1if not modules then modules = { } end modules ['font-ota'] = {
2 version = 1.001,
3 comment = "companion to font-ini.mkiv",
4 author = "Hans Hagen, PRAGMA-ADE, Hasselt NL",
5 copyright = "PRAGMA ADE / ConTeXt Development Team",
6 license = "see context related readme files"
7}
8
9
10
11local type = type
12local setmetatableindex = table.setmetatableindex
13
14if not trackers then trackers = { register = function() end } end
15
16
17
18local fonts, nodes, node = fonts, nodes, node
19
20local allocate = utilities.storage.allocate
21
22local otf = fonts.handlers.otf
23
24local analyzers = fonts.analyzers
25local initializers = allocate()
26local methods = allocate()
27
28analyzers.initializers = initializers
29analyzers.methods = methods
30
31local nuts = nodes.nuts
32local tonut = nuts.tonut
33
34local getnext = nuts.getnext
35local getprev = nuts.getprev
36local getprev = nuts.getprev
37local getprop = nuts.getprop
38local setprop = nuts.setprop
39local getchar = nuts.getchar
40local ischar = nuts.ischar
41local isnextchar = nuts.isnextchar
42
43
44local endofmath = nuts.endofmath
45
46local nodecodes = nodes.nodecodes
47local disc_code <const> = nodecodes.disc
48local math_code <const> = nodecodes.math
49
50local fontdata = fonts.hashes.identifiers
51local descriptiondata = fonts.hashes.descriptions
52local categories = characters and characters.categories or { }
53local chardata = characters and characters.data
54
55local otffeatures = fonts.constructors.features.otf
56local registerotffeature = otffeatures.register
57
58
59
60
61local setstate = nuts.setstate
62local getstate = nuts.getstate
63
64local classifiers = characters.classifiers
65
66
67
68local s_init = 1 local s_rphf = 7
69local s_medi = 2 local s_half = 8
70local s_fina = 3 local s_pref = 9
71local s_isol = 4 local s_blwf = 10
72local s_mark = 5 local s_pstf = 11
73local s_rest = 6
74
75local states = allocate {
76 init = s_init,
77 medi = s_medi,
78 med2 = s_medi,
79 fina = s_fina,
80 fin2 = s_fina,
81 fin3 = s_fina,
82 isol = s_isol,
83 mark = s_mark,
84 rest = s_rest,
85 rphf = s_rphf,
86 half = s_half,
87 pref = s_pref,
88 blwf = s_blwf,
89 pstf = s_pstf,
90}
91
92local features = allocate {
93 init = s_init,
94 medi = s_medi,
95 med2 = s_medi,
96 fina = s_fina,
97 fin2 = s_fina,
98 fin3 = s_fina,
99 isol = s_isol,
100
101
102 rphf = s_rphf,
103 half = s_half,
104 pref = s_pref,
105 blwf = s_blwf,
106 pstf = s_pstf,
107}
108
109local mappers = allocate {
110 l = s_init,
111 d = s_medi,
112 c = s_medi,
113 r = s_fina,
114 u = s_isol,
115}
116
117analyzers.states = states
118analyzers.features = features
119analyzers.useunicodemarks = false
120
121if not classifiers then
122
123
124
125 local f_arabic, l_arabic = characters.blockrange("arabic")
126 local f_syriac, l_syriac = characters.blockrange("syriac")
127 local f_mandiac, l_mandiac = characters.blockrange("mandiac")
128 local f_nko, l_nko = characters.blockrange("nko")
129 local f_ext_a, l_ext_a = characters.blockrange("arabicextendeda")
130
131 classifiers = setmetatableindex(function(t,k)
132 if type(k) == "number" then
133 local c = chardata[k]
134 local v = false
135 if c then
136 local arabic = c.arabic
137 if arabic then
138 v = mappers[arabic]
139 if not v then
140 log.report("analyze","error in mapping arabic %C",k)
141
142 v = false
143 end
144 elseif (k >= f_arabic and k <= l_arabic) or
145 (k >= f_syriac and k <= l_syriac) or
146 (k >= f_mandiac and k <= l_mandiac) or
147 (k >= f_nko and k <= l_nko) or
148 (k >= f_ext_a and k <= l_ext_a) then
149 if categories[k] == "mn" then
150 v = s_mark
151 else
152 v = s_rest
153 end
154 end
155 end
156 t[k] = v
157 return v
158 end
159 end)
160
161 characters.classifiers = classifiers
162
163end
164
165
166
167
168local is_letter = characters.is_letter
169local is_mark = characters.is_mark
170
171function analyzers.setstate(head,font)
172 local useunicodemarks = analyzers.useunicodemarks
173 local descriptions = descriptiondata[font]
174 local first = false
175 local last = false
176 local current = head
177 local done = false
178
179 while current do
180
181 local nxt, char, id = isnextchar(current,font)
182 if char then
183 if not getstate(current) then
184
185
186
187 if is_mark[char] then
188 setstate(current,s_mark)
189 elseif is_letter[char] then
190 if first then
191 setstate(current,s_medi)
192 else
193 setstate(current,s_init)
194 first = current
195 done = true
196 end
197 last = current
198 else
199 goto PICKUP
200 end
201 goto NEXT
202
203 end
204 ::PICKUP::
205 if first then
206 setstate(last,first == last and s_isol or s_fina)
207 end
208 first = false
209 elseif char == false then
210
211 if first then
212 setstate(last,first == last and s_isol or s_fina)
213 end
214 first = false
215 if id == math_code then
216
217 nxt = getnext(endofmath(current))
218 end
219 elseif id == disc_code then
220
221
222
223 setstate(current,s_medi)
224 last = current
225 else
226 if first then
227 setstate(last,first == last and s_isol or s_fina)
228 end
229 first = false
230 if id == math_code then
231 current = endofmath(current)
232 end
233 end
234 ::NEXT::
235
236 current = nxt
237 end
238 if first then
239 setstate(last,first == last and s_isol or s_fina)
240 end
241 return head, done
242end
243
244
245
246
247
248local function analyzeinitializer(tfmdata,value)
249 local script, language = otf.scriptandlanguage(tfmdata)
250 local action = initializers[script]
251 if not action then
252
253 elseif type(action) == "function" then
254 return action(tfmdata,value)
255 else
256 local action = action[language]
257 if action then
258 return action(tfmdata,value)
259 end
260 end
261end
262
263local function analyzeprocessor(head,font,attr)
264 local tfmdata = fontdata[font]
265 local script, language = otf.scriptandlanguage(tfmdata,attr)
266 local action = methods[script]
267 if not action then
268
269 elseif type(action) == "function" then
270 return action(head,font,attr)
271 else
272 action = action[language]
273 if action then
274 return action(head,font,attr)
275 end
276 end
277 return head, false
278end
279
280registerotffeature {
281 name = "analyze",
282 description = "analysis of character classes",
283 default = true,
284 initializers = {
285 node = analyzeinitializer,
286 },
287 processors = {
288 position = 1,
289 node = analyzeprocessor,
290 }
291}
292
293
294
295methods.latn = analyzers.setstate
296
297
298local arab_warned = { }
299
300local function warning(current,what)
301 local char = getchar(current)
302 if not arab_warned[char] then
303 log.report("analyze","arab: character %C has no %a class",char,what)
304 arab_warned[char] = true
305 end
306end
307
308
309
310function methods.arab(head,font,attr)
311 local first, last, c_first, c_last
312 local current = head
313 local done = false
314 current = tonut(current)
315 while current do
316
317 local nxt, char, id = isnextchar(current,font)
318 if char and not getstate(current) then
319 done = true
320 local classifier = classifiers[char]
321 if not classifier then
322 if last then
323 if c_last == s_medi or c_last == s_fina then
324 setstate(last,s_fina)
325 else
326 warning(last,"fina")
327 setstate(last,s_error)
328 end
329 first = nil
330 last = nil
331 elseif first then
332 if c_first == s_medi or c_first == s_fina then
333 setstate(first,s_isol)
334 else
335 warning(first,"isol")
336 setstate(first,s_error)
337 end
338 first = nil
339 end
340 elseif classifier == s_mark then
341 setstate(current,s_mark)
342 elseif classifier == s_isol then
343 if last then
344 if c_last == s_medi or c_last == s_fina then
345 setstate(last,s_fina)
346 else
347 warning(last,"fina")
348 setstate(last,s_error)
349 end
350 first = nil
351 last = nil
352 elseif first then
353 if c_first == s_medi or c_first == s_fina then
354 setstate(first,s_isol)
355 else
356 warning(first,"isol")
357 setstate(first,s_error)
358 end
359 first = nil
360 end
361 setstate(current,s_isol)
362 elseif classifier == s_medi then
363 if first then
364 last = current
365 c_last = classifier
366 setstate(current,s_medi)
367 else
368 setstate(current,s_init)
369 first = current
370 c_first = classifier
371 end
372 elseif classifier == s_fina then
373 if last then
374 if getstate(last) ~= s_init then
375 setstate(last,s_medi)
376 end
377 setstate(current,s_fina)
378 first, last = nil, nil
379 elseif first then
380
381
382
383
384 setstate(current,s_fina)
385 first = nil
386 else
387 setstate(current,s_isol)
388 end
389 else
390 setstate(current,s_rest)
391 if last then
392 if c_last == s_medi or c_last == s_fina then
393 setstate(last,s_fina)
394 else
395 warning(last,"fina")
396 setstate(last,s_error)
397 end
398 first = nil
399 last = nil
400 elseif first then
401 if c_first == s_medi or c_first == s_fina then
402 setstate(first,s_isol)
403 else
404 warning(first,"isol")
405 setstate(first,s_error)
406 end
407 first = nil
408 end
409 end
410 else
411 if last then
412 if c_last == s_medi or c_last == s_fina then
413 setstate(last,s_fina)
414 else
415 warning(last,"fina")
416 setstate(last,s_error)
417 end
418 first = nil
419 last = nil
420 elseif first then
421 if c_first == s_medi or c_first == s_fina then
422 setstate(first,s_isol)
423 else
424 warning(first,"isol")
425 setstate(first,s_error)
426 end
427 first = nil
428 end
429 if id == math_code then
430
431 nxt = getnext(endofmath(current))
432 end
433 end
434
435 current = nxt
436 end
437 if last then
438 if c_last == s_medi or c_last == s_fina then
439 setstate(last,s_fina)
440 else
441 warning(last,"fina")
442 setstate(last,s_error)
443 end
444 elseif first then
445 if c_first == s_medi or c_first == s_fina then
446 setstate(first,s_isol)
447 else
448 warning(first,"isol")
449 setstate(first,s_error)
450 end
451 end
452 return head, done
453end
454
455methods.syrc = methods.arab
456methods.mand = methods.arab
457methods.nko = methods.arab
458
459
460
461do
462
463
464
465
466 local joining = setmetatableindex(function(t,k)
467 if type(k) == "number" then
468 local c = chardata[k]
469 local v = false
470 if c then
471 local mongolian = c.mongolian
472
473 v = mongolian
474 end
475 t[k] = v
476 return v
477 end
478 end)
479
480 function methods.mong(head,font,attr)
481 local first, last
482 local current = head
483 local done = false
484 local prevjoin = nil
485 local prestate = nil
486 current = tonut(current)
487
488 local function wrapup()
489 if last then
490 if last ~= first then
491 local s = getstate(last)
492 if s == s_medi then
493 setstate(last,s_fina)
494 elseif s == s_init then
495 setstate(last,s_isol)
496 end
497 end
498 last = nil
499 first = nil
500 prevjoin = nil
501 prestate = nil
502 end
503 end
504
505 while current do
506
507 local nxt, char, id = isnextchar(current,font)
508 if char and not getstate(current) then
509 local currjoin = joining[char]
510 done = true
511 if not last then
512 setstate(current,s_isol)
513 prevjoin = currjoin
514 first = current
515 last = current
516 prevstate = s_isol
517 elseif currjoin == "t" then
518
519 last = current
520 elseif prevjoin == "d" or prevjoin == "jc" or prevjoin == "l" then
521 if currjoin == "d" or prevjoin == "jc" or prevjoin == "r" then
522 local s = getstate(last)
523 if s == s_isol then
524 setstate(last,s_init)
525 elseif s == s_fina then
526 setstate(last,s_medi)
527 end
528 setstate(current,s_fina)
529 prevstate = s_fina
530 elseif prevjoin == "nj" or prevjoin == "l" then
531 local s = getstate(last)
532 if s == s_medi then
533 setstate(last,s_fina)
534 elseif s == s_init then
535 setstate(last,s_isol)
536 end
537 setstate(current,s_isol)
538 prevstate = s_isol
539 end
540 prevjoin = currjoin
541 last = current
542 elseif prevjoin == "nj" or prevjoin == "r" then
543 if s == s_medi then
544 setstate(last,s_fina)
545 elseif s == s_init then
546 setstate(last,s_isol)
547 end
548 setstate(current,s_isol)
549 prevjoin = currjoin
550 prevstate = s_isol
551 last = current
552 elseif last then
553 wrapup()
554 end
555 else
556 if last then
557 wrapup()
558 end
559 if id == math_code then
560
561 nxt = getnext(endofmath(current))
562 end
563 end
564
565 current = nxt
566 end
567 if last then
568 wrapup()
569 end
570 return head, done
571 end
572
573end
574
575directives.register("otf.analyze.useunicodemarks",function(v)
576 analyzers.useunicodemarks = v
577end)
578 |
