update finkinfo.
[chise/omega.git] / outCMAP
diff --git a/outCMAP b/outCMAP
index 28c4606..16880b9 100755 (executable)
--- a/outCMAP
+++ b/outCMAP
@@ -3,6 +3,7 @@
 use strict;
 use vars qw($opt_in_cs $opt_out_cs $opt_help $usage
            $in_cs $out_cs
+           $i @chars
            $char $char_id $out_char $omegadb_home
            $ids $ids_argc %ids $idsdb
            $idsdata_file $ids_start $font_start
@@ -16,6 +17,31 @@ use Getopt::Long;
 use utf8;
 use Chise_utils ':all';
 
+my $strictly_forbidden_after = '「【『[(〈“‘‘(〔{《{\[\(\x{3016}{「';
+#       \x{3016} | # white 【
+
+my $forbidden_after = "\x{0000}";
+
+# ¥¥$$〒♯##¢¢££@@§
+my $slightly_forbidden_after = '¥¥$$〒♯##¢¢££@@§';
+
+# $strictly_forbidden_before
+# All these characters are allowed to protrude
+# in the right margin
+my $strictly_forbidden_before=
+    '!,.:;?、。!,.:;?。\)#}’”〉》」』】〕\x{3017})]}}」\]';
+#       \x{3017} | # white 】
+
+my $forbidden_before
+    = 'ー々ぁぃぅぇぉゃゅょっゎァィゥェォャュョッヮヵヶ';
+
+my $slightly_forbidden_before
+    = '\x{000a}\#\-‐−‰′″℃゛゜ゝゞヽヾ"%-゙゚';
+
+my $asian = '\x{1100}-\x{11FF}\x{2E80}-\x{D7AF}\x{F900}-\x{FAFF}\x{FE30}-\x{FE4F}\x{FF00}-';
+
+my $space = '\x{0020}\x{0009}\x{000A}\x{000C}\x{000D}';
+
 $useGT=1;
 $useHZK=0;
 $useCDP=0;
@@ -33,7 +59,7 @@ if($perl58){
     binmode(STDOUT, ':encoding(utf8)');
 }
 
-$omegadb_home="$HOME/.chise";
+$omegadb_home="/usr/local/lib/chise/omega/db";
 
 &GetOptions("in=s"=>\$opt_in_cs,
            "i=s"=>\$opt_in_cs,
@@ -98,18 +124,22 @@ while(<>){
                               and $perl58);
     s/(amp.+?;)/&tex_de_er($1)/ge;
 #    s/(&.+?;)/&tex_de_er($1)/ge;
-    while(m/(.)/g){
-       $char=&get_char_in_utf8mcs($1,$in_cs);
+#    s/^(.*)$/&add_break($1)/e;
+    @chars=split(//);
+    for($i=0;$i<=$#chars;$i++){
+#    while(m/(.)/g){
+       $char=&get_char_in_utf8mcs($chars[$i],$in_cs);
        $char_id=unpack("U",$char);
        if($ids_argc>0){
+           # It's in IDS.
            ($ids,$ids_argc)=&ids_rest($ids,$ids_argc,$char);
            if($ids_argc==0){
                if(($char_id=&get_char_id_for_ids($ids))
                   and(($out_char=&get_output_char($char_id,$out_cs)))){
-                   print $out_char;
+                   print $out_char,&add_break($i);
                }else{
-                   print &replace_ids($ids) if($perl56);
-                   print encode('utf8', &replace_ids($ids)) if($perl58);
+                   print &replace_ids($ids),&add_break($i) if($perl56);
+                   print encode('utf8', &replace_ids($ids)),&add_break($i) if($perl58);
                }
                $ids="";
            }
@@ -122,7 +152,7 @@ while(<>){
                next;
            }
            if(($out_char=&get_output_char($char_id,$out_cs))){
-               print $out_char;
+               print $out_char,&add_break($i);
            }elsif($char_id >= 0x20000 && $char_id <=0x2a6df){
                unless(defined($ids{$char}) and $ids{$char}[1]>=0){
                    $ids{$char}[0]=$font_start;
@@ -135,10 +165,10 @@ while(<>){
                }
                print "{\\fontencoding{OT1}\\fontfamily{" .
                    sprintf("chise%03d",$ids{$char}[0]) .
-                   "}\\selectfont\\char$ids{$char}[1]}";
+                   "}\\selectfont\\char$ids{$char}[1]}",&add_break($i);
                next;
            }else{
-               print &replace_ids(&get_ids($char));
+               print &replace_ids(&get_ids($char)),&add_break($i);
            }
        }
     }
@@ -177,6 +207,48 @@ sub tex_de_er{
     }
 }
 
+sub add_break{
+    my($i)=@_;
+
+    if($i<($#chars-1)){
+       if(($chars[$i+1]=~m/[$strictly_forbidden_before]/o)
+          and($chars[$i+2]=~m/[$strictly_forbidden_before]/o)){
+           return "\\CJKunbreakablekernone ";
+       }elsif(($chars[$i+1]=~m/[$strictly_forbidden_before]/o)
+              and($chars[$i+2]=~m/[^$strictly_forbidden_before]/o)){
+           return "\\CJKunbreakablekernone \\CJKprotrude ";
+       }
+    }
+    if(($i<$#chars)
+       and($chars[$i+1]=~m/[$strictly_forbidden_before]/o)){
+       return "\\CJKunbreakablekernone ";
+    }
+    if($chars[$i]=~m/[$strictly_forbidden_after]/o){
+       return "\\CJKunbreakablekernone ";
+    }
+    if(($i<$#chars)
+       and($chars[$i+1]=~m/[$forbidden_before]/o)){
+       return "\\CJKunbreakablekerntwo ";
+
+    }
+    if($chars[$i]=~m/[$forbidden_after]/o){
+       return "\\CJKunbreakablekerntwo ";
+    }
+    if(($i<$#chars)
+       and($chars[$i+1]=~m/[$slightly_forbidden_before]/o)){
+       return "\\CJKunbreakablekernthree ";
+    }
+    if($chars[$i]=~m/[$slightly_forbidden_after]/o){
+       return "\\CJKunbreakablekernthree ";
+    }
+    if($chars[$i]=~m/[$asian]/o){
+       return "\\CJKbreakablekern ";
+    }
+    if(($i<$#chars)and($chars[$i+1]=~m/[$asian]/o)){
+       return "\\CJKbreakablekern ";
+    }
+}
+
 sub ids_rest{
     my($ids,$ids_argc,$char)=@_;
     my($argc);
@@ -337,6 +409,7 @@ sub get_macro_for_GT{
     }
     if($gt){
        return "{\\fontencoding{OT1}\\fontfamily{".sprintf("gt%02d",$GT)."}\\selectfont\\char".($gt|0x8080)."}";
+#      return "\\GT{".sprintf("gt%02d",$GT)."}{\\char".($gt|0x8080)."}";
     }else{
        return undef;
     }