digraphs: mkdigraphlist: redetermine source class

[sheet.git] / tools / mkdigraphlist

diff --git a/tools/mkdigraphlist b/tools/mkdigraphlist
index 0714a2b1b64e15ff37e31a7faca36a6088b5d749..e472ce0e0d01a41b5e0f27ef5f16f927c51c4d78 100755 (executable)
--- a/tools/mkdigraphlist
+++ b/tools/mkdigraphlist
@@ -6,30 +6,21 @@ use utf8;
 
 use open OUT => ':utf8', ':std';
 
-our $VERSION = '1.03';
+our $VERSION = '1.04';
 
-# create a hash of desired input
-my $di = do 'data/digraphs-rfc.inc.pl'
+# import and combine various digraph data
+my $rfc = do 'data/digraphs-rfc.inc.pl'
        or die "error reading digraphs include: ", $@ // $!;
-
-# personal addendums
 my $extra = do 'data/digraphs-shiar.inc.pl'
        or warn "could not include shiar proposals: ", $@ // $!;
 my $vim = do 'data/digraphs-vim.inc.pl'
        or warn "could not include vim extensions ", $@ // $!;
-$di = { %{$vim // {}}, %{$di}, %{$extra // {}} };
+my $di = { %{$vim // {}}, %{$rfc}, %{$extra // {}} };
 
 # optionally get unicode character information
 my $uninfo = do 'unicode-char.inc.pl'
        or warn "could not include unicode details: ", $@ // $!;
 
-# convert info hashes into arrays of strings to output in display order
-for my $row (values %{$uninfo}) {
-       my ($class, $name, $di, $html, $string) = @{$row};
-       $row = [$name, $class];
-       push @{$row}, '', $string if defined $string;
-}
-
 # output perl code of hash
 # (assume no backslashes or curlies, so we can just q{} w/o escaping)
 print "# automatically generated by $0\n";
@@ -41,7 +32,13 @@ printf '(map {$_=>0} qw{%s}),'."\n", join(' ',
 );
 printf "q{%s}=>[%s],\n", $_, join(',',
        $di->{$_},   # original code point
-       (map {"'$_'"} @{ $uninfo->{ chr $di->{$_} } // [] }),  # optional additional arguments
+       map {"'$_'"}
+       $uninfo->{ chr $di->{$_} }->[1],  # name
+       $vim->{$_}
+               ? $rfc->{$_} ? 'u-di' : 'u-prop'  # vim+rfc or vim-only
+               :              'u-prop ex',  # neither
+       $uninfo->{ chr $di->{$_} }->[0] =~ s/ u-di| u-prop| ex//gr,  # class
+       $uninfo->{ chr $di->{$_} }->[4] // (),  # string
 ) for sort keys %{$di};
 print "}\n";
 
@@ -65,11 +62,12 @@ The value can either be a scalar string containing another
 digraph which can be considered identical (usually inverted),
 or an array ref containing at least the resulting character's
 Unicode code point value.  If available, the following UCD data
-is appended: character name, category, script, and output string.
+is appended: character name, usage classes, unicode classes,
+and replacement output string.
 For example:
 
  +{
-   AE => [198, 'LATIN CAPITAL LETTER AE', 'Lu Xl', 'Latin'],
+   AE => [198, 'LATIN CAPITAL LETTER AE', 'u-di', 'Latin Lu Xl u-v11'],
    EA => 'AE',
   }