1 ;; BENG-OTF.flt -- Font Layout Table for Bengali OpenType font
3 ;; National Institute of Advanced Industrial Science and Technology (AIST)
4 ;; Registration Number H15PRO112
6 ;; This file is part of the m17n database; a sub-part of the m17n
9 ;; The m17n library is free software; you can redistribute it and/or
10 ;; modify it under the terms of the GNU Lesser General Public License
11 ;; as published by the Free Software Foundation; either version 2.1 of
12 ;; the License, or (at your option) any later version.
14 ;; The m17n library is distributed in the hope that it will be useful,
15 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
17 ;; Lesser General Public License for more details.
19 ;; You should have received a copy of the GNU Lesser General Public
20 ;; License along with the m17n library; if not, write to the Free
21 ;; Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA
26 ;;; For Bengali OpenType fonts to draw the Bengali script. Tested with
27 ;;; MuktiNarrow.ttf <http://www.nongnu.org/freebangfont/index.html>
29 ;;; LikhanNormal.otf <http:http://www.stat.wisc.edu/~deepayan/Bengali/WebPage/Font/fonts.html>
32 ;; C: consonant (excluding B, Y and R)
33 ;; B: consonant BA (below)
34 ;; Y: consonant YA (post)
35 ;; R: consonant RA (reph, below)
41 ;; t: MATRA (two-part)
42 ;; A: vowel modifier (above)
43 ;; a: vowel modifier (post)
44 ;; V: independent vowel
45 ;; N: ZWNJ (ZERO WIDTH NON-JOINER)
46 ;; J: ZWJ (ZERO WIDTH JOINER)
51 (0x0964 0x0965 ?E) ; DANDA, DOUBLE DANDA
52 (0x0980 0x09FF ?E) ; ELSE
53 (0x0981 ?A) ; SIGN CANDRABINDU (above)
54 (0x0982 0x0983 ?a) ; SIGN ANUSWAR, VISARGA (post)
55 (0x0985 0x098C ?V) ; LETTER A .. VOCALIC L
56 (0x098F 0x0990 ?V) ; LETTER E .. AI
57 (0x0993 0x0994 ?V) ; LETTER O .. AU
58 (0x0995 0x09B9 ?C) ; LETTER KA .. HA
59 (0x09AC ?B) ; LETTER BA
60 (0x09AF ?Y) ; LETTER YA
61 (0x09B0 ?R) ; LETTER RA
62 (0x09BC ?n) ; SIGN NUKTA
63 (0x09BE ?p) ; VOWEL SIGN AA (post)
64 (0x09BF ?m) ; VOWEL SIGN I (pre)
65 (0x09C0 ?p) ; VOWEL SIGN II (post)
66 (0x09C1 0x09C4 ?b) ; VOWEL SIGN U, UU, R, RR (below)
67 (0x09C7 0x09C8 ?m) ; VOWEL SIGN E, AI (pre)
68 (0x09CB 0x09CC ?t) ; VOWEL SIGN O, AU (two-part)
69 (0x09CD ?H) ; SIGN VIRAMA (HASANT)
70 (0x09D7 ?p) ; AU LENGTH MARK
71 (0x09DC 0x09DD ?C) ; LETTER RRA, RHA
72 (0x09DF ?C) ; LETTER YYA
73 (0x09E0 0x09E1 ?V) ; LETTER VOCALIC RR, LL
74 (0x09E2 0x09E3 ?b) ; VOWEL SIGN L .. LL (below)
75 (0x09F0 0x09F1 ?C) ; LETTER RR', RR'' (assamese)
76 (0x09FE ?x) ; mark #1 (internal use)
77 (0x09FF ?y) ; mark #2 (internal use)
80 ;; Step 1 : Syllable identification. Recognised syllables are quoted
81 ;; by the pseudo character, which is generated by the command "|" and
82 ;; has the category " " (space).
86 ;; Case A-C are for those syllables that end with an explicit vowel
87 ;; mark and/or a vowel modifier. They are divided into three cases
88 ;; for the readability of regular expression. The leading
89 ;; consonant-Hasant repetition is analysed for reordering in the
90 ;; next step. Two-part vowel, if any, is split for
93 ;; Case A : A syllable ending with a vowel modifier.
95 ("(RH)?(([CRBY]n?H[NJ]?)*([CRBY]n?))([mbp]*)(t)?([Aa])"
104 ;; Case B : A syllable ending with a two-part vowel.
106 ("(RH)?(([CRBY]n?H[NJ]?)*([CRBY]n?))(t)"
113 ;; Case C : A syllable ending with other vowel. Note that a
114 ;; two-part vowel may be expressed with two vowel marks for
115 ;; backward compatibility.
117 ("(RH)?(([CRBY]n?H[NJ]?)*([CRBY]n?))([mbp]+)"
124 ;; Case D : Ya-phalaa. Reorder H and Y for the next step.
125 ;; The web page "Unicode FAQ for Indic Scripts and Languages"
126 ;; <http://www.unicode.org/faq/indic.html> says "it should be
127 ;; permissible for the Ya-phalla to be consistently formed by "ZWNJ
136 ;; Case E : No explicit vowel nor modifier. If the syllable ends
137 ;; with a consonant, analyse it for reordering in the next step.
138 ;; Otherwise, just identify the syllable without changing anything.
140 ("(RH)?(([CRBY]n?H[NJ]?)*[CRBY]n?)(HN|HJ|H)?"
147 ;; Case F : Syllables that begin with an independent vowel. An
148 ;; optional HYp sequence appears when this syllable represents the
149 ;; sound "a" in English "bat" (see the FAQ above). If it appears,
150 ;; we reorder the H and Y for the next step.
152 < | (1 =) (2 ("HY(p)" 0x09AF 0x09CD (1 =))) (3 =) | >)
157 ;; Set mark #1 (x) at the position where below consonants begin, and
158 ;; mark #2 (y) at the position to which below and above signs will be
164 ("([CRBY]n?(H[NJ]?Cn?)*)(H)(([RB]H)*)(Y)"
165 (1 = *) ; prebase & base
167 (4 = *) ; below consonants
170 (3 =)) ; moved HASANT
171 ;; Ending with R or B.
173 ("([CRBY]n?(H[NJ]?Cn?)*)(H)(([RB]H)*[RB])"
174 (1 = *) ; prebase & base
176 (4 = *) ; below consonants
184 ;; Split two-part dependent vowel signs for canonicalisation.
187 ((0x09CB) 0x09C7 0x09BE)
188 ((0x09CC) 0x09C7 0x09D7)))
191 ;; Step 2 : Move Reph and Matra if necessary. From now on, we care
192 ;; only for those syllables that have been identified in Step 1.
196 ;; Special case: a single consonant and a Halant.
205 ;; This is the most generic pattern. It follows Cases A, B, C and
206 ;; E in Step 1. Now Mark #1 is used to indicate the critical part
207 ;; that requires pre-base substitution in the following steps.
209 ;; 1 2 3 4 5 6 7 8 9 10
210 (" (RH)?([^ xy]+)x([^ y]*)y(YH)?(m)?(b)?(p)?(A)?(a)?(HN|HJ|H)? "
213 ;; Actually, the nukt feature is not necessary for Bengali because
214 ;; all the necessary Nukta forms are precomposed in the Unicode
215 ;; standard. Even if a Nukta consonant is given in the form of
216 ;; the combination of the base consonant and a Nukta sign, we can
217 ;; safely perform the composition here because it does not affect
218 ;; surrounding letters in the syllable. The Akhand ligature
219 ;; operation is also applied here, before applying the half form
220 ;; operation because the Mukti font generates Akhand ligatures
221 ;; directly from the "C H C" sequence, not via the half form.
222 0x09FE ; begin Cpre & Cbase
223 (2 otf:beng=nukt,akhn) ; {Cpre + H} + Cbase
224 0x09FE ; end Cpre & Cbase
225 (3 otf:beng=blwf) ; {Cbelow + H}
227 (1 otf:beng=rphf) ; [Reph]
229 (4 otf:beng=pstf) ; [Cpost + H]
232 (10 = *) ; optional HASANT
235 ;; Syllables that begin with an independent vowel (following up
236 ;; Step 1, Case F). If a YH sequence exist, it is changed to the
237 ;; post-base form. Syllables of this type do not require further
246 ;; Ya-phalaa (following up Step 1, Case D). Remove N and change YH
247 ;; to the post base form. Syllables of this type do not require
248 ;; further modification.
258 ;; Step 3 : Now only those syllables that contain the pseudo character
259 ;; x require pre-base substition. This is the most complicated part
262 ;; If the sequence "C1 H C2" makes ligature L12, L12 replaces the
263 ;; original sequence.
265 ;; To test the availability of such a ligature, we try to generate it
266 ;; using the pre-base substitute feature, then see whether succeeded
267 ;; or not. In the case of failure, the pre-base feature does not
268 ;; change the original sequence.
270 ;; To create a ligature, the "C1 H" part must be first converted into
271 ;; the half form of C1. Creating the half form of a consonant always
274 ;; ligature(half(C1,H),C2)
275 ;; ==> ligature(C1half,C2)
279 ;; If the ligature is not available, the "C1 H" part must be converted
280 ;; into the _Halant_ (not half) form of C1. However, there is no way
281 ;; to reconvert C1half into C1halant nor to revert back to "C1 H".
282 ;; Thus we duplicate the critical part in two different forms so that
283 ;; we can select the appropriate one in the next step. The pseudo
284 ;; character x is used to indicate the boundaries.
286 ;; ... C1 H C2 ... ==> ... x C1halant C2 x L12 x ...
288 ;; If the length of the L12 part is one, ligature generation was
289 ;; successful. In this case we wipe out the duplicated C1halant and
290 ;; C2. Otherwise we remove L12.
292 ;; In very few cases (I found only one in the Mukti font), the "C1 H"
293 ;; part need to be converted into C1halant to make a ligature with C2.
294 ;; So when try to generate a ligature form, we apply the GSUB features
295 ;; "half", "haln" and "pres" in this order.
298 ;; C: consonant (excluding B, Y and R)
300 ;; N: ZWNJ (ZERO WIDTH NON-JOINER)
301 ;; J: ZWJ (ZERO WIDTH JOINER)
306 (0x0964 0x0965 ?E) ; DANDA, DOUBLE DANDA
307 (0x0980 0x09FF ?E) ; ELSE
308 (0x09CD ?H) ; SIGN VIRAMA (HASANT)
309 (0x0995 ?K) ; LETTER KA
310 (0x09B7 ?S) ; LETTER SSA
311 (0x09A3 ?M) ; LETTER NNA
312 (0x09AE ?M) ; LETTER MA
313 (0x09FE ?x) ; mark #1 (internal use)
320 ;; One pre-base and base.
322 (" ([^x ]*)x((.H)([^NJ]))(H)?x([^ ]*) "
326 (3 otf:beng=haln) ; C1halant
329 (2 otf:beng=half,haln,pres) ; ligature result
335 ;; One pre-base with ZWJ. According to the Unicode FAQ, the half
336 ;; form is forced in this case. So we fake as if ligature
337 ;; generation was failed.
338 (" ([^x ]*)x(.H)J(.)?x([^ ]*) "
342 (2 otf:beng=half) ; C1half
345 0x09FD ; pseudo result
346 0x09FD ; pseudo result
351 ;; One pre-base possibly with ZWNJ. Similar to above.
352 (" ([^x ]*)x(.H)N?(.)?x([^ ]*) "
356 (2 otf:beng=haln) ; C1halant
359 0x09FD ; pseudo result
360 0x09FD ; pseudo result
365 ;; Standalone base. There is nothing more to do.
366 (" ([^x ]*)x(.)x([^ ]*) "
373 ;; KA-SSA-NNA and KA-SSA-MA are the only pre-base ligatures that
374 ;; consist of three consonants.
376 (" ([^x ]*)x((KH)(SH)(M))(H)?x([^ ]*) "
380 (3 otf:beng=haln) ; KAhalant
381 (4 otf:beng=haln) ; SSAhalant
384 (2 otf:beng=half,haln,pres) ; ligature result
390 ;; Two or more pre-bases plus base. Give up. Convert all
391 ;; pre-bases into halant form.
393 (" ([^x ]*)x(([^x]H[JN]?)+)([^x])?x([^ ]*) "
397 (2 force-haln) ; halant forms
400 0x09FD ; pseudo result
401 0x09FD ; pseudo result
409 ;; This is to remove ZWNJ and ZWJ. The half-form-force-effect of ZWJ
410 ;; is ignored. Sorry.
420 ;; Step 4 : Select the appropriate representation. Only those
421 ;; syllables that contain the virtual character x require
426 ;; Only one glyph in the ligature section (between the second and
427 ;; the third x). It means a ligature was successfully generated.
428 ;; C1halant and C2 (between the first and second x) are removed.
429 (" ([^x ]*)x[^x]+x(.)x([^ ]*) "
436 ;; Otherwise halant and base forms are used. The failed ligature
438 (" ([^x ]*)x([^x]+)x[^x]+x([^ ]*) "
445 ;; No need to care the other cases.
449 ;; Step 5 : Select appropriate glyph variants for fine adjustments.
450 ;; Now the syllable boundary marks are removed so that the final step
451 ;; can find word boundaries.
456 (1 otf:beng=blws,abvs,psts,vatu))
462 ;; Step 6 : Word initial substitute. As the syllable boundaries have
463 ;; been eliminated in the previous step, this rule is applied to a run
464 ;; of Bengali glyphs, i.e. word by word. We finally apply the init
465 ;; feature to the word initial gylphs to get the final result.