path: root/Porting/sync-with-cpan

#!/usr/bin/env perl

=head1 NAME

Porting/sync-with-cpan - Synchronize with CPAN distributions

=head1 SYNOPSIS

    sh ./Configure
    perl Porting/sync-with-cpan <module>

where C<module> is the name it appears in the C<%Modules> hash
of F<Porting/Maintainers.pl>

=head1 DESCRIPTION

Script to help out with syncing cpan distros.

Does the following:

=over 4

=item *

Fetches the package list from CPAN. Finds the current version of the given
package. [1]

=item *

Downloads the relevant tarball; unpacks the tarball. [1]

=item *

Clean out the old directory (C<git clean -dfx>)

=item *

Moves the old directory out of the way, moves the new directory in place.

=item *

Restores any F<.gitignore> file.

=item *

Removes files from C<@IGNORE> and C<EXCLUDED>

=item *

C<git add> any new files.

=item *

C<git rm> any files that are gone.

=item *

Remove the +x bit on files in F<t/>

=item *

Remove the +x bit on files that don't have it enabled in the current dir

=item *

Restore files mentioned in C<CUSTOMIZED>

=item *

Updates the contents of F<MANIFEST>

=item *

Runs a C<make> (assumes a configure has been run)

=item *

Cleans up

=item *

Runs tests for the package

=item *

Runs the porting tests

=back

[1]  If the C<--tarball> option is given, then CPAN is not consulted.
C<--tarball> should be the path to the tarball; the version is extracted
from the filename -- but can be overwritten by the C<--version> option.

=head1 OPTIONS

=over 4

=item C<--jobs> I<N>

When running C<make>, pass a C<< -jI<N> >> option to it.

=back

=head1 TODO

=over 4

=item *

Update F<Porting/Maintainers.pl>

=item *

Optional, run a full test suite

=item *

Handle complicated C<FILES>

=back

This is an initial version; no attempt has been made yet to make this
portable. It shells out instead of trying to find a Perl solution.
In particular, it assumes git, perl, and make
to be available.

=cut


package Maintainers;

use 5.010;

use strict;
use warnings;
use Getopt::Long;
use Archive::Tar;
use File::Basename qw( basename );
use File::Path qw( remove_tree );
use File::Find;
use File::Spec::Functions qw( tmpdir rel2abs );
use Config qw( %Config );

$| = 1;

use constant WIN32 => $^O eq 'MSWin32';

die "This does not look like a top level directory"
     unless -d "cpan" && -d "Porting";

# Check that there's a Makefile, if needed; otherwise, we'll do most of our
# work only to fail when we try to run make, and the user will have to
# either unpick everything we've done, or do the rest manually.
die "Please run Configure before using $0\n"
    if !WIN32 && !-f "Makefile";

our @IGNORABLE;
our %Modules;

use autodie;

require "./Porting/Maintainers.pl";

my $MAKE_LOG = 'make.log';

my %IGNORABLE    = map {$_ => 1} @IGNORABLE;

my $tmpdir = tmpdir();

my $package      = "02packages.details.txt";
my $package_url  = "http://www.cpan.org/modules/$package";
my $package_file = "$tmpdir/$package"; # this is a cache

my @problematic = (
    # no current entries as of perl-5.37.9 (Dec 2022)
);


sub usage
{
    my $err = shift and select STDERR;
    print "Usage: $0 <module-or-dist> [args]\n";
    exit $err;
}

GetOptions ('tarball=s'  =>  \my $tarball,
            'version=s'  =>  \my $version,
            'jobs=i'     =>  \my $make_jobs,
             force       =>  \my $force,
             help        =>  sub { usage 0; },
             ) or  die "Failed to parse arguments";

usage 1 unless @ARGV == 1 || @ARGV == 2;

sub find_type_f {
    my @res;
    find( { no_chdir => 1, wanted => sub {
        my $file= $File::Find::name;
        return unless -f $file;
        push @res, $file
    }}, @_ );
    @res
};

# Equivalent of `chmod a-x`
sub de_exec {
    my ($filename) = @_;
    my $mode = (stat $filename)[2] & 0777;
    if ($mode & 0111) { # exec-bit set
        chmod $mode & 0666, $filename;
    }
}

# Equivalent of `chmod +w`
sub make_writable {
    my ($filename) = @_;
    my $mode = (stat $filename)[2] & 0777;
    if (!($mode & 0222)) { # not writable
        chmod $mode | (0222 & ~umask), $filename;
    }
}

sub make {
    my @args= @_;
    unshift @args, "-j$make_jobs" if defined $make_jobs;
    if (WIN32) {
        chdir "Win32";
        system "$Config{make} @args> ..\\$MAKE_LOG 2>&1"
            and die "Running make failed, see $MAKE_LOG";
        chdir '..';
    } else {
        system "$Config{make} @args> $MAKE_LOG 2>&1"
            and die "Running make failed, see $MAKE_LOG";
    };
};

my ($module)  = shift;

my $info = $Modules{$module};
if (!$info) {
    # Maybe the user said "Test-Simple" instead of "Test::Simple", or
    # "IO::Compress" instead of "IO-Compress". See if we can fix it up.
    my $guess = $module;
    s/-/::/g or s/::/-/g for $guess;
    $info = $Modules{$guess} or die <<"EOF";
Cannot find module $module.
The available options are listed in the %Modules hash in Porting/Maintainers.pl
EOF
    say "Guessing you meant $guess instead of $module";
    $module = $guess;
}

if ($info->{CUSTOMIZED}) {
    print <<"EOF";
$module has a CUSTOMIZED entry in Porting/Maintainers.pl.

This program's behaviour is to copy every CUSTOMIZED file into the version
of the module being imported. But that might not be the right thing: in some
cases, the new CPAN version will supersede whatever changes had previously
been made in blead, so it would be better to import the new CPAN files.

If you've checked that the CUSTOMIZED versions are still correct, you can
proceed now. Otherwise, you should abort and investigate the situation. If
the blead customizations are no longer needed, delete the CUSTOMIZED entry
for $module in Porting/Maintainers.pl (and you'll also need to regenerate
t/porting/customized.dat in that case; see t/porting/customized.t).

EOF
    print "Hit return to continue; ^C to abort "; <STDIN>;
}

my  $distribution = $$info {DISTRIBUTION};

my @files         = glob $$info {FILES};
if (!-d $files [0] || grep { $_ eq $module } @problematic) {
    say "This looks like a setup $0 cannot handle (yet)";
    unless ($force) {
        say "Will not continue without a --force option";
        exit 1;
    }
    say "--force is in effect, so we'll soldier on. Wish me luck!";
}

use Cwd 'cwd';
my $orig_pwd = cwd();

chdir "cpan";

my  $pkg_dir      = $files[0];
    $pkg_dir      =~ s!.*/!!;

my ($old_version) = $distribution =~ /-([0-9.]+(?:-TRIAL[0-9]*)?)\.tar\.gz/;

sub wget {
    my ($url, $saveas) = @_;
    my $ht_res;
    eval {
        require IO::Socket::SSL;
        require Net::SSLeay;
        require HTTP::Tiny;
        my $http = HTTP::Tiny->new();
        $ht_res  = $http->mirror( $url => $saveas );
        1;
    } or
       # Try harder to download the file
       # Some system do not have wget.  Fall back to curl if we do not
       # have it.  On Windows, `which wget` is not going to work, so
       # just use wget, as this script has always done.
       WIN32 || -x substr(`which wget`, 0, -1)
         ? system wget => $url, '-qO', $saveas
         : system curl => $url, '-sSo', $saveas;

    # We were able to use HTTP::Tiny and it didn't have fatal errors,
    # but we failed the request
    if ( $ht_res && ! $ht_res->{'success'} ) {
        die "Cannot retrieve file: $url\n" .
            sprintf "Status: %s\nReason: %s\nContent: %s\n",
            map $_ // '(unavailable)', @{$ht_res}{qw< status reason content >};
    }
}

#
# Find the information from CPAN.
#
my $new_file;
my $new_version;
if (defined $tarball) {
    $tarball = rel2abs( $tarball, $orig_pwd ) ;
    die "Tarball $tarball does not exist\n" if !-e $tarball;
    die "Tarball $tarball is not a plain file\n" if !-f _;
    $new_file     = $tarball;
    $new_version  = $version // ($new_file =~ /-([0-9._]+(?:-TRIAL[0-9]*)?)\.tar\.gz/) [0];
    die "Blead and that tarball both have version $new_version of $module\n"
        if $new_version eq $old_version;
}
else {
    #
    # Poor man's cache
    #
    unless (-f $package_file && -M $package_file < 1) {
        wget $package_url, $package_file;
    }

    my $cpan_mod = $info->{MAIN_MODULE} // $module;
    open my $fh, '<', $package_file;
    (my $new_line) = grep {/^\Q$cpan_mod\E /} <$fh> # Yes, this needs a lot of memory
                     or die "Cannot find $cpan_mod on CPAN\n";
    (undef, $new_version, my $new_path) = split ' ', $new_line;
    if (defined $version) {
        $new_path =~ s/-$new_version\./-$version\./;
        $new_version = $version;
    }
    $new_file = (split '/', $new_path) [-1];

    die "The latest version of $module is $new_version, but blead already has it\n"
        if $new_version eq $old_version;

    my $url = "https://cpan.metacpan.org/authors/id/$new_path";
    say "Fetching $url";
    #
    # Fetch the new distro
    #
    wget $url, $new_file;
}

my  $old_dir      = "$pkg_dir-$old_version";

say "Cleaning out old directory";
system git => 'clean', '-dfxq', $pkg_dir;

say "Unpacking $new_file";
Archive::Tar->extract_archive( $new_file );

(my $new_dir = basename($new_file)) =~ s/\.tar\.gz//;
# ensure 'make' will update all files
my $t= time;
for my $file (find_type_f($new_dir)) {
    make_writable($file); # for convenience if the user later edits it
    utime($t,$t,$file);
};

say "Renaming directories";
rename $pkg_dir => $old_dir;

say "Creating new package directory";
mkdir $pkg_dir;

say "Populating new package directory";
my $map = $$info {MAP};
my @EXCLUDED_QR;
my %EXCLUDED_QQ;
if ($$info {EXCLUDED}) {
    foreach my $entry (@{$$info {EXCLUDED}}) {
        if (ref $entry) {push @EXCLUDED_QR => $entry}
        else            {$EXCLUDED_QQ {$entry} = 1}
    }
}

FILE: for my $file ( find_type_f( $new_dir )) {
    my $old_file = $file;
    $file =~ s{^\Q$new_dir\E/}{};

    next if $EXCLUDED_QQ{$file};
    for my $qr (@EXCLUDED_QR) {
        next FILE if $file =~ $qr;
    }

    if ( $map ) {
        for my $key ( sort { length $b <=> length $a } keys %$map ) {
            my $val = $map->{$key};
            last if $file =~ s/^$key/$val/;
        }
    }
    else {
        $file = $files[0] . '/' . $file;
    }

    if ( $file =~ m{^cpan/} ) {
        $file =~ s{^cpan/}{};
    }
    else {
        $file = '../' . $file;
    }

    my $prefix = '';
    my @parts = split '/', $file;
    pop @parts;
    for my $part (@parts) {
        $prefix .= '/' if $prefix;
        $prefix .= $part;
        mkdir $prefix unless -d $prefix;
    }

    rename $old_file => $file;
}
remove_tree( $new_dir );

if (-f "$old_dir/.gitignore") {
    say "Restoring .gitignore";
    system git => 'checkout', "$pkg_dir/.gitignore";
}

my @new_files = find_type_f( $pkg_dir );
@new_files = grep {$_ ne $pkg_dir} @new_files;
s!^[^/]+/!! for @new_files;
my %new_files = map {$_ => 1} @new_files;

my @old_files = find_type_f( $old_dir );
@old_files = grep {$_ ne $old_dir} @old_files;
s!^[^/]+/!! for @old_files;
my %old_files = map {$_ => 1} @old_files;

my @delete;
my @commit;
my @gone;
FILE:
foreach my $file (@new_files) {
    next if -d "$pkg_dir/$file";   # Ignore directories.
    next if $old_files {$file};    # It's already there.
    if ($IGNORABLE {$file}) {
        push @delete => $file;
        next;
    }
    push @commit => $file;
}
foreach my $file (@old_files) {
    next if -d "$old_dir/$file";
    next if $new_files {$file};
    push @gone => $file;
}

#
# Find all files with an exec bit
#
my @exec = find_type_f( $pkg_dir );
my @de_exec;
foreach my $file (@exec) {
    # Remove leading dir
    $file =~ s!^[^/]+/!!;
    if ($file =~ m!^t/!) {
        push @de_exec => $file;
        next;
    }
    # Check to see if the file exists; if it doesn't and doesn't have
    # the exec bit, remove it.
    if ($old_files {$file}) {
        unless (-x "$old_dir/$file") {
            push @de_exec => $file;
        }
    }
}

#
# No need to change the +x bit on files that will be deleted.
#
if (@de_exec && @delete) {
    my %delete = map {+"$pkg_dir/$_" => 1} @delete;
    @de_exec = grep {!$delete {$_}} @de_exec;
}

#
# Mustn't change the +x bit on files that are whitelisted
#
if (@de_exec) {
    my %permitted = map { (my $x = $_) =~ tr/\n//d; $x => 1 } grep !/^#/,
        do { local @ARGV = '../Porting/exec-bit.txt'; <> };
    @de_exec = grep !$permitted{"cpan/$pkg_dir/$_"}, @de_exec;
}

say "unlink $pkg_dir/$_" for @delete;
say "git add $pkg_dir/$_" for @commit;
say "git rm -f $pkg_dir/$_" for @gone;
say "chmod a-x $pkg_dir/$_" for @de_exec;

print "Hit return to continue; ^C to abort "; <STDIN>;

unlink "$pkg_dir/$_"                      for @delete;
system git   => 'add', "$pkg_dir/$_"      for @commit;
system git   => 'rm', '-f', "$pkg_dir/$_" for @gone;
de_exec( "$pkg_dir/$_" )                  for @de_exec;

#
# Restore anything that is customized.
# We don't really care whether we've deleted the file - since we
# do a git restore, it's going to be resurrected if necessary.
#
if ($$info {CUSTOMIZED}) {
    say "Restoring customized files";
    foreach my $file (@{$$info {CUSTOMIZED}}) {
        system git => "checkout", "$pkg_dir/$file";
    }
}

chdir "..";
if (@commit || @gone) {
    say "Fixing MANIFEST";
    my $MANIFEST     = "MANIFEST";
    my $MANIFEST_NEW = "$MANIFEST.new";

    open my $orig, "<", $MANIFEST
        or die "Failed to open $MANIFEST for reading: $!\n";
    open my $new, ">", $MANIFEST_NEW
        or die "Failed to open $MANIFEST_NEW for writing: $!\n";
    my %gone = map +("cpan/$pkg_dir/$_" => 1), @gone;
    while (my $line = <$orig>) {
        my ($file) = $line =~ /^(\S+)/
            or die "Can't parse MANIFEST line: $line";
        print $new $line if !$gone{$file};
    }

    say $new "cpan/$pkg_dir/$_" for @commit;

    close $new or die "Can't close $MANIFEST: $!\n";

    system $^X => "Porting/manisort", '--quiet', "--output=$MANIFEST", $MANIFEST_NEW;
    unlink $MANIFEST_NEW
        or die "Can't delete temporary $MANIFEST_NEW: $!\n";
}


print "Running a make and saving its output to $MAKE_LOG ... ";
# Prepare for running (selected) tests
make 'test-prep';
print "done\n";

# The build system installs code from CPAN dists into the lib/ directory,
# creating directories as needed. This means that the cleaning-related rules
# in the Makefile need to know which directories to clean up. The Makefile
# is generated by Configure from Makefile.SH, so *that* file needs the list
# of directories. regen/lib_cleanup.pl is capable of automatically updating
# the contents of Makefile.SH (and win32/Makefile, which needs similar but
# not identical lists of directories), so we can just run that (using the
# newly-built Perl, as is done with the regen programs run by "make regen").
#
# We do this if any files at all have been added or deleted, regardless of
# whether those changes result in any directories being added or deleted,
# because the alternative would be to replicate the regen/lib_cleanup.pl
# logic here. That's fine, because regen/lib_cleanup.pl is idempotent if run
# repeatedly.
if (@commit || @gone) {
    say "Running regen/lib_cleanup.pl to handle potential added/deleted dirs";
    my $exe_dir = WIN32 ? ".\\" : './';
    system "${exe_dir}perl$Config{_exe}", "-Ilib", "regen/lib_cleanup.pl"
        and die "regen/lib_cleanup.pl failed\n";
}

#
# Must clean up, or else t/porting/FindExt.t will fail.
# Note that we can always retrieve the original directory with a git checkout.
#
print "About to clean up; hit return or abort (^C) "; <STDIN>;

remove_tree( "cpan/$old_dir" );
unlink "cpan/$new_file" unless $tarball;

#
# Run the tests. First the test belonging to the module, followed by the
# tests in t/porting
#
chdir "t";
say "Running module tests";
my @test_files = grep { /\.t$/ } find_type_f( "../cpan/$pkg_dir" );
my $exe_dir = WIN32 ? "..\\" : './';
my $output = `${exe_dir}perl$Config{_exe} TEST @test_files`;
unless ($output =~ /All tests successful/) {
    say $output;
    exit 1;
}

print "Running tests in t/porting ";
my @tests = glob 'porting/*.t';
chomp @tests;
my @failed;
foreach my $t (@tests) {
    my @not = grep {!/# TODO/ }
              grep { /^not/ }
              `${exe_dir}perl -I../lib -I.. $t`;
    print @not ? '!' : '.';
    push @failed => $t if @not;
}
print "\n";
say "Failed tests: @failed" if @failed;


chdir '..';

open my $Maintainers_pl, '<', 'Porting/Maintainers.pl';
open my $new_Maintainers_pl, '>', 'Maintainers.pl';

my $found;
my $in_mod_section;
while (<$Maintainers_pl>) {
    if (!$found) {
        if ($in_mod_section) {
            if (/DISTRIBUTION/) {
                if (s/\Q$old_version/$new_version/) {
                    $found = 1;
                }
            }

            if (/^    \}/) {
                $in_mod_section = 0;
            }
        }

        if (/\Q$module/) {
            $in_mod_section = 1;
        }
    }

    print $new_Maintainers_pl $_;
}

if ($found) {
    say "Successfully updated Maintainers.pl";
    unlink 'Porting/Maintainers.pl';
    rename 'Maintainers.pl' => 'Porting/Maintainers.pl';
    chmod 0755 => 'Porting/Maintainers.pl';
}
else {
    say "Could not update Porting/Maintainers.pl.";
    say "Make sure you update this by hand before committing.";
}

print <<"EOF";

=======================================================================

$module is now at version $new_version
Next, you should run "make minitest" and then "make test".

Minitest uses miniperl, which does not support XS modules. The full test
suite uses perl, which does. Minitest can fail - e.g. if a cpan module
has added an XS dependency - even if the full test suite passes just fine.

Hopefully all will complete successfully, but if not, you can make any
changes you need to get the tests to pass. Don't forget that you'll need
a "CUSTOMIZED" entry in Porting/Maintainers.pl if you change any of the
files under cpan/$pkg_dir.

Once all tests pass, you can "git add -u" and "git commit" the changes
with a message along the lines of "Update Foo::Bar to v1.234".

EOF

__END__