Resync with mainline prior to post-5.6.0 updates

[p5sagit/p5-mst-13.2.git] / lib / unicode / mktables.PL

diff --git a/lib/unicode/mktables.PL b/lib/unicode/mktables.PL
index cef6936..4f705a4 100755 (executable)
--- a/lib/unicode/mktables.PL
+++ b/lib/unicode/mktables.PL
@@ -11,9 +11,10 @@ mkdir "To", 0777;
 @todo = (
 # typical
 
-    ['IsWord',  '$cat =~ /^L[ulo]|^Nd/ or $code eq "005F"',    ''],
-    ['IsAlnum', '$cat =~ /^L[ulo]|^Nd/',       ''],
-    ['IsAlpha',  '$cat =~ /^L[ulo]/',  ''],
+    ['IsWord',  '$cat =~ /^L[ulot]|^Nd/ or $code eq "005F"',   ''],
+    ['IsAlnum', '$cat =~ /^L[ulot]|^Nd/',      ''],
+    ['IsAlpha',  '$cat =~ /^L[ulot]/', ''],
+    # XXX broken: recursive definition (/\s/ will look up IsSpace in future)
     ['IsSpace',  '$cat =~ /^Z/ or $code lt "0020" and chr(hex $code) =~ /^\s/',        ''],
     ['IsDigit',  '$cat =~ /^Nd$/',     ''],
     ['IsUpper',  '$cat =~ /^Lu$/',     ''],
@@ -166,6 +167,40 @@ mkdir "To", 0777;
     ['IsSylWA',        '$syl eq "A"',          ''],
     ['IsSylWE',        '$syl eq "E"',          ''],
     ['IsSylWC',        '$syl eq "C"',          ''],
+    
+# Line break properties - Normative
+
+    ['IsLbrkBK','$brk eq "BK"',                ''],    # Mandatory Break
+    ['IsLbrkCR','$brk eq "CR"',                ''],    # Carriage Return
+    ['IsLbrkLF','$brk eq "LF"',                ''],    # Line Feed
+    ['IsLbrkCM','$brk eq "CM"',                ''],    # Attached Characters and Combining Marks
+    ['IsLbrkSG','$brk eq "SG"',                ''],    # Surrogates
+    ['IsLbrkGL','$brk eq "GL"',                ''],    # Non-breaking (Glue)
+    ['IsLbrkCB','$brk eq "CB"',                ''],    # Contingent Break Opportunity
+    ['IsLbrkSP','$brk eq "SP"',                ''],    # Space
+    ['IsLbrkZW','$brk eq "ZW"',                ''],    # Zero Width Space
+
+# Line break properties - Informative
+    ['IsLbrkXX','$brk eq "XX"',                ''],    # Unknown
+    ['IsLbrkOP','$brk eq "OP"',                ''],    # Opening Punctuation
+    ['IsLbrkCL','$brk eq "CL"',                ''],    # Closing Punctuation
+    ['IsLbrkQU','$brk eq "QU"',                ''],    # Ambiguous Quotation
+    ['IsLbrkNS','$brk eq "NS"',                ''],    # Non Starter
+    ['IsLbrkEX','$brk eq "EX"',                ''],    # Exclamation/Interrogation
+    ['IsLbrkSY','$brk eq "SY"',                ''],    # Symbols Allowing Breaks
+    ['IsLbrkIS','$brk eq "IS"',                ''],    # Infix Separator (Numeric)
+    ['IsLbrkPR','$brk eq "PR"',                ''],    # Prefix (Numeric)
+    ['IsLbrkPO','$brk eq "PO"',                ''],    # Postfix (Numeric)
+    ['IsLbrkNU','$brk eq "NU"',                ''],    # Numeric
+    ['IsLbrkAL','$brk eq "AL"',                ''],    # Ordinary Alphabetic and Symbol Characters
+    ['IsLbrkID','$brk eq "ID"',                ''],    # Ideographic
+    ['IsLbrkIN','$brk eq "IN"',                ''],    # Inseparable
+    ['IsLbrkHY','$brk eq "HY"',                ''],    # Hyphen
+    ['IsLbrkBB','$brk eq "BB"',                ''],    # Break Opportunity Before
+    ['IsLbrkBA','$brk eq "BA"',                ''],    # Break Opportunity After
+    ['IsLbrkSA','$brk eq "SA"',                ''],    # Complex Context (South East Asian)
+    ['IsLbrkAI','$brk eq "AI"',                ''],    # Ambiguous (Alphabetic or Ideographic)
+    ['IsLbrkB2','$brk eq "B2"',                ''],    # Break Opportunity Before and After
 );
 
 # This is not written for speed...
@@ -257,6 +292,11 @@ sub proplist {
 
        $split = '($code, $short, $syl) = split(/; */); $code =~ s/^U\+//;';
     }
+    elsif ($table =~ /^IsLbrk/) {
+       open(UD, "LineBrk.txt") or warn "Can't open $table: $!";
+
+       $split = '($code, $brk, $name) = split(/;/);';
+    }
     else {
        open(UD, $UnicodeData) or warn "Can't open $UnicodeData: $!";