diff options
Diffstat (limited to 'lib/unicore')
| -rw-r--r-- | lib/unicore/mktables | 82 | ||||
| -rw-r--r-- | lib/unicore/uni_keywords.pl | 2 |
2 files changed, 50 insertions, 34 deletions
diff --git a/lib/unicore/mktables b/lib/unicore/mktables index dc0d7db10b..f11e466823 100644 --- a/lib/unicore/mktables +++ b/lib/unicore/mktables @@ -13989,6 +13989,51 @@ END ); } } # End of loop through aliases for this property + + + # Properties that have sets of values for some characters are now + # converted. For example, the Script_Extensions property started out + # as a clone of the Script property. But processing its data file + # caused some elements to be replaced with different data. (These + # elements were for the Common and Inherited properties.) This data + # is a qw() list of all the scripts that the code points in the given + # range are in. An example line is: + # + # 060C ; Arab Syrc Thaa # Po ARABIC COMMA + # + # Code executed earlier has created a new match table named "Arab Syrc + # Thaa" which contains 060C. (The cloned table started out with this + # code point mapping to "Common".) Now we add 060C to each of the + # Arab, Syrc, and Thaa match tables. Then we delete the now spurious + # "Arab Syrc Thaa" match table. This is repeated for all these tables + # and ranges. The map data is retained in the map table for + # reference, but the spurious match tables are deleted. + my $format = $property->format; + if (defined $format && $format eq $STRING_WHITE_SPACE_LIST) { + foreach my $table ($property->tables) { + + # Space separates the entries which should go in multiple + # tables + next unless $table->name =~ /\s/; + + # The list of the entries, hence the names of the tables that + # everything in this combo table should be added to. + my @list = split /\s+/, $table->name; + + # Add the entries from the combo table to each individual + # table + foreach my $individual (@list) { + my $existing_table = $property->table($individual); + + # This should only be necessary if this particular entry + # occurs only in combo with others. + $existing_table = $property->add_match_table($individual) + unless defined $existing_table; + $existing_table += $table; + } + $property->delete_match_table($table); + } + } } # End of loop through all Unicode properties. # Fill in the mappings that Unicode doesn't completely furnish. First the @@ -14077,40 +14122,11 @@ END Initialize => property_ref('cf'), ); - # The Script_Extensions property started out as a clone of the Script - # property. But processing its data file caused some elements to be - # replaced with different data. (These elements were for the Common and - # Inherited properties.) This data is a qw() list of all the scripts that - # the code points in the given range are in. An example line is: - # 060C ; Arab Syrc Thaa # Po ARABIC COMMA - # - # The code above has created a new match table named "Arab Syrc Thaa" - # which contains 060C. (The cloned table started out with this code point - # mapping to "Common".) Now we add 060C to each of the Arab, Syrc, and - # Thaa match tables. Then we delete the now spurious "Arab Syrc Thaa" - # match table. This is repeated for all these tables and ranges. The map - # data is retained in the map table for reference, but the spurious match - # tables are deleted. - + # Mark the scx table as the parent of the corresponding sc table for those + # which are identical. This causes the pod for the script table to refer + # to the corresponding scx one. This is done after everything, so as to + # wait until the tables are stabilized before checking for equivalency. if (defined $scx) { - foreach my $table ($scx->tables) { - next unless $table->name =~ /\s/; # All the new and only the new - # tables have a space in their - # names - my @scripts = split /\s+/, $table->name; - foreach my $script (@scripts) { - my $script_table = $scx->table($script); - $script_table += $table; - } - $scx->delete_match_table($table); - } - - # Mark the scx table as the parent of the corresponding sc table for - # those which are identical. This causes the pod for the script table - # to refer to the corresponding scx one. - # - # This has to be in a separate loop from above, so as to wait until - # the tables are stabilized before checking for equivalency. if (defined $pod_directory) { foreach my $table ($scx->tables) { my $plain_sc_equiv = $script->table($table->name); diff --git a/lib/unicore/uni_keywords.pl b/lib/unicore/uni_keywords.pl index 63c3dd9b11..894ede5df4 100644 --- a/lib/unicore/uni_keywords.pl +++ b/lib/unicore/uni_keywords.pl @@ -1280,7 +1280,7 @@ # 0fea35394151afefbb4121b6380db1b480be6f9bafb4eba3382dc292dcf68526 lib/unicore/extracted/DLineBreak.txt # 6d4a8c945dd7db83ed617cbb7d937de7f4ecf016ff22970d846e996a7c9a2a5d lib/unicore/extracted/DNumType.txt # 5b7c14380d5cceeaffcfbc18db1ed936391d2af2d51f5a41f1a17b692c77e59b lib/unicore/extracted/DNumValues.txt -# e3cbd908e180d39554e2c7351cebfd939a567c796c79e2e4a71ec2af874adfa3 lib/unicore/mktables +# eefd34f5f03191cb297d62f66273ca68efc972703f5323543427a9b285f28657 lib/unicore/mktables # 50b85a67451145545a65cea370dab8d3444fbfe07e9c34cef560c5b7da9d3eef lib/unicore/version # 2680b9254eb236c5c090f11b149605043e8c8433661b96efc4a42fb4709342a5 regen/charset_translations.pl # 6bbad21de0848e0236b02f34f5fa0edd3cdae9ba8173cc9469a5513936b9e728 regen/mk_PL_charclass.pl |