAT|O_RDWR unless defined $opts{mode}; + $opts{rdonly} = (($opts{mode} & O_ACCMODE) == O_RDONLY); + $opts{sawlastrec} = undef; + + my $fh; + + if (UNIVERSAL::isa($file, 'GLOB')) { + # We use 1 here on the theory that some systems + # may not indicate failure if we use 0. + # MSWin32 does not indicate failure with 0, but I don't know if + # it will indicate failure with 1 or not. + unless (seek $file, 1, SEEK_SET) { + croak "$pack: your filehandle does not appear to be seekable"; + } + seek $file, 0, SEEK_SET; # put it back + $fh = $file; # setting binmode is the user's problem + } elsif (ref $file) { + croak "usage: tie \@array, $pack, filename, [option => value]..."; + } else { + # $fh = \do { local *FH }; # XXX this is buggy + if ($] < 5.006) { + # perl 5.005 and earlier don't autovivify filehandles + require Symbol; + $fh = Symbol::gensym(); + } + sysopen $fh, $file, $opts{mode}, 0666 or return; + binmode $fh; + ++$opts{ourfh}; + } + { my $ofh = select $fh; $| = 1; select $ofh } # autoflush on write + if (defined $opts{discipline} && $] >= 5.006) { + # This avoids a compile-time warning under 5.005 + eval 'binmode($fh, $opts{discipline})'; + croak $@ if $@ =~ /unknown discipline/i; + die if $@; + } + $opts{fh} = $fh; + + bless \%opts => $pack; +} + +sub FETCH { + my ($self, $n) = @_; + my $rec; + + # check the defer buffer + $rec = $self->{deferred}{$n} if exists $self->{deferred}{$n}; + $rec = $self->_fetch($n) unless defined $rec; + + # inlined _chomp1 + substr($rec, - $self->{recseplen}) = "" + if defined $rec && $self->{autochomp}; + $rec; +} + +# Chomp many records in-place; return nothing useful +sub _chomp { + my $self = shift; + return unless $self->{autochomp}; + if ($self->{autochomp}) { + for (@_) { + next unless defined; + substr($_, - $self->{recseplen}) = ""; + } + } +} + +# Chomp one record in-place; return modified record +sub _chomp1 { + my ($self, $rec) = @_; + return $rec unless $self->{autochomp}; + return unless defined $rec; + substr($rec, - $self->{recseplen}) = ""; + $rec; +} + +sub _fetch { + my ($self, $n) = @_; + + # check the record cache + { my $cached = $self->{cache}->lookup($n); + return $cached if defined $cached; + } + + if ($#{$self->{offsets}} < $n) { + return if $self->{eof}; # request for record beyond end of file + my $o = $self->_fill_offsets_to($n); + # If it's still undefined, there is no such record, so return 'undef' + return unless defined $o; + } + + my $fh = $self->{FH}; + $self->_seek($n); # we can do this now that offsets is populated + my $rec = $self->_read_record; + +# If we happen to have just read the first record, check to see if +# the length of the record matches what 'tell' says. If not, Tie::File +# won't work, and should drop dead. +# +# if ($n == 0 && defined($rec) && tell($self->{fh}) != length($rec)) { +# if (defined $self->{discipline}) { +# croak "I/O discipline $self->{discipline} not supported"; +# } else { +# croak "File encoding not supported"; +# } +# } + + $self->{cache}->insert($n, $rec) if defined $rec && not $self->{flushing}; + $rec; +} + +sub STORE { + my ($self, $n, $rec) = @_; + die "STORE called from _check_integrity!" if $DIAGNOSTIC; + + $self->_fixrecs($rec); + + if ($self->{autodefer}) { + $self->_annotate_ad_history($n); + } + + return $self->_store_deferred($n, $rec) if $self->_is_deferring; + + + # We need this to decide whether the new record will fit + # It incidentally populates the offsets table + # Note we have to do this before we alter the cache + # 20020324 Wait, but this DOES alter the cache. TODO BUG? + my $oldrec = $self->_fetch($n); + + if (not defined $oldrec) { + # We're storing a record beyond the end of the file + $self->_extend_file_to($n+1); + $oldrec = $self->{recsep}; + } +# return if $oldrec eq $rec; # don't bother + my $len_diff = length($rec) - length($oldrec); + + # length($oldrec) here is not consistent with text mode TODO XXX BUG + $self->_mtwrite($rec, $self->{offsets}[$n], length($oldrec)); + $self->_oadjust([$n, 1, $rec]); + $self->{cache}->update($n, $rec); +} + +sub _store_deferred { + my ($self, $n, $rec) = @_; + $self->{cache}->remove($n); + my $old_deferred = $self->{deferred}{$n}; + + if (defined $self->{deferred_max} && $n > $self->{deferred_max}) { + $self->{deferred_max} = $n; + } + $self->{deferred}{$n} = $rec; + + my $len_diff = length($rec); + $len_diff -= length($old_deferred) if defined $old_deferred; + $self->{deferred_s} += $len_diff; + $self->{cache}->adj_limit(-$len_diff); + if ($self->{deferred_s} > $self->{dw_size}) { + $self->_flush; + } elsif ($self->_cache_too_full) { + $self->_cache_flush; + } +} + +# Remove a single record from the deferred-write buffer without writing it +# The record need not be present +sub _delete_deferred { + my ($self, $n) = @_; + my $rec = delete $self->{deferred}{$n}; + return unless defined $rec; + + if (defined $self->{deferred_max} + && $n == $self->{deferred_max}) { + undef $self->{deferred_max}; + } + + $self->{deferred_s} -= length $rec; + $self->{cache}->adj_limit(length $rec); +} + +sub FETCHSIZE { + my $self = shift; + my $n = $self->{eof} ? $#{$self->{offsets}} : $self->_fill_offsets; + + my $top_deferred = $self->_defer_max; + $n = $top_deferred+1 if defined $top_deferred && $n < $top_deferred+1; + $n; +} + +sub STORESIZE { + my ($self, $len) = @_; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('STORESIZE'); + } + + my $olen = $self->FETCHSIZE; + return if $len == $olen; # Woo-hoo! + + # file gets longer + if ($len > $olen) { + if ($self->_is_deferring) { + for ($olen .. $len-1) { + $self->_store_deferred($_, $self->{recsep}); + } + } else { + $self->_extend_file_to($len); + } + return; + } + + # file gets shorter + if ($self->_is_deferring) { + # TODO maybe replace this with map-plus-assignment? + for (grep $_ >= $len, keys %{$self->{deferred}}) { + $self->_delete_deferred($_); + } + $self->{deferred_max} = $len-1; + } + + $self->_seek($len); + $self->_chop_file; + $#{$self->{offsets}} = $len; +# $self->{offsets}[0] = 0; # in case we just chopped this + + $self->{cache}->remove(grep $_ >= $len, $self->{cache}->ckeys); +} + +### OPTIMIZE ME +### It should not be necessary to do FETCHSIZE +### Just seek to the end of the file. +sub PUSH { + my $self = shift; + $self->SPLICE($self->FETCHSIZE, scalar(@_), @_); + + # No need to return: + # $self->FETCHSIZE; # because av.c takes care of this for me +} + +sub POP { + my $self = shift; + my $size = $self->FETCHSIZE; + return if $size == 0; +# print STDERR "# POPPITY POP POP POP\n"; + scalar $self->SPLICE($size-1, 1); +} + +sub SHIFT { + my $self = shift; + scalar $self->SPLICE(0, 1); +} + +sub UNSHIFT { + my $self = shift; + $self->SPLICE(0, 0, @_); + # $self->FETCHSIZE; # av.c takes care of this for me +} + +sub CLEAR { + my $self = shift; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('CLEAR'); + } + + $self->_seekb(0); + $self->_chop_file; + $self->{cache}->set_limit($self->{memory}); + $self->{cache}->empty; + @{$self->{offsets}} = (0); + %{$self->{deferred}}= (); + $self->{deferred_s} = 0; + $self->{deferred_max} = -1; +} + +sub EXTEND { + my ($self, $n) = @_; + + # No need to pre-extend anything in this case + return if $self->_is_deferring; + + $self->_fill_offsets_to($n); + $self->_extend_file_to($n); +} + +sub DELETE { + my ($self, $n) = @_; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('DELETE'); + } + + my $lastrec = $self->FETCHSIZE-1; + my $rec = $self->FETCH($n); + $self->_delete_deferred($n) if $self->_is_deferring; + if ($n == $lastrec) { + $self->_seek($n); + $self->_chop_file; + $#{$self->{offsets}}--; + $self->{cache}->remove($n); + # perhaps in this case I should also remove trailing null records? + # 20020316 + # Note that delete @a[-3..-1] deletes the records in the wrong order, + # so we only chop the very last one out of the file. We could repair this + # by tracking deleted records inside the object. + } elsif ($n < $lastrec) { + $self->STORE($n, ""); + } + $rec; +} + +sub EXISTS { + my ($self, $n) = @_; + return 1 if exists $self->{deferred}{$n}; + $n < $self->FETCHSIZE; +} + +sub SPLICE { + my $self = shift; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('SPLICE'); + } + + $self->_flush if $self->_is_deferring; # move this up? + if (wantarray) { + $self->_chomp(my @a = $self->_splice(@_)); + @a; + } else { + $self->_chomp1(scalar $self->_splice(@_)); + } +} + +sub DESTROY { + my $self = shift; + $self->flush if $self->_is_deferring; + $self->{cache}->delink if defined $self->{cache}; # break circular link + if ($self->{fh} and $self->{ourfh}) { + delete $self->{ourfh}; + close delete $self->{fh}; + } +} + +sub _splice { + my ($self, $pos, $nrecs, @data) = @_; + my @result; + + $pos = 0 unless defined $pos; + + # Deal with negative and other out-of-range positions + # Also set default for $nrecs + { + my $oldsize = $self->FETCHSIZE; + $nrecs = $oldsize unless defined $nrecs; + my $oldpos = $pos; + + if ($pos < 0) { + $pos += $oldsize; + if ($pos < 0) { + croak "Modification of non-creatable array value attempted, subscript $oldpos"; + } + } + + if ($pos > $oldsize) { + return unless @data; + $pos = $oldsize; # This is what perl does for normal arrays + } + + # The manual is very unclear here + if ($nrecs < 0) { + $nrecs = $oldsize - $pos + $nrecs; + $nrecs = 0 if $nrecs < 0; + } + + # nrecs is too big---it really means "until the end" + # 20030507 + if ($nrecs + $pos > $oldsize) { + $nrecs = $oldsize - $pos; + } + } + + $self->_fixrecs(@data); + my $data = join '', @data; + my $datalen = length $data; + my $oldlen = 0; + + # compute length of data being removed + for ($pos .. $pos+$nrecs-1) { + last unless defined $self->_fill_offsets_to($_); + my $rec = $self->_fetch($_); + last unless defined $rec; + push @result, $rec; + + # Why don't we just use length($rec) here? + # Because that record might have come from the cache. _splice + # might have been called to flush out the deferred-write records, + # and in this case length($rec) is the length of the record to be + # *written*, not the length of the actual record in the file. But + # the offsets are still true. 20020322 + $oldlen += $self->{offsets}[$_+1] - $self->{offsets}[$_] + if defined $self->{offsets}[$_+1]; + } + $self->_fill_offsets_to($pos+$nrecs); + + # Modify the file + $self->_mtwrite($data, $self->{offsets}[$pos], $oldlen); + # Adjust the offsets table + $self->_oadjust([$pos, $nrecs, @data]); + + { # Take this read cache stuff out into a separate function + # You made a half-attempt to put it into _oadjust. + # Finish something like that up eventually. + # STORE also needs to do something similarish + + # update the read cache, part 1 + # modified records + for ($pos .. $pos+$nrecs-1) { + my $new = $data[$_-$pos]; + if (defined $new) { + $self->{cache}->update($_, $new); + } else { + $self->{cache}->remove($_); + } + } + + # update the read cache, part 2 + # moved records - records past the site of the change + # need to be renumbered + # Maybe merge this with the previous block? + { + my @oldkeys = grep $_ >= $pos + $nrecs, $self->{cache}->ckeys; + my @newkeys = map $_-$nrecs+@data, @oldkeys; + $self->{cache}->rekey(\@oldkeys, \@newkeys); + } + + # Now there might be too much data in the cache, if we spliced out + # some short records and spliced in some long ones. If so, flush + # the cache. + $self->_cache_flush; + } + + # Yes, the return value of 'splice' *is* actually this complicated + wantarray ? @result : @result ? $result[-1] : undef; +} + + +# write data into the file +# $data is the data to be written. +# it should be written at position $pos, and should overwrite +# exactly $len of the following bytes. +# Note that if length($data) > $len, the subsequent bytes will have to +# be moved up, and if length($data) < $len, they will have to +# be moved down +sub _twrite { + my ($self, $data, $pos, $len) = @_; + + unless (defined $pos) { + die "\$pos was undefined in _twrite"; + } + + my $len_diff = length($data) - $len; + + if ($len_diff == 0) { # Woo-hoo! + my $fh = $self->{fh}; + $self->_seekb($pos); + $self->_write_record($data); + return; # well, that was easy. + } + + # the two records are of different lengths + # our strategy here: rewrite the tail of the file, + # reading ahead one buffer at a time + # $bufsize is required to be at least as large as the data we're overwriting + my $bufsize = _bufsize($len_diff); + my ($writepos, $readpos) = ($pos, $pos+$len); + my $next_block; + my $more_data; + + # Seems like there ought to be a way to avoid the repeated code + # and the special case here. The read(1) is also a little weird. + # Think about this. + do { + $self->_seekb($readpos); + my $br = read $self->{fh}, $next_block, $bufsize; + $more_data = read $self->{fh}, my($dummy), 1; + $self->_seekb($writepos); + $self->_write_record($data); + $readpos += $br; + $writepos += length $data; + $data = $next_block; + } while $more_data; + $self->_seekb($writepos); + $self->_write_record($next_block); + + # There might be leftover data at the end of the file + $self->_chop_file if $len_diff < 0; +} + +# _iwrite(D, S, E) +# Insert text D at position S. +# Let C = E-S-|D|. If C < 0; die. +# Data in [S,S+C) is copied to [S+D,S+D+C) = [S+D,E). +# Data in [S+C = E-D, E) is returned. Data in [E, oo) is untouched. +# +# In a later version, don't read the entire intervening area into +# memory at once; do the copying block by block. +sub _iwrite { + my $self = shift; + my ($D, $s, $e) = @_; + my $d = length $D; + my $c = $e-$s-$d; + local *FH = $self->{fh}; + confess "Not enough space to insert $d bytes between $s and $e" + if $c < 0; + confess "[$s,$e) is an invalid insertion range" if $e < $s; + + $self->_seekb($s); + read FH, my $buf, $e-$s; + + $D .= substr($buf, 0, $c, ""); + + $self->_seekb($s); + $self->_write_record($D); + + return $buf; +} + +# Like _twrite, but the data-pos-len triple may be repeated; you may +# write several chunks. All the writing will be done in +# one pass. Chunks SHALL be in ascending order and SHALL NOT overlap. +sub _mtwrite { + my $self = shift; + my $unwritten = ""; + my $delta = 0; + + @_ % 3 == 0 + or die "Arguments to _mtwrite did not come in groups of three"; + + while (@_) { + my ($data, $pos, $len) = splice @_, 0, 3; + my $end = $pos + $len; # The OLD end of the segment to be replaced + $data = $unwritten . $data; + $delta -= length($unwritten); + $unwritten = ""; + $pos += $delta; # This is where the data goes now + my $dlen = length $data; + $self->_seekb($pos); + if ($len >= $dlen) { # the data will fit + $self->_write_record($data); + $delta += ($dlen - $len); # everything following moves down by this much + $data = ""; # All the data in the buffer has been written + } else { # won't fit + my $writable = substr($data, 0, $len - $delta, ""); + $self->_write_record($writable); + $delta += ($dlen - $len); # everything following moves down by this much + } + + # At this point we've written some but maybe not all of the data. + # There might be a gap to close up, or $data might still contain a + # bunch of unwritten data that didn't fit. + my $ndlen = length $data; + if ($delta == 0) { + $self->_write_record($data); + } elsif ($delta < 0) { + # upcopy (close up gap) + if (@_) { + $self->_upcopy($end, $end + $delta, $_[1] - $end); + } else { + $self->_upcopy($end, $end + $delta); + } + } else { + # downcopy (insert data that didn't fit; replace this data in memory + # with _later_ data that doesn't fit) + if (@_) { + $unwritten = $self->_downcopy($data, $end, $_[1] - $end); + } else { + # Make the file longer to accommodate the last segment that doesn' + $unwritten = $self->_downcopy($data, $end); + } + } + } +} + +# Copy block of data of length $len from position $spos to position $dpos +# $dpos must be <= $spos +# +# If $len is undefined, go all the way to the end of the file +# and then truncate it ($spos - $dpos bytes will be removed) +sub _upcopy { + my $blocksize = 8192; + my ($self, $spos, $dpos, $len) = @_; + if ($dpos > $spos) { + die "source ($spos) was upstream of destination ($dpos) in _upcopy"; + } elsif ($dpos == $spos) { + return; + } + + while (! defined ($len) || $len > 0) { + my $readsize = ! defined($len) ? $blocksize + : $len > $blocksize ? $blocksize + : $len; + + my $fh = $self->{fh}; + $self->_seekb($spos); + my $bytes_read = read $fh, my($data), $readsize; + $self->_seekb($dpos); + if ($data eq "") { + $self->_chop_file; + last; + } + $self->_write_record($data); + $spos += $bytes_read; + $dpos += $bytes_read; + $len -= $bytes_read if defined $len; + } +} + +# Write $data into a block of length $len at position $pos, +# moving everything in the block forwards to make room. +# Instead of writing the last length($data) bytes from the block +# (because there isn't room for them any longer) return them. +# +# Undefined $len means 'until the end of the file' +sub _downcopy { + my $blocksize = 8192; + my ($self, $data, $pos, $len) = @_; + my $fh = $self->{fh}; + + while (! defined $len || $len > 0) { + my $readsize = ! defined($len) ? $blocksize + : $len > $blocksize? $blocksize : $len; + $self->_seekb($pos); + read $fh, my($old), $readsize; + my $last_read_was_short = length($old) < $readsize; + $data .= $old; + my $writable; + if ($last_read_was_short) { + # If last read was short, then $data now contains the entire rest + # of the file, so there's no need to write only one block of it + $writable = $data; + $data = ""; + } else { + $writable = substr($data, 0, $readsize, ""); + } + last if $writable eq ""; + $self->_seekb($pos); + $self->_write_record($writable); + last if $last_read_was_short && $data eq ""; + $len -= $readsize if defined $len; + $pos += $readsize; + } + return $data; +} + +# Adjust the object data structures following an '_mtwrite' +# Arguments are +# [$pos, $nrecs, @length] items +# indicating that $nrecs records were removed at $recpos (a record offset) +# and replaced with records of length @length... +# Arguments guarantee that $recpos is strictly increasing. +# No return value +sub _oadjust { + my $self = shift; + my $delta = 0; + my $delta_recs = 0; + my $prev_end = -1; + my %newkeys; + + for (@_) { + my ($pos, $nrecs, @data) = @$_; + $pos += $delta_recs; + + # Adjust the offsets of the records after the previous batch up + # to the first new one of this batch + for my $i ($prev_end+2 .. $pos - 1) { + $self->{offsets}[$i] += $delta; + $newkey{$i} = $i + $delta_recs; + } + + $prev_end = $pos + @data - 1; # last record moved on this pass + + # Remove the offsets for the removed records; + # replace with the offsets for the inserted records + my @newoff = ($self->{offsets}[$pos] + $delta); + for my $i (0 .. $#data) { + my $newlen = length $data[$i]; + push @newoff, $newoff[$i] + $newlen; + $delta += $newlen; + } + + for my $i ($pos .. $pos+$nrecs-1) { + last if $i+1 > $#{$self->{offsets}}; + my $oldlen = $self->{offsets}[$i+1] - $self->{offsets}[$i]; + $delta -= $oldlen; + } + +# # also this data has changed, so update it in the cache +# for (0 .. $#data) { +# $self->{cache}->update($pos + $_, $data[$_]); +# } +# if ($delta_recs) { +# my @oldkeys = grep $_ >= $pos + @data, $self->{cache}->ckeys; +# my @newkeys = map $_ + $delta_recs, @oldkeys; +# $self->{cache}->rekey(\@oldkeys, \@newkeys); +# } + + # replace old offsets with new + splice @{$self->{offsets}}, $pos, $nrecs+1, @newoff; + # What if we just spliced out the end of the offsets table? + # shouldn't we clear $self->{eof}? Test for this XXX BUG TODO + + $delta_recs += @data - $nrecs; # net change in total number of records + } + + # The trailing records at the very end of the file + if ($delta) { + for my $i ($prev_end+2 .. $#{$self->{offsets}}) { + $self->{offsets}[$i] += $delta; + } + } + + # If we scrubbed out all known offsets, regenerate the trivial table + # that knows that the file does indeed start at 0. + $self->{offsets}[0] = 0 unless @{$self->{offsets}}; + # If the file got longer, the offsets table is no longer complete + # $self->{eof} = 0 if $delta_recs > 0; + + # Now there might be too much data in the cache, if we spliced out + # some short records and spliced in some long ones. If so, flush + # the cache. + $self->_cache_flush; +} + +# If a record does not already end with the appropriate terminator +# string, append one. +sub _fixrecs { + my $self = shift; + for (@_) { + $_ = "" unless defined $_; + $_ .= $self->{recsep} + unless substr($_, - $self->{recseplen}) eq $self->{recsep}; + } +} + + +################################################################ +# +# Basic read, write, and seek +# + +# seek to the beginning of record #$n +# Assumes that the offsets table is already correctly populated +# +# Note that $n=-1 has a special meaning here: It means the start of +# the last known record; this may or may not be the very last record +# in the file, depending on whether the offsets table is fully populated. +# +sub _seek { + my ($self, $n) = @_; + my $o = $self->{offsets}[$n]; + defined($o) + or confess("logic error: undefined offset for record $n"); + seek $self->{fh}, $o, SEEK_SET + or confess "Couldn't seek filehandle: $!"; # "Should never happen." +} + +# seek to byte $b in the file +sub _seekb { + my ($self, $b) = @_; + seek $self->{fh}, $b, SEEK_SET + or die "Couldn't seek filehandle: $!"; # "Should never happen." +} + +# populate the offsets table up to the beginning of record $n +# return the offset of record $n +sub _fill_offsets_to { + my ($self, $n) = @_; + + return $self->{offsets}[$n] if $self->{eof}; + + my $fh = $self->{fh}; + local *OFF = $self->{offsets}; + my $rec; + + until ($#OFF >= $n) { + $self->_seek(-1); # tricky -- see comment at _seek + $rec = $self->_read_record; + if (defined $rec) { + push @OFF, int(tell $fh); # Tels says that int() saves memory here + } else { + $self->{eof} = 1; + return; # It turns out there is no such record + } + } + + # we have now read all the records up to record n-1, + # so we can return the offset of record n + $OFF[$n]; +} + +sub _fill_offsets { + my ($self) = @_; + + my $fh = $self->{fh}; + local *OFF = $self->{offsets}; + + $self->_seek(-1); # tricky -- see comment at _seek + + # Tels says that inlining read_record() would make this loop + # five times faster. 20030508 + while ( defined $self->_read_record()) { + # int() saves us memory here + push @OFF, int(tell $fh); + } + + $self->{eof} = 1; + $#OFF; +} + +# assumes that $rec is already suitably terminated +sub _write_record { + my ($self, $rec) = @_; + my $fh = $self->{fh}; + local $\ = ""; + print $fh $rec + or die "Couldn't write record: $!"; # "Should never happen." +# $self->{_written} += length($rec); +} + +sub _read_record { + my $self = shift; + my $rec; + { local $/ = $self->{recsep}; + my $fh = $self->{fh}; + $rec = <$fh>; + } + return unless defined $rec; + if (substr($rec, -$self->{recseplen}) ne $self->{recsep}) { + # improperly terminated final record --- quietly fix it. +# my $ac = substr($rec, -$self->{recseplen}); +# $ac =~ s/\n/\\n/g; + $self->{sawlastrec} = 1; + unless ($self->{rdonly}) { + local $\ = ""; + my $fh = $self->{fh}; + print $fh $self->{recsep}; + } + $rec .= $self->{recsep}; + } +# $self->{_read} += length($rec) if defined $rec; + $rec; +} + +sub _rw_stats { + my $self = shift; + @{$self}{'_read', '_written'}; +} + +################################################################ +# +# Read cache management + +sub _cache_flush { + my ($self) = @_; + $self->{cache}->reduce_size_to($self->{memory} - $self->{deferred_s}); +} + +sub _cache_too_full { + my $self = shift; + $self->{cache}->bytes + $self->{deferred_s} >= $self->{memory}; +} + +################################################################ +# +# File custodial services +# + + +# We have read to the end of the file and have the offsets table +# entirely populated. Now we need to write a new record beyond +# the end of the file. We prepare for this by writing +# empty records into the file up to the position we want +# +# assumes that the offsets table already contains the offset of record $n, +# if it exists, and extends to the end of the file if not. +sub _extend_file_to { + my ($self, $n) = @_; + $self->_seek(-1); # position after the end of the last record + my $pos = $self->{offsets}[-1]; + + # the offsets table has one entry more than the total number of records + my $extras = $n - $#{$self->{offsets}}; + + # Todo : just use $self->{recsep} x $extras here? + while ($extras-- > 0) { + $self->_write_record($self->{recsep}); + push @{$self->{offsets}}, int(tell $self->{fh}); + } +} + +# Truncate the file at the current position +sub _chop_file { + my $self = shift; + truncate $self->{fh}, tell($self->{fh}); +} + + +# compute the size of a buffer suitable for moving +# all the data in a file forward $n bytes +# ($n may be negative) +# The result should be at least $n. +sub _bufsize { + my $n = shift; + return 8192 if $n <= 0; + my $b = $n & ~8191; + $b += 8192 if $n & 8191; + $b; +} + +################################################################ +# +# Miscellaneous public methods +# + +# Lock the file +sub flock { + my ($self, $op) = @_; + unless (@_ <= 3) { + my $pack = ref $self; + croak "Usage: $pack\->flock([OPERATION])"; + } + my $fh = $self->{fh}; + $op = LOCK_EX unless defined $op; + my $locked = flock $fh, $op; + + if ($locked && ($op & (LOCK_EX | LOCK_SH))) { + # If you're locking the file, then presumably it's because + # there might have been a write access by another process. + # In that case, the read cache contents and the offsets table + # might be invalid, so discard them. 20030508 + $self->{offsets} = [0]; + $self->{cache}->empty; + } + + $locked; +} + +# Get/set autochomp option +sub autochomp { + my $self = shift; + if (@_) { + my $old = $self->{autochomp}; + $self->{autochomp} = shift; + $old; + } else { + $self->{autochomp}; + } +} + +# Get offset table entries; returns offset of nth record +sub offset { + my ($self, $n) = @_; + + if ($#{$self->{offsets}} < $n) { + return if $self->{eof}; # request for record beyond the end of file + my $o = $self->_fill_offsets_to($n); + # If it's still undefined, there is no such record, so return 'undef' + return unless defined $o; + } + + $self->{offsets}[$n]; +} + +sub discard_offsets { + my $self = shift; + $self->{offsets} = [0]; +} + +################################################################ +# +# Matters related to deferred writing +# + +# Defer writes +sub defer { + my $self = shift; + $self->_stop_autodeferring; + @{$self->{ad_history}} = (); + $self->{defer} = 1; +} + +# Flush deferred writes +# +# This could be better optimized to write the file in one pass, instead +# of one pass per block of records. But that will require modifications +# to _twrite, so I should have a good _twrite test suite first. +sub flush { + my $self = shift; + + $self->_flush; + $self->{defer} = 0; +} + +sub _old_flush { + my $self = shift; + my @writable = sort {$a<=>$b} (keys %{$self->{deferred}}); + + while (@writable) { + # gather all consecutive records from the front of @writable + my $first_rec = shift @writable; + my $last_rec = $first_rec+1; + ++$last_rec, shift @writable while @writable && $last_rec == $writable[0]; + --$last_rec; + $self->_fill_offsets_to($last_rec); + $self->_extend_file_to($last_rec); + $self->_splice($first_rec, $last_rec-$first_rec+1, + @{$self->{deferred}}{$first_rec .. $last_rec}); + } + + $self->_discard; # clear out defered-write-cache +} + +sub _flush { + my $self = shift; + my @writable = sort {$a<=>$b} (keys %{$self->{deferred}}); + my @args; + my @adjust; + + while (@writable) { + # gather all consecutive records from the front of @writable + my $first_rec = shift @writable; + my $last_rec = $first_rec+1; + ++$last_rec, shift @writable while @writable && $last_rec == $writable[0]; + --$last_rec; + my $end = $self->_fill_offsets_to($last_rec+1); + if (not defined $end) { + $self->_extend_file_to($last_rec); + $end = $self->{offsets}[$last_rec]; + } + my ($start) = $self->{offsets}[$first_rec]; + push @args, + join("", @{$self->{deferred}}{$first_rec .. $last_rec}), # data + $start, # position + $end-$start; # length + push @adjust, [$first_rec, # starting at this position... + $last_rec-$first_rec+1, # this many records... + # are replaced with these... + @{$self->{deferred}}{$first_rec .. $last_rec}, + ]; + } + + $self->_mtwrite(@args); # write multiple record groups + $self->_discard; # clear out defered-write-cache + $self->_oadjust(@adjust); +} + +# Discard deferred writes and disable future deferred writes +sub discard { + my $self = shift; + $self->_discard; + $self->{defer} = 0; +} + +# Discard deferred writes, but retain old deferred writing mode +sub _discard { + my $self = shift; + %{$self->{deferred}} = (); + $self->{deferred_s} = 0; + $self->{deferred_max} = -1; + $self->{cache}->set_limit($self->{memory}); +} + +# Deferred writing is enabled, either explicitly ($self->{defer}) +# or automatically ($self->{autodeferring}) +sub _is_deferring { + my $self = shift; + $self->{defer} || $self->{autodeferring}; +} + +# The largest record number of any deferred record +sub _defer_max { + my $self = shift; + return $self->{deferred_max} if defined $self->{deferred_max}; + my $max = -1; + for my $key (keys %{$self->{deferred}}) { + $max = $key if $key > $max; + } + $self->{deferred_max} = $max; + $max; +} + +################################################################ +# +# Matters related to autodeferment +# + +# Get/set autodefer option +sub autodefer { + my $self = shift; + if (@_) { + my $old = $self->{autodefer}; + $self->{autodefer} = shift; + if ($old) { + $self->_stop_autodeferring; + @{$self->{ad_history}} = (); + } + $old; + } else { + $self->{autodefer}; + } +} + +# The user is trying to store record #$n Record that in the history, +# and then enable (or disable) autodeferment if that seems useful. +# Note that it's OK for $n to be a non-number, as long as the function +# is prepared to deal with that. Nobody else looks at the ad_history. +# +# Now, what does the ad_history mean, and what is this function doing? +# Essentially, the idea is to enable autodeferring when we see that the +# user has made three consecutive STORE calls to three consecutive records. +# ("Three" is actually ->{autodefer_threshhold}.) +# A STORE call for record #$n inserts $n into the autodefer history, +# and if the history contains three consecutive records, we enable +# autodeferment. An ad_history of [X, Y] means that the most recent +# STOREs were for records X, X+1, ..., Y, in that order. +# +# Inserting a nonconsecutive number erases the history and starts over. +# +# Performing a special operation like SPLICE erases the history. +# +# There's one special case: CLEAR means that CLEAR was just called. +# In this case, we prime the history with [-2, -1] so that if the next +# write is for record 0, autodeferring goes on immediately. This is for +# the common special case of "@a = (...)". +# +sub _annotate_ad_history { + my ($self, $n) = @_; + return unless $self->{autodefer}; # feature is disabled + return if $self->{defer}; # already in explicit defer mode + return unless $self->{offsets}[-1] >= $self->{autodefer_filelen_threshhold}; + + local *H = $self->{ad_history}; + if ($n eq 'CLEAR') { + @H = (-2, -1); # prime the history with fake records + $self->_stop_autodeferring; + } elsif ($n =~ /^\d+$/) { + if (@H == 0) { + @H = ($n, $n); + } else { # @H == 2 + if ($H[1] == $n-1) { # another consecutive record + $H[1]++; + if ($H[1] - $H[0] + 1 >= $self->{autodefer_threshhold}) { + $self->{autodeferring} = 1; + } + } else { # nonconsecutive- erase and start over + @H = ($n, $n); + $self->_stop_autodeferring; + } + } + } else { # SPLICE or STORESIZE or some such + @H = (); + $self->_stop_autodeferring; + } +} + +# If autodeferring was enabled, cut it out and discard the history +sub _stop_autodeferring { + my $self = shift; + if ($self->{autodeferring}) { + $self->_flush; + } + $self->{autodeferring} = 0; +} + +################################################################ + + +# This is NOT a method. It is here for two reasons: +# 1. To factor a fairly complicated block out of the constructor +# 2. To provide access for the test suite, which need to be sure +# files are being written properly. +sub _default_recsep { + my $recsep = $/; + if ($^O eq 'MSWin32') { # Dos too? + # Windows users expect files to be terminated with \r\n + # But $/ is set to \n instead + # Note that this also transforms \n\n into \r\n\r\n. + # That is a feature. + $recsep =~ s/\n/\r\n/g; + } + $recsep; +} + +# Utility function for _check_integrity +sub _ci_warn { + my $msg = shift; + $msg =~ s/\n/\\n/g; + $msg =~ s/\r/\\r/g; + print "# $msg\n"; +} + +# Given a file, make sure the cache is consistent with the +# file contents and the internal data structures are consistent with +# each other. Returns true if everything checks out, false if not +# +# The $file argument is no longer used. It is retained for compatibility +# with the existing test suite. +sub _check_integrity { + my ($self, $file, $warn) = @_; + my $rsl = $self->{recseplen}; + my $rs = $self->{recsep}; + my $good = 1; + local *_; # local $_ does not work here + local $DIAGNOSTIC = 1; + + if (not defined $rs) { + _ci_warn("recsep is undef!"); + $good = 0; + } elsif ($rs eq "") { + _ci_warn("recsep is empty!"); + $good = 0; + } elsif ($rsl != length $rs) { + my $ln = length $rs; + _ci_warn("recsep <$rs> has length $ln, should be $rsl"); + $good = 0; + } + + if (not defined $self->{offsets}[0]) { + _ci_warn("offset 0 is missing!"); + $good = 0; + + } elsif ($self->{offsets}[0] != 0) { + _ci_warn("rec 0: offset <$self->{offsets}[0]> s/b 0!"); + $good = 0; + } + + my $cached = 0; + { + local *F = $self->{fh}; + seek F, 0, SEEK_SET; + local $. = 0; + local $/ = $rs; + + while (<F>) { + my $n = $. - 1; + my $cached = $self->{cache}->_produce($n); + my $offset = $self->{offsets}[$.]; + my $ao = tell F; + if (defined $offset && $offset != $ao) { + _ci_warn("rec $n: offset <$offset> actual <$ao>"); + $good = 0; + } + if (defined $cached && $_ ne $cached && ! $self->{deferred}{$n}) { + $good = 0; + _ci_warn("rec $n: cached <$cached> actual <$_>"); + } + if (defined $cached && substr($cached, -$rsl) ne $rs) { + $good = 0; + _ci_warn("rec $n in the cache is missing the record separator"); + } + if (! defined $offset && $self->{eof}) { + $good = 0; + _ci_warn("The offset table was marked complete, but it is missing element $."); + } + } + if (@{$self->{offsets}} > $.+1) { + $good = 0; + my $n = @{$self->{offsets}}; + _ci_warn("The offset table has $n items, but the file has only $."); + } + + my $deferring = $self->_is_deferring; + for my $n ($self->{cache}->ckeys) { + my $r = $self->{cache}->_produce($n); + $cached += length($r); + next if $n+1 <= $.; # checked this already + _ci_warn("spurious caching of record $n"); + $good = 0; + } + my $b = $self->{cache}->bytes; + if ($cached != $b) { + _ci_warn("cache size is $b, should be $cached"); + $good = 0; + } + } + + # That cache has its own set of tests + $good = 0 unless $self->{cache}->_check_integrity; + + # Now let's check the deferbuffer + # Unless deferred writing is enabled, it should be empty + if (! $self->_is_deferring && %{$self->{deferred}}) { + _ci_warn("deferred writing disabled, but deferbuffer nonempty"); + $good = 0; + } + + # Any record in the deferbuffer should *not* be present in the readcache + my $deferred_s = 0; + while (my ($n, $r) = each %{$self->{deferred}}) { + $deferred_s += length($r); + if (defined $self->{cache}->_produce($n)) { + _ci_warn("record $n is in the deferbuffer *and* the readcache"); + $good = 0; + } + if (substr($r, -$rsl) ne $rs) { + _ci_warn("rec $n in the deferbuffer is missing the record separator"); + $good = 0; + } + } + + # Total size of deferbuffer should match internal total + if ($deferred_s != $self->{deferred_s}) { + _ci_warn("buffer size is $self->{deferred_s}, should be $deferred_s"); + $good = 0; + } + + # Total size of deferbuffer should not exceed the specified limit + if ($deferred_s > $self->{dw_size}) { + _ci_warn("buffer size is $self->{deferred_s} which exceeds the limit of $self->{dw_size}"); + $good = 0; + } + + # Total size of cached data should not exceed the specified limit + if ($deferred_s + $cached > $self->{memory}) { + my $total = $deferred_s + $cached; + _ci_warn("total stored data size is $total which exceeds the limit of $self->{memory}"); + $good = 0; + } + + # Stuff related to autodeferment + if (!$self->{autodefer} && @{$self->{ad_history}}) { + _ci_warn("autodefer is disabled, but ad_history is nonempty"); + $good = 0; + } + if ($self->{autodeferring} && $self->{defer}) { + _ci_warn("both autodeferring and explicit deferring are active"); + $good = 0; + } + if (@{$self->{ad_history}} == 0) { + # That's OK, no additional tests required + } elsif (@{$self->{ad_history}} == 2) { + my @non_number = grep !/^-?\d+$/, @{$self->{ad_history}}; + if (@non_number) { + my $msg; + { local $" = ')('; + $msg = "ad_history contains non-numbers (@{$self->{ad_history}})"; + } + _ci_warn($msg); + $good = 0; + } elsif ($self->{ad_history}[1] < $self->{ad_history}[0]) { + _ci_warn("ad_history has nonsensical values @{$self->{ad_history}}"); + $good = 0; + } + } else { + _ci_warn("ad_history has bad length <@{$self->{ad_history}}>"); + $good = 0; + } + + $good; +} + +################################################################ +# +# Tie::File::Cache +# +# Read cache + +package Tie::File::Cache; +$Tie::File::Cache::VERSION = $Tie::File::VERSION; +use Carp ':DEFAULT', 'confess'; + +sub HEAP () { 0 } +sub HASH () { 1 } +sub MAX () { 2 } +sub BYTES() { 3 } +#sub STAT () { 4 } # Array with request statistics for each record +#sub MISS () { 5 } # Total number of cache misses +#sub REQ () { 6 } # Total number of cache requests +use strict 'vars'; + +sub new { + my ($pack, $max) = @_; + local *_; + croak "missing argument to ->new" unless defined $max; + my $self = []; + bless $self => $pack; + @$self = (Tie::File::Heap->new($self), {}, $max, 0); + $self; +} + +sub adj_limit { + my ($self, $n) = @_; + $self->[MAX] += $n; +} + +sub set_limit { + my ($self, $n) = @_; + $self->[MAX] = $n; +} + +# For internal use only +# Will be called by the heap structure to notify us that a certain +# piece of data has moved from one heap element to another. +# $k is the hash key of the item +# $n is the new index into the heap at which it is stored +# If $n is undefined, the item has been removed from the heap. +sub _heap_move { + my ($self, $k, $n) = @_; + if (defined $n) { + $self->[HASH]{$k} = $n; + } else { + delete $self->[HASH]{$k}; + } +} + +sub insert { + my ($self, $key, $val) = @_; + local *_; + croak "missing argument to ->insert" unless defined $key; + unless (defined $self->[MAX]) { + confess "undefined max" ; + } + confess "undefined val" unless defined $val; + return if length($val) > $self->[MAX]; + +# if ($self->[STAT]) { +# $self->[STAT][$key] = 1; +# return; +# } + + my $oldnode = $self->[HASH]{$key}; + if (defined $oldnode) { + my $oldval = $self->[HEAP]->set_val($oldnode, $val); + $self->[BYTES] -= length($oldval); + } else { + $self->[HEAP]->insert($key, $val); + } + $self->[BYTES] += length($val); + $self->flush if $self->[BYTES] > $self->[MAX]; +} + +sub expire { + my $self = shift; + my $old_data = $self->[HEAP]->popheap; + return unless defined $old_data; + $self->[BYTES] -= length $old_data; + $old_data; +} + +sub remove { + my ($self, @keys) = @_; + my @result; + +# if ($self->[STAT]) { +# for my $key (@keys) { +# $self->[STAT][$key] = 0; +# } +# return; +# } + + for my $key (@keys) { + next unless exists $self->[HASH]{$key}; + my $old_data = $self->[HEAP]->remove($self->[HASH]{$key}); + $self->[BYTES] -= length $old_data; + push @result, $old_data; + } + @result; +} + +sub lookup { + my ($self, $key) = @_; + local *_; + croak "missing argument to ->lookup" unless defined $key; + +# if ($self->[STAT]) { +# $self->[MISS]++ if $self->[STAT][$key]++ == 0; +# $self->[REQ]++; +# my $hit_rate = 1 - $self->[MISS] / $self->[REQ]; +# # Do some testing to determine this threshhold +# $#$self = STAT - 1 if $hit_rate > 0.20; +# } + + if (exists $self->[HASH]{$key}) { + $self->[HEAP]->lookup($self->[HASH]{$key}); + } else { + return; + } +} + +# For internal use only +sub _produce { + my ($self, $key) = @_; + my $loc = $self->[HASH]{$key}; + return unless defined $loc; + $self->[HEAP][$loc][2]; +} + +# For internal use only +sub _promote { + my ($self, $key) = @_; + $self->[HEAP]->promote($self->[HASH]{$key}); +} + +sub empty { + my ($self) = @_; + %{$self->[HASH]} = (); + $self->[BYTES] = 0; + $self->[HEAP]->empty; +# @{$self->[STAT]} = (); +# $self->[MISS] = 0; +# $self->[REQ] = 0; +} + +sub is_empty { + my ($self) = @_; + keys %{$self->[HASH]} == 0; +} + +sub update { + my ($self, $key, $val) = @_; + local *_; + croak "missing argument to ->update" unless defined $key; + if (length($val) > $self->[MAX]) { + my ($oldval) = $self->remove($key); + $self->[BYTES] -= length($oldval) if defined $oldval; + } elsif (exists $self->[HASH]{$key}) { + my $oldval = $self->[HEAP]->set_val($self->[HASH]{$key}, $val); + $self->[BYTES] += length($val); + $self->[BYTES] -= length($oldval) if defined $oldval; + } else { + $self->[HEAP]->insert($key, $val); + $self->[BYTES] += length($val); + } + $self->flush; +} + +sub rekey { + my ($self, $okeys, $nkeys) = @_; + local *_; + my %map; + @map{@$okeys} = @$nkeys; + croak "missing argument to ->rekey" unless defined $nkeys; + croak "length mismatch in ->rekey arguments" unless @$nkeys == @$okeys; + my %adjusted; # map new keys to heap indices + # You should be able to cut this to one loop TODO XXX + for (0 .. $#$okeys) { + $adjusted{$nkeys->[$_]} = delete $self->[HASH]{$okeys->[$_]}; + } + while (my ($nk, $ix) = each %adjusted) { + # @{$self->[HASH]}{keys %adjusted} = values %adjusted; + $self->[HEAP]->rekey($ix, $nk); + $self->[HASH]{$nk} = $ix; + } +} + +sub ckeys { + my $self = shift; + my @a = keys %{$self->[HASH]}; + @a; +} + +# Return total amount of cached data +sub bytes { + my $self = shift; + $self->[BYTES]; +} + +# Expire oldest item from cache until cache size is smaller than $max +sub reduce_size_to { + my ($self, $max) = @_; + until ($self->[BYTES] <= $max) { + # Note that Tie::File::Cache::expire has been inlined here + my $old_data = $self->[HEAP]->popheap; + return unless defined $old_data; + $self->[BYTES] -= length $old_data; + } +} + +# Why not just $self->reduce_size_to($self->[MAX])? +# Try this when things stabilize TODO XXX +# If the cache is too full, expire the oldest records +sub flush { + my $self = shift; + $self->reduce_size_to($self->[MAX]) if $self->[BYTES] > $self->[MAX]; +} + +# For internal use only +sub _produce_lru { + my $self = shift; + $self->[HEAP]->expire_order; +} + +BEGIN { *_ci_warn = \&Tie::File::_ci_warn } + +sub _check_integrity { # For CACHE + my $self = shift; + my $good = 1; + + # Test HEAP + $self->[HEAP]->_check_integrity or $good = 0; + + # Test HASH + my $bytes = 0; + for my $k (keys %{$self->[HASH]}) { + if ($k ne '0' && $k !~ /^[1-9][0-9]*$/) { + $good = 0; + _ci_warn "Cache hash key <$k> is non-numeric"; + } + + my $h = $self->[HASH]{$k}; + if (! defined $h) { + $good = 0; + _ci_warn "Heap index number for key $k is undefined"; + } elsif ($h == 0) { + $good = 0; + _ci_warn "Heap index number for key $k is zero"; + } else { + my $j = $self->[HEAP][$h]; + if (! defined $j) { + $good = 0; + _ci_warn "Heap contents key $k (=> $h) are undefined"; + } else { + $bytes += length($j->[2]); + if ($k ne $j->[1]) { + $good = 0; + _ci_warn "Heap contents key $k (=> $h) is $j->[1], should be $k"; + } + } + } + } + + # Test BYTES + if ($bytes != $self->[BYTES]) { + $good = 0; + _ci_warn "Total data in cache is $bytes, expected $self->[BYTES]"; + } + + # Test MAX + if ($bytes > $self->[MAX]) { + $good = 0; + _ci_warn "Total data in cache is $bytes, exceeds maximum $self->[MAX]"; + } + + return $good; +} + +sub delink { + my $self = shift; + $self->[HEAP] = undef; # Bye bye heap +} + +################################################################ +# +# Tie::File::Heap +# +# Heap data structure for use by cache LRU routines + +package Tie::File::Heap; +use Carp ':DEFAULT', 'confess'; +$Tie::File::Heap::VERSION = $Tie::File::Cache::VERSION; +sub SEQ () { 0 }; +sub KEY () { 1 }; +sub DAT () { 2 }; + +sub new { + my ($pack, $cache) = @_; + die "$pack: Parent cache object $cache does not support _heap_move method" + unless eval { $cache->can('_heap_move') }; + my $self = [[0,$cache,0]]; + bless $self => $pack; +} + +# Allocate a new sequence number, larger than all previously allocated numbers +sub _nseq { + my $self = shift; + $self->[0][0]++; +} + +sub _cache { + my $self = shift; + $self->[0][1]; +} + +sub _nelts { + my $self = shift; + $self->[0][2]; +} + +sub _nelts_inc { + my $self = shift; + ++$self->[0][2]; +} + +sub _nelts_dec { + my $self = shift; + --$self->[0][2]; +} + +sub is_empty { + my $self = shift; + $self->_nelts == 0; +} + +sub empty { + my $self = shift; + $#$self = 0; + $self->[0][2] = 0; + $self->[0][0] = 0; # might as well reset the sequence numbers +} + +# notify the parent cache object that we moved something +sub _heap_move { + my $self = shift; + $self->_cache->_heap_move(@_); +} + +# Insert a piece of data into the heap with the indicated sequence number. +# The item with the smallest sequence number is always at the top. +# If no sequence number is specified, allocate a new one and insert the +# item at the bottom. +sub insert { + my ($self, $key, $data, $seq) = @_; + $seq = $self->_nseq unless defined $seq; + $self->_insert_new([$seq, $key, $data]); +} + +# Insert a new, fresh item at the bottom of the heap +sub _insert_new { + my ($self, $item) = @_; + my $i = @$self; + $i = int($i/2) until defined $self->[$i/2]; + $self->[$i] = $item; + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + $self->_nelts_inc; +} + +# Insert [$data, $seq] pair at or below item $i in the heap. +# If $i is omitted, default to 1 (the top element.) +sub _insert { + my ($self, $item, $i) = @_; +# $self->_check_loc($i) if defined $i; + $i = 1 unless defined $i; + until (! defined $self->[$i]) { + if ($self->[$i][SEQ] > $item->[SEQ]) { # inserted item is older + ($self->[$i], $item) = ($item, $self->[$i]); + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + } + # If either is undefined, go that way. Otherwise, choose at random + my $dir; + $dir = 0 if !defined $self->[2*$i]; + $dir = 1 if !defined $self->[2*$i+1]; + $dir = int(rand(2)) unless defined $dir; + $i = 2*$i + $dir; + } + $self->[$i] = $item; + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + $self->_nelts_inc; +} + +# Remove the item at node $i from the heap, moving child items upwards. +# The item with the smallest sequence number is always at the top. +# Moving items upwards maintains this condition. +# Return the removed item. Return undef if there was no item at node $i. +sub remove { + my ($self, $i) = @_; + $i = 1 unless defined $i; + my $top = $self->[$i]; + return unless defined $top; + while (1) { + my $ii; + my ($L, $R) = (2*$i, 2*$i+1); + + # If either is undefined, go the other way. + # Otherwise, go towards the smallest. + last unless defined $self->[$L] || defined $self->[$R]; + $ii = $R if not defined $self->[$L]; + $ii = $L if not defined $self->[$R]; + unless (defined $ii) { + $ii = $self->[$L][SEQ] < $self->[$R][SEQ] ? $L : $R; + } + + $self->[$i] = $self->[$ii]; # Promote child to fill vacated spot + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + $i = $ii; # Fill new vacated spot + } + $self->[0][1]->_heap_move($top->[KEY], undef); + undef $self->[$i]; + $self->_nelts_dec; + return $top->[DAT]; +} + +sub popheap { + my $self = shift; + $self->remove(1); +} + +# set the sequence number of the indicated item to a higher number +# than any other item in the heap, and bubble the item down to the +# bottom. +sub promote { + my ($self, $n) = @_; +# $self->_check_loc($n); + $self->[$n][SEQ] = $self->_nseq; + my $i = $n; + while (1) { + my ($L, $R) = (2*$i, 2*$i+1); + my $dir; + last unless defined $self->[$L] || defined $self->[$R]; + $dir = $R unless defined $self->[$L]; + $dir = $L unless defined $self->[$R]; + unless (defined $dir) { + $dir = $self->[$L][SEQ] < $self->[$R][SEQ] ? $L : $R; + } + @{$self}[$i, $dir] = @{$self}[$dir, $i]; + for ($i, $dir) { + $self->[0][1]->_heap_move($self->[$_][KEY], $_) if defined $self->[$_]; + } + $i = $dir; + } +} + +# Return item $n from the heap, promoting its LRU status +sub lookup { + my ($self, $n) = @_; +# $self->_check_loc($n); + my $val = $self->[$n]; + $self->promote($n); + $val->[DAT]; +} + + +# Assign a new value for node $n, promoting it to the bottom of the heap +sub set_val { + my ($self, $n, $val) = @_; +# $self->_check_loc($n); + my $oval = $self->[$n][DAT]; + $self->[$n][DAT] = $val; + $self->promote($n); + return $oval; +} + +# The hask key has changed for an item; +# alter the heap's record of the hash key +sub rekey { + my ($self, $n, $new_key) = @_; +# $self->_check_loc($n); + $self->[$n][KEY] = $new_key; +} + +sub _check_loc { + my ($self, $n) = @_; + unless (1 || defined $self->[$n]) { + confess "_check_loc($n) failed"; + } +} + +BEGIN { *_ci_warn = \&Tie::File::_ci_warn } + +sub _check_integrity { + my $self = shift; + my $good = 1; + my %seq; + + unless (eval {$self->[0][1]->isa("Tie::File::Cache")}) { + _ci_warn "Element 0 of heap corrupt"; + $good = 0; + } + $good = 0 unless $self->_satisfies_heap_condition(1); + for my $i (2 .. $#{$self}) { + my $p = int($i/2); # index of parent node + if (defined $self->[$i] && ! defined $self->[$p]) { + _ci_warn "Element $i of heap defined, but parent $p isn't"; + $good = 0; + } + + if (defined $self->[$i]) { + if ($seq{$self->[$i][SEQ]}) { + my $seq = $self->[$i][SEQ]; + _ci_warn "Nodes $i and $seq{$seq} both have SEQ=$seq"; + $good = 0; + } else { + $seq{$self->[$i][SEQ]} = $i; + } + } + } + + return $good; +} + +sub _satisfies_heap_condition { + my $self = shift; + my $n = shift || 1; + my $good = 1; + for (0, 1) { + my $c = $n*2 + $_; + next unless defined $self->[$c]; + if ($self->[$n][SEQ] >= $self->[$c]) { + _ci_warn "Node $n of heap does not predate node $c"; + $good = 0 ; + } + $good = 0 unless $self->_satisfies_heap_condition($c); + } + return $good; +} + +# Return a list of all the values, sorted by expiration order +sub expire_order { + my $self = shift; + my @nodes = sort {$a->[SEQ] <=> $b->[SEQ]} $self->_nodes; + map { $_->[KEY] } @nodes; +} + +sub _nodes { + my $self = shift; + my $i = shift || 1; + return unless defined $self->[$i]; + ($self->[$i], $self->_nodes($i*2), $self->_nodes($i*2+1)); +} + +"Cogito, ergo sum."; # don't forget to return a true value from the file + +__END__ + +=head1 NAME + +Tie::File - Access the lines of a disk file via a Perl array + +=head1 SYNOPSIS + + # This file documents Tie::File version 0.98 + use Tie::File; + + tie @array, 'Tie::File', filename or die ...; + + $array[13] = 'blah'; # line 13 of the file is now 'blah' + print $array[42]; # display line 42 of the file + + $n_recs = @array; # how many records are in the file? + $#array -= 2; # chop two records off the end + + + for (@array) { + s/PERL/Perl/g; # Replace PERL with Perl everywhere in the file + } + + # These are just like regular push, pop, unshift, shift, and splice + # Except that they modify the file in the way you would expect + + push @array, new recs...; + my $r1 = pop @array; + unshift @array, new recs...; + my $r2 = shift @array; + @old_recs = splice @array, 3, 7, new recs...; + + untie @array; # all finished + + +=head1 DESCRIPTION + +C<Tie::File> represents a regular text file as a Perl array. Each +element in the array corresponds to a record in the file. The first +line of the file is element 0 of the array; the second line is element +1, and so on. + +The file is I<not> loaded into memory, so this will work even for +gigantic files. + +Changes to the array are reflected in the file immediately. + +Lazy people and beginners may now stop reading the manual. + +=head2 C<recsep> + +What is a 'record'? By default, the meaning is the same as for the +C<E<lt>...E<gt>> operator: It's a string terminated by C<$/>, which is +probably C<"\n">. (Minor exception: on DOS and Win32 systems, a +'record' is a string terminated by C<"\r\n">.) You may change the +definition of "record" by supplying the C<recsep> option in the C<tie> +call: + + tie @array, 'Tie::File', $file, recsep => 'es'; + +This says that records are delimited by the string C<es>. If the file +contained the following data: + + Curse these pesky flies!\n + +then the C<@array> would appear to have four elements: + + "Curse th" + "e p" + "ky fli" + "!\n" + +An undefined value is not permitted as a record separator. Perl's +special "paragraph mode" semantics (E<agrave> la C<$/ = "">) are not +emulated. + +Records read from the tied array do not have the record separator +string on the end; this is to allow + + $array[17] .= "extra"; + +to work as expected. + +(See L<"autochomp">, below.) Records stored into the array will have +the record separator string appended before they are written to the +file, if they don't have one already. For example, if the record +separator string is C<"\n">, then the following two lines do exactly +the same thing: + + $array[17] = "Cherry pie"; + $array[17] = "Cherry pie\n"; + +The result is that the contents of line 17 of the file will be +replaced with "Cherry pie"; a newline character will separate line 17 +from line 18. This means that this code will do nothing: + + chomp $array[17]; + +Because the C<chomp>ed value will have the separator reattached when +it is written back to the file. There is no way to create a file +whose trailing record separator string is missing. + +Inserting records that I<contain> the record separator string is not +supported by this module. It will probably produce a reasonable +result, but what this result will be may change in a future version. +Use 'splice' to insert records or to replace one record with several. + +=head2 C<autochomp> + +Normally, array elements have the record separator removed, so that if +the file contains the text + + Gold + Frankincense + Myrrh + +the tied array will appear to contain C<("Gold", "Frankincense", +"Myrrh")>. If you set C<autochomp> to a false value, the record +separator will not be removed. If the file above was tied with + + tie @gifts, "Tie::File", $gifts, autochomp => 0; + +then the array C<@gifts> would appear to contain C<("Gold\n", +"Frankincense\n", "Myrrh\n")>, or (on Win32 systems) C<("Gold\r\n", +"Frankincense\r\n", "Myrrh\r\n")>. + +=head2 C<mode> + +Normally, the specified file will be opened for read and write access, +and will be created if it does not exist. (That is, the flags +C<O_RDWR | O_CREAT> are supplied in the C<open> call.) If you want to +change this, you may supply alternative flags in the C<mode> option. +See L<Fcntl> for a listing of available flags. +For example: + + # open the file if it exists, but fail if it does not exist + use Fcntl 'O_RDWR'; + tie @array, 'Tie::File', $file, mode => O_RDWR; + + # create the file if it does not exist + use Fcntl 'O_RDWR', 'O_CREAT'; + tie @array, 'Tie::File', $file, mode => O_RDWR | O_CREAT; + + # open an existing file in read-only mode + use Fcntl 'O_RDONLY'; + tie @array, 'Tie::File', $file, mode => O_RDONLY; + +Opening the data file in write-only or append mode is not supported. + +=head2 C<memory> + +This is an upper limit on the amount of memory that C<Tie::File> will +consume at any time while managing the file. This is used for two +things: managing the I<read cache> and managing the I<deferred write +buffer>. + +Records read in from the file are cached, to avoid having to re-read +them repeatedly. If you read the same record twice, the first time it +will be stored in memory, and the second time it will be fetched from +the I<read cache>. The amount of data in the read cache will not +exceed the value you specified for C<memory>. If C<Tie::File> wants +to cache a new record, but the read cache is full, it will make room +by expiring the least-recently visited records from the read cache. + +The default memory limit is 2Mib. You can adjust the maximum read +cache size by supplying the C<memory> option. The argument is the +desired cache size, in bytes. + + # I have a lot of memory, so use a large cache to speed up access + tie @array, 'Tie::File', $file, memory => 20_000_000; + +Setting the memory limit to 0 will inhibit caching; records will be +fetched from disk every time you examine them. + +The C<memory> value is not an absolute or exact limit on the memory +used. C<Tie::File> objects contains some structures besides the read +cache and the deferred write buffer, whose sizes are not charged +against C<memory>. + +The cache itself consumes about 310 bytes per cached record, so if +your file has many short records, you may want to decrease the cache +memory limit, or else the cache overhead may exceed the size of the +cached data. + + +=head2 C<dw_size> + +(This is an advanced feature. Skip this section on first reading.) + +If you use deferred writing (See L<"Deferred Writing">, below) then +data you write into the array will not be written directly to the +file; instead, it will be saved in the I<deferred write buffer> to be +written out later. Data in the deferred write buffer is also charged +against the memory limit you set with the C<memory> option. + +You may set the C<dw_size> option to limit the amount of data that can +be saved in the deferred write buffer. This limit may not exceed the +total memory limit. For example, if you set C<dw_size> to 1000 and +C<memory> to 2500, that means that no more than 1000 bytes of deferred +writes will be saved up. The space available for the read cache will +vary, but it will always be at least 1500 bytes (if the deferred write +buffer is full) and it could grow as large as 2500 bytes (if the +deferred write buffer is empty.) + +If you don't specify a C<dw_size>, it defaults to the entire memory +limit. + +=head2 Option Format + +C<-mode> is a synonym for C<mode>. C<-recsep> is a synonym for +C<recsep>. C<-memory> is a synonym for C<memory>. You get the +idea. + +=head1 Public Methods + +The C<tie> call returns an object, say C<$o>. You may call + + $rec = $o->FETCH($n); + $o->STORE($n, $rec); + +to fetch or store the record at line C<$n>, respectively; similarly +the other tied array methods. (See L<perltie> for details.) You may +also call the following methods on this object: + +=head2 C<flock> + + $o->flock(MODE) + +will lock the tied file. C<MODE> has the same meaning as the second +argument to the Perl built-in C<flock> function; for example +C<LOCK_SH> or C<LOCK_EX | LOCK_NB>. (These constants are provided by +the C<use Fcntl ':flock'> declaration.) + +C<MODE> is optional; the default is C<LOCK_EX>. + +C<Tie::File> maintains an internal table of the byte offset of each +record it has seen in the file. + +When you use C<flock> to lock the file, C<Tie::File> assumes that the +read cache is no longer trustworthy, because another process might +have modified the file since the last time it was read. Therefore, a +successful call to C<flock> discards the contents of the read cache +and the internal record offset table. + +C<Tie::File> promises that the following sequence of operations will +be safe: + + my $o = tie @array, "Tie::File", $filename; + $o->flock; + +In particular, C<Tie::File> will I<not> read or write the file during +the C<tie> call. (Exception: Using C<mode =E<gt> O_TRUNC> will, of +course, erase the file during the C<tie> call. If you want to do this +safely, then open the file without C<O_TRUNC>, lock the file, and use +C<@array = ()>.) + +The best way to unlock a file is to discard the object and untie the +array. It is probably unsafe to unlock the file without also untying +it, because if you do, changes may remain unwritten inside the object. +That is why there is no shortcut for unlocking. If you really want to +unlock the file prematurely, you know what to do; if you don't know +what to do, then don't do it. + +All the usual warnings about file locking apply here. In particular, +note that file locking in Perl is B<advisory>, which means that +holding a lock will not prevent anyone else from reading, writing, or +erasing the file; it only prevents them from getting another lock at +the same time. Locks are analogous to green traffic lights: If you +have a green light, that does not prevent the idiot coming the other +way from plowing into you sideways; it merely guarantees to you that +the idiot does not also have a green light at the same time. + +=head2 C<autochomp> + + my $old_value = $o->autochomp(0); # disable autochomp option + my $old_value = $o->autochomp(1); # enable autochomp option + + my $ac = $o->autochomp(); # recover current value + +See L<"autochomp">, above. + +=head2 C<defer>, C<flush>, C<discard>, and C<autodefer> + +See L<"Deferred Writing">, below. + +=head2 C<offset> + + $off = $o->offset($n); + +This method returns the byte offset of the start of the C<$n>th record +in the file. If there is no such record, it returns an undefined +value. + +=head1 Tying to an already-opened filehandle + +If C<$fh> is a filehandle, such as is returned by C<IO::File> or one +of the other C<IO> modules, you may use: + + tie @array, 'Tie::File', $fh, ...; + +Similarly if you opened that handle C<FH> with regular C<open> or +C<sysopen>, you may use: + + tie @array, 'Tie::File', \*FH, ...; + +Handles that were opened write-only won't work. Handles that were +opened read-only will work as long as you don't try to modify the +array. Handles must be attached to seekable sources of data---that +means no pipes or sockets. If C<Tie::File> can detect that you +supplied a non-seekable handle, the C<tie> call will throw an +exception. (On Unix systems, it can detect this.) + +Note that Tie::File will only close any filehandles that it opened +internally. If you passed it a filehandle as above, you "own" the +filehandle, and are responsible for closing it after you have untied +the @array. + +=head1 Deferred Writing + +(This is an advanced feature. Skip this section on first reading.) + +Normally, modifying a C<Tie::File> array writes to the underlying file +immediately. Every assignment like C<$a[3] = ...> rewrites as much of +the file as is necessary; typically, everything from line 3 through +the end will need to be rewritten. This is the simplest and most +transparent behavior. Performance even for large files is reasonably +good. + +However, under some circumstances, this behavior may be excessively +slow. For example, suppose you have a million-record file, and you +want to do: + + for (@FILE) { + $_ = "> $_"; + } + +The first time through the loop, you will rewrite the entire file, +from line 0 through the end. The second time through the loop, you +will rewrite the entire file from line 1 through the end. The third +time through the loop, you will rewrite the entire file from line 2 to +the end. And so on. + +If the performance in such cases is unacceptable, you may defer the +actual writing, and then have it done all at once. The following loop +will perform much better for large files: + + (tied @a)->defer; + for (@a) { + $_ = "> $_"; + } + (tied @a)->flush; + +If C<Tie::File>'s memory limit is large enough, all the writing will +done in memory. Then, when you call C<-E<gt>flush>, the entire file +will be rewritten in a single pass. + +(Actually, the preceding discussion is something of a fib. You don't +need to enable deferred writing to get good performance for this +common case, because C<Tie::File> will do it for you automatically +unless you specifically tell it not to. See L<"autodeferring">, +below.) + +Calling C<-E<gt>flush> returns the array to immediate-write mode. If +you wish to discard the deferred writes, you may call C<-E<gt>discard> +instead of C<-E<gt>flush>. Note that in some cases, some of the data +will have been written already, and it will be too late for +C<-E<gt>discard> to discard all the changes. Support for +C<-E<gt>discard> may be withdrawn in a future version of C<Tie::File>. + +Deferred writes are cached in memory up to the limit specified by the +C<dw_size> option (see above). If the deferred-write buffer is full +and you try to write still more deferred data, the buffer will be +flushed. All buffered data will be written immediately, the buffer +will be emptied, and the now-empty space will be used for future +deferred writes. + +If the deferred-write buffer isn't yet full, but the total size of the +buffer and the read cache would exceed the C<memory> limit, the oldest +records will be expired from the read cache until the total size is +under the limit. + +C<push>, C<pop>, C<shift>, C<unshift>, and C<splice> cannot be +deferred. When you perform one of these operations, any deferred data +is written to the file and the operation is performed immediately. +This may change in a future version. + +If you resize the array with deferred writing enabled, the file will +be resized immediately, but deferred records will not be written. +This has a surprising consequence: C<@a = (...)> erases the file +immediately, but the writing of the actual data is deferred. This +might be a bug. If it is a bug, it will be fixed in a future version. + +=head2 Autodeferring + +C<Tie::File> tries to guess when deferred writing might be helpful, +and to turn it on and off automatically. + + for (@a) { + $_ = "> $_"; + } + +In this example, only the first two assignments will be done +immediately; after this, all the changes to the file will be deferred +up to the user-specified memory limit. + +You should usually be able to ignore this and just use the module +without thinking about deferring. However, special applications may +require fine control over which writes are deferred, or may require +that all writes be immediate. To disable the autodeferment feature, +use + + (tied @o)->autodefer(0); + +or + + tie @array, 'Tie::File', $file, autodefer => 0; + + +Similarly, C<-E<gt>autodefer(1)> re-enables autodeferment, and +C<-E<gt>autodefer()> recovers the current value of the autodefer setting. + + +=head1 CONCURRENT ACCESS TO FILES + +Caching and deferred writing are inappropriate if you want the same +file to be accessed simultaneously from more than one process. Other +optimizations performed internally by this module are also +incompatible with concurrent access. A future version of this module will +support a C<concurrent =E<gt> 1> option that enables safe concurrent access. + +Previous versions of this documentation suggested using C<memory +=E<gt> 0> for safe concurrent access. This was mistaken. Tie::File +will not support safe concurrent access before version 0.96. + +=head1 CAVEATS + +(That's Latin for 'warnings'.) + +=over 4 + +=item * + +Reasonable effort was made to make this module efficient. Nevertheless, +changing the size of a record in the middle of a large file will +always be fairly slow, because everything after the new record must be +moved. + +=item * + +The behavior of tied arrays is not precisely the same as for regular +arrays. For example: + + # This DOES print "How unusual!" + undef $a[10]; print "How unusual!\n" if defined $a[10]; + +C<undef>-ing a C<Tie::File> array element just blanks out the +corresponding record in the file. When you read it back again, you'll +get the empty string, so the supposedly-C<undef>'ed value will be +defined. Similarly, if you have C<autochomp> disabled, then + + # This DOES print "How unusual!" if 'autochomp' is disabled + undef $a[10]; + print "How unusual!\n" if $a[10]; + +Because when C<autochomp> is disabled, C<$a[10]> will read back as +C<"\n"> (or whatever the record separator string is.) + +There are other minor differences, particularly regarding C<exists> +and C<delete>, but in general, the correspondence is extremely close. + +=item * + +I have supposed that since this module is concerned with file I/O, +almost all normal use of it will be heavily I/O bound. This means +that the time to maintain complicated data structures inside the +module will be dominated by the time to actually perform the I/O. +When there was an opportunity to spend CPU time to avoid doing I/O, I +usually tried to take it. + +=item * + +You might be tempted to think that deferred writing is like +transactions, with C<flush> as C<commit> and C<discard> as +C<rollback>, but it isn't, so don't. + +=item * + +There is a large memory overhead for each record offset and for each +cache entry: about 310 bytes per cached data record, and about 21 bytes per offset table entry. + +The per-record overhead will limit the maximum number of records you +can access per file. Note that I<accessing> the length of the array +via C<$x = scalar @tied_file> accesses B<all> records and stores their +offsets. The same for C<foreach (@tied_file)>, even if you exit the +loop early. + +=back + +=head1 SUBCLASSING + +This version promises absolutely nothing about the internals, which +may change without notice. A future version of the module will have a +well-defined and stable subclassing API. + +=head1 WHAT ABOUT C<DB_File>? + +People sometimes point out that L<DB_File> will do something similar, +and ask why C<Tie::File> module is necessary. + +There are a number of reasons that you might prefer C<Tie::File>. +A list is available at C<http://perl.plover.com/TieFile/why-not-DB_File>. + +=head1 AUTHOR + +Mark Jason Dominus + +To contact the author, send email to: C<mjd-perl-tiefile+@plover.com> + +To receive an announcement whenever a new version of this module is +released, send a blank email message to +C<mjd-perl-tiefile-subscribe@plover.com>. + +The most recent version of this module, including documentation and +any news of importance, will be available at + + http://perl.plover.com/TieFile/ + + +=head1 LICENSE + +C<Tie::File> version 0.96 is copyright (C) 2003 Mark Jason Dominus. + +This library is free software; you may redistribute it and/or modify +it under the same terms as Perl itself. + +These terms are your choice of any of (1) the Perl Artistic Licence, +or (2) version 2 of the GNU General Public License as published by the +Free Software Foundation, or (3) any later version of the GNU General +Public License. + +This library is distributed in the hope that it will be useful, +but WITHOUT ANY WARRANTY; without even the implied warranty of +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +GNU General Public License for more details. + +You should have received a copy of the GNU General Public License +along with this library program; it should be in the file C<COPYING>. +If not, write to the Free Software Foundation, Inc., 51 Franklin Street, +Fifth Floor, Boston, MA 02110-1301, USA + +For licensing inquiries, contact the author at: + + Mark Jason Dominus + 255 S. Warnock St. + Philadelphia, PA 19107 + +=head1 WARRANTY + +C<Tie::File> version 0.98 comes with ABSOLUTELY NO WARRANTY. +For details, see the license. + +=head1 THANKS + +Gigantic thanks to Jarkko Hietaniemi, for agreeing to put this in the +core when I hadn't written it yet, and for generally being helpful, +supportive, and competent. (Usually the rule is "choose any one.") +Also big thanks to Abhijit Menon-Sen for all of the same things. + +Special thanks to Craig Berry and Peter Prymmer (for VMS portability +help), Randy Kobes (for Win32 portability help), Clinton Pierce and +Autrijus Tang (for heroic eleventh-hour Win32 testing above and beyond +the call of duty), Michael G Schwern (for testing advice), and the +rest of the CPAN testers (for testing generally). + +Special thanks to Tels for suggesting several speed and memory +optimizations. + +Additional thanks to: +Edward Avis / +Mattia Barbon / +Tom Christiansen / +Gerrit Haase / +Gurusamy Sarathy / +Jarkko Hietaniemi (again) / +Nikola Knezevic / +John Kominetz / +Nick Ing-Simmons / +Tassilo von Parseval / +H. Dieter Pearcey / +Slaven Rezic / +Eric Roode / +Peter Scott / +Peter Somu / +Autrijus Tang (again) / +Tels (again) / +Juerd Waalboer / +Todd Rinaldo + +=head1 TODO + +More tests. (Stuff I didn't think of yet.) + +Paragraph mode? + +Fixed-length mode. Leave-blanks mode. + +Maybe an autolocking mode? + +For many common uses of the module, the read cache is a liability. +For example, a program that inserts a single record, or that scans the +file once, will have a cache hit rate of zero. This suggests a major +optimization: The cache should be initially disabled. Here's a hybrid +approach: Initially, the cache is disabled, but the cache code +maintains statistics about how high the hit rate would be *if* it were +enabled. When it sees the hit rate get high enough, it enables +itself. The STAT comments in this code are the beginning of an +implementation of this. + +Record locking with fcntl()? Then the module might support an undo +log and get real transactions. What a tour de force that would be. + +Keeping track of the highest cached record. This would allow reads-in-a-row +to skip the cache lookup faster (if reading from 1..N with empty cache at +start, the last cached value will be always N-1). + +More tests. + +=cut + diff --git a/dist/Tie-File/t/00_version.t b/dist/Tie-File/t/00_version.t new file mode 100644 index 0000000000..84f326d750 --- /dev/null +++ b/dist/Tie-File/t/00_version.t @@ -0,0 +1,22 @@ +#!/usr/bin/perl + +print "1..1\n"; + +my $testversion = "0.98"; +use Tie::File; + +if ($Tie::File::VERSION != $testversion) { + print STDERR " + +*** WHOA THERE!!! *** + +You seem to be running version $Tie::File::VERSION of the module +against version $testversion of the test suite! + +None of the other test results will be reliable. +"; + exit 1; +} + +print "ok 1\n"; + diff --git a/dist/Tie-File/t/01_gen.t b/dist/Tie-File/t/01_gen.t new file mode 100644 index 0000000000..202b09c76a --- /dev/null +++ b/dist/Tie-File/t/01_gen.t @@ -0,0 +1,165 @@ +#!/usr/bin/perl + +$| = 1; +my $file = "tf$$.txt"; +1 while unlink $file; + +print "1..75\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +my $o = tie @a, 'Tie::File', $file, autochomp => 0, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$: = $o->{recsep}; + +# 3-5 create +$a[0] = 'rec0'; +check_contents("rec0"); + +# 6-11 append +$a[1] = 'rec1'; +check_contents("rec0", "rec1"); +$a[2] = 'rec2'; +check_contents("rec0", "rec1", "rec2"); + +# 12-20 same-length alterations +$a[0] = 'new0'; +check_contents("new0", "rec1", "rec2"); +$a[1] = 'new1'; +check_contents("new0", "new1", "rec2"); +$a[2] = 'new2'; +check_contents("new0", "new1", "new2"); + +# 21-35 lengthening alterations +$a[0] = 'long0'; +check_contents("long0", "new1", "new2"); +$a[1] = 'long1'; +check_contents("long0", "long1", "new2"); +$a[2] = 'long2'; +check_contents("long0", "long1", "long2"); +$a[1] = 'longer1'; +check_contents("long0", "longer1", "long2"); +$a[0] = 'longer0'; +check_contents("longer0", "longer1", "long2"); + +# 36-50 shortening alterations, including truncation +$a[0] = 'short0'; +check_contents("short0", "longer1", "long2"); +$a[1] = 'short1'; +check_contents("short0", "short1", "long2"); +$a[2] = 'short2'; +check_contents("short0", "short1", "short2"); +$a[1] = 'sh1'; +check_contents("short0", "sh1", "short2"); +$a[0] = 'sh0'; +check_contents("sh0", "sh1", "short2"); + +# (51-56) file with holes +$a[4] = 'rec4'; +check_contents("sh0", "sh1", "short2", "", "rec4"); +$a[3] = 'rec3'; +check_contents("sh0", "sh1", "short2", "rec3", "rec4"); + +# (57-59) zero out file +@a = (); +check_contents(); + +# (60-62) insert into the middle of an empty file +$a[3] = "rec3"; +check_contents("", "", "", "rec3"); + +# (63-68) 20020326 You thought there would be a bug in STORE where if +# a cached record was false, STORE wouldn't see it at all. But you +# forgot that records always come back from the cache with the record +# separator attached, so they are unlikely to be false. The only +# really weird case is when the cached record is empty and the record +# separator is "0". Test that in 09_gen_rs.t. +$a[1] = "0"; +check_contents("", "0", "", "rec3"); +$a[1] = "whoops"; +check_contents("", "whoops", "", "rec3"); + +# (69-72) make sure that undefs are treated correctly---they should +# be converted to empty records, and should not raise any warnings. +# (Some of these failed in 0.90. The change to _fixrec fixed them.) +# 20020331 +{ + my $good = 1; my $warn; + # If any of these raise warnings, we have a problem. + local $SIG{__WARN__} = sub { $good = 0; $warn = shift(); ctrlfix($warn)}; + local $^W = 1; + @a = (1); + $a[0] = undef; + print $good ? "ok $N\n" : "not ok $N # $warn\n"; + $N++; $good = 1; + print defined($a[0]) ? "ok $N\n" : "not ok $N\n"; + $N++; $good = 1; + $a[3] = '3'; + print defined($a[1]) ? "ok $N\n" : "not ok $N\n"; + $N++; $good = 1; + undef $a[3]; + print $good ? "ok $N\n" : "not ok $N # $warn\n"; + $N++; $good = 1; +} + +# (73-75) What if the user has tampered with $\ ? +{ { local $\ = "stop messing with the funny variables!"; + @a = (0..2); + } + check_contents(0..2); +} + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $:, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; +# my $open = open FH, "< $file"; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + my $msg; + for (0.. $#c) { + my $aa = $a[$_]; + unless ($aa eq "$c[$_]$:") { + $msg = "expected <$c[$_]$:>, got <$aa>"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # $msg\n"; + $N++; + + print $o->_check_integrity($file, $ENV{INTEGRITY}) + ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/02_fetchsize.t b/dist/Tie-File/t/02_fetchsize.t new file mode 100644 index 0000000000..12d2b51cba --- /dev/null +++ b/dist/Tie-File/t/02_fetchsize.t @@ -0,0 +1,53 @@ +#!/usr/bin/perl + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec1$:rec2$:rec3$:"; + +print "1..6\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + + +my $o = tie @a, 'Tie::File', $file, autochomp => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$: = $o->{recsep}; + +my $n; + +# 3 test array element count +$n = @a; +print $n == 3 ? "ok $N\n" : "not ok $N # n=$n\n"; +$N++; + +# 4 same thing again +$n = @a; +print $n == 3 ? "ok $N\n" : "not ok $N # n=$n\n"; +$N++; + +# 5 test $#a notation +$n = $#a; +print $n == 2 ? "ok $N\n" : "not ok $N # n=$n\n"; +$N++; + +# 6 test looping over array elements +my $q; +for (@a) { $q .= $_ } +print $q eq $data ? "ok $N\n" : "not ok $N # n=$n\n"; +$N++; + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/03_longfetch.t b/dist/Tie-File/t/03_longfetch.t new file mode 100644 index 0000000000..7d5a3886fe --- /dev/null +++ b/dist/Tie-File/t/03_longfetch.t @@ -0,0 +1,54 @@ +#!/usr/bin/perl +# +# Make sure we can fetch a record in the middle of the file +# before we've ever looked at any records before it +# +# Make sure fetching past the end of the file returns the undefined value +# +# (tests _fill_offsets_to() ) +# + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; + +print "1..8\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + + +my $o = tie @a, 'Tie::File', $file, autochomp => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$: = $o->{recsep}; + +my $n; + +# 3-5 +for (2, 1, 0) { + my $rec = $a[$_]; + print $rec eq "rec$_$:" ? "ok $N\n" : "not ok $N # rec=<$rec> ?\n"; + $N++; +} + +# 6-8 +for (3, 4, 6) { + my $rec = $a[$_]; + print ((not defined $rec) ? "ok $N\n" : "not ok $N # rec=<$rec> is defined\n"); + $N++; +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/04_splice.t b/dist/Tie-File/t/04_splice.t new file mode 100644 index 0000000000..b3880b758c --- /dev/null +++ b/dist/Tie-File/t/04_splice.t @@ -0,0 +1,264 @@ +#!/usr/bin/perl + +# +# Check SPLICE function's effect on the file +# (07_rv_splice.t checks its return value) +# +# Each call to 'check_contents' actually performs two tests. +# First, it calls the tied object's own 'check_integrity' method, +# which makes sure that the contents of the read cache and offset tables +# accurately reflect the contents of the file. +# Then, it checks the actual contents of the file against the expected +# contents. + + +$| = 1; +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; +print "1..118\n"; + +init_file($data); + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; # partial credit just for showing up + +my $o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$: = $o->{recsep}; +my $n; + +# (3-22) splicing at the beginning +splice(@a, 0, 0, "rec4"); +check_contents("rec4$:$data"); +splice(@a, 0, 1, "rec5"); # same length +check_contents("rec5$:$data"); +splice(@a, 0, 1, "record5"); # longer +check_contents("record5$:$data"); + +splice(@a, 0, 1, "r5"); # shorter +check_contents("r5$:$data"); +splice(@a, 0, 1); # removal +check_contents("$data"); +splice(@a, 0, 0); # no-op +check_contents("$data"); +splice(@a, 0, 0, 'r7', 'rec8'); # insert more than one +check_contents("r7$:rec8$:$data"); +splice(@a, 0, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("rec7$:record8$:rec9$:$data"); + +splice(@a, 0, 3, 'record9', 'rec10'); # delete more than insert +check_contents("record9$:rec10$:$data"); +splice(@a, 0, 2); # delete more than one +check_contents("$data"); + + +# (23-42) splicing in the middle +splice(@a, 1, 0, "rec4"); +check_contents("rec0$:rec4$:rec1$:rec2$:"); +splice(@a, 1, 1, "rec5"); # same length +check_contents("rec0$:rec5$:rec1$:rec2$:"); +splice(@a, 1, 1, "record5"); # longer +check_contents("rec0$:record5$:rec1$:rec2$:"); + +splice(@a, 1, 1, "r5"); # shorter +check_contents("rec0$:r5$:rec1$:rec2$:"); +splice(@a, 1, 1); # removal +check_contents("$data"); +splice(@a, 1, 0); # no-op +check_contents("$data"); +splice(@a, 1, 0, 'r7', 'rec8'); # insert more than one +check_contents("rec0$:r7$:rec8$:rec1$:rec2$:"); +splice(@a, 1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("rec0$:rec7$:record8$:rec9$:rec1$:rec2$:"); + +splice(@a, 1, 3, 'record9', 'rec10'); # delete more than insert +check_contents("rec0$:record9$:rec10$:rec1$:rec2$:"); +splice(@a, 1, 2); # delete more than one +check_contents("$data"); + +# (43-62) splicing at the end +splice(@a, 3, 0, "rec4"); +check_contents("$ {data}rec4$:"); +splice(@a, 3, 1, "rec5"); # same length +check_contents("$ {data}rec5$:"); +splice(@a, 3, 1, "record5"); # longer +check_contents("$ {data}record5$:"); + +splice(@a, 3, 1, "r5"); # shorter +check_contents("$ {data}r5$:"); +splice(@a, 3, 1); # removal +check_contents("$data"); +splice(@a, 3, 0); # no-op +check_contents("$data"); +splice(@a, 3, 0, 'r7', 'rec8'); # insert more than one +check_contents("$ {data}r7$:rec8$:"); +splice(@a, 3, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("$ {data}rec7$:record8$:rec9$:"); + +splice(@a, 3, 3, 'record9', 'rec10'); # delete more than insert +check_contents("$ {data}record9$:rec10$:"); +splice(@a, 3, 2); # delete more than one +check_contents("$data"); + +# (63-82) splicing with negative subscript +splice(@a, -1, 0, "rec4"); +check_contents("rec0$:rec1$:rec4$:rec2$:"); +splice(@a, -1, 1, "rec5"); # same length +check_contents("rec0$:rec1$:rec4$:rec5$:"); +splice(@a, -1, 1, "record5"); # longer +check_contents("rec0$:rec1$:rec4$:record5$:"); + +splice(@a, -1, 1, "r5"); # shorter +check_contents("rec0$:rec1$:rec4$:r5$:"); +splice(@a, -1, 1); # removal +check_contents("rec0$:rec1$:rec4$:"); +splice(@a, -1, 0); # no-op +check_contents("rec0$:rec1$:rec4$:"); +splice(@a, -1, 0, 'r7', 'rec8'); # insert more than one +check_contents("rec0$:rec1$:r7$:rec8$:rec4$:"); +splice(@a, -1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("rec0$:rec1$:r7$:rec8$:rec7$:record8$:rec9$:"); + +splice(@a, -3, 3, 'record9', 'rec10'); # delete more than insert +check_contents("rec0$:rec1$:r7$:rec8$:record9$:rec10$:"); +splice(@a, -4, 3); # delete more than one +check_contents("rec0$:rec1$:rec10$:"); + +# (83-84) scrub it all out +splice(@a, 0, 3); +check_contents(""); + +# (85-86) put some back in +splice(@a, 0, 0, "rec0", "rec1"); +check_contents("rec0$:rec1$:"); + +# (87-88) what if we remove too many records? +splice(@a, 0, 17); +check_contents(""); + +# (89-92) In the past, splicing past the end was not correctly detected +# (1.14) +splice(@a, 89, 3); +check_contents(""); +splice(@a, @a, 3); +check_contents(""); + +# (93-96) Also we did not emulate splice's freaky behavior when inserting +# past the end of the array (1.14) +splice(@a, 89, 0, "I", "like", "pie"); +check_contents("I$:like$:pie$:"); +splice(@a, 89, 0, "pie pie pie"); +check_contents("I$:like$:pie$:pie pie pie$:"); + +# (97) Splicing with too large a negative number should be fatal +# This test ignored because it causes 5.6.1 and 5.7.3 to dump core +# It also garbles the stack under 5.005_03 (20020401) +# NOT MY FAULT +if ($] > 5.007003) { + eval { splice(@a, -7, 0) }; + print $@ =~ /^Modification of non-creatable array value attempted, subscript -7/ + ? "ok $N\n" : "not ok $N \# \$\@ was '$@'\n"; +} else { + print "ok $N \# skipped (versions through 5.7.3 dump core here.)\n"; +} +$N++; + +# (98-101) Test default arguments +splice @a, 0, 0, (0..11); +splice @a, 4; +check_contents("0$:1$:2$:3$:"); +splice @a; +check_contents(""); + +# (102-103) I think there's a bug here---it will fail to clear the EOF flag +@a = (0..11); +splice @a, -1, 1000; +check_contents("0$:1$:2$:3$:4$:5$:6$:7$:8$:9$:10$:"); + +# (104-106) make sure that undefs are treated correctly---they should +# be converted to empty records, and should not raise any warnings. +# (Some of these failed in 0.90. The change to _fixrec fixed them.) +# 20020331 +{ + my $good = 1; my $warn; + # If any of these raise warnings, we have a problem. + local $SIG{__WARN__} = sub { $good = 0; $warn = shift(); ctrlfix($warn)}; + local $^W = 1; + @a = (1); + splice @a, 1, 0, undef, undef, undef; + print $good ? "ok $N\n" : "not ok $N # $warn\n"; + $N++; $good = 1; + print defined($a[2]) ? "ok $N\n" : "not ok $N\n"; + $N++; $good = 1; + my @r = splice @a, 2; + print defined($r[0]) ? "ok $N\n" : "not ok $N\n"; + $N++; $good = 1; +} + +# (107-118) splice with negative length was treated wrong +# 20020402 Reported by Juerd Waalboer +@a = (0..8) ; +splice @a, 0, -3; +check_contents("6$:7$:8$:"); +@a = (0..8) ; +splice @a, 1, -3; +check_contents("0$:6$:7$:8$:"); +@a = (0..8) ; +splice @a, 7, -3; +check_contents("0$:1$:2$:3$:4$:5$:6$:7$:8$:"); +@a = (0..2) ; +splice @a, 0, -3; +check_contents("0$:1$:2$:"); +@a = (0..2) ; +splice @a, 1, -3; +check_contents("0$:1$:2$:"); +@a = (0..2) ; +splice @a, 7, -3; +check_contents("0$:1$:2$:"); + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +use POSIX 'SEEK_SET'; +sub check_contents { + my $x = shift; + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/05_size.t b/dist/Tie-File/t/05_size.t new file mode 100644 index 0000000000..44c69f910f --- /dev/null +++ b/dist/Tie-File/t/05_size.t @@ -0,0 +1,129 @@ +#!/usr/bin/perl +# +# Check FETCHSIZE and SETSIZE functions +# PUSH POP SHIFT UNSHIFT +# + +use POSIX 'SEEK_SET'; + +my $file = "tf$$.txt"; +my ($o, $n); + +print "1..16\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +# 2-3 FETCHSIZE 0-length file +open F, "> $file" or die $!; +binmode F; +close F; +$o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$: = $o->{recsep}; + +$n = @a; +print $n == 0 ? "ok $N\n" : "not ok $N # $n, s/b 0\n"; +$N++; + +# Reset everything +undef $o; +untie @a; + +my $data = "rec0$:rec1$:rec2$:"; +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + +$o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# 4-5 FETCHSIZE positive-length file +$n = @a; +print $n == 3 ? "ok $N\n" : "not ok $N # $n, s/b 0\n"; +$N++; + +# STORESIZE +# (6-7) Make it longer: +populate(); +$#a = 4; +check_contents("$data$:$:"); + +# (8-9) Make it longer again: +populate(); +$#a = 6; +check_contents("$data$:$:$:$:"); + +# (10-11) Make it shorter: +populate(); +$#a = 4; +check_contents("$data$:$:"); + +# (12-13) Make it shorter again: +populate(); +$#a = 2; +check_contents($data); + +# (14-15) Get rid of it completely: +populate(); +$#a = -1; +check_contents(''); + +# (16) 20020324 I have an idea that shortening the array will not +# expunge a cached record at the end if one is present. +$o->defer; +$a[3] = "record"; +my $r = $a[3]; +$#a = -1; +$r = $a[3]; +print (! defined $r ? "ok $N\n" : "not ok $N \# was <$r>; should be UNDEF\n"); +# Turns out not to be the case---STORESIZE explicitly removes them later +# 20020326 Well, but happily, this test did fail today. + +# In the past, there was a bug in STORESIZE that it didn't correctly +# remove deleted records from the cache. This wasn't detected +# because these tests were all done with an empty cache. populate() +# will ensure that the cache is fully populated. +sub populate { + my $z; + $z = $a[$_] for 0 .. $#a; +} + +sub check_contents { + my $x = shift; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N \# integrity\n"; + $N++; +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/06_fixrec.t b/dist/Tie-File/t/06_fixrec.t new file mode 100644 index 0000000000..bf24be1300 --- /dev/null +++ b/dist/Tie-File/t/06_fixrec.t @@ -0,0 +1,53 @@ +#!/usr/bin/perl + +use POSIX 'SEEK_SET'; +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); + +print "1..5\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +my $o = tie @a, 'Tie::File', $file, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$a[0] = 'rec0'; +check_contents("rec0$:"); +$a[1] = "rec1$:"; +check_contents("rec0$:rec1$:"); +$a[2] = "rec2$:$:"; # should we detect this? +check_contents("rec0$:rec1$:rec2$:$:"); + +sub check_contents { + my $x = shift; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + my $msg = "not ok $N # expected <$x>, got <$a>"; + ctrlfix($msg); + print "$msg\n"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/07_rv_splice.t b/dist/Tie-File/t/07_rv_splice.t new file mode 100644 index 0000000000..e5c09b1a48 --- /dev/null +++ b/dist/Tie-File/t/07_rv_splice.t @@ -0,0 +1,205 @@ +#!/usr/bin/perl +# +# Check SPLICE function's return value +# (04_splice.t checks its effect on the file) +# + + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; + +print "1..56\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; # partial credit just for showing up + +init_file($data); + +my $o = tie @a, 'Tie::File', $file, autochomp => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +my $n; + +# (3-12) splicing at the beginning +@r = splice(@a, 0, 0, "rec4"); +check_result(); +@r = splice(@a, 0, 1, "rec5"); # same length +check_result("rec4"); +@r = splice(@a, 0, 1, "record5"); # longer +check_result("rec5"); + +@r = splice(@a, 0, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, 0, 1); # removal +check_result("r5"); +@r = splice(@a, 0, 0); # no-op +check_result(); +@r = splice(@a, 0, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 0, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 0, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 0, 2); # delete more than one +check_result('record9', 'rec10'); + + +# (13-22) splicing in the middle +@r = splice(@a, 1, 0, "rec4"); +check_result(); +@r = splice(@a, 1, 1, "rec5"); # same length +check_result('rec4'); +@r = splice(@a, 1, 1, "record5"); # longer +check_result('rec5'); + +@r = splice(@a, 1, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, 1, 1); # removal +check_result("r5"); +@r = splice(@a, 1, 0); # no-op +check_result(); +@r = splice(@a, 1, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 1, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 1, 2); # delete more than one +check_result('record9','rec10'); + +# (23-32) splicing at the end +@r = splice(@a, 3, 0, "rec4"); +check_result(); +@r = splice(@a, 3, 1, "rec5"); # same length +check_result('rec4'); +@r = splice(@a, 3, 1, "record5"); # longer +check_result('rec5'); + +@r = splice(@a, 3, 1, "r5"); # shorter +check_result('record5'); +@r = splice(@a, 3, 1); # removal +check_result('r5'); +@r = splice(@a, 3, 0); # no-op +check_result(); +@r = splice(@a, 3, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 3, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 3, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 3, 2); # delete more than one +check_result('record9', 'rec10'); + +# (33-42) splicing with negative subscript +@r = splice(@a, -1, 0, "rec4"); +check_result(); +@r = splice(@a, -1, 1, "rec5"); # same length +check_result('rec2'); +@r = splice(@a, -1, 1, "record5"); # longer +check_result("rec5"); + +@r = splice(@a, -1, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, -1, 1); # removal +check_result("r5"); +@r = splice(@a, -1, 0); # no-op +check_result(); +@r = splice(@a, -1, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, -1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('rec4'); + +@r = splice(@a, -3, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, -4, 3); # delete more than one +check_result('r7', 'rec8', 'record9'); + +# (43) scrub it all out +@r = splice(@a, 0, 3); +check_result('rec0', 'rec1', 'rec10'); + +# (44) put some back in +@r = splice(@a, 0, 0, "rec0", "rec1"); +check_result(); + +# (45) what if we remove too many records? +@r = splice(@a, 0, 17); +check_result('rec0', 'rec1'); + +# (46-48) Now check the scalar context return +splice(@a, 0, 0, qw(I like pie)); +my $r; +$r = splice(@a, 0, 0); +print !defined($r) ? "ok $N\n" : "not ok $N \# return should have been undef, was <$r>\n"; +$N++; + +$r = splice(@a, 2, 1); +print $r eq "pie$:" ? "ok $N\n" : "not ok $N \# return should have been 'pie\\n', was <$r>\n"; +$N++; + +$r = splice(@a, 0, 2); +print $r eq "like$:" ? "ok $N\n" : "not ok $N \# return should have been 'like\\n', was <$r>\n"; +$N++; + +# (49-50) Test default arguments +splice @a, 0, 0, (0..11); +@r = splice @a, 4; +check_result(4..11); +@r = splice @a; +check_result(0..3); + +# (51-56) splice with negative length was treated wrong +# 20020402 Reported by Juerd Waalboer +@a = (0..8) ; +@r = splice @a, 0, -3; +check_result(0..5); +@a = (0..8) ; +@r = splice @a, 1, -3; +check_result(1..5); +@a = (0..8) ; +@r = splice @a, 7, -3; +check_result(); +@a = (0..2) ; +@r = splice @a, 0, -3; +check_result(); +@a = (0..2) ; +@r = splice @a, 1, -3; +check_result(); +@a = (0..2) ; +@r = splice @a, 7, -3; +check_result(); + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +# actual results are in @r. +# expected results are in @_ +sub check_result { + my @x = @_; + s/$:$// for @r; + my $good = 1; + $good = 0 unless @r == @x; + for my $i (0 .. $#r) { + $good = 0 unless $r[$i] eq $x[$i]; + } + print $good ? "ok $N\n" : "not ok $N \# was (@r); should be (@x)\n"; + $N++; +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/08_ro.t b/dist/Tie-File/t/08_ro.t new file mode 100644 index 0000000000..5fd8933bf8 --- /dev/null +++ b/dist/Tie-File/t/08_ro.t @@ -0,0 +1,86 @@ +#!/usr/bin/perl +# +# Make sure it works to open the file in read-only mode +# + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); + +print "1..13\n"; + +my $N = 1; +use Tie::File; +use Fcntl 'O_RDONLY'; +print "ok $N\n"; $N++; + +my @items = qw(Gold Frankincense Myrrh Ivory Apes Peacocks); +init_file(join $:, @items, ''); + +my $o = tie @a, 'Tie::File', $file, mode => O_RDONLY, autochomp => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$#a == $#items ? print "ok $N\n" : print "not ok $N\n"; +$N++; + +for my $i (0..$#items) { + ("$items[$i]$:" eq $a[$i]) ? print "ok $N\n" : print "not ok $N\n"; + $N++; +} + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +undef $o; untie @a; +my $badrec = "Malformed"; +# (10-13) When a record lacks the record seprator, we sneakily try +# to fix it. How does that work when the file is read-only? +if (setup_badly_terminated_file(4)) { + my $good = 1; + my $warn; + local $SIG{__WARN__} = sub { $good = 0; ctrlfix($warn = shift); }; + local $^W = 1; + my $o = tie @a, 'Tie::File', $file, mode => O_RDONLY, autochomp => 0 + or die "Couldn't tie $file: $!"; + + print $a[0] eq "Malformed$:" ? "ok $N\n" : "not ok $N\n"; $N++; + print $good ? "ok $N\n" : "not ok $N # $warn\n"; $good = 1; $N++; + print $a[0] eq "Malformed$:" ? "ok $N\n" : "not ok $N\n"; $N++; + print $good ? "ok $N\n" : "not ok $N # $warn\n"; $good = 1; $N++; +} + +sub setup_badly_terminated_file { + my $NTESTS = shift; + open F, "> $file" or die "Couldn't open $file: $!"; + binmode F; + print F $badrec; + close F; + unless (-s $file == length $badrec) { + for (1 .. $NTESTS) { + print "ok $N \# skipped - can't create improperly terminated file\n"; + $N++; + } + return; + } + return 1; +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/09_gen_rs.t b/dist/Tie-File/t/09_gen_rs.t new file mode 100644 index 0000000000..e590210335 --- /dev/null +++ b/dist/Tie-File/t/09_gen_rs.t @@ -0,0 +1,225 @@ +#!/usr/bin/perl + +my $file = "tf$$.txt"; + +print "1..59\n"; + +use Fcntl 'O_RDONLY'; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +$RECSEP = 'blah'; +my $o = tie @a, 'Tie::File', $file, + recsep => $RECSEP, autochomp => 0, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + + +# 3-4 create +$a[0] = 'rec0'; +check_contents("rec0"); + +# 5-8 append +$a[1] = 'rec1'; +check_contents("rec0", "rec1"); +$a[2] = 'rec2'; +check_contents("rec0", "rec1", "rec2"); + +# 9-14 same-length alterations +$a[0] = 'new0'; +check_contents("new0", "rec1", "rec2"); +$a[1] = 'new1'; +check_contents("new0", "new1", "rec2"); +$a[2] = 'new2'; +check_contents("new0", "new1", "new2"); + +# 15-24 lengthening alterations +$a[0] = 'long0'; +check_contents("long0", "new1", "new2"); +$a[1] = 'long1'; +check_contents("long0", "long1", "new2"); +$a[2] = 'long2'; +check_contents("long0", "long1", "long2"); +$a[1] = 'longer1'; +check_contents("long0", "longer1", "long2"); +$a[0] = 'longer0'; +check_contents("longer0", "longer1", "long2"); + +# 25-34 shortening alterations, including truncation +$a[0] = 'short0'; +check_contents("short0", "longer1", "long2"); +$a[1] = 'short1'; +check_contents("short0", "short1", "long2"); +$a[2] = 'short2'; +check_contents("short0", "short1", "short2"); +$a[1] = 'sh1'; +check_contents("short0", "sh1", "short2"); +$a[0] = 'sh0'; +check_contents("sh0", "sh1", "short2"); + +# (35-38) file with holes +$a[4] = 'rec4'; +check_contents("sh0", "sh1", "short2", "", "rec4"); +$a[3] = 'rec3'; +check_contents("sh0", "sh1", "short2", "rec3", "rec4"); + +# (39-40) zero out file +@a = (); +check_contents(); + +# (41-42) insert into the middle of an empty file +$a[3] = "rec3"; +check_contents("", "", "", "rec3"); + +# (43-47) 20020326 You thought there would be a bug in STORE where if +# a cached record was false, STORE wouldn't see it at all. Yup, there is, +# and adding the appropriate defined() test fixes the problem. +undef $o; untie @a; 1 while unlink $file; +$RECSEP = '0'; +$o = tie @a, 'Tie::File', $file, + recsep => $RECSEP, autochomp => 0, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; +$#a = 2; +my $z = $a[1]; # caches "0" +$a[2] = "oops"; +check_contents("", "", "oops"); +$a[1] = "bah"; +check_contents("", "bah", "oops"); +undef $o; untie @a; + +# (48-56) 20020331 Make sure we correctly handle the case where the final +# record of the file is not properly terminated, Through version 0.90, +# we would mangle the file. +my $badrec = "Malformed"; +$: = $RECSEP = Tie::File::_default_recsep(); +# (48-50) +if (setup_badly_terminated_file(3)) { + $o = tie @a, 'Tie::File', $file, + recsep => $RECSEP, autochomp => 0, autodefer => 0 + or die "Couldn't tie file: $!"; + my $z = $a[0]; + print $z eq "$badrec$:" ? "ok $N\n" : + "not ok $N \# got $z, expected $badrec\n"; + $N++; + push @a, "next"; + check_contents($badrec, "next"); +} +# (51-52) +if (setup_badly_terminated_file(2)) { + $o = tie @a, 'Tie::File', $file, + recsep => $RECSEP, autochomp => 0, autodefer => 0 + or die "Couldn't tie file: $!"; + splice @a, 1, 0, "x", "y"; + check_contents($badrec, "x", "y"); +} +# (53-56) +if (setup_badly_terminated_file(4)) { + $o = tie @a, 'Tie::File', $file, + recsep => $RECSEP, autochomp => 0, autodefer => 0 + or die "Couldn't tie file: $!"; + my @r = splice @a, 0, 1, "x", "y"; + my $n = @r; + print $n == 1 ? "ok $N\n" : "not ok $N \# expected 1 elt, got $n\n"; + $N++; + print $r[0] eq "$badrec$:" ? "ok $N\n" + : "not ok $N \# expected <$badrec>, got <$r[0]>\n"; + $N++; + check_contents("x", "y"); +} + +# (57-58) 20020402 The modification would have failed if $\ were set wrong. +# I hate $\. +if (setup_badly_terminated_file(2)) { + $o = tie @a, 'Tie::File', $file, + recsep => $RECSEP, autochomp => 0, autodefer => 0 + or die "Couldn't tie file: $!"; + { local $\ = "I hate \$\\."; + my $z = $a[0]; + } + check_contents($badrec); +} + +# (59) 20030527 Tom Christiansen pointed out that FETCH returns the wrong +# data on the final record of an unterminated file if the file is opened +# in read-only mode. Note that the $#a is necessary here. +# There's special-case code to fix the final record when it is read normally. +# But the $#a forces it to be read from the cache, which skips the +# termination. +$badrec = "world${RECSEP}hello"; +if (setup_badly_terminated_file(1)) { + tie(@a, "Tie::File", $file, mode => O_RDONLY, recsep => $RECSEP) + or die "Couldn't tie file: $!"; + my $z = $#a; + $z = $a[1]; + print $z eq "hello" ? "ok $N\n" : + "not ok $N \# got $z, expected hello\n"; + $N++; +} + +sub setup_badly_terminated_file { + my $NTESTS = shift; + open F, "> $file" or die "Couldn't open $file: $!"; + binmode F; + print F $badrec; + close F; + unless (-s $file == length $badrec) { + for (1 .. $NTESTS) { + print "ok $N \# skipped - can't create improperly terminated file\n"; + $N++; + } + return; + } + return 1; +} + + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $RECSEP, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + my $msg = "# expected <$x>, got <$a>"; + ctrlfix($msg); + print "not ok $N $msg\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + for (0.. $#c) { + unless ($a[$_] eq "$c[$_]$RECSEP") { + $msg = "expected $c[$_]$RECSEP, got $a[$_]"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # fetch $msg\n"; + $N++; +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/10_splice_rs.t b/dist/Tie-File/t/10_splice_rs.t new file mode 100644 index 0000000000..50b8b0a7ee --- /dev/null +++ b/dist/Tie-File/t/10_splice_rs.t @@ -0,0 +1,215 @@ +#!/usr/bin/perl +# +# Check SPLICE function's effect on the file +# (07_rv_splice.t checks its return value) +# +# Each call to 'check_contents' actually performs two tests. +# First, it calls the tied object's own 'check_integrity' method, +# which makes sure that the contents of the read cache and offset tables +# accurately reflect the contents of the file. +# Then, it checks the actual contents of the file against the expected +# contents. + +use POSIX 'SEEK_SET'; + +my $file = "tf$$.txt"; +my $data = "rec0blahrec1blahrec2blah"; + +print "1..101\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; # partial credit just for showing up + +init_file($data); + +my $o = tie @a, 'Tie::File', $file, recsep => 'blah'; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +my $n; + +# (3-22) splicing at the beginning +splice(@a, 0, 0, "rec4"); +check_contents("rec4blah$data"); +splice(@a, 0, 1, "rec5"); # same length +check_contents("rec5blah$data"); +splice(@a, 0, 1, "record5"); # longer +check_contents("record5blah$data"); + +splice(@a, 0, 1, "r5"); # shorter +check_contents("r5blah$data"); +splice(@a, 0, 1); # removal +check_contents("$data"); +splice(@a, 0, 0); # no-op +check_contents("$data"); +splice(@a, 0, 0, 'r7', 'rec8'); # insert more than one +check_contents("r7blahrec8blah$data"); +splice(@a, 0, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("rec7blahrecord8blahrec9blah$data"); + +splice(@a, 0, 3, 'record9', 'rec10'); # delete more than insert +check_contents("record9blahrec10blah$data"); +splice(@a, 0, 2); # delete more than one +check_contents("$data"); + + +# (23-42) splicing in the middle +splice(@a, 1, 0, "rec4"); +check_contents("rec0blahrec4blahrec1blahrec2blah"); +splice(@a, 1, 1, "rec5"); # same length +check_contents("rec0blahrec5blahrec1blahrec2blah"); +splice(@a, 1, 1, "record5"); # longer +check_contents("rec0blahrecord5blahrec1blahrec2blah"); + +splice(@a, 1, 1, "r5"); # shorter +check_contents("rec0blahr5blahrec1blahrec2blah"); +splice(@a, 1, 1); # removal +check_contents("$data"); +splice(@a, 1, 0); # no-op +check_contents("$data"); +splice(@a, 1, 0, 'r7', 'rec8'); # insert more than one +check_contents("rec0blahr7blahrec8blahrec1blahrec2blah"); +splice(@a, 1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("rec0blahrec7blahrecord8blahrec9blahrec1blahrec2blah"); + +splice(@a, 1, 3, 'record9', 'rec10'); # delete more than insert +check_contents("rec0blahrecord9blahrec10blahrec1blahrec2blah"); +splice(@a, 1, 2); # delete more than one +check_contents("$data"); + +# (43-62) splicing at the end +splice(@a, 3, 0, "rec4"); +check_contents("$ {data}rec4blah"); +splice(@a, 3, 1, "rec5"); # same length +check_contents("$ {data}rec5blah"); +splice(@a, 3, 1, "record5"); # longer +check_contents("$ {data}record5blah"); + +splice(@a, 3, 1, "r5"); # shorter +check_contents("$ {data}r5blah"); +splice(@a, 3, 1); # removal +check_contents("$data"); +splice(@a, 3, 0); # no-op +check_contents("$data"); +splice(@a, 3, 0, 'r7', 'rec8'); # insert more than one +check_contents("$ {data}r7blahrec8blah"); +splice(@a, 3, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("$ {data}rec7blahrecord8blahrec9blah"); + +splice(@a, 3, 3, 'record9', 'rec10'); # delete more than insert +check_contents("$ {data}record9blahrec10blah"); +splice(@a, 3, 2); # delete more than one +check_contents("$data"); + +# (63-82) splicing with negative subscript +splice(@a, -1, 0, "rec4"); +check_contents("rec0blahrec1blahrec4blahrec2blah"); +splice(@a, -1, 1, "rec5"); # same length +check_contents("rec0blahrec1blahrec4blahrec5blah"); +splice(@a, -1, 1, "record5"); # longer +check_contents("rec0blahrec1blahrec4blahrecord5blah"); + +splice(@a, -1, 1, "r5"); # shorter +check_contents("rec0blahrec1blahrec4blahr5blah"); +splice(@a, -1, 1); # removal +check_contents("rec0blahrec1blahrec4blah"); +splice(@a, -1, 0); # no-op +check_contents("rec0blahrec1blahrec4blah"); +splice(@a, -1, 0, 'r7', 'rec8'); # insert more than one +check_contents("rec0blahrec1blahr7blahrec8blahrec4blah"); +splice(@a, -1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_contents("rec0blahrec1blahr7blahrec8blahrec7blahrecord8blahrec9blah"); + +splice(@a, -3, 3, 'record9', 'rec10'); # delete more than insert +check_contents("rec0blahrec1blahr7blahrec8blahrecord9blahrec10blah"); +splice(@a, -4, 3); # delete more than one +check_contents("rec0blahrec1blahrec10blah"); + +# (83-84) scrub it all out +splice(@a, 0, 3); +check_contents(""); + +# (85-86) put some back in +splice(@a, 0, 0, "rec0", "rec1"); +check_contents("rec0blahrec1blah"); + +# (87-88) what if we remove too many records? +splice(@a, 0, 17); +check_contents(""); + +# (89-92) In the past, splicing past the end was not correctly detected +# (0.14) +splice(@a, 89, 3); +check_contents(""); +splice(@a, @a, 3); +check_contents(""); + +# (93-96) Also we did not emulate splice's freaky behavior when inserting +# past the end of the array (1.14) +splice(@a, 89, 0, "I", "like", "pie"); +check_contents("Iblahlikeblahpieblah"); +splice(@a, 89, 0, "pie pie pie"); +check_contents("Iblahlikeblahpieblahpie pie pieblah"); + +# (97) Splicing with too large a negative number should be fatal +# This test ignored because it causes 5.6.1 and 5.7.3 to dump core +# It also garbles the stack under 5.005_03 (20020401) +# NOT MY FAULT +if ($] > 5.007003) { + eval { splice(@a, -7, 0) }; + print $@ =~ /^Modification of non-creatable array value attempted, subscript -7/ + ? "ok $N\n" : "not ok $N \# \$\@ was '$@'\n"; +} else { + print "ok $N \# skipped (versions through 5.7.3 dump core here.)\n"; +} +$N++; + +# (98-101) Test default arguments +splice @a, 0, 0, (0..11); +splice @a, 4; +check_contents("0blah1blah2blah3blah"); +splice @a; +check_contents(""); + + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +sub check_contents { + my $x = shift; + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix(my $msg = "# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/11_rv_splice_rs.t b/dist/Tie-File/t/11_rv_splice_rs.t new file mode 100644 index 0000000000..ae1053802a --- /dev/null +++ b/dist/Tie-File/t/11_rv_splice_rs.t @@ -0,0 +1,182 @@ +#!/usr/bin/perl +# +# Check SPLICE function's return value +# (04_splice.t checks its effect on the file) +# + +my $file = "tf$$.txt"; +my $data = "rec0blahrec1blahrec2blah"; + +print "1..50\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; # partial credit just for showing up + +init_file($data); + +my $o = tie @a, 'Tie::File', $file, autochomp => 0, recsep => 'blah'; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +my $n; + +# (3-12) splicing at the beginning +@r = splice(@a, 0, 0, "rec4"); +check_result(); +@r = splice(@a, 0, 1, "rec5"); # same length +check_result("rec4"); +@r = splice(@a, 0, 1, "record5"); # longer +check_result("rec5"); + +@r = splice(@a, 0, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, 0, 1); # removal +check_result("r5"); +@r = splice(@a, 0, 0); # no-op +check_result(); +@r = splice(@a, 0, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 0, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 0, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 0, 2); # delete more than one +check_result('record9', 'rec10'); + + +# (13-22) splicing in the middle +@r = splice(@a, 1, 0, "rec4"); +check_result(); +@r = splice(@a, 1, 1, "rec5"); # same length +check_result('rec4'); +@r = splice(@a, 1, 1, "record5"); # longer +check_result('rec5'); + +@r = splice(@a, 1, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, 1, 1); # removal +check_result("r5"); +@r = splice(@a, 1, 0); # no-op +check_result(); +@r = splice(@a, 1, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 1, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 1, 2); # delete more than one +check_result('record9','rec10'); + +# (23-32) splicing at the end +@r = splice(@a, 3, 0, "rec4"); +check_result(); +@r = splice(@a, 3, 1, "rec5"); # same length +check_result('rec4'); +@r = splice(@a, 3, 1, "record5"); # longer +check_result('rec5'); + +@r = splice(@a, 3, 1, "r5"); # shorter +check_result('record5'); +@r = splice(@a, 3, 1); # removal +check_result('r5'); +@r = splice(@a, 3, 0); # no-op +check_result(); +@r = splice(@a, 3, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 3, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 3, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 3, 2); # delete more than one +check_result('record9', 'rec10'); + +# (33-42) splicing with negative subscript +@r = splice(@a, -1, 0, "rec4"); +check_result(); +@r = splice(@a, -1, 1, "rec5"); # same length +check_result('rec2'); +@r = splice(@a, -1, 1, "record5"); # longer +check_result("rec5"); + +@r = splice(@a, -1, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, -1, 1); # removal +check_result("r5"); +@r = splice(@a, -1, 0); # no-op +check_result(); +@r = splice(@a, -1, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, -1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('rec4'); + +@r = splice(@a, -3, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, -4, 3); # delete more than one +check_result('r7', 'rec8', 'record9'); + +# (43) scrub it all out +@r = splice(@a, 0, 3); +check_result('rec0', 'rec1', 'rec10'); + +# (44) put some back in +@r = splice(@a, 0, 0, "rec0", "rec1"); +check_result(); + +# (45) what if we remove too many records? +@r = splice(@a, 0, 17); +check_result('rec0', 'rec1'); + +# (46-48) Now check the scalar context return +splice(@a, 0, 0, qw(I like pie)); +my $r; +$r = splice(@a, 0, 0); +print !defined($r) ? "ok $N\n" : "not ok $N \# return should have been undef\n"; +$N++; + +$r = splice(@a, 2, 1); +print $r eq "pieblah" ? "ok $N\n" : "not ok $N \# return should have been 'pie'\n"; +$N++; + +$r = splice(@a, 0, 2); +print $r eq "likeblah" ? "ok $N\n" : "not ok $N \# return should have been 'like'\n"; +$N++; + +# (49-50) Test default arguments +splice @a, 0, 0, (0..11); +@r = splice @a, 4; +check_result(4..11); +@r = splice @a; +check_result(0..3); + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +# actual results are in @r. +# expected results are in @_ +sub check_result { + my @x = @_; + s/blah$// for @r; + my $good = 1; + $good = 0 unless @r == @x; + for my $i (0 .. $#r) { + $good = 0 unless $r[$i] eq $x[$i]; + } + print $good ? "ok $N\n" : "not ok $N \# was (@r); should be (@x)\n"; + $N++; +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/12_longfetch_rs.t b/dist/Tie-File/t/12_longfetch_rs.t new file mode 100644 index 0000000000..6f1905d6af --- /dev/null +++ b/dist/Tie-File/t/12_longfetch_rs.t @@ -0,0 +1,41 @@ +#!/usr/bin/perl +# +# Make sure we can fetch a record in the middle of the file +# before we've ever looked at any records before it +# +# (tests _fill_offsets_to() ) +# + +my $file = "tf$$.txt"; +my $data = "rec0blahrec1blahrec2blah"; + +print "1..5\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + + +my $o = tie @a, 'Tie::File', $file, autochomp => 0, recsep => 'blah'; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +my $n; + +# 3-5 +for (2, 1, 0) { + print $a[$_] eq "rec${_}blah" ? "ok $N\n" : "not ok $N # rec=$a[$_] ?\n"; + $N++; +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/13_size_rs.t b/dist/Tie-File/t/13_size_rs.t new file mode 100644 index 0000000000..a2a8d53bdd --- /dev/null +++ b/dist/Tie-File/t/13_size_rs.t @@ -0,0 +1,95 @@ +#!/usr/bin/perl +# +# Check FETCHSIZE and SETSIZE functions +# PUSH POP SHIFT UNSHIFT +# + +use POSIX 'SEEK_SET'; + +my $file = "tf$$.txt"; +my $data = "rec0blahrec1blahrec2blah"; +my ($o, $n); + +print "1..10\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +# 2-3 FETCHSIZE 0-length file +open F, "> $file" or die $!; +close F; +$o = tie @a, 'Tie::File', $file, recsep => 'blah'; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; +$n = @a; +print $n == 0 ? "ok $N\n" : "not ok $N # $n, s/b 0\n"; +$N++; + +# Reset everything +undef $o; +untie @a; + +# 4-5 FETCHSIZE positive-length file +open F, "> $file" or die $!; +print F $data; +close F; +$o = tie @a, 'Tie::File', $file, recsep => 'blah'; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; +$n = @a; +print $n == 3 ? "ok $N\n" : "not ok $N # $n, s/b 0\n"; +$N++; + +# STORESIZE +# 6 Make it longer: +$#a = 4; +check_contents("${data}blahblah"); + +# 7 Make it longer again: +$#a = 6; +check_contents("${data}blahblahblahblah"); + +# 8 Make it shorter: +$#a = 4; +check_contents("${data}blahblah"); + +# 9 Make it shorter again: +$#a = 2; +check_contents($data); + +# 10 Get rid of it completely: +$#a = -1; +check_contents(''); + + +sub check_contents { + my $x = shift; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix(my $msg = "# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/14_lock.t b/dist/Tie-File/t/14_lock.t new file mode 100644 index 0000000000..cab48125b0 --- /dev/null +++ b/dist/Tie-File/t/14_lock.t @@ -0,0 +1,50 @@ +#!/usr/bin/perl +# +# Check flock() feature +# +# This isn't a real test; it just checks to make sure we can call the method. +# It doesn't even check to make sure that the default behavior +# (LOCK_EX) is occurring. This is because I don't know how to write a good +# portable test for flocking. I checked the Perl core distribution, +# and found that Perl doesn't test flock either! + +BEGIN { + eval { flock STDOUT, 0 }; + if ($@ && $@ =~ /unimplemented/) { + print "1..0\n"; + exit; + } +} + +use Fcntl ':flock'; # This works at least back to 5.004_04 + +my $file = "tf$$.txt"; +my ($o, $n); +my @a; + +print "1..4\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +# 2-4 Who the heck knows? +open F, "> $file" or die $!; +close F; +$o = tie @a, 'Tie::File', $file, recsep => 'blah'; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +print $o->flock() ? "ok $N\n" : "not ok $N\n"; +$N++; + +print $o->flock(LOCK_UN) ? "ok $N\n" : "not ok $N\n"; +$N++; + + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/15_pushpop.t b/dist/Tie-File/t/15_pushpop.t new file mode 100644 index 0000000000..4b6d1bc959 --- /dev/null +++ b/dist/Tie-File/t/15_pushpop.t @@ -0,0 +1,135 @@ +#!/usr/bin/perl +# +# Check PUSH, POP, SHIFT, and UNSHIFT +# +# Each call to 'check_contents' actually performs two tests. +# First, it calls the tied object's own 'check_integrity' method, +# which makes sure that the contents of the read cache and offset tables +# accurately reflect the contents of the file. +# Then, it checks the actual contents of the file against the expected +# contents. + +use POSIX 'SEEK_SET'; + +my $file = "tf$$.txt"; +1 while unlink $file; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; + +print "1..38\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; # partial credit just for showing up + +my $o = tie @a, 'Tie::File', $file, autochomp => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; +my ($n, @r); + + +# (3-11) PUSH tests +$n = push @a, "rec0", "rec1", "rec2"; +check_contents($data); +print $n == 3 ? "ok $N\n" : "not ok $N # size is $n, should be 3\n"; +$N++; + +$n = push @a, "rec3", "rec4$:"; +check_contents("$ {data}rec3$:rec4$:"); +print $n == 5 ? "ok $N\n" : "not ok $N # size is $n, should be 5\n"; +$N++; + +# Trivial push +$n = push @a, (); +check_contents("$ {data}rec3$:rec4$:"); +print $n == 5 ? "ok $N\n" : "not ok $N # size is $n, should be 5\n"; +$N++; + +# (12-20) POP tests +$n = pop @a; +check_contents("$ {data}rec3$:"); +print $n eq "rec4$:" ? "ok $N\n" : "not ok $N # last rec is $n, should be rec4\n"; +$N++; + +# Presumably we have already tested this to death +splice(@a, 1, 3); +$n = pop @a; +check_contents(""); +print $n eq "rec0$:" ? "ok $N\n" : "not ok $N # last rec is $n, should be rec0\n"; +$N++; + +$n = pop @a; +check_contents(""); +print ! defined $n ? "ok $N\n" : "not ok $N # last rec should be undef, is $n\n"; +$N++; + + +# (21-29) UNSHIFT tests +$n = unshift @a, "rec0", "rec1", "rec2"; +check_contents($data); +print $n == 3 ? "ok $N\n" : "not ok $N # size is $n, should be 3\n"; +$N++; + +$n = unshift @a, "rec3", "rec4$:"; +check_contents("rec3$:rec4$:$data"); +print $n == 5 ? "ok $N\n" : "not ok $N # size is $n, should be 5\n"; +$N++; + +# Trivial unshift +$n = unshift @a, (); +check_contents("rec3$:rec4$:$data"); +print $n == 5 ? "ok $N\n" : "not ok $N # size is $n, should be 5\n"; +$N++; + +# (30-38) SHIFT tests +$n = shift @a; +check_contents("rec4$:$data"); +print $n eq "rec3$:" ? "ok $N\n" : "not ok $N # last rec is $n, should be rec3\n"; +$N++; + +# Presumably we have already tested this to death +splice(@a, 1, 3); +$n = shift @a; +check_contents(""); +print $n eq "rec4$:" ? "ok $N\n" : "not ok $N # last rec is $n, should be rec4\n"; +$N++; + +$n = shift @a; +check_contents(""); +print ! defined $n ? "ok $N\n" : "not ok $N # last rec should be undef, is $n\n"; +$N++; + + +sub check_contents { + my $x = shift; + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix(my $msg = "# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/16_handle.t b/dist/Tie-File/t/16_handle.t new file mode 100644 index 0000000000..f799496be1 --- /dev/null +++ b/dist/Tie-File/t/16_handle.t @@ -0,0 +1,160 @@ +#!/usr/bin/perl +# +# Basic operation, initializing the object from an already-open handle +# instead of from a filename + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); + +if ($^O =~ /vms/i) { + print "1..0\n"; + exit; +} + +print "1..39\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +use Fcntl 'O_CREAT', 'O_RDWR'; +sysopen F, $file, O_CREAT | O_RDWR + or die "Couldn't create temp file $file: $!; aborting"; +binmode F; + +my $o = tie @a, 'Tie::File', \*F, autochomp => 0, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# 3-4 create +$a[0] = 'rec0'; +check_contents("rec0"); + +# 5-8 append +$a[1] = 'rec1'; +check_contents("rec0", "rec1"); +$a[2] = 'rec2'; +check_contents("rec0", "rec1", "rec2"); + +# 9-14 same-length alterations +$a[0] = 'new0'; +check_contents("new0", "rec1", "rec2"); +$a[1] = 'new1'; +check_contents("new0", "new1", "rec2"); +$a[2] = 'new2'; +check_contents("new0", "new1", "new2"); + +# 15-24 lengthening alterations +$a[0] = 'long0'; +check_contents("long0", "new1", "new2"); +$a[1] = 'long1'; +check_contents("long0", "long1", "new2"); +$a[2] = 'long2'; +check_contents("long0", "long1", "long2"); +$a[1] = 'longer1'; +check_contents("long0", "longer1", "long2"); +$a[0] = 'longer0'; +check_contents("longer0", "longer1", "long2"); + +# 25-38 shortening alterations, including truncation +$a[0] = 'short0'; +check_contents("short0", "longer1", "long2"); +$a[1] = 'short1'; +check_contents("short0", "short1", "long2"); +$a[2] = 'short2'; +check_contents("short0", "short1", "short2"); +$a[1] = 'sh1'; +check_contents("short0", "sh1", "short2"); +$a[0] = 'sh0'; +check_contents("sh0", "sh1", "short2"); + +# file with holes +$a[4] = 'rec4'; +check_contents("sh0", "sh1", "short2", "", "rec4"); +$a[3] = 'rec3'; +check_contents("sh0", "sh1", "short2", "rec3", "rec4"); + +close F; +undef $o; +untie @a; + +# (39) Does it correctly detect a non-seekable handle? +{ if ($^O =~ /^(MSWin32|dos|beos)$/) { + print "ok $N # skipped ($^O has broken pipe semantics)\n"; + last; + } + if ($] < 5.006) { + print "ok $N # skipped - 5.005_03 panics after this test\n"; + last; + } + my $pipe_succeeded = eval {pipe *R, *W}; + if ($@) { + chomp $@; + print "ok $N # skipped (no pipes: $@)\n"; + last; + } elsif (! $pipe_succeeded) { + print "ok $N # skipped (pipe call failed: $!)\n"; + last; + } + close R; + $o = eval {tie @a, 'Tie::File', \*W}; + if ($@) { + if ($@ =~ /filehandle does not appear to be seekable/) { + print "ok $N\n"; + } else { + chomp $@; + print "not ok $N \# \$\@ is $@\n"; + } + } else { + print "not ok $N \# passing pipe to TIEARRAY didn't abort program\n"; + } + $N++; +} + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $:, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; +# my $open = open FH, "< $file"; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix(my $msg = "# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + my $msg; + for (0.. $#c) { + unless ($a[$_] eq "$c[$_]$:") { + $msg = "expected $c[$_]$:, got $a[$_]"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # $msg\n"; + $N++; +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + + diff --git a/dist/Tie-File/t/17_misc_meth.t b/dist/Tie-File/t/17_misc_meth.t new file mode 100644 index 0000000000..020774bb91 --- /dev/null +++ b/dist/Tie-File/t/17_misc_meth.t @@ -0,0 +1,158 @@ +#!/usr/bin/perl +# +# Check miscellaneous tied-array interface methods +# EXTEND, CLEAR, DELETE, EXISTS +# + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +1 while unlink $file; + +print "1..35\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +my $o = tie @a, 'Tie::File', $file, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (3-8) EXTEND +$o->EXTEND(3); +check_contents("$:$:$:"); +$o->EXTEND(4); +check_contents("$:$:$:$:"); +$o->EXTEND(3); +check_contents("$:$:$:$:"); + +# (9-10) CLEAR +@a = (); +check_contents(""); + +# (11-20) EXISTS +if ($] >= 5.006) { + eval << 'TESTS'; +print !exists $a[0] ? "ok $N\n" : "not ok $N\n"; +$N++; +$a[0] = "I like pie."; +print exists $a[0] ? "ok $N\n" : "not ok $N\n"; +$N++; +print !exists $a[1] ? "ok $N\n" : "not ok $N\n"; +$N++; +$a[2] = "GIVE ME PIE"; +print exists $a[0] ? "ok $N\n" : "not ok $N\n"; +$N++; +# exists $a[1] is not defined by this module under these circumstances +print exists $a[1] ? "ok $N\n" : "ok $N\n"; +$N++; +print exists $a[2] ? "ok $N\n" : "not ok $N\n"; +$N++; +print exists $a[-1] ? "ok $N\n" : "not ok $N\n"; +$N++; +print exists $a[-2] ? "ok $N\n" : "not ok $N\n"; +$N++; +print exists $a[-3] ? "ok $N\n" : "not ok $N\n"; +$N++; +print !exists $a[-4] ? "ok $N\n" : "not ok $N\n"; +$N++; +TESTS + } else { # perl 5.005 doesn't have exists $array[1] + for (11..20) { + print "ok $_ \# skipped (no exists for arrays)\n"; + $N++; + } + } + +my $del; + +# (21-35) DELETE +if ($] >= 5.006) { + eval << 'TESTS'; +$del = delete $a[0]; +check_contents("$:$:GIVE ME PIE$:"); +# 20020317 Through 0.20, the 'delete' function returned the wrong values. +expect($del, "I like pie."); +$del = delete $a[2]; +check_contents("$:$:"); +expect($del, "GIVE ME PIE"); +$del = delete $a[0]; +check_contents("$:$:"); +expect($del, ""); +$del = delete $a[1]; +check_contents("$:"); +expect($del, ""); + +# 20020317 Through 0.20, we had a bug where deleting an element past the +# end of the array would actually extend the array to that length. +$del = delete $a[4]; +check_contents("$:"); +expect($del, undef); + + + +TESTS + } else { # perl 5.005 doesn't have delete $array[1] + for (21..35) { + print "ok $_ \# skipped (no delete for arrays)\n"; + $N++; + } + } + +use POSIX 'SEEK_SET'; +sub check_contents { + my $x = shift; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix(my $msg = "# expected <$x>, got <$a>"); + print "not ok $N # $msg\n"; + } + $N++; + print $o->_check_integrity($file, $ENV{INTEGRITY}) ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub expect { + if (@_ == 1) { + print $_[0] ? "ok $N\n" : "not ok $N\n"; + } elsif (@_ == 2) { + my ($a, $x) = @_; + if (! defined($a) && ! defined($x)) { print "ok $N\n" } + elsif ( defined($a) && ! defined($x)) { + ctrlfix(my $msg = "expected UNDEF, got <$a>"); + print "not ok $N \# $msg\n"; + } + elsif (! defined($a) && defined($x)) { + ctrlfix(my $msg = "expected <$x>, got UNDEF"); + print "not ok $N \# $msg\n"; + } elsif ($a eq $x) { print "ok $N\n" } + else { + ctrlfix(my $msg = "expected <$x>, got <$a>"); + print "not ok $N \# $msg\n"; + } + } else { + die "expect() got ", scalar(@_), " args, should have been 1 or 2"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + + diff --git a/dist/Tie-File/t/18_rs_fixrec.t b/dist/Tie-File/t/18_rs_fixrec.t new file mode 100644 index 0000000000..3c2a807e64 --- /dev/null +++ b/dist/Tie-File/t/18_rs_fixrec.t @@ -0,0 +1,53 @@ +#!/usr/bin/perl + +use POSIX 'SEEK_SET'; +my $file = "tf$$.txt"; +$/ = "blah"; + +print "1..5\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +my $o = tie @a, 'Tie::File', $file, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$a[0] = 'rec0'; +check_contents("rec0blah"); +$a[1] = "rec1blah"; +check_contents("rec0blahrec1blah"); +$a[2] = "rec2blahblah"; # should we detect this? +check_contents("rec0blahrec1blahrec2blahblah"); + +sub check_contents { + my $x = shift; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + my $msg = "not ok $N # expected <$x>, got <$a>"; + ctrlfix($msg); + print "$msg\n"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/19_cache.t b/dist/Tie-File/t/19_cache.t new file mode 100644 index 0000000000..81c693263e --- /dev/null +++ b/dist/Tie-File/t/19_cache.t @@ -0,0 +1,205 @@ +#!/usr/bin/perl +# +# Tests for various caching errors +# + +$|=1; +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = join $:, "rec0" .. "rec9", ""; +my $V = $ENV{INTEGRITY}; # Verbose integrity checking? + +print "1..55\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + +my $o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (3) Through 0.18, this 'splice' call would corrupt the cache. +my @z = @a; # force cache to contain all ten records +splice @a, 0, 0, "x"; +print $o->_check_integrity($file, $V) ? "ok $N\n" : "not ok $N\n"; +$N++; + +# Here we redo *all* the splice tests, with populate() +# calls before each one, to make sure that splice() does not botch the cache. + +# (4-14) splicing at the beginning +check(); +splice(@a, 0, 0, "rec4"); +check(); +splice(@a, 0, 1, "rec5"); # same length +check(); +splice(@a, 0, 1, "record5"); # longer +check(); +splice(@a, 0, 1, "r5"); # shorter +check(); +splice(@a, 0, 1); # removal +check(); +splice(@a, 0, 0); # no-op +check(); + +splice(@a, 0, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, 0, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, 0, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, 0, 2); # delete more than one +check(); + + +# (15-24) splicing in the middle +splice(@a, 1, 0, "rec4"); +check(); +splice(@a, 1, 1, "rec5"); # same length +check(); +splice(@a, 1, 1, "record5"); # longer +check(); +splice(@a, 1, 1, "r5"); # shorter +check(); +splice(@a, 1, 1); # removal +check(); +splice(@a, 1, 0); # no-op +check(); + +splice(@a, 1, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, 1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, 1, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, 1, 2); # delete more than one +check(); + +# (25-34) splicing at the end +splice(@a, 3, 0, "rec4"); +check(); +splice(@a, 3, 1, "rec5"); # same length +check(); +splice(@a, 3, 1, "record5"); # longer +check(); +splice(@a, 3, 1, "r5"); # shorter +check(); +splice(@a, 3, 1); # removal +check(); +splice(@a, 3, 0); # no-op +check(); + +splice(@a, 3, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, 3, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, 3, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, 3, 2); # delete more than one +check(); + +# (35-44) splicing with negative subscript +splice(@a, -1, 0, "rec4"); +check(); +splice(@a, -1, 1, "rec5"); # same length +check(); +splice(@a, -1, 1, "record5"); # longer +check(); +splice(@a, -1, 1, "r5"); # shorter +check(); +splice(@a, -1, 1); # removal +check(); +splice(@a, -1, 0); # no-op +check(); + +splice(@a, -1, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, -1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, -3, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, -4, 3); # delete more than one +check(); + +# (45) scrub it all out +splice(@a, 0, 3); +check(); + +# (46) put some back in +splice(@a, 0, 0, "rec0", "rec1"); +check(); + +# (47) what if we remove too many records? +splice(@a, 0, 17); +check(); + +# (48-49) In the past, splicing past the end was not correctly detected +# (1.14) +splice(@a, 89, 3); +check(); +splice(@a, @a, 3); +check(); + +# (50-51) Also we did not emulate splice's freaky behavior when inserting +# past the end of the array (1.14) +splice(@a, 89, 0, "I", "like", "pie"); +check(); +splice(@a, 89, 0, "pie pie pie"); +check(); + +# (52-54) Test default arguments +splice @a, 0, 0, (0..11); +check(); +splice @a, 4; +check(); +splice @a; +check(); + +# (55) This was broken on 20030507 when you moved the cache management +# stuff out of _oadjust back into _splice without also putting it back +# into _store. +@a = (0..11); +check(); + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +sub check { + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + repopulate(); +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +sub repopulate { + $o->{cache}->empty; + my @z = @a; # refill the cache with correct data +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + + + diff --git a/dist/Tie-File/t/20_cache_full.t b/dist/Tie-File/t/20_cache_full.t new file mode 100644 index 0000000000..8b3bf0b2e0 --- /dev/null +++ b/dist/Tie-File/t/20_cache_full.t @@ -0,0 +1,228 @@ +#!/usr/bin/perl +# +# Tests for various caching errors +# + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = join $:, "record0" .. "record9", ""; +my $V = $ENV{INTEGRITY}; # Verbose integrity checking? + +print "1..111\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + +# Limit cache size to 30 bytes +my $MAX = 30; +# -- that's enough space for 3 records, but not 4, on both \n and \r\n systems +my $o = tie @a, 'Tie::File', $file, memory => $MAX, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (3-5) Let's see if data was properly expired from the cache +my @z = @a; # force cache to contain all ten records +# It should now contain only the *last* three records, 7, 8, and 9 +{ + my $x = "7 8 9"; + my $a = join " ", sort $o->{cache}->ckeys; + if ($a eq $x) { print "ok $N\n" } + else { print "not ok $N # cache keys were <$a>; expected <$x>\n" } + $N++; +} +check(); + +# Here we redo *all* the splice tests, with populate() +# calls before each one, to make sure that splice() does not botch the cache. + +# (6-25) splicing at the beginning +splice(@a, 0, 0, "rec4"); +check(); +splice(@a, 0, 1, "rec5"); # same length +check(); +splice(@a, 0, 1, "record5"); # longer +check(); +splice(@a, 0, 1, "r5"); # shorter +check(); +splice(@a, 0, 1); # removal +check(); +splice(@a, 0, 0); # no-op +check(); + +splice(@a, 0, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, 0, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, 0, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, 0, 2); # delete more than one +check(); + + +# (26-45) splicing in the middle +splice(@a, 1, 0, "rec4"); +check(); +splice(@a, 1, 1, "rec5"); # same length +check(); +splice(@a, 1, 1, "record5"); # longer +check(); +splice(@a, 1, 1, "r5"); # shorter +check(); +splice(@a, 1, 1); # removal +check(); +splice(@a, 1, 0); # no-op +check(); + +splice(@a, 1, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, 1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, 1, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, 1, 2); # delete more than one +check(); + +# (46-65) splicing at the end +splice(@a, 3, 0, "rec4"); +check(); +splice(@a, 3, 1, "rec5"); # same length +check(); +splice(@a, 3, 1, "record5"); # longer +check(); +splice(@a, 3, 1, "r5"); # shorter +check(); +splice(@a, 3, 1); # removal +check(); +splice(@a, 3, 0); # no-op +check(); + +splice(@a, 3, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, 3, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, 3, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, 3, 2); # delete more than one +check(); + +# (66-85) splicing with negative subscript +splice(@a, -1, 0, "rec4"); +check(); +splice(@a, -1, 1, "rec5"); # same length +check(); +splice(@a, -1, 1, "record5"); # longer +check(); +splice(@a, -1, 1, "r5"); # shorter +check(); +splice(@a, -1, 1); # removal +check(); +splice(@a, -1, 0); # no-op +check(); + +splice(@a, -1, 0, 'r7', 'rec8'); # insert more than one +check(); +splice(@a, -1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check(); +splice(@a, -3, 3, 'record9', 'rec10'); # delete more than insert +check(); +splice(@a, -4, 3); # delete more than one +check(); + +# (86-87) scrub it all out +splice(@a, 0, 3); +check(); + +# (88-89) put some back in +splice(@a, 0, 0, "rec0", "rec1"); +check(); + +# (90-91) what if we remove too many records? +splice(@a, 0, 17); +check(); + +# (92-95) In the past, splicing past the end was not correctly detected +# (1.14) +splice(@a, 89, 3); +check(); +splice(@a, @a, 3); +check(); + +# (96-99) Also we did not emulate splice's freaky behavior when inserting +# past the end of the array (1.14) +splice(@a, 89, 0, "I", "like", "pie"); +check(); +splice(@a, 89, 0, "pie pie pie"); +check(); + +# (100-105) Test default arguments +splice @a, 0, 0, (0..11); +check(); +splice @a, 4; +check(); +splice @a; +check(); + +# (106-111) One last set of tests. I don't know what state the cache +# is in now. But if I read any three records, those three records are +# what should be in the cache, and nothing else. +@a = "record0" .. "record9"; +check(); # In 0.18 #107 fails here--STORE was not flushing the cache when + # replacing an old cached record with a longer one +for (5, 6, 1) { my $z = $a[$_] } +{ + my $x = "5 6 1"; + my $a = join " ", $o->{cache}->_produce_lru; + if ($a eq $x) { print "ok $N\n" } + else { print "not ok $N # LRU was <$a>; expected <$x>\n" } + $N++; + $x = "1 5 6"; + $a = join " ", sort $o->{cache}->ckeys; + if ($a eq $x) { print "ok $N\n" } + else { print "not ok $N # cache keys were <$a>; expected <$x>\n" } + $N++; +} +check(); + + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +sub check { + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + + my $b = $o->{cache}->bytes; + print $b <= $MAX + ? "ok $N\n" + : "not ok $N # $b bytes cached, should be <= $MAX\n"; + $N++; +} + + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + + + diff --git a/dist/Tie-File/t/21_win32.t b/dist/Tie-File/t/21_win32.t new file mode 100644 index 0000000000..d06854441b --- /dev/null +++ b/dist/Tie-File/t/21_win32.t @@ -0,0 +1,61 @@ +#!/usr/bin/perl +# +# Formerly, on a Win32 system, Tie::File would create files with +# \n-terminated records instead of \r\n-terminated. The tests never +# picked this up because they were using $/ everywhere, and $/ is \n +# on windows systems. +# +# These tests (Win32 only) make sure that the file had \r\n as it should. + +my $file = "tf$$.txt"; + +unless ($^O =~ /^(MSWin32|dos)$/) { + print "1..0\n"; + exit; +} + + +print "1..3\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +my $o = tie @a, 'Tie::File', $file, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +my $n; + +# (3) Make sure that on Win32 systems, the file is written with \r\n by default +@a = qw(fish dog carrot); +undef $o; +untie @a; +open F, "< $file" or die "Couldn't open file $file: $!"; +binmode F; +my $a = do {local $/ ; <F> }; +my $x = "fish\r\ndog\r\ncarrot\r\n" ; +if ($a eq $x) { + print "ok $N\n"; +} else { + ctrlfix(my $msg = "expected <$x>, got <$a>"); + print "not ok $N # $msg\n"; +} + +close F; + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + + + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/22_autochomp.t b/dist/Tie-File/t/22_autochomp.t new file mode 100644 index 0000000000..dee07a8ec8 --- /dev/null +++ b/dist/Tie-File/t/22_autochomp.t @@ -0,0 +1,175 @@ +#!/usr/bin/perl + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); + +print "1..71\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +my $o = tie @a, 'Tie::File', $file, autochomp => 1, autodefer => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# 3-5 create +$a[0] = 'rec0'; +check_contents("rec0"); + +# 6-11 append +$a[1] = 'rec1'; +check_contents("rec0", "rec1"); +$a[2] = 'rec2'; +check_contents("rec0", "rec1", "rec2"); + +# 12-20 same-length alterations +$a[0] = 'new0'; +check_contents("new0", "rec1", "rec2"); +$a[1] = 'new1'; +check_contents("new0", "new1", "rec2"); +$a[2] = 'new2'; +check_contents("new0", "new1", "new2"); + +# 21-35 lengthening alterations +$a[0] = 'long0'; +check_contents("long0", "new1", "new2"); +$a[1] = 'long1'; +check_contents("long0", "long1", "new2"); +$a[2] = 'long2'; +check_contents("long0", "long1", "long2"); +$a[1] = 'longer1'; +check_contents("long0", "longer1", "long2"); +$a[0] = 'longer0'; +check_contents("longer0", "longer1", "long2"); + +# 36-50 shortening alterations, including truncation +$a[0] = 'short0'; +check_contents("short0", "longer1", "long2"); +$a[1] = 'short1'; +check_contents("short0", "short1", "long2"); +$a[2] = 'short2'; +check_contents("short0", "short1", "short2"); +$a[1] = 'sh1'; +check_contents("short0", "sh1", "short2"); +$a[0] = 'sh0'; +check_contents("sh0", "sh1", "short2"); + +# (51-56) file with holes +$a[4] = 'rec4'; +check_contents("sh0", "sh1", "short2", "", "rec4"); +$a[3] = 'rec3'; +check_contents("sh0", "sh1", "short2", "rec3", "rec4"); + +# (57-59) zero out file +@a = (); +check_contents(); + +# (60-62) insert into the middle of an empty file +$a[3] = "rec3"; +check_contents("", "", "", "rec3"); + +# (63-68) Test the ->autochomp() method +@a = qw(Gold Frankincense Myrrh); +my $ac; +$ac = $o->autochomp(); +expect($ac); +# See if that accidentally changed it +$ac = $o->autochomp(); +expect($ac); +# Now clear it +$ac = $o->autochomp(0); +expect($ac); +expect(join("-", @a), "Gold$:-Frankincense$:-Myrrh$:"); +# Now set it again +$ac = $o->autochomp(1); +expect(!$ac); +expect(join("-", @a), "Gold-Frankincense-Myrrh"); + +# (69) Does 'splice' work correctly with autochomp? +my @sr; +@sr = splice @a, 0, 2; +expect(join("-", @sr), "Gold-Frankincense"); + +# (70-71) Didn't you forget that fetch may return an unchomped cached record? +$a1 = $a[0]; # populate cache +$a2 = $a[0]; +expect($a1, "Myrrh"); +expect($a2, "Myrrh"); +# Actually no, you didn't---_fetch might return such a record, but +# the chomping is done by FETCH. + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $:, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; +# my $open = open FH, "< $file"; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + my $msg; + for (0.. $#c) { + my $aa = $a[$_]; + unless ($aa eq $c[$_]) { + $msg = "expected <$c[$_]>, got <$aa>"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # $msg\n"; + $N++; + + print $o->_check_integrity($file, $ENV{INTEGRITY}) + ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub expect { + if (@_ == 1) { + print $_[0] ? "ok $N\n" : "not ok $N\n"; + } elsif (@_ == 2) { + my ($a, $x) = @_; + if (! defined($a) && ! defined($x)) { print "ok $N\n" } + elsif ( defined($a) && ! defined($x)) { + ctrlfix(my $msg = "expected UNDEF, got <$a>"); + print "not ok $N \# $msg\n"; + } + elsif (! defined($a) && defined($x)) { + ctrlfix(my $msg = "expected <$x>, got UNDEF"); + print "not ok $N \# $msg\n"; + } elsif ($a eq $x) { print "ok $N\n" } + else { + ctrlfix(my $msg = "expected <$x>, got <$a>"); + print "not ok $N \# $msg\n"; + } + } else { + die "expect() got ", scalar(@_), " args, should have been 1 or 2"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/23_rv_ac_splice.t b/dist/Tie-File/t/23_rv_ac_splice.t new file mode 100644 index 0000000000..be229574f9 --- /dev/null +++ b/dist/Tie-File/t/23_rv_ac_splice.t @@ -0,0 +1,182 @@ +#!/usr/bin/perl +# +# Check SPLICE function's return value when autochoping is now +# (07_rv_splice.t checks it aith autochomping off) +# + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; + +print "1..50\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; # partial credit just for showing up + +init_file($data); + +my $o = tie @a, 'Tie::File', $file, autochomp => 1; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +my $n; + +# (3-12) splicing at the beginning +@r = splice(@a, 0, 0, "rec4"); +check_result(); +@r = splice(@a, 0, 1, "rec5"); # same length +check_result("rec4"); +@r = splice(@a, 0, 1, "record5"); # longer +check_result("rec5"); + +@r = splice(@a, 0, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, 0, 1); # removal +check_result("r5"); +@r = splice(@a, 0, 0); # no-op +check_result(); +@r = splice(@a, 0, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 0, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 0, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 0, 2); # delete more than one +check_result('record9', 'rec10'); + + +# (13-22) splicing in the middle +@r = splice(@a, 1, 0, "rec4"); +check_result(); +@r = splice(@a, 1, 1, "rec5"); # same length +check_result('rec4'); +@r = splice(@a, 1, 1, "record5"); # longer +check_result('rec5'); + +@r = splice(@a, 1, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, 1, 1); # removal +check_result("r5"); +@r = splice(@a, 1, 0); # no-op +check_result(); +@r = splice(@a, 1, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 1, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 1, 2); # delete more than one +check_result('record9','rec10'); + +# (23-32) splicing at the end +@r = splice(@a, 3, 0, "rec4"); +check_result(); +@r = splice(@a, 3, 1, "rec5"); # same length +check_result('rec4'); +@r = splice(@a, 3, 1, "record5"); # longer +check_result('rec5'); + +@r = splice(@a, 3, 1, "r5"); # shorter +check_result('record5'); +@r = splice(@a, 3, 1); # removal +check_result('r5'); +@r = splice(@a, 3, 0); # no-op +check_result(); +@r = splice(@a, 3, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, 3, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('r7', 'rec8'); + +@r = splice(@a, 3, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, 3, 2); # delete more than one +check_result('record9', 'rec10'); + +# (33-42) splicing with negative subscript +@r = splice(@a, -1, 0, "rec4"); +check_result(); +@r = splice(@a, -1, 1, "rec5"); # same length +check_result('rec2'); +@r = splice(@a, -1, 1, "record5"); # longer +check_result("rec5"); + +@r = splice(@a, -1, 1, "r5"); # shorter +check_result("record5"); +@r = splice(@a, -1, 1); # removal +check_result("r5"); +@r = splice(@a, -1, 0); # no-op +check_result(); +@r = splice(@a, -1, 0, 'r7', 'rec8'); # insert more than one +check_result(); +@r = splice(@a, -1, 2, 'rec7', 'record8', 'rec9'); # insert more than delete +check_result('rec4'); + +@r = splice(@a, -3, 3, 'record9', 'rec10'); # delete more than insert +check_result('rec7', 'record8', 'rec9'); +@r = splice(@a, -4, 3); # delete more than one +check_result('r7', 'rec8', 'record9'); + +# (43) scrub it all out +@r = splice(@a, 0, 3); +check_result('rec0', 'rec1', 'rec10'); + +# (44) put some back in +@r = splice(@a, 0, 0, "rec0", "rec1"); +check_result(); + +# (45) what if we remove too many records? +@r = splice(@a, 0, 17); +check_result('rec0', 'rec1'); + +# (46-48) Now check the scalar context return +splice(@a, 0, 0, qw(I like pie)); +my $r; +$r = splice(@a, 0, 0); +print !defined($r) ? "ok $N\n" : "not ok $N \# return should have been undef, was <$r>\n"; +$N++; + +$r = splice(@a, 2, 1); +print $r eq "pie" ? "ok $N\n" : "not ok $N \# return should have been 'pie', was <$r>\n"; +$N++; + +$r = splice(@a, 0, 2); +print $r eq "like" ? "ok $N\n" : "not ok $N \# return should have been 'like', was <$r>\n"; +$N++; + +# (49-50) Test default arguments +splice @a, 0, 0, (0..11); +@r = splice @a, 4; +check_result(4..11); +@r = splice @a; +check_result(0..3); + +sub init_file { + my $data = shift; + open F, "> $file" or die $!; + binmode F; + print F $data; + close F; +} + +# actual results are in @r. +# expected results are in @_ +sub check_result { + my @x = @_; + my $good = 1; + $good = 0 unless @r == @x; + for my $i (0 .. $#r) { + $good = 0 unless $r[$i] eq $x[$i]; + } + print $good ? "ok $N\n" : "not ok $N \# was (@r); should be (@x)\n"; + $N++; +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/24_cache_loop.t b/dist/Tie-File/t/24_cache_loop.t new file mode 100644 index 0000000000..0bc66bee2b --- /dev/null +++ b/dist/Tie-File/t/24_cache_loop.t @@ -0,0 +1,55 @@ +#!/usr/bin/perl +# +# Tests for various caching errors +# + +use Config; +my $file = "tf$$.txt"; +unless ($Config{d_alarm}) { + print "1..0\n"; exit; +} + +$: = Tie::File::_default_recsep(); +my $data = join $:, "record0" .. "record9", ""; +my $V = $ENV{INTEGRITY}; # Verbose integrity checking? + +print "1..3\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + +# Limit cache size to 30 bytes +my $MAX = 30; +# -- that's enough space for 3 records, but not 4, on both \n and \r\n systems +my $o = tie @a, 'Tie::File', $file, memory => $MAX, autodefer => 1; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (3) In 0.50 this goes into an infinite loop. Explanation: +# +# Suppose you overfill the defer buffer by so much that the memory +# limit is also exceeded. You'll go into _splice to prepare to +# write out the defer buffer, and _splice will call _fetch, which +# will then try to flush the read cache---but the read cache is +# already empty, so you're stuck in an infinite loop. +# +# Five seconds should be plenty of time for it to complete if it works. +alarm 5 unless $^P; +@a = "record0" .. "record9"; +print "ok 3\n"; +alarm 0; + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + + + diff --git a/dist/Tie-File/t/25_gen_nocache.t b/dist/Tie-File/t/25_gen_nocache.t new file mode 100644 index 0000000000..78e5506215 --- /dev/null +++ b/dist/Tie-File/t/25_gen_nocache.t @@ -0,0 +1,137 @@ +#!/usr/bin/perl +# +# Regular read-write tests with caching disabled +# (Same as 01_gen.t) +# +my $file = "tf$$.txt"; + +print "1..68\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +my $o = tie @a, 'Tie::File', $file, autochomp => 0, autodefer => 0, memory => 0; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +$: = $o->{recsep}; + +# 3-5 create +$a[0] = 'rec0'; +check_contents("rec0"); + +# 6-11 append +$a[1] = 'rec1'; +check_contents("rec0", "rec1"); +$a[2] = 'rec2'; +check_contents("rec0", "rec1", "rec2"); + +# 12-20 same-length alterations +$a[0] = 'new0'; +check_contents("new0", "rec1", "rec2"); +$a[1] = 'new1'; +check_contents("new0", "new1", "rec2"); +$a[2] = 'new2'; +check_contents("new0", "new1", "new2"); + +# 21-35 lengthening alterations +$a[0] = 'long0'; +check_contents("long0", "new1", "new2"); +$a[1] = 'long1'; +check_contents("long0", "long1", "new2"); +$a[2] = 'long2'; +check_contents("long0", "long1", "long2"); +$a[1] = 'longer1'; +check_contents("long0", "longer1", "long2"); +$a[0] = 'longer0'; +check_contents("longer0", "longer1", "long2"); + +# 36-50 shortening alterations, including truncation +$a[0] = 'short0'; +check_contents("short0", "longer1", "long2"); +$a[1] = 'short1'; +check_contents("short0", "short1", "long2"); +$a[2] = 'short2'; +check_contents("short0", "short1", "short2"); +$a[1] = 'sh1'; +check_contents("short0", "sh1", "short2"); +$a[0] = 'sh0'; +check_contents("sh0", "sh1", "short2"); + +# (51-56) file with holes +$a[4] = 'rec4'; +check_contents("sh0", "sh1", "short2", "", "rec4"); +$a[3] = 'rec3'; +check_contents("sh0", "sh1", "short2", "rec3", "rec4"); + +# (57-59) zero out file +@a = (); +check_contents(); + +# (60-62) insert into the middle of an empty file +$a[3] = "rec3"; +check_contents("", "", "", "rec3"); + +# (63-68) 20020326 You thought there would be a bug in STORE where if +# a cached record was false, STORE wouldn't see it at all. But you +# forgot that records always come back from the cache with the record +# separator attached, so they are unlikely to be false. The only +# really weird case is when the cached record is empty and the record +# separator is "0". Test that in 09_gen_rs.t. +$a[1] = "0"; +check_contents("", "0", "", "rec3"); +$a[1] = "whoops"; +check_contents("", "whoops", "", "rec3"); + + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $:, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; +# my $open = open FH, "< $file"; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + my $msg; + for (0.. $#c) { + my $aa = $a[$_]; + unless ($aa eq "$c[$_]$:") { + $msg = "expected <$c[$_]$:>, got <$aa>"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # $msg\n"; + $N++; + + print $o->_check_integrity($file, $ENV{INTEGRITY}) + ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/26_twrite.t b/dist/Tie-File/t/26_twrite.t new file mode 100644 index 0000000000..e2a925f4e0 --- /dev/null +++ b/dist/Tie-File/t/26_twrite.t @@ -0,0 +1,359 @@ +#!/usr/bin/perl +# +# Unit tests of _twrite function +# +# _twrite($self, $data, $pos, $len) +# +# 't' here is for 'tail'. This writes $data at absolute position $pos +# in the file, overwriting exactly $len of the bytes at that position. +# Everything else is moved down or up, dependong on whether +# length($data) > $len or length($data) < $len. +# $len == 0 is a pure insert; $len == length($data) is a simple overwrite. +# + +my $file = "tf$$.txt"; + +print "1..181\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +$: = Tie::File::_default_recsep(); + +# (2) Peter Scott sent this one. It fails in 0.51 and works in 0.90 +# <4.3.2.7.2.20020331102819.00b913d0@shell2.webquarry.com> +# +# The problem was premature termination in the inner loop +# because you had $more_data scoped *inside* the block instead of outside. +# 20020331 +open F, "> $file" or die "Couldn't open $file: $!"; +binmode F; +for (1..100) { + print F "$_ ", 'a'x150, $: ; +} +close F; +# The file is now 15292 characters long on Unix, 15392 on Win32 +die -s $file unless -s $file == 15292 + 100 * length($:); + +tie my @lines, 'Tie::File', $file or die $!; +push @lines, "1001 ".('a' x 100); +splice @lines, 0, 1; +untie @lines; + +my $s = -s $file; +my $x = 15292 - 152 + 105 + 100*length($:); +print $s == $x + ? "ok $N\n" : "not ok $N # expected $x, got $s\n"; +$N++; + +my @subtests = qw(x <x x> x><x <x> <x><x x><x> <x><x> <x><x><x> 0); + +# (3-73) These were generated by 'gentests.pl' to cover all possible cases +# (I hope) +# Legend: +# x: data is entirely contained within one block +# x>: data runs from the middle to the end of the block +# <x: data runs from the start to the middle of the block +# <x>: data occupies precisely one block +# x><x: data overlaps one block boundary +# <x><x: data runs from the start of one block into the middle of the next +# x><x>: data runs from the middle of one block to the end of the next +# <x><x>: data occupies two blocks exactly +# <x><x><x>: data occupies three blocks exactly +# 0: data is null +# +# For each possible alignment of the old and new data, we investigate +# up to three situations: old data is shorter, old and new data are the +# same length, and new data is shorter. +# +# try($pos, $old, $new) means to run a test where the data starts at +# position $pos, the old data has length $old, +# and the new data has length $new. +try( 9659, 6635, 6691); # old=x , new=x ; old < new +try( 8605, 2394, 2394); # old=x , new=x ; old = new +try( 9768, 1361, 664); # old=x , new=x ; old > new +try( 9955, 6429, 6429); # old=x> , new=x ; old = new +try(10550, 5834, 4123); # old=x> , new=x ; old > new +try(14580, 6158, 851); # old=x><x , new=x ; old > new +try(13442, 11134, 1572); # old=x><x> , new=x ; old > new +try( 8394, 0, 5742); # old=0 , new=x ; old < new +try( 8192, 2819, 6738); # old=<x , new=<x ; old < new +try( 8192, 514, 514); # old=<x , new=<x ; old = new +try( 8192, 2196, 858); # old=<x , new=<x ; old > new +try( 8192, 8192, 8192); # old=<x> , new=<x ; old = new +try( 8192, 8192, 1290); # old=<x> , new=<x ; old > new +try( 8192, 10575, 6644); # old=<x><x , new=<x ; old > new +try( 8192, 16384, 5616); # old=<x><x> , new=<x ; old > new +try( 8192, 24576, 6253); # old=<x><x><x>, new=<x ; old > new +try( 8192, 0, 6870); # old=0 , new=<x ; old < new +try( 8478, 6259, 7906); # old=x , new=x> ; old < new +try( 9965, 6419, 6419); # old=x> , new=x> ; old = new +try(16059, 6102, 325); # old=x><x , new=x> ; old > new +try( 9503, 15073, 6881); # old=x><x> , new=x> ; old > new +try( 9759, 0, 6625); # old=0 , new=x> ; old < new +try( 8525, 2081, 8534); # old=x , new=x><x ; old < new +try(15550, 834, 1428); # old=x> , new=x><x ; old < new +try(14966, 1668, 3479); # old=x><x , new=x><x ; old < new +try(16316, 1605, 1605); # old=x><x , new=x><x ; old = new +try(16093, 4074, 993); # old=x><x , new=x><x ; old > new +try(14739, 9837, 9837); # old=x><x> , new=x><x ; old = new +try(14071, 10505, 7344); # old=x><x> , new=x><x ; old > new +try(12602, 0, 8354); # old=0 , new=x><x ; old < new +try( 8192, 2767, 8192); # old=<x , new=<x> ; old < new +try( 8192, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 8192, 14817, 8192); # old=<x><x , new=<x> ; old > new +try( 8192, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 8192, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 8192, 0, 8192); # old=0 , new=<x> ; old < new +try( 8192, 6532, 10882); # old=<x , new=<x><x ; old < new +try( 8192, 8192, 16044); # old=<x> , new=<x><x ; old < new +try( 8192, 9555, 11020); # old=<x><x , new=<x><x ; old < new +try( 8192, 9001, 9001); # old=<x><x , new=<x><x ; old = new +try( 8192, 11760, 10274); # old=<x><x , new=<x><x ; old > new +try( 8192, 16384, 10781); # old=<x><x> , new=<x><x ; old > new +try( 8192, 24576, 9284); # old=<x><x><x>, new=<x><x ; old > new +try( 8192, 0, 12488); # old=0 , new=<x><x ; old < new +try( 8222, 6385, 16354); # old=x , new=x><x> ; old < new +try(13500, 2884, 11076); # old=x> , new=x><x> ; old < new +try(14069, 4334, 10507); # old=x><x , new=x><x> ; old < new +try(14761, 9815, 9815); # old=x><x> , new=x><x> ; old = new +try(10469, 0, 14107); # old=0 , new=x><x> ; old < new +try( 8192, 4181, 16384); # old=<x , new=<x><x> ; old < new +try( 8192, 8192, 16384); # old=<x> , new=<x><x> ; old < new +try( 8192, 12087, 16384); # old=<x><x , new=<x><x> ; old < new +try( 8192, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 8192, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 8192, 0, 16384); # old=0 , new=<x><x> ; old < new +try( 8192, 4968, 24576); # old=<x , new=<x><x><x>; old < new +try( 8192, 8192, 24576); # old=<x> , new=<x><x><x>; old < new +try( 8192, 14163, 24576); # old=<x><x , new=<x><x><x>; old < new +try( 8192, 16384, 24576); # old=<x><x> , new=<x><x><x>; old < new +try( 8192, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 8192, 0, 24576); # old=0 , new=<x><x><x>; old < new +try( 8771, 776, 0); # old=x , new=0 ; old > new +try( 8192, 2813, 0); # old=<x , new=0 ; old > new +try(13945, 2439, 0); # old=x> , new=0 ; old > new +try(14493, 6090, 0); # old=x><x , new=0 ; old > new +try( 8192, 8192, 0); # old=<x> , new=0 ; old > new +try( 8192, 10030, 0); # old=<x><x , new=0 ; old > new +try(14983, 9593, 0); # old=x><x> , new=0 ; old > new +try( 8192, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 8192, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(10489, 0, 0); # old=0 , new=0 ; old = new + +# (74-114) +# These tests all take place at the start of the file +try( 0, 771, 1593); # old=<x , new=<x ; old < new +try( 0, 4868, 4868); # old=<x , new=<x ; old = new +try( 0, 147, 118); # old=<x , new=<x ; old > new +try( 0, 8192, 8192); # old=<x> , new=<x ; old = new +try( 0, 8192, 4574); # old=<x> , new=<x ; old > new +try( 0, 11891, 1917); # old=<x><x , new=<x ; old > new +try( 0, 16384, 5155); # old=<x><x> , new=<x ; old > new +try( 0, 24576, 2953); # old=<x><x><x>, new=<x ; old > new +try( 0, 0, 1317); # old=0 , new=<x ; old < new +try( 0, 5609, 8192); # old=<x , new=<x> ; old < new +try( 0, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 0, 11083, 8192); # old=<x><x , new=<x> ; old > new +try( 0, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 0, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 0, 0, 8192); # old=0 , new=<x> ; old < new +try( 0, 6265, 9991); # old=<x , new=<x><x ; old < new +try( 0, 8192, 16119); # old=<x> , new=<x><x ; old < new +try( 0, 10218, 11888); # old=<x><x , new=<x><x ; old < new +try( 0, 14126, 14126); # old=<x><x , new=<x><x ; old = new +try( 0, 12002, 9034); # old=<x><x , new=<x><x ; old > new +try( 0, 16384, 13258); # old=<x><x> , new=<x><x ; old > new +try( 0, 24576, 14367); # old=<x><x><x>, new=<x><x ; old > new +try( 0, 0, 10881); # old=0 , new=<x><x ; old < new +try( 0, 6448, 16384); # old=<x , new=<x><x> ; old < new +try( 0, 8192, 16384); # old=<x> , new=<x><x> ; old < new +try( 0, 15082, 16384); # old=<x><x , new=<x><x> ; old < new +try( 0, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 0, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 0, 0, 16384); # old=0 , new=<x><x> ; old < new +try( 0, 2421, 24576); # old=<x , new=<x><x><x>; old < new +try( 0, 8192, 24576); # old=<x> , new=<x><x><x>; old < new +try( 0, 11655, 24576); # old=<x><x , new=<x><x><x>; old < new +try( 0, 16384, 24576); # old=<x><x> , new=<x><x><x>; old < new +try( 0, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 0, 0, 24576); # old=0 , new=<x><x><x>; old < new +try( 0, 6530, 0); # old=<x , new=0 ; old > new +try( 0, 8192, 0); # old=<x> , new=0 ; old > new +try( 0, 14707, 0); # old=<x><x , new=0 ; old > new +try( 0, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 0, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try( 0, 0, 0); # old=0 , new=0 ; old = new + +# (115-141) +# These tests all take place at the end of the file +$FLEN = 40960; # Force the file to be exactly 40960 bytes long +try(32768, 8192, 8192); # old=<x> , new=<x ; old = new +try(32768, 8192, 4026); # old=<x> , new=<x ; old > new +try(24576, 16384, 1917); # old=<x><x> , new=<x ; old > new +try(16384, 24576, 3818); # old=<x><x><x>, new=<x ; old > new +try(40960, 0, 2779); # old=0 , new=<x ; old < new +try(32768, 8192, 8192); # old=<x> , new=<x> ; old = new +try(24576, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try(16384, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try(40960, 0, 8192); # old=0 , new=<x> ; old < new +try(32768, 8192, 10724); # old=<x> , new=<x><x ; old < new +try(24576, 16384, 12221); # old=<x><x> , new=<x><x ; old > new +try(16384, 24576, 15030); # old=<x><x><x>, new=<x><x ; old > new +try(40960, 0, 11752); # old=0 , new=<x><x ; old < new +try(32768, 8192, 16384); # old=<x> , new=<x><x> ; old < new +try(24576, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try(16384, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try(40960, 0, 16384); # old=0 , new=<x><x> ; old < new +try(32768, 8192, 24576); # old=<x> , new=<x><x><x>; old < new +try(24576, 16384, 24576); # old=<x><x> , new=<x><x><x>; old < new +try(16384, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try(40960, 0, 24576); # old=0 , new=<x><x><x>; old < new +try(35973, 4987, 0); # old=x> , new=0 ; old > new +try(32768, 8192, 0); # old=<x> , new=0 ; old > new +try(29932, 11028, 0); # old=x><x> , new=0 ; old > new +try(24576, 16384, 0); # old=<x><x> , new=0 ; old > new +try(16384, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(40960, 0, 0); # old=0 , new=0 ; old = new + +# (142-181) +# These tests all take place at the end of the file +$FLEN = 42000; # Force the file to be exactly 42000 bytes long +try(41275, 725, 4059); # old=x , new=x ; old < new +try(41683, 317, 317); # old=x , new=x ; old = new +try(41225, 775, 405); # old=x , new=x ; old > new +try(35709, 6291, 284); # old=x><x , new=x ; old > new +try(42000, 0, 2434); # old=0 , new=x ; old < new +try(40960, 1040, 1608); # old=<x , new=<x ; old < new +try(40960, 1040, 1040); # old=<x , new=<x ; old = new +try(40960, 1040, 378); # old=<x , new=<x ; old > new +try(32768, 9232, 5604); # old=<x><x , new=<x ; old > new +try(42000, 0, 6637); # old=0 , new=<x ; old < new +try(41022, 978, 8130); # old=x , new=x> ; old < new +try(39994, 2006, 966); # old=x><x , new=x> ; old > new +try(42000, 0, 7152); # old=0 , new=x> ; old < new +try(41613, 387, 10601); # old=x , new=x><x ; old < new +try(38460, 3540, 3938); # old=x><x , new=x><x ; old < new +try(36725, 5275, 5275); # old=x><x , new=x><x ; old = new +try(37990, 4010, 3199); # old=x><x , new=x><x ; old > new +try(42000, 0, 9189); # old=0 , new=x><x ; old < new +try(40960, 1040, 8192); # old=<x , new=<x> ; old < new +try(32768, 9232, 8192); # old=<x><x , new=<x> ; old > new +try(42000, 0, 8192); # old=0 , new=<x> ; old < new +try(40960, 1040, 11778); # old=<x , new=<x><x ; old < new +try(32768, 9232, 13792); # old=<x><x , new=<x><x ; old < new +try(32768, 9232, 9232); # old=<x><x , new=<x><x ; old = new +try(32768, 9232, 8795); # old=<x><x , new=<x><x ; old > new +try(42000, 0, 8578); # old=0 , new=<x><x ; old < new +try(41531, 469, 15813); # old=x , new=x><x> ; old < new +try(39618, 2382, 9534); # old=x><x , new=x><x> ; old < new +try(42000, 0, 15344); # old=0 , new=x><x> ; old < new +try(40960, 1040, 16384); # old=<x , new=<x><x> ; old < new +try(32768, 9232, 16384); # old=<x><x , new=<x><x> ; old < new +try(42000, 0, 16384); # old=0 , new=<x><x> ; old < new +try(40960, 1040, 24576); # old=<x , new=<x><x><x>; old < new +try(32768, 9232, 24576); # old=<x><x , new=<x><x><x>; old < new +try(42000, 0, 24576); # old=0 , new=<x><x><x>; old < new +try(41500, 500, 0); # old=x , new=0 ; old > new +try(40960, 1040, 0); # old=<x , new=0 ; old > new +try(35272, 6728, 0); # old=x><x , new=0 ; old > new +try(32768, 9232, 0); # old=<x><x , new=0 ; old > new +try(42000, 0, 0); # old=0 , new=0 ; old = new + +sub try { + my ($pos, $len, $newlen) = @_; + open F, "> $file" or die "Couldn't open file $file: $!"; + binmode F; + + # The record has exactly 17 characters. This will help ensure that + # even if _twrite screws up, the data doesn't coincidentally + # look good because the remainder accidentally lines up. + my $d = substr("0123456789abcdef$:", -17); + my $recs = defined($FLEN) ? + int($FLEN/length($d))+1 : # enough to make up at least $FLEN + int(8192*5/length($d))+1; # at least 5 blocks' worth + my $oldfile = $d x $recs; + my $flen = defined($FLEN) ? $FLEN : $recs * 17; + substr($oldfile, $FLEN) = "" if defined $FLEN; # truncate + print F $oldfile; + close F; + + die "wrong length!" unless -s $file == $flen; + + my $newdata = "-" x $newlen; + my $expected = $oldfile; + substr($expected, $pos, $len) = $newdata; + + my $o = tie my @lines, 'Tie::File', $file or die $!; + $o->_twrite($newdata, $pos, $len); + undef $o; untie @lines; + + open F, "< $file" or die "Couldn't open file $file: $!"; + binmode F; + my $actual; + { local $/; + $actual = <F>; + } + close F; + + my ($alen, $xlen) = (length $actual, length $expected); + unless ($alen == $xlen) { + print "# try(@_) expected file length $xlen, actual $alen!\n"; + } + print $actual eq $expected ? "ok $N\n" : "not ok $N\n"; + $N++; +} + + + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $:, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; +# my $open = open FH, "< $file"; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + my $msg; + for (0.. $#c) { + my $aa = $a[$_]; + unless ($aa eq "$c[$_]$:") { + $msg = "expected <$c[$_]$:>, got <$aa>"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # $msg\n"; + $N++; + + print $o->_check_integrity($file, $ENV{INTEGRITY}) + ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/27_iwrite.t b/dist/Tie-File/t/27_iwrite.t new file mode 100644 index 0000000000..db591a81ba --- /dev/null +++ b/dist/Tie-File/t/27_iwrite.t @@ -0,0 +1,235 @@ +#!/usr/bin/perl +# +# Unit tests of _iwrite function +# +# _iwrite($self, $data, $start, $end) +# +# 'i' here is for 'insert'. This writes $data at absolute position $start +# in the file, copying the data at that position downwards--- +# but only down to position $end. Data at or past $end is not moved +# or even examined. Since there isn't enough room for the full copy +# (Because we inserted $data at the beginning) we copy as much as possible +# and return a string containing the remainder. + +my $file = "tf$$.txt"; +$| = 1; + +print "1..203\n"; + +my $N = 1; +my $oldfile; +use Tie::File; +print "ok $N\n"; $N++; + +$: = Tie::File::_default_recsep(); + +$FLEN = 40970; # Use files of this length +$oldfile = mkrand($FLEN); +print "# MOF tests\n"; +# (2-85) These were generated by 'gentests.pl' to cover all possible cases +# (I hope) +# Legend: +# x: data is entirely contained within one block +# x>: data runs from the middle to the end of the block +# <x: data runs from the start to the middle of the block +# <x>: data occupies precisely one block +# x><x: data overlaps one block boundary +# <x><x: data runs from the start of one block into the middle of the next +# x><x>: data runs from the middle of one block to the end of the next +# <x><x>: data occupies two blocks exactly +# <x><x><x>: data occupies three blocks exactly +# 0: data is null +# +# For each possible alignment of the old and new data, we investigate +# up to three situations: old data is shorter, old and new data are the +# same length, and new data is shorter. +# +# try($pos, $old, $new) means to run a test where the area being +# written into starts at position $pos, the area being written into +# has length $old, and and the new data has length $new. +try( 8605, 2394, 2394); # old=x , new=x ; old = new +try( 9768, 1361, 664); # old=x , new=x ; old > new +try( 9955, 6429, 6429); # old=x> , new=x ; old = new +try(10550, 5834, 4123); # old=x> , new=x ; old > new +try(14580, 6158, 851); # old=x><x , new=x ; old > new +try(13442, 11134, 1572); # old=x><x> , new=x ; old > new +try( 8192, 514, 514); # old=<x , new=<x ; old = new +try( 8192, 2196, 858); # old=<x , new=<x ; old > new +try( 8192, 8192, 8192); # old=<x> , new=<x ; old = new +try( 8192, 8192, 1290); # old=<x> , new=<x ; old > new +try( 8192, 10575, 6644); # old=<x><x , new=<x ; old > new +try( 8192, 16384, 5616); # old=<x><x> , new=<x ; old > new +try( 8192, 24576, 6253); # old=<x><x><x>, new=<x ; old > new +try( 9965, 6419, 6419); # old=x> , new=x> ; old = new +try(16059, 6102, 325); # old=x><x , new=x> ; old > new +try( 9503, 15073, 6881); # old=x><x> , new=x> ; old > new +try(16316, 1605, 1605); # old=x><x , new=x><x ; old = new +try(16093, 4074, 993); # old=x><x , new=x><x ; old > new +try(14739, 9837, 9837); # old=x><x> , new=x><x ; old = new +try(14071, 10505, 7344); # old=x><x> , new=x><x ; old > new +try( 8192, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 8192, 14817, 8192); # old=<x><x , new=<x> ; old > new +try( 8192, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 8192, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 8192, 9001, 9001); # old=<x><x , new=<x><x ; old = new +try( 8192, 11760, 10274); # old=<x><x , new=<x><x ; old > new +try( 8192, 16384, 10781); # old=<x><x> , new=<x><x ; old > new +try( 8192, 24576, 9284); # old=<x><x><x>, new=<x><x ; old > new +try(14761, 9815, 9815); # old=x><x> , new=x><x> ; old = new +try( 8192, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 8192, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 8192, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 8771, 776, 0); # old=x , new=0 ; old > new +try( 8192, 2813, 0); # old=<x , new=0 ; old > new +try(13945, 2439, 0); # old=x> , new=0 ; old > new +try(14493, 6090, 0); # old=x><x , new=0 ; old > new +try( 8192, 8192, 0); # old=<x> , new=0 ; old > new +try( 8192, 10030, 0); # old=<x><x , new=0 ; old > new +try(14983, 9593, 0); # old=x><x> , new=0 ; old > new +try( 8192, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 8192, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(10489, 0, 0); # old=0 , new=0 ; old = new + +print "# SOF tests\n"; +# (86-133) +# These tests all take place at the start of the file +try( 0, 4868, 4868); # old=<x , new=<x ; old = new +try( 0, 147, 118); # old=<x , new=<x ; old > new +try( 0, 8192, 8192); # old=<x> , new=<x ; old = new +try( 0, 8192, 4574); # old=<x> , new=<x ; old > new +try( 0, 11891, 1917); # old=<x><x , new=<x ; old > new +try( 0, 16384, 5155); # old=<x><x> , new=<x ; old > new +try( 0, 24576, 2953); # old=<x><x><x>, new=<x ; old > new +try( 0, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 0, 11083, 8192); # old=<x><x , new=<x> ; old > new +try( 0, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 0, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 0, 14126, 14126); # old=<x><x , new=<x><x ; old = new +try( 0, 12002, 9034); # old=<x><x , new=<x><x ; old > new +try( 0, 16384, 13258); # old=<x><x> , new=<x><x ; old > new +try( 0, 24576, 14367); # old=<x><x><x>, new=<x><x ; old > new +try( 0, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 0, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 0, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 0, 6530, 0); # old=<x , new=0 ; old > new +try( 0, 8192, 0); # old=<x> , new=0 ; old > new +try( 0, 14707, 0); # old=<x><x , new=0 ; old > new +try( 0, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 0, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try( 0, 0, 0); # old=0 , new=0 ; old = new + +print "# EOF tests 1\n"; +# (134-169) +# These tests all take place at the end of the file +$FLEN = 40960; # Force the file to be exactly 40960 bytes long +$oldfile = mkrand($FLEN); +try(32768, 8192, 8192); # old=<x> , new=<x ; old = new +try(32768, 8192, 4026); # old=<x> , new=<x ; old > new +try(24576, 16384, 1917); # old=<x><x> , new=<x ; old > new +try(16384, 24576, 3818); # old=<x><x><x>, new=<x ; old > new +try(32768, 8192, 8192); # old=<x> , new=<x> ; old = new +try(24576, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try(16384, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try(24576, 16384, 12221); # old=<x><x> , new=<x><x ; old > new +try(16384, 24576, 15030); # old=<x><x><x>, new=<x><x ; old > new +try(24576, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try(16384, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try(16384, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try(35973, 4987, 0); # old=x> , new=0 ; old > new +try(32768, 8192, 0); # old=<x> , new=0 ; old > new +try(29932, 11028, 0); # old=x><x> , new=0 ; old > new +try(24576, 16384, 0); # old=<x><x> , new=0 ; old > new +try(16384, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(40960, 0, 0); # old=0 , new=0 ; old = new + +print "# EOF tests 2\n"; +# (170-203) +# These tests all take place at the end of the file +$FLEN = 42000; # Force the file to be exactly 42000 bytes long +$oldfile = mkrand($FLEN); +try(41683, 317, 317); # old=x , new=x ; old = new +try(41225, 775, 405); # old=x , new=x ; old > new +try(35709, 6291, 284); # old=x><x , new=x ; old > new +try(40960, 1040, 1040); # old=<x , new=<x ; old = new +try(40960, 1040, 378); # old=<x , new=<x ; old > new +try(32768, 9232, 5604); # old=<x><x , new=<x ; old > new +try(39994, 2006, 966); # old=x><x , new=x> ; old > new +try(36725, 5275, 5275); # old=x><x , new=x><x ; old = new +try(37990, 4010, 3199); # old=x><x , new=x><x ; old > new +try(32768, 9232, 8192); # old=<x><x , new=<x> ; old > new +try(32768, 9232, 9232); # old=<x><x , new=<x><x ; old = new +try(32768, 9232, 8795); # old=<x><x , new=<x><x ; old > new +try(41500, 500, 0); # old=x , new=0 ; old > new +try(40960, 1040, 0); # old=<x , new=0 ; old > new +try(35272, 6728, 0); # old=x><x , new=0 ; old > new +try(32768, 9232, 0); # old=<x><x , new=0 ; old > new +try(42000, 0, 0); # old=0 , new=0 ; old = new + +sub mkrand { + my $len = shift; + srand $len; + my @c = ('a' .. 'z', 'A' .. 'Z', 0..9, $:); + my $d = ""; + $d .= $c[rand @c] until length($d) >= $len; + substr($d, $len) = ""; # chop it off to the proper length + $d; +} + +sub try { + my ($s, $len, $newlen) = @_; + my $e = $s + $len; + + open F, "> $file" or die "Couldn't open file $file: $!"; + binmode F; + + print F $oldfile; + close F; + + die "wrong length!" unless -s $file == $FLEN; + + my $newdata = "-" x $newlen; + my $expected = $oldfile; + + my $expected_return = substr($expected, $e - $newlen, $newlen, ""); + substr($expected, $s, 0, $newdata); + + my $o = tie my @lines, 'Tie::File', $file or die $!; + my $actual_return = $o->_iwrite($newdata, $s, $e); + undef $o; untie @lines; + + open F, "< $file" or die "Couldn't open file $file: $!"; + binmode F; + my $actual; + { local $/; + $actual = <F>; + } + close F; + + my ($alen, $xlen) = (length $actual, length $expected); + unless ($alen == $xlen) { + print "# try(@_) expected file length $xlen, actual $alen!\n"; + } + print $actual eq $expected ? "ok $N\n" : "not ok $N\n"; + $N++; + + if (! defined $actual_return && ! defined $expected_return) { + print "ok $N\n"; + } elsif (! defined $actual_return || ! defined $expected_return) { + print "not ok $N\n"; + } else { + print $actual_return eq $expected_return ? "ok $N\n" : "not ok $N\n"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/28_mtwrite.t b/dist/Tie-File/t/28_mtwrite.t new file mode 100644 index 0000000000..50e306d3b6 --- /dev/null +++ b/dist/Tie-File/t/28_mtwrite.t @@ -0,0 +1,295 @@ +#!/usr/bin/perl +# +# Unit tests of _mtwrite function +# +# _mtwrite($self, $d1, $s1, $l1, $d2, $s2, $l2, ...) +# +# 'm' here is for 'multiple'. This writes data $d1 at position $s1 +# over a block of space $l1, moving subsequent data up or down as necessary. + +my $file = "tf$$.txt"; +$| = 1; + +print "1..2252\n"; + +my $N = 1; +my $oldfile; +use Tie::File; +print "ok $N\n"; $N++; + +$: = Tie::File::_default_recsep(); + +# Only these are used for the triple-region tests +@BASE_TRIES = ( + [10, 20, 30], + [10, 30, 20], + [100, 30, 20], + [100, 20, 30], + [100, 40, 20], + [100, 20, 40], + [200, 20, 30], + [200, 30, 20], + [200, 20, 60], + [200, 60, 20], + ); + +@TRIES = @BASE_TRIES; + +$FLEN = 40970; # Use files of this length +$oldfile = mkrand($FLEN); +print "# MOF tests\n"; +# These were generated by 'gentests.pl' to cover all possible cases +# (I hope) +# Legend: +# x: data is entirely contained within one block +# x>: data runs from the middle to the end of the block +# <x: data runs from the start to the middle of the block +# <x>: data occupies precisely one block +# x><x: data overlaps one block boundary +# <x><x: data runs from the start of one block into the middle of the next +# x><x>: data runs from the middle of one block to the end of the next +# <x><x>: data occupies two blocks exactly +# <x><x><x>: data occupies three blocks exactly +# 0: data is null +# +# For each possible alignment of the old and new data, we investigate +# up to three situations: old data is shorter, old and new data are the +# same length, and new data is shorter. +# +# try($pos, $old, $new) means to run a test where the area being +# written into starts at position $pos, the area being written into +# has length $old, and and the new data has length $new. +try( 8605, 2394, 2394); # old=x , new=x ; old = new +try( 9768, 1361, 664); # old=x , new=x ; old > new +try( 9955, 6429, 6429); # old=x> , new=x ; old = new +try(10550, 5834, 4123); # old=x> , new=x ; old > new +try(14580, 6158, 851); # old=x><x , new=x ; old > new +try(13442, 11134, 1572); # old=x><x> , new=x ; old > new +try( 8192, 514, 514); # old=<x , new=<x ; old = new +try( 8192, 2196, 858); # old=<x , new=<x ; old > new +try( 8192, 8192, 8192); # old=<x> , new=<x ; old = new +try( 8192, 8192, 1290); # old=<x> , new=<x ; old > new +try( 8192, 10575, 6644); # old=<x><x , new=<x ; old > new +try( 8192, 16384, 5616); # old=<x><x> , new=<x ; old > new +try( 8192, 24576, 6253); # old=<x><x><x>, new=<x ; old > new +try( 9965, 6419, 6419); # old=x> , new=x> ; old = new +try(16059, 6102, 325); # old=x><x , new=x> ; old > new +try( 9503, 15073, 6881); # old=x><x> , new=x> ; old > new +try(16316, 1605, 1605); # old=x><x , new=x><x ; old = new +try(16093, 4074, 993); # old=x><x , new=x><x ; old > new +try(14739, 9837, 9837); # old=x><x> , new=x><x ; old = new +try(14071, 10505, 7344); # old=x><x> , new=x><x ; old > new +try( 8192, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 8192, 14817, 8192); # old=<x><x , new=<x> ; old > new +try( 8192, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 8192, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 8192, 9001, 9001); # old=<x><x , new=<x><x ; old = new +try( 8192, 11760, 10274); # old=<x><x , new=<x><x ; old > new +try( 8192, 16384, 10781); # old=<x><x> , new=<x><x ; old > new +try( 8192, 24576, 9284); # old=<x><x><x>, new=<x><x ; old > new +try(14761, 9815, 9815); # old=x><x> , new=x><x> ; old = new +try( 8192, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 8192, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 8192, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 8771, 776, 0); # old=x , new=0 ; old > new +try( 8192, 2813, 0); # old=<x , new=0 ; old > new +try(13945, 2439, 0); # old=x> , new=0 ; old > new +try(14493, 6090, 0); # old=x><x , new=0 ; old > new +try( 8192, 8192, 0); # old=<x> , new=0 ; old > new +try( 8192, 10030, 0); # old=<x><x , new=0 ; old > new +try(14983, 9593, 0); # old=x><x> , new=0 ; old > new +try( 8192, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 8192, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(10489, 0, 0); # old=0 , new=0 ; old = new + +print "# SOF tests\n"; +# These tests all take place at the start of the file +try( 0, 4868, 4868); # old=<x , new=<x ; old = new +try( 0, 147, 118); # old=<x , new=<x ; old > new +try( 0, 8192, 8192); # old=<x> , new=<x ; old = new +try( 0, 8192, 4574); # old=<x> , new=<x ; old > new +try( 0, 11891, 1917); # old=<x><x , new=<x ; old > new +try( 0, 16384, 5155); # old=<x><x> , new=<x ; old > new +try( 0, 24576, 2953); # old=<x><x><x>, new=<x ; old > new +try( 0, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 0, 11083, 8192); # old=<x><x , new=<x> ; old > new +try( 0, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 0, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 0, 14126, 14126); # old=<x><x , new=<x><x ; old = new +try( 0, 12002, 9034); # old=<x><x , new=<x><x ; old > new +try( 0, 16384, 13258); # old=<x><x> , new=<x><x ; old > new +try( 0, 24576, 14367); # old=<x><x><x>, new=<x><x ; old > new +try( 0, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 0, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 0, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 0, 6530, 0); # old=<x , new=0 ; old > new +try( 0, 8192, 0); # old=<x> , new=0 ; old > new +try( 0, 14707, 0); # old=<x><x , new=0 ; old > new +try( 0, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 0, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try( 0, 0, 0); # old=0 , new=0 ; old = new + +print "# EOF tests 1\n"; +# These tests all take place at the end of the file +$FLEN = 40960; # Force the file to be exactly 40960 bytes long +$oldfile = mkrand($FLEN); +try(32768, 8192, 8192); # old=<x> , new=<x ; old = new +try(32768, 8192, 4026); # old=<x> , new=<x ; old > new +try(24576, 16384, 1917); # old=<x><x> , new=<x ; old > new +try(16384, 24576, 3818); # old=<x><x><x>, new=<x ; old > new +try(32768, 8192, 8192); # old=<x> , new=<x> ; old = new +try(24576, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try(16384, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try(24576, 16384, 12221); # old=<x><x> , new=<x><x ; old > new +try(16384, 24576, 15030); # old=<x><x><x>, new=<x><x ; old > new +try(24576, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try(16384, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try(16384, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try(35973, 4987, 0); # old=x> , new=0 ; old > new +try(32768, 8192, 0); # old=<x> , new=0 ; old > new +try(29932, 11028, 0); # old=x><x> , new=0 ; old > new +try(24576, 16384, 0); # old=<x><x> , new=0 ; old > new +try(16384, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(40960, 0, 0); # old=0 , new=0 ; old = new + +print "# EOF tests 2\n"; +# These tests all take place at the end of the file +$FLEN = 42000; # Force the file to be exactly 42000 bytes long +$oldfile = mkrand($FLEN); +try(41683, 317, 317); # old=x , new=x ; old = new +try(41225, 775, 405); # old=x , new=x ; old > new +try(35709, 6291, 284); # old=x><x , new=x ; old > new +try(40960, 1040, 1040); # old=<x , new=<x ; old = new +try(40960, 1040, 378); # old=<x , new=<x ; old > new +try(32768, 9232, 5604); # old=<x><x , new=<x ; old > new +try(39994, 2006, 966); # old=x><x , new=x> ; old > new +try(36725, 5275, 5275); # old=x><x , new=x><x ; old = new +try(37990, 4010, 3199); # old=x><x , new=x><x ; old > new +try(32768, 9232, 8192); # old=<x><x , new=<x> ; old > new +try(32768, 9232, 9232); # old=<x><x , new=<x><x ; old = new +try(32768, 9232, 8795); # old=<x><x , new=<x><x ; old > new +try(41500, 500, 0); # old=x , new=0 ; old > new +try(40960, 1040, 0); # old=<x , new=0 ; old > new +try(35272, 6728, 0); # old=x><x , new=0 ; old > new +try(32768, 9232, 0); # old=<x><x , new=0 ; old > new +try(42000, 0, 0); # old=0 , new=0 ; old = new + +# Now the REAL tests +# Make sure mtwrite can properly write sequences of several intervals +# The intervals tested above were accumulated into @TRIES. +# try_all_doubles() tries every possible sensible pair of those intervals. +# try_all_triples() tries every possible sensible group of +# tree intervals from the more restrictive set @BASE_TRIES. +$FLEN = 40970; +$oldfile = mkrand($FLEN); +try_all_doubles(); +try_all_triples(); + +sub mkrand { + my $len = shift; + srand $len; + my @c = ('a' .. 'z', 'A' .. 'Z', 0..9, $:); + my $d = ""; + $d .= $c[rand @c] until length($d) >= $len; + substr($d, $len) = ""; # chop it off to the proper length + $d; +} + +sub try { + push @TRIES, [@_] if @_ == 3; + + open F, "> $file" or die "Couldn't open file $file: $!"; + binmode F; + print F $oldfile; + close F; + die "wrong length!" unless -s $file == $FLEN; + + my @mt_args; + my $expected = $oldfile; + { my @a = @_; + my $c = "a"; + while (@a) { + my ($s, $len, $newlen) = splice @a, -3; + my $newdata = $c++ x $newlen; + substr($expected, $s, $len, $newdata); + unshift @mt_args, $newdata, $s, $len; + } + } + + my $o = tie my @lines, 'Tie::File', $file or die $!; + my $actual_return = $o->_mtwrite(@mt_args); + undef $o; untie @lines; + + open F, "< $file" or die "Couldn't open file $file: $!"; + binmode F; + my $actual; + { local $/; + $actual = <F>; + } + close F; + + my ($alen, $xlen) = (length $actual, length $expected); + unless ($alen == $xlen) { + print "# try(@_) expected file length $xlen, actual $alen!\n"; + } + print $actual eq $expected ? "ok $N\n" : "not ok $N\n"; + $N++; + +# if (! defined $actual_return && ! defined $expected_return) { +# print "ok $N\n"; +# } elsif (! defined $actual_return || ! defined $expected_return) { +# print "not ok $N\n"; +# } else { +# print $actual_return eq $expected_return ? "ok $N\n" : "not ok $N\n"; +# } +# $N++; +} + +sub try_all_doubles { + print "# Trying double regions.\n"; + for my $a (@TRIES) { + next if $a->[0] + $a->[1] >= $FLEN; + next if $a->[0] + $a->[2] >= $FLEN; + for my $b (@TRIES) { + next if $b->[0] + $b->[1] >= $FLEN; + next if $b->[0] + $b->[2] >= $FLEN; + + next if $b->[0] < $a->[0] + $a->[1]; # Overlapping regions + try(@$a, @$b); + } + } +} + +sub try_all_triples { + print "# Trying triple regions.\n"; + for my $a (@BASE_TRIES) { + next if $a->[0] + $a->[1] >= $FLEN; + next if $a->[0] + $a->[2] >= $FLEN; + for my $b (@BASE_TRIES) { + next if $b->[0] + $b->[1] >= $FLEN; + next if $b->[0] + $b->[2] >= $FLEN; + + next if $b->[0] < $a->[0] + $a->[1]; # Overlapping regions + + for my $c (@BASE_TRIES) { + next if $c->[0] + $c->[1] >= $FLEN; + next if $c->[0] + $c->[2] >= $FLEN; + + next if $c->[0] < $b->[0] + $b->[1]; # Overlapping regions + try(@$a, @$b, @$c); + } + } + } +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/29_downcopy.t b/dist/Tie-File/t/29_downcopy.t new file mode 100644 index 0000000000..d75806d5b2 --- /dev/null +++ b/dist/Tie-File/t/29_downcopy.t @@ -0,0 +1,363 @@ +#!/usr/bin/perl +# +# Unit tests of _downcopy function +# +# _downcopy($self, $data, $pos, $len) +# Write $data into a block of length $len at position $pos, +# moving everything in the block forwards to make room. +# Instead of writing the last length($data) bytes from the block +# (because there isn't room for them any longer) return them. +# +# + +my $file = "tf$$.txt"; + +print "1..718\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +$: = Tie::File::_default_recsep(); + +my @subtests = qw(x <x x> x><x <x> <x><x x><x> <x><x> <x><x><x> 0); +print "ok $N\n"; $N++; + +# (3-144) These were generated by 'gentests.pl' to cover all possible cases +# (I hope) +# Legend: +# x: data is entirely contained within one block +# x>: data runs from the middle to the end of the block +# <x: data runs from the start to the middle of the block +# <x>: data occupies precisely one block +# x><x: data overlaps one block boundary +# <x><x: data runs from the start of one block into the middle of the next +# x><x>: data runs from the middle of one block to the end of the next +# <x><x>: data occupies two blocks exactly +# <x><x><x>: data occupies three blocks exactly +# 0: data is null +# +# For each possible alignment of the old and new data, we investigate +# up to three situations: old data is shorter, old and new data are the +# same length, and new data is shorter. +# +# try($pos, $old, $new) means to run a test where the data starts at +# position $pos, the old data has length $old, +# and the new data has length $new. +try( 9659, 6635, 6691); # old=x , new=x ; old < new +try( 8605, 2394, 2394); # old=x , new=x ; old = new +try( 9768, 1361, 664); # old=x , new=x ; old > new +try( 9955, 6429, 6429); # old=x> , new=x ; old = new +try(10550, 5834, 4123); # old=x> , new=x ; old > new +try(14580, 6158, 851); # old=x><x , new=x ; old > new +try(13442, 11134, 1572); # old=x><x> , new=x ; old > new +try( 8394, 0, 5742); # old=0 , new=x ; old < new +try( 8192, 2819, 6738); # old=<x , new=<x ; old < new +try( 8192, 514, 514); # old=<x , new=<x ; old = new +try( 8192, 2196, 858); # old=<x , new=<x ; old > new +try( 8192, 8192, 8192); # old=<x> , new=<x ; old = new +try( 8192, 8192, 1290); # old=<x> , new=<x ; old > new +try( 8192, 10575, 6644); # old=<x><x , new=<x ; old > new +try( 8192, 16384, 5616); # old=<x><x> , new=<x ; old > new +try( 8192, 24576, 6253); # old=<x><x><x>, new=<x ; old > new +try( 8192, 0, 6870); # old=0 , new=<x ; old < new +try( 8478, 6259, 7906); # old=x , new=x> ; old < new +try( 9965, 6419, 6419); # old=x> , new=x> ; old = new +try(16059, 6102, 325); # old=x><x , new=x> ; old > new +try( 9503, 15073, 6881); # old=x><x> , new=x> ; old > new +try( 9759, 0, 6625); # old=0 , new=x> ; old < new +try( 8525, 2081, 8534); # old=x , new=x><x ; old < new +try(15550, 834, 1428); # old=x> , new=x><x ; old < new +try(14966, 1668, 3479); # old=x><x , new=x><x ; old < new +try(16316, 1605, 1605); # old=x><x , new=x><x ; old = new +try(16093, 4074, 993); # old=x><x , new=x><x ; old > new +try(14739, 9837, 9837); # old=x><x> , new=x><x ; old = new +try(14071, 10505, 7344); # old=x><x> , new=x><x ; old > new +try(12602, 0, 8354); # old=0 , new=x><x ; old < new +try( 8192, 2767, 8192); # old=<x , new=<x> ; old < new +try( 8192, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 8192, 14817, 8192); # old=<x><x , new=<x> ; old > new +try( 8192, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 8192, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 8192, 0, 8192); # old=0 , new=<x> ; old < new +try( 8192, 6532, 10882); # old=<x , new=<x><x ; old < new +try( 8192, 8192, 16044); # old=<x> , new=<x><x ; old < new +try( 8192, 9555, 11020); # old=<x><x , new=<x><x ; old < new +try( 8192, 9001, 9001); # old=<x><x , new=<x><x ; old = new +try( 8192, 11760, 10274); # old=<x><x , new=<x><x ; old > new +try( 8192, 16384, 10781); # old=<x><x> , new=<x><x ; old > new +try( 8192, 24576, 9284); # old=<x><x><x>, new=<x><x ; old > new +try( 8192, 0, 12488); # old=0 , new=<x><x ; old < new +try( 8222, 6385, 16354); # old=x , new=x><x> ; old < new +try(13500, 2884, 11076); # old=x> , new=x><x> ; old < new +try(14069, 4334, 10507); # old=x><x , new=x><x> ; old < new +try(14761, 9815, 9815); # old=x><x> , new=x><x> ; old = new +try(10469, 0, 14107); # old=0 , new=x><x> ; old < new +try( 8192, 4181, 16384); # old=<x , new=<x><x> ; old < new +try( 8192, 8192, 16384); # old=<x> , new=<x><x> ; old < new +try( 8192, 12087, 16384); # old=<x><x , new=<x><x> ; old < new +try( 8192, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 8192, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 8192, 0, 16384); # old=0 , new=<x><x> ; old < new +try( 8192, 4968, 24576); # old=<x , new=<x><x><x>; old < new +try( 8192, 8192, 24576); # old=<x> , new=<x><x><x>; old < new +try( 8192, 14163, 24576); # old=<x><x , new=<x><x><x>; old < new +try( 8192, 16384, 24576); # old=<x><x> , new=<x><x><x>; old < new +try( 8192, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 8192, 0, 24576); # old=0 , new=<x><x><x>; old < new +try( 8771, 776, 0); # old=x , new=0 ; old > new +try( 8192, 2813, 0); # old=<x , new=0 ; old > new +try(13945, 2439, 0); # old=x> , new=0 ; old > new +try(14493, 6090, 0); # old=x><x , new=0 ; old > new +try( 8192, 8192, 0); # old=<x> , new=0 ; old > new +try( 8192, 10030, 0); # old=<x><x , new=0 ; old > new +try(14983, 9593, 0); # old=x><x> , new=0 ; old > new +try( 8192, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 8192, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(10489, 0, 0); # old=0 , new=0 ; old = new + +# (142-223) +# These tests all take place at the start of the file +try( 0, 771, 1593); # old=<x , new=<x ; old < new +try( 0, 4868, 4868); # old=<x , new=<x ; old = new +try( 0, 147, 118); # old=<x , new=<x ; old > new +try( 0, 8192, 8192); # old=<x> , new=<x ; old = new +try( 0, 8192, 4574); # old=<x> , new=<x ; old > new +try( 0, 11891, 1917); # old=<x><x , new=<x ; old > new +try( 0, 16384, 5155); # old=<x><x> , new=<x ; old > new +try( 0, 24576, 2953); # old=<x><x><x>, new=<x ; old > new +try( 0, 0, 1317); # old=0 , new=<x ; old < new +try( 0, 5609, 8192); # old=<x , new=<x> ; old < new +try( 0, 8192, 8192); # old=<x> , new=<x> ; old = new +try( 0, 11083, 8192); # old=<x><x , new=<x> ; old > new +try( 0, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try( 0, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try( 0, 0, 8192); # old=0 , new=<x> ; old < new +try( 0, 6265, 9991); # old=<x , new=<x><x ; old < new +try( 0, 8192, 16119); # old=<x> , new=<x><x ; old < new +try( 0, 10218, 11888); # old=<x><x , new=<x><x ; old < new +try( 0, 14126, 14126); # old=<x><x , new=<x><x ; old = new +try( 0, 12002, 9034); # old=<x><x , new=<x><x ; old > new +try( 0, 16384, 13258); # old=<x><x> , new=<x><x ; old > new +try( 0, 24576, 14367); # old=<x><x><x>, new=<x><x ; old > new +try( 0, 0, 10881); # old=0 , new=<x><x ; old < new +try( 0, 6448, 16384); # old=<x , new=<x><x> ; old < new +try( 0, 8192, 16384); # old=<x> , new=<x><x> ; old < new +try( 0, 15082, 16384); # old=<x><x , new=<x><x> ; old < new +try( 0, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try( 0, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try( 0, 0, 16384); # old=0 , new=<x><x> ; old < new +try( 0, 2421, 24576); # old=<x , new=<x><x><x>; old < new +try( 0, 8192, 24576); # old=<x> , new=<x><x><x>; old < new +try( 0, 11655, 24576); # old=<x><x , new=<x><x><x>; old < new +try( 0, 16384, 24576); # old=<x><x> , new=<x><x><x>; old < new +try( 0, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try( 0, 0, 24576); # old=0 , new=<x><x><x>; old < new +try( 0, 6530, 0); # old=<x , new=0 ; old > new +try( 0, 8192, 0); # old=<x> , new=0 ; old > new +try( 0, 14707, 0); # old=<x><x , new=0 ; old > new +try( 0, 16384, 0); # old=<x><x> , new=0 ; old > new +try( 0, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try( 0, 0, 0); # old=0 , new=0 ; old = new + +# (224-277) +# These tests all take place at the end of the file +$FLEN = 40960; # Force the file to be exactly 40960 bytes long +try(32768, 8192, 8192); # old=<x> , new=<x ; old = new +try(32768, 8192, 4026); # old=<x> , new=<x ; old > new +try(24576, 16384, 1917); # old=<x><x> , new=<x ; old > new +try(16384, 24576, 3818); # old=<x><x><x>, new=<x ; old > new +try(40960, 0, 2779); # old=0 , new=<x ; old < new +try(32768, 8192, 8192); # old=<x> , new=<x> ; old = new +try(24576, 16384, 8192); # old=<x><x> , new=<x> ; old > new +try(16384, 24576, 8192); # old=<x><x><x>, new=<x> ; old > new +try(40960, 0, 8192); # old=0 , new=<x> ; old < new +try(32768, 8192, 10724); # old=<x> , new=<x><x ; old < new +try(24576, 16384, 12221); # old=<x><x> , new=<x><x ; old > new +try(16384, 24576, 15030); # old=<x><x><x>, new=<x><x ; old > new +try(40960, 0, 11752); # old=0 , new=<x><x ; old < new +try(32768, 8192, 16384); # old=<x> , new=<x><x> ; old < new +try(24576, 16384, 16384); # old=<x><x> , new=<x><x> ; old = new +try(16384, 24576, 16384); # old=<x><x><x>, new=<x><x> ; old > new +try(40960, 0, 16384); # old=0 , new=<x><x> ; old < new +try(32768, 8192, 24576); # old=<x> , new=<x><x><x>; old < new +try(24576, 16384, 24576); # old=<x><x> , new=<x><x><x>; old < new +try(16384, 24576, 24576); # old=<x><x><x>, new=<x><x><x>; old = new +try(40960, 0, 24576); # old=0 , new=<x><x><x>; old < new +try(35973, 4987, 0); # old=x> , new=0 ; old > new +try(32768, 8192, 0); # old=<x> , new=0 ; old > new +try(29932, 11028, 0); # old=x><x> , new=0 ; old > new +try(24576, 16384, 0); # old=<x><x> , new=0 ; old > new +try(16384, 24576, 0); # old=<x><x><x>, new=0 ; old > new +try(40960, 0, 0); # old=0 , new=0 ; old = new + +# (278-357) +# These tests all take place at the end of the file +$FLEN = 42000; # Force the file to be exactly 42000 bytes long +try(41275, 725, 4059); # old=x , new=x ; old < new +try(41683, 317, 317); # old=x , new=x ; old = new +try(41225, 775, 405); # old=x , new=x ; old > new +try(35709, 6291, 284); # old=x><x , new=x ; old > new +try(42000, 0, 2434); # old=0 , new=x ; old < new +try(40960, 1040, 1608); # old=<x , new=<x ; old < new +try(40960, 1040, 1040); # old=<x , new=<x ; old = new +try(40960, 1040, 378); # old=<x , new=<x ; old > new +try(32768, 9232, 5604); # old=<x><x , new=<x ; old > new +try(42000, 0, 6637); # old=0 , new=<x ; old < new +try(41022, 978, 8130); # old=x , new=x> ; old < new +try(39994, 2006, 966); # old=x><x , new=x> ; old > new +try(42000, 0, 7152); # old=0 , new=x> ; old < new +try(41613, 387, 10601); # old=x , new=x><x ; old < new +try(38460, 3540, 3938); # old=x><x , new=x><x ; old < new +try(36725, 5275, 5275); # old=x><x , new=x><x ; old = new +try(37990, 4010, 3199); # old=x><x , new=x><x ; old > new +try(42000, 0, 9189); # old=0 , new=x><x ; old < new +try(40960, 1040, 8192); # old=<x , new=<x> ; old < new +try(32768, 9232, 8192); # old=<x><x , new=<x> ; old > new +try(42000, 0, 8192); # old=0 , new=<x> ; old < new +try(40960, 1040, 11778); # old=<x , new=<x><x ; old < new +try(32768, 9232, 13792); # old=<x><x , new=<x><x ; old < new +try(32768, 9232, 9232); # old=<x><x , new=<x><x ; old = new +try(32768, 9232, 8795); # old=<x><x , new=<x><x ; old > new +try(42000, 0, 8578); # old=0 , new=<x><x ; old < new +try(41531, 469, 15813); # old=x , new=x><x> ; old < new +try(39618, 2382, 9534); # old=x><x , new=x><x> ; old < new +try(42000, 0, 15344); # old=0 , new=x><x> ; old < new +try(40960, 1040, 16384); # old=<x , new=<x><x> ; old < new +try(32768, 9232, 16384); # old=<x><x , new=<x><x> ; old < new +try(42000, 0, 16384); # old=0 , new=<x><x> ; old < new +try(40960, 1040, 24576); # old=<x , new=<x><x><x>; old < new +try(32768, 9232, 24576); # old=<x><x , new=<x><x><x>; old < new +try(42000, 0, 24576); # old=0 , new=<x><x><x>; old < new +try(41500, 500, 0); # old=x , new=0 ; old > new +try(40960, 1040, 0); # old=<x , new=0 ; old > new +try(35272, 6728, 0); # old=x><x , new=0 ; old > new +try(32768, 9232, 0); # old=<x><x , new=0 ; old > new +try(42000, 0, 0); # old=0 , new=0 ; old = new + +sub try { + my ($pos, $len, $newlen) = @_; + open F, "> $file" or die "Couldn't open file $file: $!"; + binmode F; + + # The record has exactly 17 characters. This will help ensure that + # even if _downcoopy screws up, the data doesn't coincidentally + # look good because the remainder accidentally lines up. + my $d = substr("0123456789abcdef$:", -17); + my $recs = defined($FLEN) ? + int($FLEN/length($d))+1 : # enough to make up at least $FLEN + int(8192*5/length($d))+1; # at least 5 blocks' worth + my $oldfile = $d x $recs; + my $flen = defined($FLEN) ? $FLEN : $recs * 17; + substr($oldfile, $FLEN) = "" if defined $FLEN; # truncate + print F $oldfile; + close F; + + die "wrong length!" unless -s $file == $flen; + + my $newdata = "-" x $newlen; + my $expected = $oldfile; + my $old = defined $len ? substr($expected, $pos, $len) + : substr($expected, $pos); + $old = "$newdata$old"; + my $x_retval; + if (defined $len) { + substr($expected, $pos, $len, substr($old, 0, $len, "")); + $x_retval = $old; + } else { + substr($expected, $pos) = $old; + $x_retval = ""; + } + + my $o = tie my @lines, 'Tie::File', $file or die $!; + local $SIG{ALRM} = sub { die "Alarm clock" }; + my $a_retval = eval { alarm(5) unless $^P; $o->_downcopy($newdata, $pos, $len) }; + my $err = $@; + undef $o; untie @lines; alarm(0); + if ($err) { + if ($err =~ /^Alarm clock/) { + print "# Timeout\n"; + print "not ok $N\n"; $N++; + print "not ok $N\n"; $N++; + return; + } else { + $@ = $err; + die; + } + } + + open F, "< $file" or die "Couldn't open file $file: $!"; + binmode F; + my $actual; + { local $/; + $actual = <F>; + } + close F; + + my ($alen, $xlen) = (length $actual, length $expected); + unless ($alen == $xlen) { + my @ARGS = @_; + for (@ARGS) { $_ = "UNDEF" unless defined } + print "# try(@ARGS) expected file length $xlen, actual $alen!\n"; + } + print $actual eq $expected ? "ok $N\n" : "not ok $N\n"; + $N++; + print $a_retval eq $x_retval ? "ok $N\n" : "not ok $N\n"; + $N++; + + if (defined $len) { + try($pos, undef, $newlen); + } +} + + + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $:, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; +# my $open = open FH, "< $file"; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + my $msg; + for (0.. $#c) { + my $aa = $a[$_]; + unless ($aa eq "$c[$_]$:") { + $msg = "expected <$c[$_]$:>, got <$aa>"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # $msg\n"; + $N++; + + print $o->_check_integrity($file, $ENV{INTEGRITY}) + ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} diff --git a/dist/Tie-File/t/29a_upcopy.t b/dist/Tie-File/t/29a_upcopy.t new file mode 100644 index 0000000000..1130615f37 --- /dev/null +++ b/dist/Tie-File/t/29a_upcopy.t @@ -0,0 +1,211 @@ +#!/usr/bin/perl +# +# Unit tests of _upcopy function +# +# _upcopy($self, $source, $dest, $len) +# +# Take a block of data of leength $len at $source and copy it +# to $dest, which must be <= $source but which need not be <= $source - $len +# (That is, this will only copy a block to a position earlier in the file, +# but the source and destination regions may overlap.) + + +my $file = "tf$$.txt"; + +print "1..55\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +$: = Tie::File::_default_recsep(); + +my @subtests = qw(x <x x> x><x <x> <x><x x><x> <x><x> <x><x><x> 0); + +$FLEN = 40970; # 2410 records of 17 chars each + +# (2-7) Trivial non-moves at start of file +try(0, 0, 0); +try(0, 0, 10); +try(0, 0, 100); +try(0, 0, 1000); +try(0, 0, 10000); +try(0, 0, 20000); + +# (8-13) Trivial non-moves in middle of file +try(100, 100, 0); +try(100, 100, 10); +try(100, 100, 100); +try(100, 100, 1000); +try(100, 100, 10000); +try(100, 100, 20000); + +# (14) Trivial non-move at end of file +try($FLEN, $FLEN, 0); + +# (15-17) Trivial non-move of tail of file +try(0, 0, undef); +try(100, 100, undef); +try($FLEN, $FLEN, undef); + +# (18-24) Moves to start of file +try(100, 0, 0); +try(100, 0, 10); +try(100, 0, 100); +try(100, 0, 1000); +try(100, 0, 10000); +try(100, 0, 20000); +try(100, 0, undef); + +# (25-31) Moves in middle of file +try(200, 100, 0); +try(200, 100, 10); +try(200, 100, 100); +try(200, 100, 1000); +try(200, 100, 10000); +try(200, 100, 20000); +try(200, 100, undef); + +# (32-43) Moves from end of file +try($FLEN, 10000, 0); +try($FLEN-10, 10000, 10); +try($FLEN-100, 10000, 100); +try($FLEN-1000, 200, 1000); +try($FLEN-10000, 200, 10000); +try($FLEN-20000, 200, 20000); +try($FLEN, 10000, undef); +try($FLEN-10, 10000, undef); +try($FLEN-100, 10000, undef); +try($FLEN-1000, 200, undef); +try($FLEN-10000, 200, undef); +try($FLEN-20000, 200, undef); + +$FLEN = 40960; + +# (44-55) Moves from end of file when file ends on a block boundary +try($FLEN, 10000, 0); +try($FLEN-10, 10000, 10); +try($FLEN-100, 10000, 100); +try($FLEN-1000, 200, 1000); +try($FLEN-10000, 200, 10000); +try($FLEN-20000, 200, 20000); +try($FLEN, 10000, undef); +try($FLEN-10, 10000, undef); +try($FLEN-100, 10000, undef); +try($FLEN-1000, 200, undef); +try($FLEN-10000, 200, undef); +try($FLEN-20000, 200, undef); + +sub try { + my ($src, $dst, $len) = @_; + open F, "> $file" or die "Couldn't open file $file: $!"; + binmode F; + + # The record has exactly 17 characters. This will help ensure that + # even if _upcopy screws up, the data doesn't coincidentally + # look good because the remainder accidentally lines up. + my $d = substr("0123456789abcdef$:", -17); + my $recs = defined($FLEN) ? + int($FLEN/length($d))+1 : # enough to make up at least $FLEN + int(8192*5/length($d))+1; # at least 5 blocks' worth + my $oldfile = $d x $recs; + my $flen = defined($FLEN) ? $FLEN : $recs * 17; + substr($oldfile, $FLEN) = "" if defined $FLEN; # truncate + print F $oldfile; + close F; + + die "wrong length!" unless -s $file == $flen; + + # If len is specified, use that. If it's undef, + # then behave *as if* we had specified the whole rest of the file + my $expected = $oldfile; + if (defined $len) { + substr($expected, $dst, $len) = substr($expected, $src, $len); + } else { + substr($expected, $dst) = substr($expected, $src); + } + + my $o = tie my @lines, 'Tie::File', $file or die $!; + local $SIG{ALRM} = sub { die "Alarm clock" }; + my $a_retval = eval { alarm(5) unless $^P; $o->_upcopy($src, $dst, $len) }; + my $err = $@; + undef $o; untie @lines; alarm(0); + if ($err) { + if ($err =~ /^Alarm clock/) { + print "# Timeout\n"; + print "not ok $N\n"; $N++; + return; + } else { + $@ = $err; + die; + } + } + + open F, "< $file" or die "Couldn't open file $file: $!"; + binmode F; + my $actual; + { local $/; + $actual = <F>; + } + close F; + + my ($alen, $xlen) = (length $actual, length $expected); + unless ($alen == $xlen) { + print "# try(@_) expected file length $xlen, actual $alen!\n"; + } + print $actual eq $expected ? "ok $N\n" : "not ok $N\n"; + $N++; +} + + + +use POSIX 'SEEK_SET'; +sub check_contents { + my @c = @_; + my $x = join $:, @c, ''; + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; +# my $open = open FH, "< $file"; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix($a, $x); + print "not ok $N\n# expected <$x>, got <$a>\n"; + } + $N++; + + # now check FETCH: + my $good = 1; + my $msg; + for (0.. $#c) { + my $aa = $a[$_]; + unless ($aa eq "$c[$_]$:") { + $msg = "expected <$c[$_]$:>, got <$aa>"; + ctrlfix($msg); + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N # $msg\n"; + $N++; + + print $o->_check_integrity($file, $ENV{INTEGRITY}) + ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/30_defer.t b/dist/Tie-File/t/30_defer.t new file mode 100644 index 0000000000..063b3a7090 --- /dev/null +++ b/dist/Tie-File/t/30_defer.t @@ -0,0 +1,328 @@ +#!/usr/bin/perl +# +# Check ->defer and ->flush methods +# +# This is the old version, which you used in the past when +# there was a defer buffer separate from the read cache. +# There isn't any longer. +# + +use POSIX 'SEEK_SET'; +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; +my ($o, $n); + +print "1..79\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; +$o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (3-6) Deferred storage +$o->defer; +$a[3] = "rec3"; +check_contents($data); # nothing written yet +$a[4] = "rec4"; +check_contents($data); # nothing written yet + +# (7-8) Flush +$o->flush; +check_contents($data . "rec3$:rec4$:"); # now it's written + +# (9-12) Deferred writing disabled? +$a[3] = "rec9"; +check_contents("${data}rec9$:rec4$:"); +$a[4] = "rec8"; +check_contents("${data}rec9$:rec8$:"); + +# (13-18) Now let's try two batches of records +$#a = 2; +$o->defer; +$a[0] = "record0"; +check_contents($data); # nothing written yet +$a[2] = "record2"; +check_contents($data); # nothing written yet +$o->flush; +check_contents("record0$:rec1$:record2$:"); + +# (19-22) Deferred writing past the end of the file +$o->defer; +$a[4] = "record4"; +check_contents("record0$:rec1$:record2$:"); +$o->flush; +check_contents("record0$:rec1$:record2$:$:record4$:"); + + +# (23-26) Now two long batches +$o->defer; +for (0..2, 4..6) { + $a[$_] = "r$_"; +} +check_contents("record0$:rec1$:record2$:$:record4$:"); +$o->flush; +check_contents(join $:, "r0".."r2", "", "r4".."r6", ""); + +# (27-30) Now let's make sure that discarded writes are really discarded +# We have a 2Mib buffer here, so we can be sure that we aren't accidentally +# filling it up +$o->defer; +for (0, 3, 7) { + $a[$_] = "discarded$_"; +} +check_contents(join $:, "r0".."r2", "", "r4".."r6", ""); +$o->discard; +check_contents(join $:, "r0".."r2", "", "r4".."r6", ""); + +################################################################ +# +# Now we're going to test the results of a small memory limit +# +# +undef $o; untie @a; +$data = join "$:", map("record$_", 0..7), ""; # records are 8 or 9 bytes long +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; + +# Limit cache+buffer size to 47 bytes +my $MAX = 47; +# -- that's enough space for 5 records, but not 6, on both \n and \r\n systems +my $BUF = 20; +# -- that's enough space for 2 records, but not 3, on both \n and \r\n systems +$o = tie @a, 'Tie::File', $file, memory => $MAX, dw_size => $BUF; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (31-32) Fill up the read cache +my @z; +@z = @a; +# the cache now contains records 3,4,5,6,7. +check_caches({map(($_ => "record$_$:"), 3..7)}, + {}); + +# (33-44) See if overloading the defer starts by flushing the read cache +# and then flushes out the defer +$o->defer; +$a[0] = "recordA"; # That should flush record 3 from the cache +check_caches({map(($_ => "record$_$:"), 4..7)}, + {0 => "recordA$:"}); +check_contents($data); + +$a[1] = "recordB"; # That should flush record 4 from the cache +check_caches({map(($_ => "record$_$:"), 5..7)}, + {0 => "recordA$:", + 1 => "recordB$:"}); +check_contents($data); + +$a[2] = "recordC"; # That should flush the whole darn defer +# This shouldn't change the cache contents +check_caches({map(($_ => "record$_$:"), 5..7)}, + {}); # URRRP +check_contents(join("$:", qw(recordA recordB recordC + record3 record4 record5 record6 record7)) . "$:"); + +$a[3] = "recordD"; # even though we flushed, deferring is STILL ENABLED +check_caches({map(($_ => "record$_$:"), 5..7)}, + {3 => "recordD$:"}); +check_contents(join("$:", qw(recordA recordB recordC + record3 record4 record5 record6 record7)) . "$:"); + +# Check readcache-deferbuffer interactions + +# (45-47) This should remove outdated data from the read cache +$a[5] = "recordE"; +check_caches({6 => "record6$:", 7 => "record7$:"}, + {3 => "recordD$:", 5 => "recordE$:"}); +check_contents(join("$:", qw(recordA recordB recordC + record3 record4 record5 record6 record7)) . "$:"); + +# (48-51) This should read back out of the defer buffer +# without adding anything to the read cache +my $z; +$z = $a[5]; +print $z eq "recordE" ? "ok $N\n" : "not ok $N\n"; $N++; +check_caches({6 => "record6$:", 7 => "record7$:"}, + {3 => "recordD$:", 5 => "recordE$:"}); +check_contents(join("$:", qw(recordA recordB recordC + record3 record4 record5 record6 record7)) . "$:"); + +# (52-55) This should repopulate the read cache with a new record +$z = $a[0]; +print $z eq "recordA" ? "ok $N\n" : "not ok $N\n"; $N++; +check_caches({0 => "recordA$:", 6 => "record6$:", 7 => "record7$:"}, + {3 => "recordD$:", 5 => "recordE$:"}); +check_contents(join("$:", qw(recordA recordB recordC + record3 record4 record5 record6 record7)) . "$:"); + +# (56-59) This should flush the LRU record from the read cache +$z = $a[4]; +print $z eq "record4" ? "ok $N\n" : "not ok $N\n"; $N++; +check_caches({7 => "record7$:", 0 => "recordA$:", 4 => "record4$:"}, + {3 => "recordD$:", 5 => "recordE$:"}); +check_contents(join("$:", qw(recordA recordB recordC + record3 record4 record5 record6 record7)) . "$:"); + +# (60-63) This should FLUSH the deferred buffer +$z = splice @a, 3, 1, "recordZ"; +print $z eq "recordD" ? "ok $N\n" : "not ok $N\n"; $N++; +check_caches({7 => "record7$:", 0 => "recordA$:", 4 => "record4$:", 3 => "recordZ$:"}, + {}); +check_contents(join("$:", qw(recordA recordB recordC + recordZ record4 recordE record6 record7)) . "$:"); + +# (64-66) We should STILL be in deferred writing mode +$a[5] = "recordX"; +check_caches({7 => "record7$:", 0 => "recordA$:", 4 => "record4$:", 3 => "recordZ$:"}, + {5 => "recordX$:"}); +check_contents(join("$:", qw(recordA recordB recordC + recordZ record4 recordE record6 record7)) . "$:"); + +# Fill up the defer buffer again +$a[4] = "recordP"; +# (67-69) This should OVERWRITE the existing deferred record +# and NOT flush the buffer +$a[5] = "recordQ"; +check_caches({7 => "record7$:", 0 => "recordA$:", 3 => "recordZ$:"}, + {5 => "recordQ$:", 4 => "recordP$:"}); +check_contents(join("$:", qw(recordA recordB recordC + recordZ record4 recordE record6 record7)) . "$:"); + +# (70-72) Discard should just dump the whole deferbuffer +$o->discard; +check_caches({7 => "record7$:", 0 => "recordA$:", 3 => "recordZ$:"}, + {}); +check_contents(join("$:", qw(recordA recordB recordC + recordZ record4 recordE record6 record7)) . "$:"); + +# (73-75) NOW we are out of deferred writing mode +$a[0] = "recordF"; +check_caches({7 => "record7$:", 0 => "recordF$:", 3 => "recordZ$:"}, + {}); +check_contents(join("$:", qw(recordF recordB recordC + recordZ record4 recordE record6 record7)) . "$:"); + +# (76-79) Last call--untying the array should flush the deferbuffer +$o->defer; +$a[0] = "flushed"; +check_caches({7 => "record7$:", 3 => "recordZ$:"}, + {0 => "flushed$:" }); +check_contents(join("$:", qw(recordF recordB recordC + recordZ record4 recordE record6 record7)) . "$:"); +undef $o; +untie @a; +# (79) We can't use check_contents any more, because the object is dead +open F, "< $file" or die; +binmode F; +{ local $/ ; $z = <F> } +close F; +my $x = join("$:", qw(flushed recordB recordC + recordZ record4 recordE record6 record7)) . "$:"; +if ($z eq $x) { + print "ok $N\n"; +} else { + my $msg = ctrlfix("expected <$x>, got <$z>"); + print "not ok $N \# $msg\n"; +} +$N++; + +################################################################ + + +sub check_caches { + my ($xcache, $xdefer) = @_; + +# my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); +# print $integrity ? "ok $N\n" : "not ok $N\n"; +# $N++; + + my $good = 1; + + # Copy the contents of the cache into a regular hash + my %cache; + for my $k ($o->{cache}->ckeys) { + $cache{$k} = $o->{cache}->_produce($k); + } + + $good &&= hash_equal(\%cache, $xcache, "true cache", "expected cache"); + $good &&= hash_equal($o->{deferred}, $xdefer, "true defer", "expected defer"); + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub hash_equal { + my ($a, $b, $ha, $hb) = @_; + $ha = 'first hash' unless defined $ha; + $hb = 'second hash' unless defined $hb; + + my $good = 1; + my %b_seen; + + for my $k (keys %$a) { + if (! exists $b->{$k}) { + print ctrlfix("# Key $k is in $ha but not $hb"), "\n"; + $good = 0; + } elsif ($b->{$k} ne $a->{$k}) { + print ctrlfix("# Key $k is <$a->{$k}> in $ha but <$b->{$k}> in $hb"), "\n"; + $b_seen{$k} = 1; + $good = 0; + } else { + $b_seen{$k} = 1; + } + } + + for my $k (keys %$b) { + unless ($b_seen{$k}) { + print ctrlfix("# Key $k is in $hb but not $ha"), "\n"; + $good = 0; + } + } + + $good; +} + + +sub check_contents { + my $x = shift; + + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + my $msg = ctrlfix("# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; +} + +sub ctrlfix { + local $_ = shift; + s/\n/\\n/g; + s/\r/\\r/g; + $_; +} + +END { + undef $o; + untie @a if tied @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/31_autodefer.t b/dist/Tie-File/t/31_autodefer.t new file mode 100644 index 0000000000..ea929a4097 --- /dev/null +++ b/dist/Tie-File/t/31_autodefer.t @@ -0,0 +1,182 @@ +#!/usr/bin/perl +# +# Check behavior of 'autodefer' feature +# Mostly this isn't implemented yet +# This file is primarily here to make sure that the promised ->autodefer +# method doesn't croak. +# + +use POSIX 'SEEK_SET'; + +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; +my ($o, $n, @a); + +print "1..65\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; +$o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# I am an undocumented feature +$o->{autodefer_filelen_threshhold} = 0; +# Normally autodeferring only works on large files. This disables that. + +# (3-22) Deferred storage +$a[3] = "rec3"; +check_autodeferring('OFF'); +$a[4] = "rec4"; +check_autodeferring('OFF'); +$a[5] = "rec5"; +check_autodeferring('ON'); +check_contents($data . "rec3$:rec4$:"); # only the first two were written +$a[6] = "rec6"; +check_autodeferring('ON'); +check_contents($data . "rec3$:rec4$:"); # still nothing written +$a[7] = "rec7"; +check_autodeferring('ON'); +check_contents($data . "rec3$:rec4$:"); # still nothing written +$a[0] = "recX"; +check_autodeferring('OFF'); +check_contents("recX$:rec1$:rec2$:rec3$:rec4$:rec5$:rec6$:rec7$:"); +$a[1] = "recY"; +check_autodeferring('OFF'); +check_contents("recX$:recY$:rec2$:rec3$:rec4$:rec5$:rec6$:rec7$:"); +$a[2] = "recZ"; # it kicks in here +check_autodeferring('ON'); +check_contents("recX$:recY$:rec2$:rec3$:rec4$:rec5$:rec6$:rec7$:"); + +# (23-26) Explicitly enabling deferred writing deactivates autodeferring +$o->defer; +check_autodeferring('OFF'); +check_contents("recX$:recY$:recZ$:rec3$:rec4$:rec5$:rec6$:rec7$:"); +$o->discard; +check_autodeferring('OFF'); + +# (27-32) Now let's try the CLEAR special case +@a = ("r0" .. "r4"); +check_autodeferring('ON'); +# The file was extended to the right length, but nothing was actually written. +check_contents("$:$:$:$:$:"); +$a[2] = "fish"; +check_autodeferring('OFF'); +check_contents("r0$:r1$:fish$:r3$:r4$:"); + +# (33-47) Now let's try the originally intended application: a 'for' loop. +my $it = 0; +for (@a) { + $_ = "##$_"; + if ($it == 0) { + check_autodeferring('OFF'); + check_contents("##r0$:r1$:fish$:r3$:r4$:"); + } elsif ($it == 1) { + check_autodeferring('OFF'); + check_contents("##r0$:##r1$:fish$:r3$:r4$:"); + } else { + check_autodeferring('ON'); + check_contents("##r0$:##r1$:fish$:r3$:r4$:"); + } + $it++; +} + +# (48-56) Autodeferring should not become active during explicit defer mode +$o->defer(); # This should flush the pending autodeferred records + # and deactivate autodeferring +check_autodeferring('OFF'); +check_contents("##r0$:##r1$:##fish$:##r3$:##r4$:"); +@a = ("s0" .. "s4"); +check_autodeferring('OFF'); +check_contents(""); +$o->flush; +check_autodeferring('OFF'); +check_contents("s0$:s1$:s2$:s3$:s4$:"); + +undef $o; untie @a; + +# Limit cache+buffer size to 47 bytes +my $MAX = 47; +# -- that's enough space for 5 records, but not 6, on both \n and \r\n systems +my $BUF = 20; +# -- that's enough space for 2 records, but not 3, on both \n and \r\n systems +# Re-tie the object for more tests +$o = tie @a, 'Tie::File', $file, autodefer => 0; +die $! unless $o; +# I am an undocumented feature +$o->{autodefer_filelen_threshhold} = 0; +# Normally autodeferring only works on large files. This disables that. + +# (57-59) Did the autodefer => 0 option work? +# (If it doesn't, a whole bunch of the other test files will fail.) +@a = (0..3); +check_autodeferring('OFF'); +check_contents(join("$:", qw(0 1 2 3), "")); + +# (60-62) Does the ->autodefer method work? +$o->autodefer(1); +@a = (10..13); +check_autodeferring('ON'); +check_contents("$:$:$:$:"); # This might be unfortunate. + +# (63-65) Does the ->autodefer method work? +$o->autodefer(0); +check_autodeferring('OFF'); +check_contents(join("$:", qw(10 11 12 13), "")); + + +sub check_autodeferring { + my ($x) = shift; + my $a = $o->{autodeferring} ? 'ON' : 'OFF'; + if ($x eq $a) { + print "ok $N\n"; + } else { + print "not ok $N \# Autodeferring was $a, expected it to be $x\n"; + } + $N++; +} + + +sub check_contents { + my $x = shift; +# for (values %{$o->{cache}}) { +# print "# cache=$_"; +# } + + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + ctrlfix(my $msg = "# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; +} + +sub ctrlfix { + for (@_) { + s/\n/\\n/g; + s/\r/\\r/g; + } +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/32_defer_misc.t b/dist/Tie-File/t/32_defer_misc.t new file mode 100644 index 0000000000..e0e3f15bb8 --- /dev/null +++ b/dist/Tie-File/t/32_defer_misc.t @@ -0,0 +1,232 @@ +#!/usr/bin/perl +# +# Check interactions of deferred writing +# with miscellaneous methods like DELETE, EXISTS, +# FETCHSIZE, STORESIZE, CLEAR, EXTEND +# + +use POSIX 'SEEK_SET'; +my $file = "tf$$.txt"; +$: = Tie::File::_default_recsep(); +my $data = "rec0$:rec1$:rec2$:"; +my ($o, $n); + +print "1..53\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; +$o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (3-6) EXISTS +if ($] >= 5.006) { + eval << 'TESTS'; +$o->defer; +expect(not exists $a[4]); +$a[4] = "rec4"; +expect(exists $a[4]); +check_contents($data); # nothing written yet +$o->discard; +TESTS +} else { + for (3..6) { + print "ok $_ \# skipped (no exists for arrays)\n"; + $N++; + } +} + +# (7-10) FETCHSIZE +$o->defer; +expect($#a, 2); +$a[4] = "rec4"; +expect($#a, 4); +check_contents($data); # nothing written yet +$o->discard; + +# (11-21) STORESIZE +$o->defer; +$#a = 4; +check_contents($data); # nothing written yet +expect($#a, 4); +$o->flush; +expect($#a, 4); +check_contents("$data$:$:"); # two extra empty records + +$o->defer; +$a[4] = "rec4"; +$#a = 2; +expect($a[4], undef); +check_contents($data); # written data was unwritten +$o->flush; +check_contents($data); # nothing left to write + +# (22-28) CLEAR +$o->defer; +$a[9] = "rec9"; +check_contents($data); # nothing written yet +@a = (); +check_contents(""); # this happens right away +expect($a[9], undef); +$o->flush; +check_contents(""); # nothing left to write + +# (29-34) EXTEND +# Actually it's not real clear what these tests are for +# since EXTEND has no defined semantics +$o->defer; +@a = (0..3); +check_contents(""); # nothing happened yet +expect($a[3], "3"); +expect($a[4], undef); +$o->flush; +check_contents("0$:1$:2$:3$:"); # file now 4 records long + +# (35-53) DELETE +if ($] >= 5.006) { + eval << 'TESTS'; +my $del; +$o->defer; +$del = delete $a[2]; +check_contents("0$:1$:2$:3$:"); # nothing happened yet +expect($a[2], ""); +expect($del, "2"); +$del = delete $a[3]; # shortens file! +check_contents("0$:1$:2$:"); # deferred writes NOT flushed +expect($a[3], undef); +expect($a[2], ""); +expect($del, "3"); +$a[2] = "cookies"; +$del = delete $a[2]; # shortens file! +expect($a[2], undef); +expect($del, 'cookies'); +check_contents("0$:1$:"); +$a[0] = "crackers"; +$del = delete $a[0]; # file unchanged +expect($a[0], ""); +expect($del, 'crackers'); +check_contents("0$:1$:"); # no change yet +$o->flush; +check_contents("$:1$:"); # record 0 is NOT 'cookies'; +TESTS +} else { + for (35..53) { + print "ok $_ \# skipped (no delete for arrays)\n"; + $N++; + } +} + +################################################################ + + +sub check_caches { + my ($xcache, $xdefer) = @_; + +# my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); +# print $integrity ? "ok $N\n" : "not ok $N\n"; +# $N++; + + my $good = 1; + $good &&= hash_equal($o->{cache}, $xcache, "true cache", "expected cache"); + $good &&= hash_equal($o->{deferred}, $xdefer, "true defer", "expected defer"); + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub hash_equal { + my ($a, $b, $ha, $hb) = @_; + $ha = 'first hash' unless defined $ha; + $hb = 'second hash' unless defined $hb; + + my $good = 1; + my %b_seen; + + for my $k (keys %$a) { + if (! exists $b->{$k}) { + print ctrlfix("# Key $k is in $ha but not $hb"), "\n"; + $good = 0; + } elsif ($b->{$k} ne $a->{$k}) { + print ctrlfix("# Key $k is <$a->{$k}> in $ha but <$b->{$k}> in $hb"), "\n"; + $b_seen{$k} = 1; + $good = 0; + } else { + $b_seen{$k} = 1; + } + } + + for my $k (keys %$b) { + unless ($b_seen{$k}) { + print ctrlfix("# Key $k is in $hb but not $ha"), "\n"; + $good = 0; + } + } + + $good; +} + + +sub check_contents { + my $x = shift; + + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + my $msg = ctrlfix("# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; +} + +sub expect { + if (@_ == 1) { + print $_[0] ? "ok $N\n" : "not ok $N\n"; + } elsif (@_ == 2) { + my ($a, $x) = @_; + if (! defined($a) && ! defined($x)) { print "ok $N\n" } + elsif ( defined($a) && ! defined($x)) { + ctrlfix(my $msg = "expected UNDEF, got <$a>"); + print "not ok $N \# $msg\n"; + } + elsif (! defined($a) && defined($x)) { + ctrlfix(my $msg = "expected <$x>, got UNDEF"); + print "not ok $N \# $msg\n"; + } elsif ($a eq $x) { print "ok $N\n" } + else { + ctrlfix(my $msg = "expected <$x>, got <$a>"); + print "not ok $N \# $msg\n"; + } + } else { + die "expect() got ", scalar(@_), " args, should have been 1 or 2"; + } + $N++; +} + +sub ctrlfix { + local $_ = shift; + s/\n/\\n/g; + s/\r/\\r/g; + $_; +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/33_defer_vs.t b/dist/Tie-File/t/33_defer_vs.t new file mode 100644 index 0000000000..071af77a68 --- /dev/null +++ b/dist/Tie-File/t/33_defer_vs.t @@ -0,0 +1,125 @@ +#!/usr/bin/perl +# +# Deferred caching of varying size records +# +# 30_defer.t always uses records that are 8 bytes long +# (9 on \r\n machines.) We might miss some sort of +# length-calculation bug as a result. This file will run some of the same +# tests, but with with varying-length records. +# + +use POSIX 'SEEK_SET'; +my $file = "tf$$.txt"; +# print "1..0\n"; exit; +$: = Tie::File::_default_recsep(); +my $data = "$:1$:22$:"; +my ($o, $n); + +print "1..30\n"; + +my $N = 1; +use Tie::File; +print "ok $N\n"; $N++; + +open F, "> $file" or die $!; +binmode F; +print F $data; +close F; +$o = tie @a, 'Tie::File', $file; +print $o ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (3-6) Deferred storage +$o->defer; +$a[3] = "333"; +check_contents($data); # nothing written yet +$a[4] = "4444"; +check_contents($data); # nothing written yet + +# (7-8) Flush +$o->flush; +check_contents($data . "333$:4444$:"); # now it's written + +# (9-12) Deferred writing disabled? +$a[3] = "999999999"; +check_contents("${data}999999999$:4444$:"); +$a[4] = "88888888"; +check_contents("${data}999999999$:88888888$:"); + +# (13-18) Now let's try two batches of records +$#a = 2; +$o->defer; +$a[0] = "55555"; +check_contents($data); # nothing written yet +$a[2] = "aaaaaaaaaa"; +check_contents($data); # nothing written yet +$o->flush; +check_contents("55555$:1$:aaaaaaaaaa$:"); + +# (19-22) Deferred writing past the end of the file +$o->defer; +$a[4] = "7777777"; +check_contents("55555$:1$:aaaaaaaaaa$:"); +$o->flush; +check_contents("55555$:1$:aaaaaaaaaa$:$:7777777$:"); + + +# (23-26) Now two long batches +$o->defer; +%l = qw(0 2 1 3 2 4 4 5 5 4 6 3); +for (0..2, 4..6) { + $a[$_] = $_ x $l{$_}; +} +check_contents("55555$:1$:aaaaaaaaaa$:$:7777777$:"); +$o->flush; +check_contents(join $:, "00", "111", "2222", "", "44444", "5555", "666", ""); + +# (27-30) Now let's make sure that discarded writes are really discarded +# We have a 2Mib buffer here, so we can be sure that we aren't accidentally +# filling it up +$o->defer; +for (0, 3, 7) { + $a[$_] = "discarded" . $_ x $_; +} +check_contents(join $:, "00", "111", "2222", "", "44444", "5555", "666", ""); +$o->discard; +check_contents(join $:, "00", "111", "2222", "", "44444", "5555", "666", ""); + +################################################################ + + +sub check_contents { + my $x = shift; + + my $integrity = $o->_check_integrity($file, $ENV{INTEGRITY}); + print $integrity ? "ok $N\n" : "not ok $N\n"; + $N++; + + local *FH = $o->{fh}; + seek FH, 0, SEEK_SET; + + my $a; + { local $/; $a = <FH> } + $a = "" unless defined $a; + if ($a eq $x) { + print "ok $N\n"; + } else { + my $msg = ctrlfix("# expected <$x>, got <$a>"); + print "not ok $N\n$msg\n"; + } + $N++; +} + +sub ctrlfix { + local $_ = shift; + s/\n/\\n/g; + s/\r/\\r/g; + $_; +} + +END { + undef $o; + untie @a; + 1 while unlink $file; +} + diff --git a/dist/Tie-File/t/40_abs_cache.t b/dist/Tie-File/t/40_abs_cache.t new file mode 100644 index 0000000000..137c9bb78d --- /dev/null +++ b/dist/Tie-File/t/40_abs_cache.t @@ -0,0 +1,281 @@ +#!/usr/bin/perl +# +# Unit tests for abstract cache implementation +# +# Test the following methods: +# * new() +# * is_empty() +# * empty() +# * lookup(key) +# * remove(key) +# * insert(key,val) +# * update(key,val) +# * rekey(okeys,nkeys) +# * expire() +# * keys() +# * bytes() +# DESTROY() +# +# 20020327 You somehow managed to miss: +# * reduce_size_to(bytes) +# + +# print "1..0\n"; exit; +print "1..42\n"; + +my ($N, @R, $Q, $ar) = (1); + +use Tie::File; +print "ok $N\n"; +$N++; + +my $h = Tie::File::Cache->new(10000) or die; +print "ok $N\n"; +$N++; + +# (3) Are all the methods there? +{ + my $good = 1; + for my $meth (qw(new is_empty empty lookup remove + insert update rekey expire ckeys bytes + set_limit adj_limit flush reduce_size_to + _produce _produce_lru )) { + unless ($h->can($meth)) { + print STDERR "# Method '$meth' is missing.\n"; + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +# (4-5) Straight insert and removal FIFO test +$ar = 'a0'; +for (1..10) { + $h->insert($_, $ar++); +} +1; +for (1..10) { + push @R, $h->expire; +} +$iota = iota('a',9); +print "@R" eq $iota + ? "ok $N\n" : "not ok $N \# expected ($iota), got (@R)\n"; +$N++; +check($h); + +# (6-7) Remove from empty heap +$n = $h->expire; +print ! defined $n ? "ok $N\n" : "not ok $N \# expected UNDEF, got $n"; +$N++; +check($h); + +# (8-9) Interleaved insert and removal +$Q = 0; +@R = (); +for my $i (1..4) { + for my $j (1..$i) { + $h->insert($Q, "b$Q"); + $Q++; + } + for my $j (1..$i) { + push @R, $h->expire; + } +} +$iota = iota('b', 9); +print "@R" eq $iota ? "ok $N\n" : "not ok $N \# expected ($iota), got (@R)\n"; +$N++; +check($h); + +# (10) It should be empty now +print $h->is_empty ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (11-12) Insert and delete +$Q = 1; +for (1..10) { + $h->insert($_, "c$Q"); + $Q++; +} +for (2, 4, 6, 8, 10) { + $h->remove($_); +} +@R = (); +push @R, $n while defined ($n = $h->expire); +print "@R" eq "c1 c3 c5 c7 c9" ? + "ok $N\n" : "not ok $N \# expected (c1 c3 c5 c7 c9), got (@R)\n"; +$N++; +check($h); + +# (13-14) Interleaved insert and delete +$Q = 1; my $QQ = 1; +@R = (); +for my $i (1..4) { + for my $j (1..$i) { + $h->insert($Q, "d$Q"); + $Q++; + } + for my $j (1..$i) { + $h->remove($QQ) if $QQ % 2 == 0; + $QQ++; + } +} +push @R, $n while defined ($n = $h->expire); +print "@R" eq "d1 d3 d5 d7 d9" ? + "ok $N\n" : "not ok $N \# expected (d1 d3 d5 d7 d9), got (@R)\n"; +$N++; +check($h); + +# (15-16) Promote +$h->empty; +$Q = 1; +for (1..10) { + $h->insert($_, "e$Q"); + unless ($h->_check_integrity) { + die "Integrity failed after inserting ($_, e$Q)\n"; + } + $Q++; +} +1; +for (2, 4, 6, 8, 10) { + $h->_promote($_); +} +@R = (); +push @R, $n while defined ($n = $h->expire); +print "@R" eq "e1 e3 e5 e7 e9 e2 e4 e6 e8 e10" ? + "ok $N\n" : + "not ok $N \# expected (e1 e3 e5 e7 e9 e2 e4 e6 e8 e10), got (@R)\n"; +$N++; +check($h); + +# (17-22) Lookup +$Q = 1; +for (1..10) { + $h->insert($_, "f$Q"); + $Q++; +} +1; +for (2, 4, 6, 4, 8) { + my $r = $h->lookup($_); + print $r eq "f$_" ? "ok $N\n" : "not ok $N \# expected f$_, got $r\n"; + $N++; +} +check($h); + +# (23) It shouldn't be empty +print ! $h->is_empty ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (24-25) Lookup should have promoted the looked-up records +@R = (); +push @R, $n while defined ($n = $h->expire); +print "@R" eq "f1 f3 f5 f7 f9 f10 f2 f6 f4 f8" ? + "ok $N\n" : + "not ok $N \# expected (f1 f3 f5 f7 f9 f10 f2 f6 f4 f8), got (@R)\n"; +$N++; +check($h); + +# (26-29) Typical 'rekey' operation +$Q = 1; +for (1..10) { + $h->insert($_, "g$Q"); + $Q++; +} +$h->rekey([6,7,8,9,10], [8,9,10,11,12]); +my %x = qw(1 g1 2 g2 3 g3 4 g4 5 g5 + 8 g6 9 g7 10 g8 11 g9 12 g10); +{ + my $good = 1; + for my $k (keys %x) { + my $v = $h->lookup($k); + $v = "UNDEF" unless defined $v; + unless ($v eq $x{$k}) { + print "# looked up $k, got $v, expected $x{$k}\n"; + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} +check($h); +{ + my $good = 1; + for my $k (6, 7) { + my $v = $h->lookup($k); + if (defined $v) { + print "# looked up $k, got $v, should have been undef\n"; + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} +check($h); + +# (30-31) ckeys +@R = sort { $a <=> $b } $h->ckeys; +print "@R" eq "1 2 3 4 5 8 9 10 11 12" ? + "ok $N\n" : + "not ok $N \# expected (1 2 3 4 5 8 9 10 11 12) got (@R)\n"; +$N++; +check($h); +1; +# (32-33) update +for (1..5, 8..12) { + $h->update($_, "h$_"); +} +@R = (); +for (sort { $a <=> $b } $h->ckeys) { + push @R, $h->lookup($_); +} +print "@R" eq "h1 h2 h3 h4 h5 h8 h9 h10 h11 h12" ? + "ok $N\n" : + "not ok $N \# expected (h1 h2 h3 h4 h5 h8 h9 h10 h11 h12) got (@R)\n"; +$N++; +check($h); + +# (34-37) bytes +my $B; +$B = $h->bytes; +print $B == 23 ? "ok $N\n" : "not ok $N \# expected 23, got $B\n"; +$N++; +check($h); +$h->update('12', "yobgorgle"); +$B = $h->bytes; +print $B == 29 ? "ok $N\n" : "not ok $N \# expected 29, got $B\n"; +$N++; +check($h); + +# (38-41) empty +$h->empty; +print $h->is_empty ? "ok $N\n" : "not ok $N\n"; +$N++; +check($h); +$n = $h->expire; +print ! defined $n ? "ok $N\n" : "not ok $N \# expected UNDEF, got $n"; +$N++; +check($h); + +# (42) very weak testing of DESTROY +undef $h; +# are we still alive? +print "ok $N\n"; +$N++; + +sub check { + my $h = shift; + print $h->_check_integrity ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +sub iota { + my ($p, $n) = @_; + my $r; + my $i = 0; + while ($i <= $n) { + $r .= "$p$i "; + $i++; + } + chop $r; + $r; +} diff --git a/dist/Tie-File/t/41_heap.t b/dist/Tie-File/t/41_heap.t new file mode 100644 index 0000000000..9e7ad2516c --- /dev/null +++ b/dist/Tie-File/t/41_heap.t @@ -0,0 +1,259 @@ +#!/usr/bin/perl +# +# Unit tests for heap implementation +# +# Test the following methods: +# new +# is_empty +# empty +# insert +# remove +# popheap +# promote +# lookup +# set_val +# rekey +# expire_order + + +# Finish these later. + +# They're nonurgent because the important heap stuff is extensively +# tested by tests 19, 20, 24, 30, 32, 33, and 40, as well as by pretty +# much everything else. +print "1..1\n"; + + +my ($N, @R, $Q, $ar) = (1); + +use Tie::File; +print "ok $N\n"; +$N++; +exit; + +__END__ + +my @HEAP_MOVE; +sub Fake::Cache::_heap_move { push @HEAP_MOVE, @_ } + +my $h = Tie::File::Heap->new(bless [] => 'Fake::Cache'); +print "ok $N\n"; +$N++; + +# (3) Are all the methods there? +{ + my $good = 1; + for my $meth (qw(new is_empty empty lookup insert remove popheap + promote set_val rekey expire_order)) { + unless ($h->can($meth)) { + print STDERR "# Method '$meth' is missing.\n"; + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +# (4) Straight insert and removal FIFO test +$ar = 'a0'; +for (1..10) { + $h->insert($_, $ar++); +} +for (1..10) { + push @R, $h->popheap; +} +$iota = iota('a',9); +print "@R" eq $iota + ? "ok $N\n" : "not ok $N \# expected ($iota), got (@R)\n"; +$N++; + +# (5) Remove from empty heap +$n = $h->popheap; +print ! defined $n ? "ok $N\n" : "not ok $N \# expected UNDEF, got $n"; +$N++; + +# (6) Interleaved insert and removal +$Q = 0; +@R = (); +for my $i (1..4) { + for my $j (1..$i) { + $h->insert($Q, "b$Q"); + $Q++; + } + for my $j (1..$i) { + push @R, $h->popheap; + } +} +$iota = iota('b', 9); +print "@R" eq $iota ? "ok $N\n" : "not ok $N \# expected ($iota), got (@R)\n"; +$N++; + +# (7) It should be empty now +print $h->is_empty ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (8) Insert and delete +$Q = 1; +for (1..10) { + $h->insert($_, "c$Q"); + $Q++; +} +for (2, 4, 6, 8, 10) { + $h->remove($_); +} +@R = (); +push @R, $n while defined ($n = $h->popheap); +print "@R" eq "c1 c3 c5 c7 c9" ? + "ok $N\n" : "not ok $N \# expected (c1 c3 c5 c7 c9), got (@R)\n"; +$N++; + +# (9) Interleaved insert and delete +$Q = 1; my $QQ = 1; +@R = (); +for my $i (1..4) { + for my $j (1..$i) { + $h->insert($Q, "d$Q"); + $Q++; + } + for my $j (1..$i) { + $h->remove($QQ) if $QQ % 2 == 0; + $QQ++; + } +} +push @R, $n while defined ($n = $h->popheap); +print "@R" eq "d1 d3 d5 d7 d9" ? + "ok $N\n" : "not ok $N \# expected (d1 d3 d5 d7 d9), got (@R)\n"; +$N++; + +# (10) Promote +$Q = 1; +for (1..10) { + $h->insert($_, "e$Q"); + $Q++; +} +for (2, 4, 6, 8, 10) { + $h->promote($_); +} +@R = (); +push @R, $n while defined ($n = $h->popheap); +print "@R" eq "e1 e3 e5 e7 e9 e2 e4 e6 e8 e10" ? + "ok $N\n" : + "not ok $N \# expected (e1 e3 e5 e7 e9 e2 e4 e6 e8 e10), got (@R)\n"; +$N++; + +# (11-15) Lookup +$Q = 1; +for (1..10) { + $h->insert($_, "f$Q"); + $Q++; +} +for (2, 4, 6, 4, 8) { + my $r = $h->lookup($_); + print $r eq "f$_" ? "ok $N\n" : "not ok $N \# expected f$_, got $r\n"; + $N++; +} + +# (16) It shouldn't be empty +print ! $h->is_empty ? "ok $N\n" : "not ok $N\n"; +$N++; + +# (17) Lookup should have promoted the looked-up records +@R = (); +push @R, $n while defined ($n = $h->popheap); +print "@R" eq "f1 f3 f5 f7 f9 f10 f2 f6 f4 f8" ? + "ok $N\n" : + "not ok $N \# expected (f1 f3 f5 f7 f9 f10 f2 f6 f4 f8), got (@R)\n"; +$N++; + +# (18-19) Typical 'rekey' operation +$Q = 1; +for (1..10) { + $h->insert($_, "g$Q"); + $Q++; +} + +$h->rekey([6,7,8,9,10], [8,9,10,11,12]); +my %x = qw(1 g1 2 g2 3 g3 4 g4 5 g5 + 8 g6 9 g7 10 g8 11 g9 12 g10); +{ + my $good = 1; + for my $k (keys %x) { + my $v = $h->lookup($k); + $v = "UNDEF" unless defined $v; + unless ($v eq $x{$k}) { + print "# looked up $k, got $v, expected $x{$k}\n"; + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} +{ + my $good = 1; + for my $k (6, 7) { + my $v = $h->lookup($k); + if (defined $v) { + print "# looked up $k, got $v, should have been undef\n"; + $good = 0; + } + } + print $good ? "ok $N\n" : "not ok $N\n"; + $N++; +} + +# (20) keys +@R = sort { $a <=> $b } $h->keys; +print "@R" eq "1 2 3 4 5 8 9 10 11 12" ? + "ok $N\n" : + "not ok $N \# expected (1 2 3 4 5 8 9 10 11 12) got (@R)\n"; +$N++; + +# (21) update +for (1..5, 8..12) { + $h->update($_, "h$_"); +} +@R = (); +for (sort { $a <=> $b } $h->keys) { + push @R, $h->lookup($_); +} +print "@R" eq "h1 h2 h3 h4 h5 h8 h9 h10 h11 h12" ? + "ok $N\n" : + "not ok $N \# expected (h1 h2 h3 h4 h5 h8 h9 h10 h11 h12) got (@R)\n"; +$N++; + +# (22-23) bytes +my $B; +$B = $h->bytes; +print $B == 23 ? "ok $N\n" : "not ok $N \# expected 23, got $B\n"; +$N++; +$h->update('12', "yobgorgle"); +$B = $h->bytes; +print $B == 29 ? "ok $N\n" : "not ok $N \# expected 29, got $B\n"; +$N++; + +# (24-25) empty +$h->empty; +print $h->is_empty ? "ok $N\n" : "not ok $N\n"; +$N++; +$n = $h->popheap; +print ! defined $n ? "ok $N\n" : "not ok $N \# expected UNDEF, got $n"; +$N++; + +# (26) very weak testing of DESTROY +undef $h; +# are we still alive? +print "ok $N\n"; +$N++; + + +sub iota { + my ($p, $n) = @_; + my $r; + my $i = 0; + while ($i <= $n) { + $r .= "$p$i "; + $i++; + } + chop $r; + $r; +} diff --git a/dist/Tie-File/t/42_offset.t b/dist/Tie-File/t/42_offset.t new file mode 100644 index 0000000000..8c628325d9 --- /dev/null +++ b/dist/Tie-File/t/42_offset.t @@ -0,0 +1,69 @@ +#!/usr/bin/perl + +# 2003-04-09 Tels: test the offset method from 0.94 + +use Test::More; +use strict; +use File::Spec; + +use POSIX 'SEEK_SET'; +my $file = "tf$$.txt"; + +BEGIN + { + $| = 1; + unshift @INC, File::Spec->catdir(File::Spec->updir, 'lib'); + chdir 't' if -d 't'; + print "# INC = @INC\n"; + + plan tests => 24; + + use_ok ('Tie::File'); + } + +$/ = "#"; # avoid problems with \n\r vs. \n + +my @a; +my $o = tie @a, 'Tie::File', $file, autodefer => 0; + +is (ref($o), 'Tie::File'); + +is ($o->offset(0), 0, 'first one always there'); +is ($o->offset(1), undef, 'no offsets yet'); + +$a[0] = 'Bourbon'; +is ($o->offset(0), 0, 'first is ok'); +is ($o->offset(1), 8, 'and second ok'); +is ($o->offset(2), undef, 'third undef'); + +$a[1] = 'makes'; +is ($o->offset(0), 0, 'first is ok'); +is ($o->offset(1), 8, 'and second ok'); +is ($o->offset(2), 14, 'and third ok'); +is ($o->offset(3), undef, 'fourth undef'); + +$a[2] = 'the baby'; +is ($o->offset(0), 0, 'first is ok'); +is ($o->offset(1), 8, 'and second ok'); +is ($o->offset(2), 14, 'and third ok'); +is ($o->offset(3), 23, 'and fourth ok'); +is ($o->offset(4), undef, 'fourth undef'); + +$a[3] = 'grin'; +is ($o->offset(0), 0, 'first is ok'); +is ($o->offset(1), 8, 'and second ok'); +is ($o->offset(2), 14, 'and third ok'); +is ($o->offset(3), 23, 'and fourth ok'); +is ($o->offset(4), 28, 'and fifth ok'); + +$a[4] = '!'; +is ($o->offset(5), 30, 'and fifth ok'); +$a[3] = 'water'; +is ($o->offset(4), 29, 'and fourth changed ok'); +is ($o->offset(5), 31, 'and fifth ok'); + +END { + undef $o; + untie @a; + 1 while unlink $file; +} |