Commit | Line | Data |
a0ed51b3 |
1 | #!../../miniperl |
2 | |
7c6f5cd2 |
3 | use bytes; |
4 | |
190eec7c |
5 | $UnicodeData = "Unicode.txt"; |
7c6f5cd2 |
6 | $SyllableData = "syllables.txt"; |
8836d2a5 |
7 | $PropData = "PropList.txt"; |
7c6f5cd2 |
8 | |
11695a73 |
9 | |
a0ed51b3 |
10 | # Note: we try to keep filenames unique within first 8 chars. Using |
11 | # subdirectories for the following helps. |
289d4f09 |
12 | mkdir "In", 0755; |
13 | mkdir "Is", 0755; |
14 | mkdir "To", 0755; |
a0ed51b3 |
15 | |
16 | @todo = ( |
17 | # typical |
18 | |
f59877d4 |
19 | # 005F: SPACING UNDERSCROE |
20 | ['IsWord', '$cat =~ /^[LMN]/ or $code eq "005F"', ''], |
21 | ['IsAlnum', '$cat =~ /^[LMN]/', ''], |
22 | ['IsAlpha', '$cat =~ /^[LM]/', ''], |
23 | # 0009: HORIZONTAL TABULATION |
24 | # 000A: LINE FEED |
25 | # 000B: VERTICAL TABULATION |
26 | # 000C: FORM FEED |
27 | # 000D: CARRIAGE RETURN |
3bec3564 |
28 | # 0020: SPACE |
f59877d4 |
29 | ['IsSpace', '$cat =~ /^Z/ || |
30 | $code =~ /^(0009|000A|000B|000C|000D)$/', ''], |
3bec3564 |
31 | ['IsSpacePerl', |
32 | '$cat =~ /^Z/ || |
33 | $code =~ /^(0009|000A|000C|000D)$/', ''], |
f5868911 |
34 | ['IsBlank', '$code =~ /^(0020|0009)$/ || |
35 | $cat =~ /^Z[^lp]$/', ''], |
b8c5462f |
36 | ['IsDigit', '$cat =~ /^Nd$/', ''], |
7c6f5cd2 |
37 | ['IsUpper', '$cat =~ /^L[ut]$/', ''], |
b8c5462f |
38 | ['IsLower', '$cat =~ /^Ll$/', ''], |
f59877d4 |
39 | ['IsASCII', '$code le "007f"', ''], |
b8c5462f |
40 | ['IsCntrl', '$cat =~ /^C/', ''], |
f59877d4 |
41 | ['IsGraph', '$cat =~ /^([LMNPS]|Co)/', ''], |
42 | ['IsPrint', '$cat =~ /^([LMNPS]|Co|Zs)/', ''], |
43 | ['IsPunct', '$cat =~ /^P/', ''], |
44 | # 003[0-9]: DIGIT ZERO..NINE, 00[46][1-6]: A..F, a..f |
b8c5462f |
45 | ['IsXDigit', '$code =~ /^00(3[0-9]|[46][1-6])$/', ''], |
46 | ['ToUpper', '$up', '$up'], |
47 | ['ToLower', '$down', '$down'], |
48 | ['ToTitle', '$title', '$title'], |
49 | ['ToDigit', '$dec ne ""', '$dec'], |
a0ed51b3 |
50 | |
51 | # Name |
52 | |
53 | ['Name', '$name', '$name'], |
54 | |
55 | # Category |
56 | |
57 | ['Category', '$cat', '$cat'], |
58 | |
59 | # Normative |
60 | |
61 | ['IsM', '$cat =~ /^M/', ''], # Mark |
62 | ['IsMn', '$cat eq "Mn"', ''], # Mark, Non-Spacing |
63 | ['IsMc', '$cat eq "Mc"', ''], # Mark, Combining |
7c6f5cd2 |
64 | ['IsMe', '$cat eq "Me"', ''], # Mark, Enclosing |
a0ed51b3 |
65 | |
66 | ['IsN', '$cat =~ /^N/', ''], # Number |
67 | ['IsNd', '$cat eq "Nd"', ''], # Number, Decimal Digit |
68 | ['IsNo', '$cat eq "No"', ''], # Number, Other |
7c6f5cd2 |
69 | ['IsNl', '$cat eq "Nl"', ''], # Number, Letter |
a0ed51b3 |
70 | |
7c6f5cd2 |
71 | ['IsZ', '$cat =~ /^Z/', ''], # Separator |
a0ed51b3 |
72 | ['IsZs', '$cat eq "Zs"', ''], # Separator, Space |
73 | ['IsZl', '$cat eq "Zl"', ''], # Separator, Line |
74 | ['IsZp', '$cat eq "Zp"', ''], # Separator, Paragraph |
75 | |
76 | ['IsC', '$cat =~ /^C/', ''], # Crazy |
77 | ['IsCc', '$cat eq "Cc"', ''], # Other, Control or Format |
78 | ['IsCo', '$cat eq "Co"', ''], # Other, Private Use |
79 | ['IsCn', '$cat eq "Cn"', ''], # Other, Not Assigned |
7c6f5cd2 |
80 | ['IsCf', '$cat eq "Cf"', ''], # Other, Format |
81 | ['IsCs', '$cat eq "Cs"', ''], # Other, Surrogate |
82 | ['IsCn', 'Unassigned Code Value',$PropData], # Other, Not Assigned |
a0ed51b3 |
83 | |
84 | # Informative |
85 | |
86 | ['IsL', '$cat =~ /^L/', ''], # Letter |
87 | ['IsLu', '$cat eq "Lu"', ''], # Letter, Uppercase |
88 | ['IsLl', '$cat eq "Ll"', ''], # Letter, Lowercase |
89 | ['IsLt', '$cat eq "Lt"', ''], # Letter, Titlecase |
90 | ['IsLm', '$cat eq "Lm"', ''], # Letter, Modifier |
91 | ['IsLo', '$cat eq "Lo"', ''], # Letter, Other |
92 | |
93 | ['IsP', '$cat =~ /^P/', ''], # Punctuation |
94 | ['IsPd', '$cat eq "Pd"', ''], # Punctuation, Dash |
95 | ['IsPs', '$cat eq "Ps"', ''], # Punctuation, Open |
96 | ['IsPe', '$cat eq "Pe"', ''], # Punctuation, Close |
97 | ['IsPo', '$cat eq "Po"', ''], # Punctuation, Other |
7c6f5cd2 |
98 | ['IsPc', '$cat eq "Pc"', ''], # Punctuation, Connector |
99 | ['IsPi', '$cat eq "Pi"', ''], # Punctuation, Initial quote |
100 | ['IsPf', '$cat eq "Pf"', ''], # Punctuation, Final quote |
a0ed51b3 |
101 | |
102 | ['IsS', '$cat =~ /^S/', ''], # Symbol |
103 | ['IsSm', '$cat eq "Sm"', ''], # Symbol, Math |
7c6f5cd2 |
104 | ['IsSk', '$cat eq "Sk"', ''], # Symbol, Modifier |
a0ed51b3 |
105 | ['IsSc', '$cat eq "Sc"', ''], # Symbol, Currency |
106 | ['IsSo', '$cat eq "So"', ''], # Symbol, Other |
107 | |
108 | # Combining class |
109 | ['CombiningClass', '$comb', '$comb'], |
110 | |
111 | # BIDIRECTIONAL PROPERTIES |
112 | |
113 | ['Bidirectional', '$bid', '$bid'], |
114 | |
115 | # Strong types: |
116 | |
117 | ['IsBidiL', '$bid eq "L"', ''], # Left-Right; Most alphabetic, |
118 | # syllabic, and logographic |
119 | # characters (e.g., CJK |
120 | # ideographs) |
121 | ['IsBidiR', '$bid eq "R"', ''], # Right-Left; Arabic, Hebrew, |
122 | # and punctuation specific to |
123 | # those scripts |
124 | |
7c6f5cd2 |
125 | ['IsBidiLRE', '$bid eq "LRE"', ''], # Left-to-Right Embedding |
126 | ['IsBidiLRO', '$bid eq "LRO"', ''], # Left-to-Right Override |
127 | ['IsBidiAL', '$bid eq "AL"', ''], # Right-to-Left Arabic |
128 | ['IsBidiRLE', '$bid eq "RLE"', ''], # Right-to-Left Embedding |
129 | ['IsBidiRLO', '$bid eq "RLO"', ''], # Right-to-Left Override |
130 | ['IsBidiPDF', '$bid eq "PDF"', ''], # Pop Directional Format |
131 | ['IsBidiNSM', '$bid eq "NSM"', ''], # Non-Spacing Mark |
132 | ['IsBidiBN', '$bid eq "BN"', ''], # Boundary Neutral |
133 | |
a0ed51b3 |
134 | # Weak types: |
135 | |
136 | ['IsBidiEN','$bid eq "EN"', ''], # European Number |
137 | ['IsBidiES','$bid eq "ES"', ''], # European Number Separator |
138 | ['IsBidiET','$bid eq "ET"', ''], # European Number Terminator |
139 | ['IsBidiAN','$bid eq "AN"', ''], # Arabic Number |
140 | ['IsBidiCS','$bid eq "CS"', ''], # Common Number Separator |
141 | |
142 | # Separators: |
143 | |
144 | ['IsBidiB', '$bid eq "B"', ''], # Block Separator |
145 | ['IsBidiS', '$bid eq "S"', ''], # Segment Separator |
146 | |
147 | # Neutrals: |
148 | |
149 | ['IsBidiWS','$bid eq "WS"', ''], # Whitespace |
150 | ['IsBidiON','$bid eq "ON"', ''], # Other Neutrals ; All other |
151 | # characters: punctuation, |
152 | # symbols |
153 | |
154 | # Decomposition |
155 | |
156 | ['Decomposition', '$decomp', '$decomp'], |
157 | ['IsDecoCanon', '$decomp && $decomp !~ /^</', ''], |
158 | ['IsDecoCompat', '$decomp =~ /^</', ''], |
159 | ['IsDCfont', '$decomp =~ /^<font>/', ''], |
160 | ['IsDCnoBreak', '$decomp =~ /^<noBreak>/', ''], |
161 | ['IsDCinitial', '$decomp =~ /^<initial>/', ''], |
f59877d4 |
162 | ['IsDCmedial', '$decomp =~ /^<medial>/', ''], |
a0ed51b3 |
163 | ['IsDCfinal', '$decomp =~ /^<final>/', ''], |
164 | ['IsDCisolated', '$decomp =~ /^<isolated>/', ''], |
165 | ['IsDCcircle', '$decomp =~ /^<circle>/', ''], |
166 | ['IsDCsuper', '$decomp =~ /^<super>/', ''], |
167 | ['IsDCsub', '$decomp =~ /^<sub>/', ''], |
168 | ['IsDCvertical', '$decomp =~ /^<vertical>/', ''], |
169 | ['IsDCwide', '$decomp =~ /^<wide>/', ''], |
170 | ['IsDCnarrow', '$decomp =~ /^<narrow>/', ''], |
171 | ['IsDCsmall', '$decomp =~ /^<small>/', ''], |
172 | ['IsDCsquare', '$decomp =~ /^<square>/', ''], |
7c6f5cd2 |
173 | ['IsDCfraction', '$decomp =~ /^<fraction>/', ''], |
a0ed51b3 |
174 | ['IsDCcompat', '$decomp =~ /^<compat>/', ''], |
175 | |
176 | # Number |
177 | |
e1b504a6 |
178 | ['Number', '$num ne ""', '$num'], |
a0ed51b3 |
179 | |
180 | # Mirrored |
181 | |
182 | ['IsMirrored', '$mir eq "Y"', ''], |
183 | |
184 | # Arabic |
185 | |
186 | ['ArabLink', '1', '$link'], |
187 | ['ArabLnkGrp', '1', '$linkgroup'], |
188 | |
189 | # Jamo |
190 | |
191 | ['JamoShort', '1', '$short'], |
499bfa7a |
192 | |
193 | # Syllables |
194 | |
7c6f5cd2 |
195 | syllable_defs(), |
196 | |
a77b4ae5 |
197 | # Line break properties - Normative |
198 | |
199 | ['IsLbrkBK','$brk eq "BK"', ''], # Mandatory Break |
200 | ['IsLbrkCR','$brk eq "CR"', ''], # Carriage Return |
201 | ['IsLbrkLF','$brk eq "LF"', ''], # Line Feed |
202 | ['IsLbrkCM','$brk eq "CM"', ''], # Attached Characters and Combining Marks |
203 | ['IsLbrkSG','$brk eq "SG"', ''], # Surrogates |
204 | ['IsLbrkGL','$brk eq "GL"', ''], # Non-breaking (Glue) |
205 | ['IsLbrkCB','$brk eq "CB"', ''], # Contingent Break Opportunity |
206 | ['IsLbrkSP','$brk eq "SP"', ''], # Space |
207 | ['IsLbrkZW','$brk eq "ZW"', ''], # Zero Width Space |
208 | |
209 | # Line break properties - Informative |
210 | ['IsLbrkXX','$brk eq "XX"', ''], # Unknown |
211 | ['IsLbrkOP','$brk eq "OP"', ''], # Opening Punctuation |
212 | ['IsLbrkCL','$brk eq "CL"', ''], # Closing Punctuation |
213 | ['IsLbrkQU','$brk eq "QU"', ''], # Ambiguous Quotation |
214 | ['IsLbrkNS','$brk eq "NS"', ''], # Non Starter |
215 | ['IsLbrkEX','$brk eq "EX"', ''], # Exclamation/Interrogation |
216 | ['IsLbrkSY','$brk eq "SY"', ''], # Symbols Allowing Breaks |
217 | ['IsLbrkIS','$brk eq "IS"', ''], # Infix Separator (Numeric) |
218 | ['IsLbrkPR','$brk eq "PR"', ''], # Prefix (Numeric) |
219 | ['IsLbrkPO','$brk eq "PO"', ''], # Postfix (Numeric) |
220 | ['IsLbrkNU','$brk eq "NU"', ''], # Numeric |
221 | ['IsLbrkAL','$brk eq "AL"', ''], # Ordinary Alphabetic and Symbol Characters |
222 | ['IsLbrkID','$brk eq "ID"', ''], # Ideographic |
223 | ['IsLbrkIN','$brk eq "IN"', ''], # Inseparable |
224 | ['IsLbrkHY','$brk eq "HY"', ''], # Hyphen |
225 | ['IsLbrkBB','$brk eq "BB"', ''], # Break Opportunity Before |
226 | ['IsLbrkBA','$brk eq "BA"', ''], # Break Opportunity After |
227 | ['IsLbrkSA','$brk eq "SA"', ''], # Complex Context (South East Asian) |
228 | ['IsLbrkAI','$brk eq "AI"', ''], # Ambiguous (Alphabetic or Ideographic) |
229 | ['IsLbrkB2','$brk eq "B2"', ''], # Break Opportunity Before and After |
a0ed51b3 |
230 | ); |
231 | |
232 | # This is not written for speed... |
233 | |
9fdf68be |
234 | my %InId; |
235 | my $InId = 0; |
236 | |
a0ed51b3 |
237 | foreach $file (@todo) { |
238 | my ($table, $wanted, $val) = @$file; |
239 | next if @ARGV and not grep { $_ eq $table } @ARGV; |
9fdf68be |
240 | print $table, "\n"; |
241 | $table =~ s/\W+//g; |
242 | if ($table =~ /^In(.+)/) { |
243 | my $id; |
244 | unless (exists $InId{$1}) { |
245 | $InId{$1} = $InId++; |
246 | } |
247 | $id = $InId{$1}; |
248 | open(OUT, ">In/$id.pl") or die "Can't create In/$id.pl: $!\n"; |
249 | print OUT "# In/$id.pl $1\n"; |
250 | } |
251 | elsif ($table =~ /^(Is|To)(.+)/) { |
a0ed51b3 |
252 | open(OUT, ">$1/$2.pl") or die "Can't create $1/$2.pl: $!\n"; |
253 | } |
254 | else { |
255 | open(OUT, ">$table.pl") or die "Can't create $table.pl: $!\n"; |
256 | } |
14055466 |
257 | print OUT <<EOH; |
258 | # !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
259 | # This file is built by $0 from e.g. $UnicodeData. |
260 | # Any changes made here will be lost! |
261 | EOH |
a0ed51b3 |
262 | print OUT <<"END"; |
263 | return <<'END'; |
264 | END |
265 | print OUT proplist($table, $wanted, $val); |
266 | print OUT "END\n"; |
267 | close OUT; |
268 | } |
269 | |
270 | # Must treat blocks specially. |
271 | |
272 | exit if @ARGV and not grep { $_ eq Block } @ARGV; |
9fdf68be |
273 | print "Blocks\n"; |
7c6f5cd2 |
274 | open(UD, 'Blocks.txt') or die "Can't open Blocks.txt: $!\n"; |
9fdf68be |
275 | open(OUT, ">Blocks.pl") or die "Can't create Blocks.pl: $!\n"; |
14055466 |
276 | print OUT <<EOH; |
277 | # !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
278 | # This file is built by $0 from e.g. $UnicodeData. |
279 | # Any changes made here will be lost! |
280 | EOH |
a0ed51b3 |
281 | print OUT <<"END"; |
282 | return <<'END'; |
283 | END |
284 | |
285 | while (<UD>) { |
286 | next if /^#/; |
287 | next if /^$/; |
288 | chomp; |
9fdf68be |
289 | ($code, $last, $name) = /^([0-9a-f]+)\.\.([0-9a-f]+); (.+)/i; |
a0ed51b3 |
290 | if ($name) { |
291 | print OUT "$code $last $name\n"; |
9fdf68be |
292 | $name =~ s/\W+//g; |
293 | my $id; |
294 | unless (exists $InId{$name}) { |
295 | $InId{$name} = $InId++; |
296 | } |
297 | $id = $InId{$name}; |
298 | open(BLOCK, ">In/$id.pl"); |
299 | print OUT "# In/$id.pl $name\n"; |
14055466 |
300 | print BLOCK <<EOH; |
301 | # !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
302 | # This file is built by $0 from e.g. $UnicodeData. |
303 | # Any changes made here will be lost! |
304 | EOH |
a0ed51b3 |
305 | print BLOCK <<"END2"; |
306 | return <<'END'; |
307 | $code $last |
308 | END |
309 | END2 |
310 | close BLOCK; |
311 | } |
312 | } |
313 | |
314 | print OUT "END\n"; |
315 | close OUT; |
316 | |
9fdf68be |
317 | open(INID, ">In.pl"); |
318 | |
319 | print INID <<EOH; |
320 | # !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
321 | # This file is built by $0 from e.g. $UnicodeData. |
322 | # Any changes made here will be lost! |
323 | %utf8::In = ( |
324 | EOH |
325 | |
326 | # Order doesn't matter but let's prettyprint anyway. |
56ca2fc0 |
327 | foreach my $in (sort { $InId{$a} <=> $InId{$b} } keys %InId) { |
f178ed66 |
328 | printf INID "%-40s => %3d,\n", "'$in'", $InId{$in}; |
9fdf68be |
329 | } |
330 | |
331 | print INID ");\n"; |
332 | |
333 | close(INID); |
334 | |
a0ed51b3 |
335 | ################################################## |
336 | |
337 | sub proplist { |
338 | my ($table, $wanted, $val) = @_; |
339 | my @wanted; |
340 | my $out; |
341 | my $split; |
342 | |
7c6f5cd2 |
343 | return listFromPropFile($wanted) if $val eq $PropData; |
344 | |
a0ed51b3 |
345 | if ($table =~ /^Arab/) { |
d357d9fe |
346 | open(UD, "ArabShap.txt") or warn "Can't open $table: $!"; |
a0ed51b3 |
347 | |
348 | $split = '($code, $name, $link, $linkgroup) = split(/; */);'; |
349 | } |
350 | elsif ($table =~ /^Jamo/) { |
505afebf |
351 | open(UD, "Jamo.txt") or warn "Can't open $table: $!"; |
a0ed51b3 |
352 | |
353 | $split = '($code, $short, $name) = split(/; */); $code =~ s/^U\+//;'; |
354 | } |
499bfa7a |
355 | elsif ($table =~ /^IsSyl/) { |
7c6f5cd2 |
356 | open(UD, $SyllableData) or warn "Can't open $table: $!"; |
499bfa7a |
357 | |
358 | $split = '($code, $short, $syl) = split(/; */); $code =~ s/^U\+//;'; |
359 | } |
a77b4ae5 |
360 | elsif ($table =~ /^IsLbrk/) { |
361 | open(UD, "LineBrk.txt") or warn "Can't open $table: $!"; |
362 | |
50fc4248 |
363 | $split = '($code, $brk, $name) = /^([0-9a-f]+);(\w+) # (.+)/i;'; |
a77b4ae5 |
364 | } |
a0ed51b3 |
365 | else { |
11695a73 |
366 | open(UD, $UnicodeData) or warn "Can't open $UnicodeData: $!"; |
a0ed51b3 |
367 | |
368 | $split = '($code, $name, $cat, $comb, $bid, $decomp, $dec, $dig, $num, $mir, $uni1, |
369 | $comment, $up, $down, $title) = split(/;/);'; |
370 | } |
371 | |
372 | if ($table =~ /^(?:To|Is)[A-Z]/) { |
373 | eval <<"END"; |
374 | while (<UD>) { |
375 | next if /^#/; |
7c6f5cd2 |
376 | next if /^\\s/; |
377 | s/\\s+\$//; |
a0ed51b3 |
378 | $split |
379 | if ($wanted) { |
380 | push(\@wanted, [hex \$code, hex $val, \$name =~ /, First>\$/]); |
381 | } |
382 | } |
383 | END |
384 | die $@ if $@; |
385 | |
386 | while (@wanted) { |
387 | $beg = shift @wanted; |
388 | $last = $beg; |
389 | while (@wanted and $wanted[0]->[0] == $last->[0] + 1 and |
390 | (not $val or $wanted[0]->[1] == $last->[1] + 1)) { |
391 | $last = shift @wanted; |
392 | } |
393 | $out .= sprintf "%04x", $beg->[0]; |
394 | if ($beg->[2]) { |
395 | $last = shift @wanted; |
396 | } |
397 | if ($beg == $last) { |
398 | $out .= "\t"; |
399 | } |
400 | else { |
401 | $out .= sprintf "\t%04x", $last->[0]; |
402 | } |
403 | $out .= sprintf "\t%04x", $beg->[1] if $val; |
404 | $out .= "\n"; |
405 | } |
406 | } |
407 | else { |
408 | eval <<"END"; |
409 | while (<UD>) { |
410 | next if /^#/; |
7c6f5cd2 |
411 | next if /^\\s*\$/; |
a0ed51b3 |
412 | chop; |
413 | $split |
414 | if ($wanted) { |
415 | push(\@wanted, [hex \$code, $val, \$name =~ /, First>\$/]); |
416 | } |
417 | } |
418 | END |
419 | die $@ if $@; |
420 | |
421 | while (@wanted) { |
422 | $beg = shift @wanted; |
423 | $last = $beg; |
424 | while (@wanted and $wanted[0]->[0] == $last->[0] + 1 and |
425 | ($wanted[0]->[1] eq $last->[1])) { |
426 | $last = shift @wanted; |
427 | } |
428 | $out .= sprintf "%04x", $beg->[0]; |
429 | if ($beg->[2]) { |
430 | $last = shift @wanted; |
431 | } |
432 | if ($beg == $last) { |
433 | $out .= "\t"; |
434 | } |
435 | else { |
436 | $out .= sprintf "\t%04x", $last->[0]; |
437 | } |
438 | $out .= sprintf "\t%s\n", $beg->[1]; |
439 | } |
440 | } |
441 | $out; |
442 | } |
11695a73 |
443 | |
7c6f5cd2 |
444 | sub listFromPropFile { |
445 | my ($wanted) = @_; |
446 | my $out; |
447 | |
448 | open (UD, $PropData) or die "Can't open $PropData: $!\n"; |
449 | local($/) = "\n" . '*' x 43 . "\n\nProperty dump for:"; # not 42? |
450 | |
451 | <UD>; |
452 | while (<UD>) { |
453 | chomp; |
454 | if (s/0x[\d\w]+\s+\((.*?)\)// and $wanted eq $1) { |
455 | s/\(\d+ chars\)//g; |
456 | s/^\s+//mg; |
457 | s/\s+$//mg; |
458 | s/\.\./\t/g; |
459 | $out = lc $_; |
460 | last; |
461 | } |
462 | } |
463 | close (UD); |
464 | "$out\n"; |
465 | } |
466 | |
467 | sub syllable_defs { |
468 | my @defs; |
469 | my %seen; |
470 | |
471 | open (SD, $SyllableData) or die "Can't open $SyllableData: $!\n"; |
472 | while (<SD>) { |
473 | next if /^\s*(#|$)/; |
474 | s/\s+$//; |
475 | ($code, $name, $syl) = split /; */; |
476 | next unless $syl; |
477 | push (@defs, ["IsSyl$syl", qq{\$syl eq "$syl"}, '']) |
478 | unless $seen{$syl}++; |
479 | } |
480 | close (SD); |
481 | return (@defs); |
482 | } |
483 | |
6dd159d1 |
484 | # eof |