PluginProbe
PDF & Print by BestWebSoft – WordPress Posts and Pages PDF Generator Plugin / 1.5
PDF & Print by BestWebSoft – WordPress Posts and Pages PDF Generator Plugin v1.5
trunk 1.5 1.6 1.7 1.7.1 1.7.2 1.7.3 1.7.4 1.7.5 1.7.6 1.7.7 1.7.8 1.7.9 1.8.0 1.8.1 1.8.2 1.8.3 1.8.4 1.8.5 1.8.6 1.8.7 1.8.8 1.8.9 1.9.0 1.9.1 All 71 releases
← All changes | mpdf/classes/indic.php +433 -1864 trunk → 1.5 View file →
@@ -1,1864 +1,433 @@
1 -<?php
2 -
3 -require_once __DIR__ . '/../MpdfException.php';
4 -
5 -class INDIC
6 -{
7 - /* FROM hb-ot-shape-complex-indic-private.hh */
8 -
9 - // indic_category
10 - const OT_X = 0;
11 - const OT_C = 1;
12 - const OT_V = 2;
13 - const OT_N = 3;
14 - const OT_H = 4;
15 - const OT_ZWNJ = 5;
16 - const OT_ZWJ = 6;
17 - const OT_M = 7; /* Matra or Dependent Vowel */
18 - const OT_SM = 8;
19 - const OT_VD = 9;
20 - const OT_A = 10;
21 - const OT_NBSP = 11;
22 - const OT_DOTTEDCIRCLE = 12; /* Not in the spec, but special in Uniscribe. /Very very/ special! */
23 - const OT_RS = 13; /* Register Shifter, used in Khmer OT spec */
24 - const OT_Coeng = 14;
25 - const OT_Repha = 15;
26 -
27 - const OT_Ra = 16; /* Not explicitly listed in the OT spec, but used in the grammar. */
28 - const OT_CM = 17;
29 -
30 - // Based on indic_category used to make string to find syllables
31 - // OT_ to string character (using e.g. OT_C from INDIC) hb-ot-shape-complex-indic-private.hh
32 - public static $indic_category_char = array(
33 - 'x',
34 - 'C',
35 - 'V',
36 - 'N',
37 - 'H',
38 - 'Z',
39 - 'J',
40 - 'M',
41 - 'S',
42 - 'v',
43 - 'A', /* Spec gives Andutta U+0952 as OT_A. However, testing shows that Uniscribe
44 - * treats U+0951..U+0952 all as OT_VD - see set_indic_properties */
45 - 's',
46 - 'D',
47 - 'F', /* Register shift Khmer only */
48 - 'G', /* Khmer only */
49 - 'r', /* 0D4E (dot reph) only one in Malayalam */
50 - 'R',
51 - 'm', /* Consonant medial only used in Indic 0A75 in Gurmukhi (0A00..0A7F) : also in Lao, Myanmar, Tai Tham, Javanese & Cham */
52 - );
53 -
54 - /* Visual positions in a syllable from left to right. */
55 - /* FROM hb-ot-shape-complex-indic-private.hh */
56 -
57 - // indic_position
58 - const POS_START = 0;
59 -
60 - const POS_RA_TO_BECOME_REPH = 1;
61 - const POS_PRE_M = 2;
62 - const POS_PRE_C = 3;
63 -
64 - const POS_BASE_C = 4;
65 - const POS_AFTER_MAIN = 5;
66 -
67 - const POS_ABOVE_C = 6;
68 -
69 - const POS_BEFORE_SUB = 7;
70 - const POS_BELOW_C = 8;
71 - const POS_AFTER_SUB = 9;
72 -
73 - const POS_BEFORE_POST = 10;
74 - const POS_POST_C = 11;
75 - const POS_AFTER_POST = 12;
76 -
77 - const POS_FINAL_C = 13;
78 - const POS_SMVD = 14;
79 -
80 - const POS_END = 15;
81 -
82 - /*
83 - * Basic features.
84 - * These features are applied in order, one at a time, after initial_reordering.
85 - */
86 - /*
87 - * Must be in the same order as the indic_features array. Ones starting with _ are F_GLOBAL
88 - * Ones without the _ are only applied where the mask says!
89 - */
90 -
91 - const _NUKT = 0;
92 - const _AKHN = 1;
93 - const RPHF = 2;
94 - const _RKRF = 3;
95 - const PREF = 4;
96 - const BLWF = 5;
97 - const HALF = 6;
98 - const ABVF = 7;
99 - const PSTF = 8;
100 - const CFAR = 9; // Khmer only
101 - const _VATU = 10;
102 - const _CJCT = 11;
103 - const INIT = 12;
104 -
105 - public static function set_indic_properties(&$info, $scriptblock)
106 - {
107 - $u = $info['uni'];
108 - $type = self::indic_get_categories($u);
109 - $cat = ($type & 0x7F);
110 - $pos = ($type >> 8);
111 -
112 - /*
113 - * Re-assign category
114 - */
115 -
116 - if ($u == 0x17D1)
117 - $cat = self::OT_X;
118 -
119 - if ($cat == self::OT_X && self::in_range($u, 0x17CB, 0x17D3)) { /* Khmer Various signs */
120 - /* These are like Top Matras. */
121 - $cat = self::OT_M;
122 - $pos = self::POS_ABOVE_C;
123 - }
124 -
125 - if ($u == 0x17C6)
126 - $cat = self::OT_N; /* Khmer Bindu doesn't like to be repositioned. */
127 -
128 - if ($u == 0x17D2)
129 - $cat = self::OT_Coeng; /* Khmer coeng */
130 -
131 - /* The spec says U+0952 is OT_A. However, testing shows that Uniscribe
132 - * treats U+0951..U+0952 all as OT_VD.
133 - * TESTS:
134 - * U+092E,U+0947,U+0952
135 - * U+092E,U+0952,U+0947
136 - * U+092E,U+0947,U+0951
137 - * U+092E,U+0951,U+0947
138 - * */
139 - //if ($u == 0x0952) $cat = self::OT_A;
140 - if (self::in_range($u, 0x0951, 0x0954))
141 - $cat = self::OT_VD;
142 -
143 - if ($u == 0x200C)
144 - $cat = self::OT_ZWNJ;
145 - else if ($u == 0x200D)
146 - $cat = self::OT_ZWJ;
147 - else if ($u == 0x25CC)
148 - $cat = self::OT_DOTTEDCIRCLE;
149 - else if ($u == 0x0A71)
150 - $cat = self::OT_SM; /* GURMUKHI ADDAK. More like consonant medial. like 0A75. */
151 -
152 - if ($cat == self::OT_Repha) {
153 - /* There are two kinds of characters marked as Repha:
154 - * - The ones that are GenCat=Mn are already positioned visually, ie. after base. (eg. Khmer)
155 - * - The ones that are GenCat=Lo is encoded logically, ie. beginning of syllable. (eg. Malayalam)
156 - *
157 - * We recategorize the first kind to look like a Nukta and attached to the base directly.
158 - */
159 - if ($info['general_category'] == UCDN::UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)
160 - $cat = self::OT_N;
161 - }
162 -
163 - /*
164 - * Re-assign position.
165 - */
166 -
167 - if ((self::FLAG($cat) & (self::FLAG(self::OT_C) | self::FLAG(self::OT_CM) | self::FLAG(self::OT_Ra) | self::FLAG(self::OT_V) | self::FLAG(self::OT_NBSP) | self::FLAG(self::OT_DOTTEDCIRCLE)))) { // = CONSONANT_FLAGS like is_consonant
168 - if ($scriptblock == UCDN::SCRIPT_KHMER)
169 - $pos = self::POS_BELOW_C; /* Khmer differs from Indic here. */
170 - else
171 - $pos = self::POS_BASE_C; /* Will recategorize later based on font lookups. */
172 -
173 - if (self::is_ra($u))
174 - $cat = self::OT_Ra;
175 - }
176 - else if ($cat == self::OT_M) {
177 - $pos = self::matra_position($u, $pos);
178 - } else if ($cat == self::OT_SM || $cat == self::OT_VD) {
179 - $pos = self::POS_SMVD;
180 - }
181 -
182 - if ($u == 0x0B01)
183 - $pos = self::POS_BEFORE_SUB; /* Oriya Bindu is BeforeSub in the spec. */
184 -
185 - $info['indic_category'] = $cat;
186 - $info['indic_position'] = $pos;
187 - }
188 -
189 - // syllable_type
190 - const CONSONANT_SYLLABLE = 0;
191 - const VOWEL_SYLLABLE = 1;
192 - const STANDALONE_CLUSTER = 2;
193 - const BROKEN_CLUSTER = 3;
194 - const NON_INDIC_CLUSTER = 4;
195 -
196 - public static function set_syllables(&$o, $s, &$broken_syllables)
197 - {
198 - $ptr = 0;
199 - $syllable_serial = 1;
200 - $broken_syllables = false;
201 -
202 - while ($ptr < strlen($s)) {
203 - $match = '';
204 - $syllable_length = 1;
205 - $syllable_type = self::NON_INDIC_CLUSTER;
206 - // CONSONANT_SYLLABLE Consonant syllable
207 - // From OT spec:
208 - if (preg_match('/^([CR]m*[N]?(H[ZJ]?|[ZJ]H))*[CR]m*[N]?[A]?(H[ZJ]?|[M]*[N]?[H]?)?[S]?[v]{0,2}/', substr($s, $ptr), $ma)) {
209 - // From HarfBuzz:
210 - //if (preg_match('/^r?([CR]J?(Z?[N]{0,2})?[ZJ]?H(J[N]?)?){0,4}[CR]J?(Z?[N]{0,2})?A?((([ZJ]?H(J[N]?)?)|HZ)|(HJ)?([ZJ]{0,3}M[N]?(H|JHJR)?){0,4})?(S[Z]?)?[v]{0,2}/', substr($s,$ptr), $ma)) {
211 - $syllable_length = strlen($ma[0]);
212 - $syllable_type = self::CONSONANT_SYLLABLE;
213 - }
214 - // VOWEL_SYLLABLE Vowel-based syllable
215 - // From OT spec:
216 - else if (preg_match('/^(RH|r)?V[N]?([ZJ]?H[CR]m*|J[CR]m*)?([M]*[N]?[H]?)?[S]?[v]{0,2}/', substr($s, $ptr), $ma)) {
217 - // From HarfBuzz:
218 - //else if (preg_match('/^(RH|r)?V(Z?[N]{0,2})?(J|([ZJ]?H(J[N]?)?[CR]J?(Z?[N]{0,2})?){0,4}((([ZJ]?H(J[N]?)?)|HZ)|(HJ)?([ZJ]{0,3}M[N]?(H|JHJR)?){0,4})?(S[Z]?)?[v]{0,2})/', substr($s,$ptr), $ma)) {
219 - $syllable_length = strlen($ma[0]);
220 - $syllable_type = self::VOWEL_SYLLABLE;
221 - }
222 -
223 - /* Apply only if it's a word start. */
224 - // STANDALONE_CLUSTER Stand Alone syllable at start of word
225 - // From OT spec:
226 - else if (($ptr == 0 ||
227 - $o[$ptr - 1]['general_category'] < UCDN::UNICODE_GENERAL_CATEGORY_LOWERCASE_LETTER ||
228 - $o[$ptr - 1]['general_category'] > UCDN::UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK
229 - ) && (preg_match('/^(RH|r)?[sD][N]?([ZJ]?H[CR]m*)?([M]*[N]?[H]?)?[S]?[v]{0,2}/', substr($s, $ptr), $ma))) {
230 - // From HarfBuzz:
231 - // && (preg_match('/^(RH|r)?[sD](Z?[N]{0,2})?(([ZJ]?H(J[N]?)?)[CR]J?(Z?[N]{0,2})?){0,4}((([ZJ]?H(J[N]?)?)|HZ)|(HJ)?([ZJ]{0,3}M[N]?(H|JHJR)?){0,4})?(S[Z]?)?[v]{0,2}/', substr($s,$ptr), $ma)) {
232 - $syllable_length = strlen($ma[0]);
233 - $syllable_type = self::STANDALONE_CLUSTER;
234 - }
235 -
236 - // BROKEN_CLUSTER syllable
237 - else if (preg_match('/^(RH|r)?[N]?([ZJ]?H[CR])?([M]*[N]?[H]?)?[S]?[v]{0,2}/', substr($s, $ptr), $ma)) {
238 - // From HarfBuzz:
239 - //else if (preg_match('/^(RH|r)?(Z?[N]{0,2})?(([ZJ]?H(J[N]?)?)[CR]J?(Z?[N]{0,2})?){0,4}((([ZJ]?H(J[N]?)?)|HZ)|(HJ)?([ZJ]{0,3}M[N]?(H|JHJR)?){0,4})(S[Z]?)?[v]{0,2}/', substr($s,$ptr), $ma)) {
240 - if (strlen($ma[0])) { // May match blank
241 - $syllable_length = strlen($ma[0]);
242 - $syllable_type = self::BROKEN_CLUSTER;
243 - $broken_syllables = true;
244 - }
245 - }
246 -
247 - for ($i = $ptr; $i < $ptr + $syllable_length; $i++) {
248 - $o[$i]['syllable'] = ($syllable_serial << 4) | $syllable_type;
249 - }
250 - $ptr += $syllable_length;
251 - $syllable_serial++;
252 - if ($syllable_serial == 16)
253 - $syllable_serial = 1;
254 - }
255 - }
256 -
257 - public static function set_syllables_sinhala(&$o, $s, &$broken_syllables)
258 - {
259 - $ptr = 0;
260 - $syllable_serial = 1;
261 - $broken_syllables = false;
262 -
263 - while ($ptr < strlen($s)) {
264 - $match = '';
265 - $syllable_length = 1;
266 - $syllable_type = self::NON_INDIC_CLUSTER;
267 - // CONSONANT_SYLLABLE Consonant syllable
268 - // From OT spec:
269 - if (preg_match('/^([CR]HJ|[CR]JH){0,8}[CR][HM]{0,3}[S]{0,1}/', substr($s, $ptr), $ma)) {
270 - $syllable_length = strlen($ma[0]);
271 - $syllable_type = self::CONSONANT_SYLLABLE;
272 - }
273 - // VOWEL_SYLLABLE Vowel-based syllable
274 - // From OT spec:
275 - else if (preg_match('/^V[S]{0,1}/', substr($s, $ptr), $ma)) {
276 - $syllable_length = strlen($ma[0]);
277 - $syllable_type = self::VOWEL_SYLLABLE;
278 - }
279 -
280 - for ($i = $ptr; $i < $ptr + $syllable_length; $i++) {
281 - $o[$i]['syllable'] = ($syllable_serial << 4) | $syllable_type;
282 - }
283 - $ptr += $syllable_length;
284 - $syllable_serial++;
285 - if ($syllable_serial == 16)
286 - $syllable_serial = 1;
287 - }
288 - }
289 -
290 - public static function set_syllables_khmer(&$o, $s, &$broken_syllables)
291 - {
292 - $ptr = 0;
293 - $syllable_serial = 1;
294 - $broken_syllables = false;
295 -
296 - while ($ptr < strlen($s)) {
297 - $match = '';
298 - $syllable_length = 1;
299 - $syllable_type = self::NON_INDIC_CLUSTER;
300 - // CONSONANT_SYLLABLE Consonant syllable
301 - if (preg_match('/^r?([CR]J?((Z?F)?[N]{0,2})?[ZJ]?G(JN?)?){0,4}[CR]J?((Z?F)?[N]{0,2})?A?((([ZJ]?G(JN?)?)|GZ)|(GJ)?([ZJ]{0,3}MN?(H|JHJR)?){0,4})?(G([CR]J?((Z?F)?[N]{0,2})?|V))?(SZ?)?[v]{0,2}/', substr($s, $ptr), $ma)) {
302 - $syllable_length = strlen($ma[0]);
303 - $syllable_type = self::CONSONANT_SYLLABLE;
304 - }
305 - // VOWEL_SYLLABLE Vowel-based syllable
306 - else if (preg_match('/^(RH|r)?V((Z?F)?[N]{0,2})?(J|([ZJ]?G(JN?)?[CR]J?((Z?F)?[N]{0,2})?){0,4}((([ZJ]?G(JN?)?)|GZ)|(GJ)?([ZJ]{0,3}MN?(H|JHJR)?){0,4})?(G([CR]J?((Z?F)?[N]{0,2})?|V))?(SZ?)?[v]{0,2})/', substr($s, $ptr), $ma)) {
307 - $syllable_length = strlen($ma[0]);
308 - $syllable_type = self::VOWEL_SYLLABLE;
309 - }
310 -
311 -
312 - // BROKEN_CLUSTER syllable
313 - else if (preg_match('/^(RH|r)?((Z?F)?[N]{0,2})?(([ZJ]?G(JN?)?)[CR]J?((Z?F)?[N]{0,2})?){0,4}((([ZJ]?G(JN?)?)|GZ)|(GJ)?([ZJ]{0,3}MN?(H|JHJR)?){0,4})(G([CR]J?((Z?F)?[N]{0,2})?|V))?(SZ?)?[v]{0,2}/', substr($s, $ptr), $ma)) {
314 - if (strlen($ma[0])) { // May match blank
315 - $syllable_length = strlen($ma[0]);
316 - $syllable_type = self::BROKEN_CLUSTER;
317 - $broken_syllables = true;
318 - }
319 - }
320 -
321 - for ($i = $ptr; $i < $ptr + $syllable_length; $i++) {
322 - $o[$i]['syllable'] = ($syllable_serial << 4) | $syllable_type;
323 - }
324 - $ptr += $syllable_length;
325 - $syllable_serial++;
326 - if ($syllable_serial == 16)
327 - $syllable_serial = 1;
328 - }
329 - }
330 -
331 - public static function initial_reordering(&$info, $GSUBdata, $broken_syllables, $indic_config, $scriptblock, $is_old_spec, $dottedcircle)
332 - {
333 -
334 - self::update_consonant_positions($info, $GSUBdata);
335 -
336 - if ($broken_syllables && $dottedcircle) {
337 - self::insert_dotted_circles($info, $dottedcircle);
338 - }
339 -
340 - $count = count($info);
341 - if (!$count)
342 - return;
343 - $last = 0;
344 - $last_syllable = $info[0]['syllable'];
345 - for ($i = 1; $i < $count; $i++) {
346 - if ($last_syllable != $info[$i]['syllable']) {
347 - self::initial_reordering_syllable($info, $GSUBdata, $indic_config, $scriptblock, $is_old_spec, $last, $i);
348 - $last = $i;
349 - $last_syllable = $info[$last]['syllable'];
350 - }
351 - }
352 - self::initial_reordering_syllable($info, $GSUBdata, $indic_config, $scriptblock, $is_old_spec, $last, $count);
353 - }
354 -
355 - public static function update_consonant_positions(&$info, $GSUBdata)
356 - {
357 - $count = count($info);
358 - for ($i = 0; $i < $count; $i++) {
359 - if ($info[$i]['indic_position'] == self::POS_BASE_C) {
360 - $c = $info[$i]['uni'];
361 - // If would substitute...
362 - if (isset($GSUBdata['pref'][$c])) {
363 - $info[$i]['indic_position'] = self::POS_POST_C;
364 - } else if (isset($GSUBdata['blwf'][$c])) {
365 - $info[$i]['indic_position'] = self::POS_BELOW_C;
366 - } else if (isset($GSUBdata['pstf'][$c])) {
367 - $info[$i]['indic_position'] = self::POS_POST_C;
368 - }
369 - }
370 - }
371 - }
372 -
373 - public static function insert_dotted_circles(&$info, $dottedcircle)
374 - {
375 - $idx = 0;
376 - $last_syllable = 0;
377 - while ($idx < count($info)) {
378 - $syllable = $info[$idx]['syllable'];
379 - $syllable_type = ($syllable & 0x0F);
380 - if ($last_syllable != $syllable && $syllable_type == self::BROKEN_CLUSTER) {
381 - $last_syllable = $syllable;
382 -
383 - $dottedcircle[0]['syllable'] = $info[$idx]['syllable'];
384 -
385 - /* Insert dottedcircle after possible Repha. */
386 - while ($idx < count($info) && $last_syllable == $info[$idx]['syllable'] && $info[$idx]['indic_category'] == self::OT_Repha)
387 - $idx++;
388 - array_splice($info, $idx, 0, $dottedcircle);
389 - } else {
390 - $idx++;
391 - }
392 - }
393 -
394 - // I am not sue how this code below got in here, since $idx should now be > count($info) and thus invalid.
395 - // In case I am missing something(!) I'll leave a warning here for now:
396 - if (isset($info[$idx])) {
397 - throw new MpdfException('Unexpected error occured in Indic processing');
398 - }
399 - // In case of final bloken cluster...
400 - //$syllable = $info[$idx]['syllable'];
401 - //$syllable_type = ($syllable & 0x0F);
402 - //if ($last_syllable != $syllable && $syllable_type == self::BROKEN_CLUSTER) {
403 - // $dottedcircle[0]['syllable'] = $info[$idx]['syllable'];
404 - // array_splice($info, $idx, 0, $dottedcircle);
405 - //}
406 - }
407 -
408 - /* Rules from:
409 - * https://www.microsoft.com/typography/otfntdev/devanot/shaping.aspx */
410 -
411 - public static function initial_reordering_syllable(&$info, $GSUBdata, $indic_config, $scriptblock, $is_old_spec, $start, $end)
412 - {
413 - /* vowel_syllable: We made the vowels look like consonants. So uses the consonant logic! */
414 - /* broken_cluster: We already inserted dotted-circles, so just call the standalone_cluster. */
415 - /* standalone_cluster: We treat NBSP/dotted-circle as if they are consonants, so we should just chain. */
416 -
417 - $syllable_type = ($info[$start]['syllable'] & 0x0F);
418 - if ($syllable_type == self::NON_INDIC_CLUSTER) {
419 - return;
420 - }
421 - if ($syllable_type == self::BROKEN_CLUSTER || $syllable_type == self::STANDALONE_CLUSTER) {
422 - //if ($uniscribe_bug_compatible) {
423 - /* For dotted-circle, this is what Uniscribe does:
424 - * If dotted-circle is the last glyph, it just does nothing.
425 - * i.e. It doesn't form Reph. */
426 - if ($info[$end - 1]['indic_category'] == self::OT_DOTTEDCIRCLE) {
427 - return;
428 - }
429 - }
430 -
431 - /* 1. Find base consonant:
432 - *
433 - * The shaping engine finds the base consonant of the syllable, using the
434 - * following algorithm: starting from the end of the syllable, move backwards
435 - * until a consonant is found that does not have a below-base or post-base
436 - * form (post-base forms have to follow below-base forms), or that is not a
437 - * pre-base reordering Ra, or arrive at the first consonant. The consonant
438 - * stopped at will be the base.
439 - *
440 - * o If the syllable starts with Ra + Halant (in a script that has Reph)
441 - * and has more than one consonant, Ra is excluded from candidates for
442 - * base consonants.
443 - */
444 -
445 - $base = $end;
446 - $has_reph = false;
447 - $limit = $start;
448 -
449 - if ($scriptblock != UCDN::SCRIPT_KHMER) {
450 - /* -> If the syllable starts with Ra + Halant (in a script that has Reph)
451 - * and has more than one consonant, Ra is excluded from candidates for
452 - * base consonants. */
453 - if (count($GSUBdata['rphf']) /* ?? $indic_plan->mask_array[RPHF] */ && $start + 3 <= $end &&
454 - (
455 - ($indic_config[4] == self::REPH_MODE_IMPLICIT && !self::is_joiner($info[$start + 2])) ||
456 - ($indic_config[4] == self::REPH_MODE_EXPLICIT && $info[$start + 2]['indic_category'] == self::OT_ZWJ)
457 - )) {
458 - /* See if it matches the 'rphf' feature. */
459 - //$glyphs = array($info[$start]['uni'], $info[$start + 1]['uni']);
460 - //if ($indic_plan->rphf->would_substitute ($glyphs, count($glyphs), true, face)) {
461 - if (isset($GSUBdata['rphf'][$info[$start]['uni']]) && self::is_halant_or_coeng($info[$start + 1])) {
462 - $limit += 2;
463 - while ($limit < $end && self::is_joiner($info[$limit]))
464 - $limit++;
465 - $base = $start;
466 - $has_reph = true;
467 - }
468 - } else if ($indic_config[4] == self::REPH_MODE_LOG_REPHA && $info[$start]['indic_category'] == self::OT_Repha) {
469 - $limit += 1;
470 - while ($limit < $end && self::is_joiner($info[$limit]))
471 - $limit++;
472 - $base = $start;
473 - $has_reph = true;
474 - }
475 - }
476 -
477 - switch ($indic_config[2]) { // base_pos
478 - case self::BASE_POS_LAST:
479 - /* -> starting from the end of the syllable, move backwards */
480 - $i = $end;
481 - $seen_below = false;
482 - do {
483 - $i--;
484 - /* -> until a consonant is found */
485 - if (self::is_consonant($info[$i])) {
486 - /* -> that does not have a below-base or post-base form
487 - * (post-base forms have to follow below-base forms), */
488 - if ($info[$i]['indic_position'] != self::POS_BELOW_C && ($info[$i]['indic_position'] != self::POS_POST_C || $seen_below)) {
489 - $base = $i;
490 - break;
491 - }
492 - if ($info[$i]['indic_position'] == self::POS_BELOW_C)
493 - $seen_below = true;
494 -
495 - /* -> or that is not a pre-base reordering Ra,
496 - *
497 - * IMPLEMENTATION NOTES:
498 - *
499 - * Our pre-base reordering Ra's are marked POS_POST_C, so will be skipped
500 - * by the logic above already.
501 - */
502 -
503 - /* -> or arrive at the first consonant. The consonant stopped at will
504 - * be the base. */
505 - $base = $i;
506 - }
507 - else {
508 - /* A ZWJ after a Halant stops the base search, and requests an explicit
509 - * half form.
510 - * [A ZWJ before a Halant, requests a subjoined form instead, and hence
511 - * search continues. This is particularly important for Bengali
512 - * sequence Ra,H,Ya that should form Ya-Phalaa by subjoining Ya] */
513 - if ($start < $i && $info[$i]['indic_category'] == self::OT_ZWJ && $info[$i - 1]['indic_category'] == self::OT_H) {
514 - if (!defined("OMIT_INDIC_FIX_1") || OMIT_INDIC_FIX_1 != 1) {
515 - $base = $i;
516 - } // INDIC_FIX_1
517 - break;
518 - }
519 - // ZKI8
520 - if ($start < $i && $info[$i]['indic_category'] == self::OT_ZWNJ) {
521 - break;
522 - }
523 - }
524 - } while ($i > $limit);
525 - break;
526 -
527 - case self::BASE_POS_FIRST:
528 - /* In scripts without half forms (eg. Khmer), the first consonant is always the base. */
529 -
530 - if (!$has_reph)
531 - $base = $limit;
532 -
533 - /* Find the last base consonant that is not blocked by ZWJ. If there is
534 - * a ZWJ right before a base consonant, that would request a subjoined form. */
535 - for ($i = $limit; $i < $end; $i++) {
536 - if (self::is_consonant($info[$i]) && $info[$i]['indic_position'] == self::POS_BASE_C) {
537 - if ($limit < $i && $info[$i - 1]['indic_category'] == self::OT_ZWJ)
538 - break;
539 - else
540 - $base = $i;
541 - }
542 - }
543 -
544 - /* Mark all subsequent consonants as below. */
545 - for ($i = $base + 1; $i < $end; $i++) {
546 - if (self::is_consonant($info[$i]) && $info[$i]['indic_position'] == self::POS_BASE_C)
547 - $info[$i]['indic_position'] = self::POS_BELOW_C;
548 - }
549 - break;
550 - //default:
551 - //assert (false);
552 - /* fallthrough */
553 - }
554 -
555 - /* -> If the syllable starts with Ra + Halant (in a script that has Reph)
556 - * and has more than one consonant, Ra is excluded from candidates for
557 - * base consonants.
558 - *
559 - * Only do this for unforced Reph. (ie. not for Ra,H,ZWJ. */
560 - if ($scriptblock != UCDN::SCRIPT_KHMER) {
561 - if ($has_reph && $base == $start && $limit - $base <= 2) {
562 - /* Have no other consonant, so Reph is not formed and Ra becomes base. */
563 - $has_reph = false;
564 - }
565 - }
566 -
567 - /* 2. Decompose and reorder Matras:
568 - *
569 - * Each matra and any syllable modifier sign in the cluster are moved to the
570 - * appropriate position relative to the consonant(s) in the cluster. The
571 - * shaping engine decomposes two- or three-part matras into their constituent
572 - * parts before any repositioning. Matra characters are classified by which
573 - * consonant in a conjunct they have affinity for and are reordered to the
574 - * following positions:
575 - *
576 - * o Before first half form in the syllable
577 - * o After subjoined consonants
578 - * o After post-form consonant
579 - * o After main consonant (for above marks)
580 - *
581 - * IMPLEMENTATION NOTES:
582 - *
583 - * The normalize() routine has already decomposed matras for us, so we don't
584 - * need to worry about that.
585 - */
586 -
587 -
588 - /* 3. Reorder marks to canonical order:
589 - *
590 - * Adjacent nukta and halant or nukta and vedic sign are always repositioned
591 - * if necessary, so that the nukta is first.
592 - *
593 - * IMPLEMENTATION NOTES:
594 - *
595 - * Use the combining Class from Unicode categories? to bubble_sort.
596 - */
597 -
598 - /* Reorder characters */
599 -
600 - for ($i = $start; $i < $base; $i++)
601 - $info[$i]['indic_position'] = min(self::POS_PRE_C, $info[$i]['indic_position']);
602 -
603 - if ($base < $end)
604 - $info[$base]['indic_position'] = self::POS_BASE_C;
605 -
606 - /* Mark final consonants. A final consonant is one appearing after a matra,
607 - * ? only in Khmer. */
608 - for ($i = $base + 1; $i < $end; $i++)
609 - if ($info[$i]['indic_category'] == self::OT_M) {
610 - for ($j = $i + 1; $j < $end; $j++)
611 - if (self::is_consonant($info[$j])) {
612 - $info[$j]['indic_position'] = self::POS_FINAL_C;
613 - break;
614 - }
615 - break;
616 - }
617 -
618 - /* Handle beginning Ra */
619 - if ($scriptblock != UCDN::SCRIPT_KHMER) {
620 - if ($has_reph)
621 - $info[$start]['indic_position'] = self::POS_RA_TO_BECOME_REPH;
622 - }
623 -
624 -
625 - /* For old-style Indic script tags, move the first post-base Halant after
626 - * last consonant. Only do this if there is *not* a Halant after last
627 - * consonant. Otherwise it becomes messy. */
628 - if ($is_old_spec) {
629 - for ($i = $base + 1; $i < $end; $i++) {
630 - if ($info[$i]['indic_category'] == self::OT_H) {
631 - for ($j = $end - 1; $j > $i; $j--) {
632 - if (self::is_consonant($info[$j]) || $info[$j]['indic_category'] == self::OT_H) {
633 - break;
634 - }
635 - }
636 - if ($info[$j]['indic_category'] != self::OT_H && $j > $i) {
637 - /* Move Halant to after last consonant. */
638 - self::_move_info_pos($info, $i, $j + 1);
639 - }
640 - break;
641 - }
642 - }
643 - }
644 -
645 - /* Attach misc marks to previous char to move with them. */
646 - $last_pos = self::POS_START;
647 - for ($i = $start; $i < $end; $i++) {
648 - if ((self::FLAG($info[$i]['indic_category']) & (self::FLAG(self::OT_ZWJ) | self::FLAG(self::OT_ZWNJ) | self::FLAG(self::OT_N) | self::FLAG(self::OT_RS) | self::FLAG(self::OT_H) | self::FLAG(self::OT_Coeng) ))) {
649 - $info[$i]['indic_position'] = $last_pos;
650 - if ($info[$i]['indic_category'] == self::OT_H && $info[$i]['indic_position'] == self::POS_PRE_M) {
651 - /*
652 - * Uniscribe doesn't move the Halant with Left Matra.
653 - * TEST: U+092B,U+093F,U+094DE
654 - * We follow. This is important for the Sinhala
655 - * U+0DDA split matra since it decomposes to U+0DD9,U+0DCA
656 - * where U+0DD9 is a left matra and U+0DCA is the virama.
657 - * We don't want to move the virama with the left matra.
658 - * TEST: U+0D9A,U+0DDA
659 - */
660 - for ($j = $i; $j > $start; $j--)
661 - if ($info[$j - 1]['indic_position'] != self::POS_PRE_M) {
662 - $info[$i]['indic_position'] = $info[$j - 1]['indic_position'];
663 - break;
664 - }
665 - }
666 - } else if ($info[$i]['indic_position'] != self::POS_SMVD) {
667 - $last_pos = $info[$i]['indic_position'];
668 - }
669 - }
670 -
671 - /* Re-attach ZWJ, ZWNJ, and halant to next char, for after-base consonants. */
672 - $last_halant = $end;
673 - for ($i = $base + 1; $i < $end; $i++) {
674 - if (self::is_halant_or_coeng($info[$i]))
675 - $last_halant = $i;
676 - else if (self::is_consonant($info[$i])) {
677 - for ($j = $last_halant; $j < $i; $j++)
678 - if ($info[$j]['indic_position'] != self::POS_SMVD)
679 - $info[$j]['indic_position'] = $info[$i]['indic_position'];
680 - }
681 - }
682 -
683 -
684 - if ($scriptblock == UCDN::SCRIPT_KHMER) {
685 - /* KHMER_FIX_2 */
686 - /* Move Coeng+RO (Halant,Ra) sequence before base consonant. */
687 - for ($i = $base + 1; $i < $end; $i++) {
688 - if (self::is_halant_or_coeng($info[$i]) && self::is_ra($info[$i + 1]['uni'])) {
689 - $info[$i]['indic_position'] = self::POS_PRE_C;
690 - $info[$i + 1]['indic_position'] = self::POS_PRE_C;
691 - break;
692 - }
693 - }
694 - }
695 -
696 -
697 - /*
698 - if (!defined("OMIT_INDIC_FIX_2") || OMIT_INDIC_FIX_2 != 1) {
699 - // INDIC_FIX_2
700 - $ZWNJ_found = false;
701 - $POST_ZWNJ_c_found = false;
702 - for ($i = $base + 1; $i < $end; $i++) {
703 - if ($info[$i]['indic_category'] == self::OT_ZWNJ) { $ZWNJ_found = true; }
704 - else if ($ZWNJ_found && $info[$i]['indic_category'] == self::OT_C) { $POST_ZWNJ_c_found = true; }
705 - else if ($POST_ZWNJ_c_found && $info[$i]['indic_position'] == self::POS_BEFORE_SUB) { $info[$i]['indic_position'] = self::POS_AFTER_SUB; }
706 - }
707 - }
708 - */
709 -
710 - /* Setup masks now */
711 - for ($i = $start; $i < $end; $i++) {
712 - $info[$i]['mask'] = 0;
713 - }
714 -
715 -
716 - if ($scriptblock == UCDN::SCRIPT_KHMER) {
717 - /* Find a Coeng+RO (Halant,Ra) sequence and mark it for pre-base processing. */
718 - $mask = self::FLAG(self::PREF);
719 - for ($i = $base; $i < $end - 1; $i++) { /* KHMER_FIX_1 From $start (not base) */
720 - if (self::is_halant_or_coeng($info[$i]) && self::is_ra($info[$i + 1]['uni'])) {
721 -
722 - $info[$i]['mask'] |= self::FLAG(self::PREF);
723 - $info[$i + 1]['mask'] |= self::FLAG(self::PREF);
724 -
725 - /* Mark the subsequent stuff with 'cfar'. Used in Khmer.
726 - * Read the feature spec.
727 - * This allows distinguishing the following cases with MS Khmer fonts:
728 - * U+1784,U+17D2,U+179A,U+17D2,U+1782 [C+Coeng+RO+Coeng+C] => Should activate CFAR
729 - * U+1784,U+17D2,U+1782,U+17D2,U+179A [C+Coeng+C+Coeng+RO] => Should NOT activate CFAR
730 - */
731 - for ($j = ($i + 2); $j < $end; $j++)
732 - $info[$j]['mask'] |= self::FLAG(self::CFAR);
733 -
734 - break;
735 - }
736 - }
737 - }
738 -
739 -
740 -
741 - /* Sit tight, rock 'n roll! */
742 - self::bubble_sort($info, $start, $end - $start);
743 -
744 - /* Find base again */
745 - $base = $end;
746 - for ($i = $start; $i < $end; $i++) {
747 - if ($info[$i]['indic_position'] == self::POS_BASE_C) {
748 - $base = $i;
749 - break;
750 - }
751 - }
752 -
753 - if ($scriptblock != UCDN::SCRIPT_KHMER) {
754 - /* Reph */
755 - for ($i = $start; $i < $end; $i++) {
756 - if ($info[$i]['indic_position'] == self::POS_RA_TO_BECOME_REPH) {
757 - $info[$i]['mask'] |= self::FLAG(self::RPHF);
758 - }
759 - }
760 -
761 - /* Pre-base */
762 - $mask = self::FLAG(self::HALF);
763 - for ($i = $start; $i < $base; $i++) {
764 - $info[$i]['mask'] |= $mask;
765 - }
766 - }
767 -
768 - /* Post-base */
769 - $mask = (self::FLAG(self::BLWF) | self::FLAG(self::ABVF) | self::FLAG(self::PSTF));
770 - for ($i = $base + 1; $i < $end; $i++) {
771 - $info[$i]['mask'] |= $mask;
772 - }
773 -
774 -
775 - if ($scriptblock != UCDN::SCRIPT_KHMER) {
776 - if (!defined("OMIT_INDIC_FIX_3") || OMIT_INDIC_FIX_3 != 1) {
777 - /* INDIC_FIX_3 */
778 - /* Find a (pre-base) Consonant, Halant,Ra sequence and mark Halant|Ra for below-base BLWF processing. */
779 - // TEST CASE &#x995;&#x9cd;&#x9b0;&#x9cd;&#x995; in FreeSans versus Vrinda
780 - if (($base - $start) >= 3) {
781 - for ($i = $start; $i < ($base - 2); $i++) {
782 - if (self::is_consonant($info[$i])) {
783 - if (self::is_halant_or_coeng($info[$i + 1]) && self::is_ra($info[$i + 2]['uni'])) {
784 - // If would substitute Halant+Ra...BLWF
785 - if (isset($GSUBdata['blwf'][$info[$i + 2]['uni']])) {
786 - $info[$i + 1]['mask'] |= self::FLAG(self::BLWF);
787 - $info[$i + 2]['mask'] |= self::FLAG(self::BLWF);
788 - }
789 - /* If would not substitute as blwf, mark Ra+Halant for RPHF using following Halant (if present) */ else if (self::is_halant_or_coeng($info[$i + 3])) {
790 - $info[$i + 2]['mask'] |= self::FLAG(self::RPHF);
791 - $info[$i + 3]['mask'] |= self::FLAG(self::RPHF);
792 - }
793 - break;
794 - }
795 - }
796 - }
797 - }
798 - }
799 - }
800 -
801 -
802 -
803 - if ($is_old_spec && $scriptblock == UCDN::SCRIPT_DEVANAGARI) {
804 - /* Old-spec eye-lash Ra needs special handling. From the spec:
805 - * "The feature 'below-base form' is applied to consonants
806 - * having below-base forms and following the base consonant.
807 - * The exception is vattu, which may appear below half forms
808 - * as well as below the base glyph. The feature 'below-base
809 - * form' will be applied to all such occurrences of Ra as well."
810 - *
811 - * Test case: U+0924,U+094D,U+0930,U+094d,U+0915
812 - * with Sanskrit 2003 font.
813 - *
814 - * However, note that Ra,Halant,ZWJ is the correct way to
815 - * request eyelash form of Ra, so we wouldbn't inhibit it
816 - * in that sequence.
817 - *
818 - * Test case: U+0924,U+094D,U+0930,U+094d,U+200D,U+0915
819 - */
820 - for ($i = $start; ($i + 1) < $base; $i++) {
821 - if ($info[$i]['indic_category'] == self::OT_Ra && $info[$i + 1]['indic_category'] == self::OT_H &&
822 - ($i + 2 == $base || $info[$i + 2]['indic_category'] != self::OT_ZWJ)) {
823 - $info[$i]['mask'] |= self::FLAG(self::BLWF);
824 - $info[$i + 1]['mask'] |= self::FLAG(self::BLWF);
825 - }
826 - }
827 - }
828 -
829 - if ($scriptblock != UCDN::SCRIPT_KHMER) {
830 - if (count($GSUBdata['pref']) && $base + 2 < $end) {
831 - /* Find a Halant,Ra sequence and mark it for pre-base processing. */
832 - for ($i = $base + 1; $i + 1 < $end; $i++) {
833 - // If old_spec find Ra-Halant...
834 - if ((isset($GSUBdata['pref'][$info[$i + 1]['uni']]) && self::is_halant_or_coeng($info[$i]) && self::is_ra($info[$i + 1]['uni']) ) ||
835 - ($is_old_spec && isset($GSUBdata['pref'][$info[$i]['uni']]) && self::is_halant_or_coeng($info[$i + 1]) && self::is_ra($info[$i]['uni']) )
836 - ) {
837 - $info[$i++]['mask'] |= self::FLAG(self::PREF);
838 - $info[$i++]['mask'] |= self::FLAG(self::PREF);
839 - break;
840 - }
841 - }
842 - }
843 - }
844 -
845 -
846 - /* Apply ZWJ/ZWNJ effects */
847 - for ($i = $start + 1; $i < $end; $i++) {
848 - if (self::is_joiner($info[$i])) {
849 - $non_joiner = ($info[$i]['indic_category'] == self::OT_ZWNJ);
850 - $j = $i;
851 - while ($j > $start) {
852 - if (defined("OMIT_INDIC_FIX_4") && OMIT_INDIC_FIX_4 == 1) {
853 - // INDIC_FIX_4 = do nothing - carry on //
854 - // ZWNJ should block H C from forming blwf post-base - need to unmask backwards beyond first consonant arrived at //
855 - if (!self::is_consonant($info[$j])) {
856 - break;
857 - }
858 - }
859 - $j--;
860 -
861 - /* ZWJ/ZWNJ should disable CJCT. They do that by simply
862 - * being there, since we don't skip them for the CJCT
863 - * feature (ie. F_MANUAL_ZWJ) */
864 -
865 - /* A ZWNJ disables HALF. */
866 - if ($non_joiner) {
867 - $info[$j]['mask'] &= ~(self::FLAG(self::HALF) | self::FLAG(self::BLWF));
868 - }
869 - }
870 - }
871 - }
872 - }
873 -
874 - public static function final_reordering(&$info, $GSUBdata, $indic_config, $scriptblock, $is_old_spec)
875 - {
876 - $count = count($info);
877 - if (!$count)
878 - return;
879 - $last = 0;
880 - $last_syllable = $info[0]['syllable'];
881 - for ($i = 1; $i < $count; $i++) {
882 - if ($last_syllable != $info[$i]['syllable']) {
883 - self::final_reordering_syllable($info, $GSUBdata, $indic_config, $scriptblock, $is_old_spec, $last, $i);
884 - $last = $i;
885 - $last_syllable = $info[$last]['syllable'];
886 - }
887 - }
888 - self::final_reordering_syllable($info, $GSUBdata, $indic_config, $scriptblock, $is_old_spec, $last, $count);
889 - }
890 -
891 - public static function final_reordering_syllable(&$info, $GSUBdata, $indic_config, $scriptblock, $is_old_spec, $start, $end)
892 - {
893 -
894 - /* 4. Final reordering:
895 - *
896 - * After the localized forms and basic shaping forms GSUB features have been
897 - * applied (see below), the shaping engine performs some final glyph
898 - * reordering before applying all the remaining font features to the entire
899 - * cluster.
900 - */
901 -
902 - /* Find base again */
903 - for ($base = $start; $base < $end; $base++)
904 - if ($info[$base]['indic_position'] >= self::POS_BASE_C) {
905 - if ($start < $base && $info[$base]['indic_position'] > self::POS_BASE_C)
906 - $base--;
907 - break;
908 - }
909 - if ($base == $end && $start < $base && $info[$base - 1]['indic_category'] != self::OT_ZWJ)
910 - $base--;
911 - while ($start < $base && isset($info[$base]) && ($info[$base]['indic_category'] == self::OT_H || $info[$base]['indic_category'] == self::OT_N))
912 - $base--;
913 -
914 -
915 - /* o Reorder matras:
916 - *
917 - * If a pre-base matra character had been reordered before applying basic
918 - * features, the glyph can be moved closer to the main consonant based on
919 - * whether half-forms had been formed. Actual position for the matra is
920 - * defined as "after last standalone halant glyph, after initial matra
921 - * position and before the main consonant". If ZWJ or ZWNJ follow this
922 - * halant, position is moved after it.
923 - */
924 -
925 -
926 - if ($start + 1 < $end && $start < $base) { /* Otherwise there can't be any pre-base matra characters. */
927 - /* If we lost track of base, alas, position before last thingy. */
928 - $new_pos = ($base == $end) ? $base - 2 : $base - 1;
929 -
930 - /* Malayalam / Tamil do not have "half" forms or explicit virama forms.
931 - * The glyphs formed by 'half' are Chillus or ligated explicit viramas.
932 - * We want to position matra after them.
933 - */
934 - if ($scriptblock != UCDN::SCRIPT_MALAYALAM && $scriptblock != UCDN::SCRIPT_TAMIL) {
935 - while ($new_pos > $start && !(self::is_one_of($info[$new_pos], (self::FLAG(self::OT_M) | self::FLAG(self::OT_H) | self::FLAG(self::OT_Coeng)))))
936 - $new_pos--;
937 -
938 - /* If we found no Halant we are done.
939 - * Otherwise only proceed if the Halant does
940 - * not belong to the Matra itself! */
941 - if (self::is_halant_or_coeng($info[$new_pos]) && $info[$new_pos]['indic_position'] != self::POS_PRE_M) {
942 - /* -> If ZWJ or ZWNJ follow this halant, position is moved after it. */
943 - if ($new_pos + 1 < $end && self::is_joiner($info[$new_pos + 1]))
944 - $new_pos++;
945 - } else
946 - $new_pos = $start; /* No move. */
947 - }
948 -
949 - if ($start < $new_pos && $info[$new_pos]['indic_position'] != self::POS_PRE_M) {
950 - /* Now go see if there's actually any matras... */
951 - for ($i = $new_pos; $i > $start; $i--)
952 - if ($info[$i - 1]['indic_position'] == self::POS_PRE_M) {
953 - $old_pos = $i - 1;
954 - //memmove (&info[$old_pos], &info[$old_pos + 1], ($new_pos - $old_pos) * sizeof ($info[0]));
955 - self::_move_info_pos($info, $old_pos, $new_pos + 1);
956 -
957 - if ($old_pos < $base && $base <= $new_pos) /* Shouldn't actually happen. */
958 - $base--;
959 - $new_pos--;
960 - }
961 - }
962 - }
963 -
964 -
965 - /* o Reorder reph:
966 - *
967 - * Reph's original position is always at the beginning of the syllable,
968 - * (i.e. it is not reordered at the character reordering stage). However,
969 - * it will be reordered according to the basic-forms shaping results.
970 - * Possible positions for reph, depending on the script, are; after main,
971 - * before post-base consonant forms, and after post-base consonant forms.
972 - */
973 -
974 - /* If there's anything after the Ra that has the REPH pos, it ought to be halant.
975 - * Which means that the font has failed to ligate the Reph. In which case, we
976 - * shouldn't move. */
977 - if ($start + 1 < $end &&
978 - $info[$start]['indic_position'] == self::POS_RA_TO_BECOME_REPH && $info[$start + 1]['indic_position'] != self::POS_RA_TO_BECOME_REPH) {
979 - $reph_pos = $indic_config[3];
980 - $skip_to_reph_step_5 = false;
981 - $skip_to_reph_move = false;
982 -
983 - /* 1. If reph should be positioned after post-base consonant forms,
984 - * proceed to step 5.
985 - */
986 - if ($reph_pos == self::REPH_POS_AFTER_POST) {
987 - $skip_to_reph_step_5 = true;
988 - }
989 -
990 - /* 2. If the reph repositioning class is not after post-base: target
991 - * position is after the first explicit halant glyph between the
992 - * first post-reph consonant and last main consonant. If ZWJ or ZWNJ
993 - * are following this halant, position is moved after it. If such
994 - * position is found, this is the target position. Otherwise,
995 - * proceed to the next step.
996 - *
997 - * Note: in old-implementation fonts, where classifications were
998 - * fixed in shaping engine, there was no case where reph position
999 - * will be found on this step.
1000 - */
1001 -
1002 - if (!$skip_to_reph_step_5) {
1003 -
1004 - $new_reph_pos = $start + 1;
1005 -
1006 - while ($new_reph_pos < $base && !self::is_halant_or_coeng($info[$new_reph_pos]))
1007 - $new_reph_pos++;
1008 -
1009 - if ($new_reph_pos < $base && self::is_halant_or_coeng($info[$new_reph_pos])) {
1010 - /* ->If ZWJ or ZWNJ are following this halant, position is moved after it. */
1011 - if ($new_reph_pos + 1 < $base && self::is_joiner($info[$new_reph_pos + 1]))
1012 - $new_reph_pos++;
1013 - $skip_to_reph_move = true;
1014 - }
1015 - }
1016 -
1017 - /* 3. If reph should be repositioned after the main consonant: find the
1018 - * first consonant not ligated with main, or find the first
1019 - * consonant that is not a potential pre-base reordering Ra.
1020 - */
1021 - if ($reph_pos == self::REPH_POS_AFTER_MAIN && !$skip_to_reph_move && !$skip_to_reph_step_5) {
1022 - $new_reph_pos = $base;
1023 - /* XXX Skip potential pre-base reordering Ra. */
1024 - while ($new_reph_pos + 1 < $end && $info[$new_reph_pos + 1]['indic_position'] <= self::POS_AFTER_MAIN)
1025 - $new_reph_pos++;
1026 - if ($new_reph_pos < $end)
1027 - $skip_to_reph_move = true;
1028 - }
1029 -
1030 - /* 4. If reph should be positioned before post-base consonant, find
1031 - * first post-base classified consonant not ligated with main. If no
1032 - * consonant is found, the target position should be before the
1033 - * first matra, syllable modifier sign or vedic sign.
1034 - */
1035 - /* This is our take on what step 4 is trying to say (and failing, BADLY). */
1036 - if ($reph_pos == self::REPH_POS_AFTER_SUB && !$skip_to_reph_move && !$skip_to_reph_step_5) {
1037 - $new_reph_pos = $base;
1038 - while ($new_reph_pos < $end && isset($info[$new_reph_pos + 1]['indic_position']) &&
1039 - !( self::FLAG($info[$new_reph_pos + 1]['indic_position']) & (self::FLAG(self::POS_POST_C) | self::FLAG(self::POS_AFTER_POST) | self::FLAG(self::POS_SMVD)))) {
1040 - $new_reph_pos++;
1041 - }
1042 - if ($new_reph_pos < $end) {
1043 - $skip_to_reph_move = true;
1044 - }
1045 - }
1046 -
1047 - /* 5. If no consonant is found in steps 3 or 4, move reph to a position
1048 - * immediately before the first post-base matra, syllable modifier
1049 - * sign or vedic sign that has a reordering class after the intended
1050 - * reph position. For example, if the reordering position for reph
1051 - * is post-main, it will skip above-base matras that also have a
1052 - * post-main position.
1053 - */
1054 - if (!$skip_to_reph_move) {
1055 - /* Copied from step 2. */
1056 - $new_reph_pos = $start + 1;
1057 - while ($new_reph_pos < $base && !self::is_halant_or_coeng($info[$new_reph_pos]))
1058 - $new_reph_pos++;
1059 -
1060 - if ($new_reph_pos < $base && self::is_halant_or_coeng($info[$new_reph_pos])) {
1061 - /* ->If ZWJ or ZWNJ are following this halant, position is moved after it. */
1062 - if ($new_reph_pos + 1 < $base && self::is_joiner($info[$new_reph_pos + 1]))
1063 - $new_reph_pos++;
1064 - $skip_to_reph_move = true;
1065 - }
1066 - }
1067 -
1068 -
1069 - /* 6. Otherwise, reorder reph to the end of the syllable.
1070 - */
1071 - if (!$skip_to_reph_move) {
1072 - $new_reph_pos = $end - 1;
1073 - while ($new_reph_pos > $start && $info[$new_reph_pos]['indic_position'] == self::POS_SMVD)
1074 - $new_reph_pos--;
1075 -
1076 - /*
1077 - * If the Reph is to be ending up after a Matra,Halant sequence,
1078 - * position it before that Halant so it can interact with the Matra.
1079 - * However, if it's a plain Consonant,Halant we shouldn't do that.
1080 - * Uniscribe doesn't do this.
1081 - * TEST: U+0930,U+094D,U+0915,U+094B,U+094D
1082 - */
1083 - //if (!$hb_options.uniscribe_bug_compatible && self::is_halant_or_coeng($info[$new_reph_pos])) {
1084 - if (self::is_halant_or_coeng($info[$new_reph_pos])) {
1085 - for ($i = $base + 1; $i < $new_reph_pos; $i++)
1086 - if ($info[$i]['indic_category'] == self::OT_M) {
1087 - /* Ok, got it. */
1088 - $new_reph_pos--;
1089 - }
1090 - }
1091 - }
1092 -
1093 -
1094 - /* Move */
1095 - self::_move_info_pos($info, $start, $new_reph_pos + 1);
1096 -
1097 - if ($start < $base && $base <= $new_reph_pos) {
1098 - $base--;
1099 - }
1100 - }
1101 -
1102 -
1103 - /* o Reorder pre-base reordering consonants:
1104 - *
1105 - * If a pre-base reordering consonant is found, reorder it according to
1106 - * the following rules:
1107 - */
1108 -
1109 -
1110 - if (count($GSUBdata['pref']) && $base + 1 < $end) { /* Otherwise there can't be any pre-base reordering Ra. */
1111 - for ($i = $base + 1; $i < $end; $i++) {
1112 - if ($info[$i]['mask'] & self::FLAG(self::PREF)) {
1113 - /* 1. Only reorder a glyph produced by substitution during application
1114 - * of the <pref> feature. (Note that a font may shape a Ra consonant with
1115 - * the feature generally but block it in certain contexts.)
1116 - */
1117 -// ??? Need to TEST if actual substitution has occurred
1118 - if ($i + 1 == $end || ($info[$i + 1]['mask'] & self::FLAG(self::PREF)) == 0) {
1119 - /*
1120 - * 2. Try to find a target position the same way as for pre-base matra.
1121 - * If it is found, reorder pre-base consonant glyph.
1122 - *
1123 - * 3. If position is not found, reorder immediately before main
1124 - * consonant.
1125 - */
1126 - $new_pos = $base;
1127 - /* Malayalam / Tamil do not have "half" forms or explicit virama forms.
1128 - * The glyphs formed by 'half' are Chillus or ligated explicit viramas.
1129 - * We want to position matra after them.
1130 - */
1131 - if ($scriptblock != UCDN::SCRIPT_MALAYALAM && $scriptblock != UCDN::SCRIPT_TAMIL) {
1132 - while ($new_pos > $start &&
1133 - !(self::is_one_of($info[$new_pos - 1], self::FLAG(self::OT_M) | self::FLAG(self::OT_H) | self::FLAG(self::OT_Coeng))))
1134 - $new_pos--;
1135 -
1136 - /* In Khmer coeng model, a V,Ra can go *after* matras. If it goes after a
1137 - * split matra, it should be reordered to *before* the left part of such matra. */
1138 - if ($new_pos > $start && $info[$new_pos - 1]['indic_category'] == self::OT_M) {
1139 - $old_pos = i;
1140 - for ($i = $base + 1; $i < $old_pos; $i++)
1141 - if ($info[$i]['indic_category'] == self::OT_M) {
1142 - $new_pos--;
1143 - break;
1144 - }
1145 - }
1146 - }
1147 -
1148 - if ($new_pos > $start && self::is_halant_or_coeng($info[$new_pos - 1])) {
1149 - /* -> If ZWJ or ZWNJ follow this halant, position is moved after it. */
1150 - if ($new_pos < $end && self::is_joiner($info[$new_pos]))
1151 - $new_pos++;
1152 - }
1153 -
1154 - $old_pos = $i;
1155 - self::_move_info_pos($info, $old_pos, $new_pos);
1156 -
1157 - if ($new_pos <= $base && $base < $old_pos)
1158 - $base++;
1159 - }
1160 -
1161 - break;
1162 - }
1163 - }
1164 - }
1165 -
1166 -
1167 - /* Apply 'init' to the Left Matra if it's a word start. */
1168 - if ($info[$start]['indic_position'] == self::POS_PRE_M &&
1169 - ($start == 0 ||
1170 - ($info[$start - 1]['general_category'] < UCDN::UNICODE_GENERAL_CATEGORY_FORMAT || $info[$start - 1]['general_category'] > UCDN::UNICODE_GENERAL_CATEGORY_NON_SPACING_MARK)
1171 - )) {
1172 - $info[$start]['mask'] |= self::FLAG(self::INIT);
1173 - }
1174 -
1175 -
1176 - /*
1177 - * Finish off and go home!
1178 - */
1179 - }
1180 -
1181 - public static function _move_info_pos(&$info, $from, $to)
1182 - {
1183 - $t = array();
1184 - $t[0] = $info[$from];
1185 - if ($from > $to) {
1186 - array_splice($info, $from, 1);
1187 - array_splice($info, $to, 0, $t);
1188 - } else {
1189 - array_splice($info, $to, 0, $t);
1190 - array_splice($info, $from, 1);
1191 - }
1192 - }
1193 -
1194 - public static $ra_chars = array(
1195 - 0x0930 => 1, /* Devanagari */
1196 - 0x09B0 => 1, /* Bengali */
1197 - 0x09F0 => 1, /* Bengali (Assamese) */
1198 - 0x0A30 => 1, /* Gurmukhi */ /* No Reph */
1199 - 0x0AB0 => 1, /* Gujarati */
1200 - 0x0B30 => 1, /* Oriya */
1201 - 0x0BB0 => 1, /* Tamil */ /* No Reph */
1202 - 0x0C30 => 1, /* Telugu */ /* Reph formed only with ZWJ */
1203 - 0x0CB0 => 1, /* Kannada */
1204 - 0x0D30 => 1, /* Malayalam */ /* No Reph, Logical Repha */
1205 - 0x0DBB => 1, /* Sinhala */ /* Reph formed only with ZWJ */
1206 - 0x179A => 1, /* Khmer */ /* No Reph, Visual Repha */
1207 - );
1208 -
1209 - public static function is_ra($u)
1210 - {
1211 - if (isset(self::$ra_chars[$u]))
1212 - return true;
1213 - return false;
1214 - }
1215 -
1216 - public static function is_one_of($info, $flags)
1217 - {
1218 - if (isset($info['is_ligature']) && $info['is_ligature'])
1219 - return false; /* If it ligated, all bets are off. */
1220 - return !!(self::FLAG($info['indic_category']) & $flags);
1221 - }
1222 -
1223 - public static function is_joiner($info)
1224 - {
1225 - return self::is_one_of($info, (self::FLAG(self::OT_ZWJ) | self::FLAG(self::OT_ZWNJ)));
1226 - }
1227 -
1228 - /* Vowels and placeholders treated as if they were consonants. */
1229 -
1230 - public static function is_consonant($info)
1231 - {
1232 - return self::is_one_of($info, (self::FLAG(self::OT_C) | self::FLAG(self::OT_CM) | self::FLAG(self::OT_Ra) | self::FLAG(self::OT_V) | self::FLAG(self::OT_NBSP) | self::FLAG(self::OT_DOTTEDCIRCLE)));
1233 - }
1234 -
1235 - public static function is_halant_or_coeng($info)
1236 - {
1237 - return self::is_one_of($info, (self::FLAG(self::OT_H) | self::FLAG(self::OT_Coeng)));
1238 - }
1239 -
1240 - // From hb-private.hh
1241 - public static function in_range($u, $lo, $hi)
1242 - {
1243 - if ((($lo ^ $hi) & $lo) == 0 && (($lo ^ $hi) & $hi) == ($lo ^ $hi) && (($lo ^ $hi) & (($lo ^ $hi) + 1)) == 0)
1244 - return ($u & ~($lo ^ $hi)) == $lo;
1245 - else
1246 - return $lo <= $u && $u <= $hi;
1247 - }
1248 -
1249 - // From hb-private.hh
1250 - public static function FLAG($x)
1251 - {
1252 - return (1 << ($x));
1253 - }
1254 -
1255 - // BELOW from hb-ot-shape-complex-indic.cc
1256 -
1257 - /*
1258 - * Indic configurations.
1259 - */
1260 -
1261 - // base_position
1262 - const BASE_POS_FIRST = 0;
1263 - const BASE_POS_LAST = 1;
1264 -
1265 - // reph_position
1266 - const REPH_POS_DEFAULT = 10; // POS_BEFORE_POST,
1267 -
1268 - const REPH_POS_AFTER_MAIN = 5; // POS_AFTER_MAIN,
1269 -
1270 - const REPH_POS_BEFORE_SUB = 7; // POS_BEFORE_SUB,
1271 - const REPH_POS_AFTER_SUB = 9; // POS_AFTER_SUB,
1272 - const REPH_POS_BEFORE_POST = 10; // POS_BEFORE_POST,
1273 - const REPH_POS_AFTER_POST = 12; // POS_AFTER_POST
1274 -
1275 - // reph_mode
1276 - const REPH_MODE_IMPLICIT = 0; /* Reph formed out of initial Ra,H sequence. */
1277 - const REPH_MODE_EXPLICIT = 1; /* Reph formed out of initial Ra,H,ZWJ sequence. */
1278 - const REPH_MODE_VIS_REPHA = 2; /* Encoded Repha character, no reordering needed. */
1279 - const REPH_MODE_LOG_REPHA = 3; /* Encoded Repha character, needs reordering. */
1280 -
1281 - /*
1282 - struct of indic_configs{
1283 - KEY - script;
1284 - 0 - has_old_spec;
1285 - 1 - virama;
1286 - 2 - base_pos;
1287 - 3 - reph_pos;
1288 - 4 - reph_mode;
1289 - };
1290 - */
1291 -
1292 - public static $indic_configs = array(/* index is SCRIPT_number from UCDN */
1293 - 9 => array(true, 0x094D, 1, 10, 0),
1294 - 10 => array(true, 0x09CD, 1, 9, 0),
1295 - 11 => array(true, 0x0A4D, 1, 7, 0),
1296 - 12 => array(true, 0x0ACD, 1, 10, 0),
1297 - 13 => array(true, 0x0B4D, 1, 5, 0),
1298 - 14 => array(true, 0x0BCD, 1, 12, 0),
1299 - 15 => array(true, 0x0C4D, 1, 12, 1),
1300 - 16 => array(true, 0x0CCD, 1, 12, 0),
1301 - 17 => array(true, 0x0D4D, 1, 5, 3),
1302 - 18 => array(false, 0x0DCA, 0, 5, 1), /* Sinhala */
1303 - 30 => array(false, 0x17D2, 0, 10, 2), /* Khmer */
1304 - 84 => array(false, 0xA9C0, 1, 10, 0), /* Javanese */
1305 - );
1306 -
1307 -
1308 -
1309 - /*
1310 -
1311 - // from "hb-ot-shape-complex-indic-table.cc"
1312 -
1313 -
1314 - const ISC_A = 0; // INDIC_SYLLABIC_CATEGORY_AVAGRAHA Avagraha
1315 - const ISC_Bi = 8; // INDIC_SYLLABIC_CATEGORY_BINDU Bindu
1316 - const ISC_C = 1; // INDIC_SYLLABIC_CATEGORY_CONSONANT Consonant
1317 - const ISC_CD = 1; // INDIC_SYLLABIC_CATEGORY_CONSONANT_DEAD Consonant_Dead
1318 - const ISC_CF = 17; // INDIC_SYLLABIC_CATEGORY_CONSONANT_FINAL Consonant_Final
1319 - const ISC_CHL = 1; // INDIC_SYLLABIC_CATEGORY_CONSONANT_HEAD_LETTER Consonant_Head_Letter
1320 - const ISC_CM = 17; // INDIC_SYLLABIC_CATEGORY_CONSONANT_MEDIAL Consonant_Medial
1321 - const ISC_CP = 11; // INDIC_SYLLABIC_CATEGORY_CONSONANT_PLACEHOLDER Consonant_Placeholder
1322 - const ISC_CR = 15; // INDIC_SYLLABIC_CATEGORY_CONSONANT_REPHA Consonant_Repha
1323 - const ISC_CS = 1; // INDIC_SYLLABIC_CATEGORY_CONSONANT_SUBJOINED Consonant_Subjoined
1324 - const ISC_ML = 0; // INDIC_SYLLABIC_CATEGORY_MODIFYING_LETTER Modifying_Letter
1325 - const ISC_N = 3; // INDIC_SYLLABIC_CATEGORY_NUKTA Nukta
1326 - const ISC_x = 0; // INDIC_SYLLABIC_CATEGORY_OTHER Other
1327 - const ISC_RS = 13; // INDIC_SYLLABIC_CATEGORY_REGISTER_SHIFTER Register_Shifter
1328 - const ISC_TL = 0; // INDIC_SYLLABIC_CATEGORY_TONE_LETTER Tone_Letter
1329 - const ISC_TM = 3; // INDIC_SYLLABIC_CATEGORY_TONE_MARK Tone_Mark
1330 - const ISC_V = 4; // INDIC_SYLLABIC_CATEGORY_VIRAMA Virama
1331 - const ISC_Vs = 8; // INDIC_SYLLABIC_CATEGORY_VISARGA Visarga
1332 - const ISC_Vo = 2; // INDIC_SYLLABIC_CATEGORY_VOWEL Vowel
1333 - const ISC_M = 7; // INDIC_SYLLABIC_CATEGORY_VOWEL_DEPENDENT Vowel_Dependent
1334 - const ISC_VI = 2; // INDIC_SYLLABIC_CATEGORY_VOWEL_INDEPENDENT Vowel_Independent
1335 -
1336 - const IMC_B = 8; // INDIC_MATRA_CATEGORY_BOTTOM Bottom
1337 - const IMC_BR = 11; // INDIC_MATRA_CATEGORY_BOTTOM_AND_RIGHT Bottom_And_Right
1338 - const IMC_I = 15; // INDIC_MATRA_CATEGORY_INVISIBLE Invisible
1339 - const IMC_L = 3; // INDIC_MATRA_CATEGORY_LEFT Left
1340 - const IMC_LR = 11; // INDIC_MATRA_CATEGORY_LEFT_AND_RIGHT Left_And_Right
1341 - const IMC_x = 15; // INDIC_MATRA_CATEGORY_NOT_APPLICABLE Not_Applicable
1342 - const IMC_O = 5; // INDIC_MATRA_CATEGORY_OVERSTRUCK Overstruck
1343 - const IMC_R = 11; // INDIC_MATRA_CATEGORY_RIGHT Right
1344 - const IMC_T = 6; // INDIC_MATRA_CATEGORY_TOP Top
1345 - const IMC_TB = 8; // INDIC_MATRA_CATEGORY_TOP_AND_BOTTOM Top_And_Bottom
1346 - const IMC_TBR = 11; // INDIC_MATRA_CATEGORY_TOP_AND_BOTTOM_AND_RIGHT Top_And_Bottom_And_Right
1347 - const IMC_TL = 6; // INDIC_MATRA_CATEGORY_TOP_AND_LEFT Top_And_Left
1348 - const IMC_TLR = 11; // INDIC_MATRA_CATEGORY_TOP_AND_LEFT_AND_RIGHT Top_And_Left_And_Right
1349 - const IMC_TR = 11; // INDIC_MATRA_CATEGORY_TOP_AND_RIGHT Top_And_Right
1350 - const IMC_VOL = 2; // INDIC_MATRA_CATEGORY_VISUAL_ORDER_LEFT Visual_Order_Left
1351 -
1352 - If in original table = _(C,x), that = ISC_C,IMC_x
1353 - Value is IMC_x << 8 (or IMC_x * 256) = 3840
1354 - plus ISC_C = 1, so = 3841
1355 -
1356 - */
1357 -
1358 - public static $indic_table = array(
1359 - /* Devanagari (0900..097F) */
1360 -
1361 - /* 0900 */ 3848, 3848, 3848, 3848, 3842, 3842, 3842, 3842,
1362 - /* 0908 */ 3842, 3842, 3842, 3842, 3842, 3842, 3842, 3842,
1363 - /* 0910 */ 3842, 3842, 3842, 3842, 3842, 3841, 3841, 3841,
1364 - /* 0918 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1365 - /* 0920 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1366 - /* 0928 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1367 - /* 0930 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1368 - /* 0938 */ 3841, 3841, 1543, 2823, 3843, 3840, 2823, 775,
1369 - /* 0940 */ 2823, 2055, 2055, 2055, 2055, 1543, 1543, 1543,
1370 - /* 0948 */ 1543, 2823, 2823, 2823, 2823, 2052, 775, 2823,
1371 - /* 0950 */ 3840, 3840, 3840, 3840, 3840, 1543, 2055, 2055,
1372 - /* 0958 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1373 - /* 0960 */ 3842, 3842, 2055, 2055, 3840, 3840, 3840, 3840,
1374 - /* 0968 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1375 - /* 0970 */ 3840, 3840, 3842, 3842, 3842, 3842, 3842, 3842,
1376 - /* 0978 */ 3840, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1377 - /* Bengali (0980..09FF) */
1378 -
1379 - /* 0980 */ 3840, 3848, 3848, 3848, 3840, 3842, 3842, 3842,
1380 - /* 0988 */ 3842, 3842, 3842, 3842, 3842, 3840, 3840, 3842,
1381 - /* 0990 */ 3842, 3840, 3840, 3842, 3842, 3841, 3841, 3841,
1382 - /* 0998 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1383 - /* 09A0 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1384 - /* 09A8 */ 3841, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1385 - /* 09B0 */ 3841, 3840, 3841, 3840, 3840, 3840, 3841, 3841,
1386 - /* 09B8 */ 3841, 3841, 3840, 3840, 3843, 3840, 2823, 775,
1387 - /* 09C0 */ 2823, 2055, 2055, 2055, 2055, 3840, 3840, 775,
1388 - /* 09C8 */ 775, 3840, 3840, 2823, 2823, 2052, 3841, 3840,
1389 - /* 09D0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 2823,
1390 - /* 09D8 */ 3840, 3840, 3840, 3840, 3841, 3841, 3840, 3841,
1391 - /* 09E0 */ 3842, 3842, 2055, 2055, 3840, 3840, 3840, 3840,
1392 - /* 09E8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1393 - /* 09F0 */ 3841, 3841, 3840, 3840, 3840, 3840, 3840, 3840,
1394 - /* 09F8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1395 - /* Gurmukhi (0A00..0A7F) */
1396 -
1397 - /* 0A00 */ 3840, 3848, 3848, 3848, 3840, 3842, 3842, 3842,
1398 - /* 0A08 */ 3842, 3842, 3842, 3840, 3840, 3840, 3840, 3842,
1399 - /* 0A10 */ 3842, 3840, 3840, 3842, 3842, 3841, 3841, 3841,
1400 - /* 0A18 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1401 - /* 0A20 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1402 - /* 0A28 */ 3841, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1403 - /* 0A30 */ 3841, 3840, 3841, 3841, 3840, 3841, 3841, 3840,
1404 - /* 0A38 */ 3841, 3841, 3840, 3840, 3843, 3840, 2823, 775,
1405 - /* 0A40 */ 2823, 2055, 2055, 3840, 3840, 3840, 3840, 1543,
1406 - /* 0A48 */ 1543, 3840, 3840, 1543, 1543, 2052, 3840, 3840,
1407 - /* 0A50 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1408 - /* 0A58 */ 3840, 3841, 3841, 3841, 3841, 3840, 3841, 3840,
1409 - /* 0A60 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1410 - /* 0A68 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1411 - /* 0A70 */ 3848, 3840, 13841, 13841, 3840, 3857, 3840, 3840,
1412 - /* 0A78 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1413 - /* Gujarati (0A80..0AFF) */
1414 -
1415 - /* 0A80 */ 3840, 3848, 3848, 3848, 3840, 3842, 3842, 3842,
1416 - /* 0A88 */ 3842, 3842, 3842, 3842, 3842, 3842, 3840, 3842,
1417 - /* 0A90 */ 3842, 3842, 3840, 3842, 3842, 3841, 3841, 3841,
1418 - /* 0A98 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1419 - /* 0AA0 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1420 - /* 0AA8 */ 3841, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1421 - /* 0AB0 */ 3841, 3840, 3841, 3841, 3840, 3841, 3841, 3841,
1422 - /* 0AB8 */ 3841, 3841, 3840, 3840, 3843, 3840, 2823, 775,
1423 - /* 0AC0 */ 2823, 2055, 2055, 2055, 2055, 1543, 3840, 1543,
1424 - /* 0AC8 */ 1543, 2823, 3840, 2823, 2823, 2052, 3840, 3840,
1425 - /* 0AD0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1426 - /* 0AD8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1427 - /* 0AE0 */ 3842, 3842, 2055, 2055, 3840, 3840, 3840, 3840,
1428 - /* 0AE8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1429 - /* 0AF0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1430 - /* 0AF8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1431 - /* Oriya (0B00..0B7F) */
1432 -
1433 - /* 0B00 */ 3840, 3848, 3848, 3848, 3840, 3842, 3842, 3842,
1434 - /* 0B08 */ 3842, 3842, 3842, 3842, 3842, 3840, 3840, 3842,
1435 - /* 0B10 */ 3842, 3840, 3840, 3842, 3842, 3841, 3841, 3841,
1436 - /* 0B18 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1437 - /* 0B20 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1438 - /* 0B28 */ 3841, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1439 - /* 0B30 */ 3841, 3840, 3841, 3841, 3840, 3841, 3841, 3841,
1440 - /* 0B38 */ 3841, 3841, 3840, 3840, 3843, 3840, 2823, 1543,
1441 - /* 0B40 */ 2823, 2055, 2055, 2055, 2055, 3840, 3840, 775,
1442 - /* 0B48 */ 1543, 3840, 3840, 2823, 2823, 2052, 3840, 3840,
1443 - /* 0B50 */ 3840, 3840, 3840, 3840, 3840, 3840, 1543, 2823,
1444 - /* 0B58 */ 3840, 3840, 3840, 3840, 3841, 3841, 3840, 3841,
1445 - /* 0B60 */ 3842, 3842, 2055, 2055, 3840, 3840, 3840, 3840,
1446 - /* 0B68 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1447 - /* 0B70 */ 3840, 3841, 3840, 3840, 3840, 3840, 3840, 3840,
1448 - /* 0B78 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1449 - /* Tamil (0B80..0BFF) */
1450 -
1451 - /* 0B80 */ 3840, 3840, 3848, 3840, 3840, 3842, 3842, 3842,
1452 - /* 0B88 */ 3842, 3842, 3842, 3840, 3840, 3840, 3842, 3842,
1453 - /* 0B90 */ 3842, 3840, 3842, 3842, 3842, 3841, 3840, 3840,
1454 - /* 0B98 */ 3840, 3841, 3841, 3840, 3841, 3840, 3841, 3841,
1455 - /* 0BA0 */ 3840, 3840, 3840, 3841, 3841, 3840, 3840, 3840,
1456 - /* 0BA8 */ 3841, 3841, 3841, 3840, 3840, 3840, 3841, 3841,
1457 - /* 0BB0 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1458 - /* 0BB8 */ 3841, 3841, 3840, 3840, 3840, 3840, 2823, 2823,
1459 - /* 0BC0 */ 1543, 2055, 2055, 3840, 3840, 3840, 775, 775,
1460 - /* 0BC8 */ 775, 3840, 2823, 2823, 2823, 1540, 3840, 3840,
1461 - /* 0BD0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 2823,
1462 - /* 0BD8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1463 - /* 0BE0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1464 - /* 0BE8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1465 - /* 0BF0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1466 - /* 0BF8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1467 - /* Telugu (0C00..0C7F) */
1468 -
1469 - /* 0C00 */ 3840, 3848, 3848, 3848, 3840, 3842, 3842, 3842,
1470 - /* 0C08 */ 3842, 3842, 3842, 3842, 3842, 3840, 3842, 3842,
1471 - /* 0C10 */ 3842, 3840, 3842, 3842, 3842, 3841, 3841, 3841,
1472 - /* 0C18 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1473 - /* 0C20 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1474 - /* 0C28 */ 3841, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1475 - /* 0C30 */ 3841, 3841, 3841, 3841, 3840, 3841, 3841, 3841,
1476 - /* 0C38 */ 3841, 3841, 3840, 3840, 3840, 3840, 1543, 1543,
1477 - /* 0C40 */ 1543, 2823, 2823, 2823, 2823, 3840, 1543, 1543,
1478 - /* 0C48 */ 2055, 3840, 1543, 1543, 1543, 1540, 3840, 3840,
1479 - /* 0C50 */ 3840, 3840, 3840, 3840, 3840, 1543, 2055, 3840,
1480 - /* 0C58 */ 3841, 3841, 3840, 3840, 3840, 3840, 3840, 3840,
1481 - /* 0C60 */ 3842, 3842, 2055, 2055, 3840, 3840, 3840, 3840,
1482 - /* 0C68 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1483 - /* 0C70 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1484 - /* 0C78 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1485 - /* Kannada (0C80..0CFF) */
1486 -
1487 - /* 0C80 */ 3840, 3840, 3848, 3848, 3840, 3842, 3842, 3842,
1488 - /* 0C88 */ 3842, 3842, 3842, 3842, 3842, 3840, 3842, 3842,
1489 - /* 0C90 */ 3842, 3840, 3842, 3842, 3842, 3841, 3841, 3841,
1490 - /* 0C98 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1491 - /* 0CA0 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1492 - /* 0CA8 */ 3841, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1493 - /* 0CB0 */ 3841, 3841, 3841, 3841, 3840, 3841, 3841, 3841,
1494 - /* 0CB8 */ 3841, 3841, 3840, 3840, 3843, 3840, 2823, 1543,
1495 - /* 0CC0 */ 2823, 2823, 2823, 2823, 2823, 3840, 1543, 2823,
1496 - /* 0CC8 */ 2823, 3840, 2823, 2823, 1543, 1540, 3840, 3840,
1497 - /* 0CD0 */ 3840, 3840, 3840, 3840, 3840, 2823, 2823, 3840,
1498 - /* 0CD8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3841, 3840,
1499 - /* 0CE0 */ 3842, 3842, 2055, 2055, 3840, 3840, 3840, 3840,
1500 - /* 0CE8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1501 - /* 0CF0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1502 - /* 0CF8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1503 - /* Malayalam (0D00..0D7F) */
1504 -
1505 - /* 0D00 */ 3840, 3840, 3848, 3848, 3840, 3842, 3842, 3842,
1506 - /* 0D08 */ 3842, 3842, 3842, 3842, 3842, 3840, 3842, 3842,
1507 - /* 0D10 */ 3842, 3840, 3842, 3842, 3842, 3841, 3841, 3841,
1508 - /* 0D18 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1509 - /* 0D20 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1510 - /* 0D28 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1511 - /* 0D30 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1512 - /* 0D38 */ 3841, 3841, 3841, 3840, 3840, 3840, 2823, 2823,
1513 - /* 0D40 */ 2823, 2823, 2823, 2055, 2055, 3840, 775, 775,
1514 - /* 0D48 */ 775, 3840, 2823, 2823, 2823, 1540, 3855, 3840,
1515 - /* 0D50 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 2823,
1516 - /* 0D58 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1517 - /* 0D60 */ 3842, 3842, 2055, 2055, 3840, 3840, 3840, 3840,
1518 - /* 0D68 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1519 - /* 0D70 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1520 - /* 0D78 */ 3840, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1521 - /* Sinhala (0D80..0DFF) */
1522 -
1523 - /* 0D80 */ 3840, 3840, 3848, 3848, 3840, 3842, 3842, 3842,
1524 - /* 0D88 */ 3842, 3842, 3842, 3842, 3842, 3842, 3842, 3842,
1525 - /* 0D90 */ 3842, 3842, 3842, 3842, 3842, 3842, 3842, 3840,
1526 - /* 0D98 */ 3840, 3840, 3841, 3841, 3841, 3841, 3841, 3841,
1527 - /* 0DA0 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1528 - /* 0DA8 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1529 - /* 0DB0 */ 3841, 3841, 3840, 3841, 3841, 3841, 3841, 3841,
1530 - /* 0DB8 */ 3841, 3841, 3841, 3841, 3840, 3841, 3840, 3840,
1531 - /* 0DC0 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3840,
1532 - /* 0DC8 */ 3840, 3840, 1540, 3840, 3840, 3840, 3840, 2823,
1533 - /* 0DD0 */ 2823, 2823, 1543, 1543, 2055, 3840, 2055, 3840,
1534 - /* 0DD8 */ 2823, 775, 1543, 775, 2823, 2823, 2823, 2823,
1535 - /* 0DE0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1536 - /* 0DE8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1537 - /* 0DF0 */ 3840, 3840, 2823, 2823, 3840, 3840, 3840, 3840,
1538 - /* 0DF8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1539 - /* Vedic Extensions (1CD0..1CFF) */
1540 -
1541 - /* 1CD0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1542 - /* 1CD8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1543 - /* 1CE0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1544 - /* 1CE8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1545 - /* 1CF0 */ 3840, 3840, 3848, 3848, 3840, 3840, 3840, 3840,
1546 - /* 1CF8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1547 - );
1548 -
1549 - public static $khmer_table = array(
1550 - /* Khmer (1780..17FF) */
1551 -
1552 - /* 1780 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1553 - /* 1788 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1554 - /* 1790 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1555 - /* 1798 */ 3841, 3841, 3841, 3841, 3841, 3841, 3841, 3841,
1556 - /* 17A0 */ 3841, 3841, 3841, 3842, 3842, 3842, 3842, 3842,
1557 - /* 17A8 */ 3842, 3842, 3842, 3842, 3842, 3842, 3842, 3842,
1558 - /* 17B0 */ 3842, 3842, 3842, 3842, 3840, 3840, 2823, 1543,
1559 - /* 17B8 */ 1543, 1543, 1543, 2055, 2055, 2055, 1543, 2823,
1560 - /* 17C0 */ 2823, 775, 775, 775, 2823, 2823, 3848, 3848,
1561 - /* 17C8 */ 2823, 3853, 3853, 3840, 3855, 3840, 3840, 3840,
1562 - /* 17D0 */ 3840, 1540, 3844, 3840, 3840, 3840, 3840, 3840,
1563 - /* 17D8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1564 - /* 17E0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1565 - /* 17E8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1566 - /* 17F0 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1567 - /* 17F8 */ 3840, 3840, 3840, 3840, 3840, 3840, 3840, 3840,
1568 - );
1569 -
1570 - // from "hb-ot-shape-complex-indic-table.cc"
1571 - public static function indic_get_categories($u)
1572 - {
1573 - if (0x0900 <= $u && $u <= 0x0DFF)
1574 - return self::$indic_table[$u - 0x0900 + 0]; // offset 0 for Most "indic"
1575 - if (0x1CD0 <= $u && $u <= 0x1D00)
1576 - return self::$indic_table[$u - 0x1CD0 + 1152]; // offset for Vedic extensions
1577 - if (0x1780 <= $u && $u <= 0x17FF)
1578 - return self::$khmer_table[$u - 0x1780]; // Khmer
1579 - if ($u == 0x00A0)
1580 - return 3851; // (ISC_CP | (IMC_x << 8))
1581 - if ($u == 0x25CC)
1582 - return 3851; // (ISC_CP | (IMC_x << 8))
1583 - return 3840; // (ISC_x | (IMC_x << 8))
1584 - }
1585 -
1586 - // BELOW from hb-ot-shape-complex-indic.cc
1587 - /*
1588 - * Indic shaper.
1589 - */
1590 -
1591 - public static function IN_HALF_BLOCK($u, $Base)
1592 - {
1593 - return (($u & ~0x7F) == $Base);
1594 - }
1595 -
1596 - public static function IS_DEVA($u)
1597 - {
1598 - return self::IN_HALF_BLOCK($u, 0x0900);
1599 - }
1600 -
1601 - public static function IS_BENG($u)
1602 - {
1603 - return self::IN_HALF_BLOCK($u, 0x0980);
1604 - }
1605 -
1606 - public static function IS_GURU($u)
1607 - {
1608 - return self::IN_HALF_BLOCK($u, 0x0A00);
1609 - }
1610 -
1611 - public static function IS_GUJR($u)
1612 - {
1613 - return self::IN_HALF_BLOCK($u, 0x0A80);
1614 - }
1615 -
1616 - public static function IS_ORYA($u)
1617 - {
1618 - return self::IN_HALF_BLOCK($u, 0x0B00);
1619 - }
1620 -
1621 - public static function IS_TAML($u)
1622 - {
1623 - return self::IN_HALF_BLOCK($u, 0x0B80);
1624 - }
1625 -
1626 - public static function IS_TELU($u)
1627 - {
1628 - return self::IN_HALF_BLOCK($u, 0x0C00);
1629 - }
1630 -
1631 - public static function IS_KNDA($u)
1632 - {
1633 - return self::IN_HALF_BLOCK($u, 0x0C80);
1634 - }
1635 -
1636 - public static function IS_MLYM($u)
1637 - {
1638 - return self::IN_HALF_BLOCK($u, 0x0D00);
1639 - }
1640 -
1641 - public static function IS_SINH($u)
1642 - {
1643 - return self::IN_HALF_BLOCK($u, 0x0D80);
1644 - }
1645 -
1646 - public static function IS_KHMR($u)
1647 - {
1648 - return self::IN_HALF_BLOCK($u, 0x1780);
1649 - }
1650 -
1651 - public static function MATRA_POS_LEFT($u)
1652 - {
1653 - return self::POS_PRE_M;
1654 - }
1655 -
1656 - public static function MATRA_POS_RIGHT($u)
1657 - {
1658 - return
1659 - (self::IS_DEVA($u) ? self::POS_AFTER_SUB :
1660 - (self::IS_BENG($u) ? self::POS_AFTER_POST :
1661 - (self::IS_GURU($u) ? self::POS_AFTER_POST :
1662 - (self::IS_GUJR($u) ? self::POS_AFTER_POST :
1663 - (self::IS_ORYA($u) ? self::POS_AFTER_POST :
1664 - (self::IS_TAML($u) ? self::POS_AFTER_POST :
1665 - (self::IS_TELU($u) ? ($u <= 0x0C42 ? self::POS_BEFORE_SUB : self::POS_AFTER_SUB) :
1666 - (self::IS_KNDA($u) ? ($u < 0x0CC3 || $u > 0xCD6 ? self::POS_BEFORE_SUB : self::POS_AFTER_SUB) :
1667 - (self::IS_MLYM($u) ? self::POS_AFTER_POST :
1668 - (self::IS_SINH($u) ? self::POS_AFTER_SUB :
1669 - (self::IS_KHMR($u) ? self::POS_AFTER_POST :
1670 - self::POS_AFTER_SUB))))))))))); /* default */
1671 - }
1672 -
1673 - public static function MATRA_POS_TOP($u)
1674 - {
1675 - return /* BENG and MLYM don't have top matras. */
1676 - (self::IS_DEVA($u) ? self::POS_AFTER_SUB :
1677 - (self::IS_GURU($u) ? self::POS_AFTER_POST : /* Deviate from spec */
1678 - (self::IS_GUJR($u) ? self::POS_AFTER_SUB :
1679 - (self::IS_ORYA($u) ? self::POS_AFTER_MAIN :
1680 - (self::IS_TAML($u) ? self::POS_AFTER_SUB :
1681 - (self::IS_TELU($u) ? self::POS_BEFORE_SUB :
1682 - (self::IS_KNDA($u) ? self::POS_BEFORE_SUB :
1683 - (self::IS_SINH($u) ? self::POS_AFTER_SUB :
1684 - (self::IS_KHMR($u) ? self::POS_AFTER_POST :
1685 - self::POS_AFTER_SUB))))))))); /* default */
1686 - }
1687 -
1688 - public static function MATRA_POS_BOTTOM($u)
1689 - {
1690 - return
1691 - (self::IS_DEVA($u) ? self::POS_AFTER_SUB :
1692 - (self::IS_BENG($u) ? self::POS_AFTER_SUB :
1693 - (self::IS_GURU($u) ? self::POS_AFTER_POST :
1694 - (self::IS_GUJR($u) ? self::POS_AFTER_POST :
1695 - (self::IS_ORYA($u) ? self::POS_AFTER_SUB :
1696 - (self::IS_TAML($u) ? self::POS_AFTER_POST :
1697 - (self::IS_TELU($u) ? self::POS_BEFORE_SUB :
1698 - (self::IS_KNDA($u) ? self::POS_BEFORE_SUB :
1699 - (self::IS_MLYM($u) ? self::POS_AFTER_POST :
1700 - (self::IS_SINH($u) ? self::POS_AFTER_SUB :
1701 - (self::IS_KHMR($u) ? self::POS_AFTER_POST :
1702 - self::POS_AFTER_SUB))))))))))); /* default */
1703 - }
1704 -
1705 - public static function matra_position($u, $side)
1706 - {
1707 - switch ($side) {
1708 - case self::POS_PRE_C: return self::MATRA_POS_LEFT($u);
1709 - case self::POS_POST_C: return self::MATRA_POS_RIGHT($u);
1710 - case self::POS_ABOVE_C: return self::MATRA_POS_TOP($u);
1711 - case self::POS_BELOW_C: return self::MATRA_POS_BOTTOM($u);
1712 - }
1713 - return $side;
1714 - }
1715 -
1716 - // vowel matras that have to be split into two parts.
1717 - // From Harfbuzz (old)
1718 - // New HarfBuzz uses /src/hb-ucdn/ucdn.c and unicodedata_db.h for full method of decomposition for all characters
1719 - // Should always fully decompose and then recompose back, but we will just do the split matras
1720 - public static function decompose_indic($ab)
1721 - {
1722 - $sub = array();
1723 - switch ($ab) {
1724 - /*
1725 - * Decompose split matras.
1726 - */
1727 - /* bengali */
1728 - case 0x9cb : $sub[0] = 0x9c7;
1729 - $sub[1] = 0x9be;
1730 - return $sub;
1731 - case 0x9cc : $sub[0] = 0x9c7;
1732 - $sub[1] = 0x9d7;
1733 - return $sub;
1734 - /* oriya */
1735 - case 0xb48 : $sub[0] = 0xb47;
1736 - $sub[1] = 0xb56;
1737 - return $sub;
1738 - case 0xb4b : $sub[0] = 0xb47;
1739 - $sub[1] = 0xb3e;
1740 - return $sub;
1741 - case 0xb4c : $sub[0] = 0xb47;
1742 - $sub[1] = 0xb57;
1743 - return $sub;
1744 - /* tamil */
1745 - case 0xbca : $sub[0] = 0xbc6;
1746 - $sub[1] = 0xbbe;
1747 - return $sub;
1748 - case 0xbcb : $sub[0] = 0xbc7;
1749 - $sub[1] = 0xbbe;
1750 - return $sub;
1751 - case 0xbcc : $sub[0] = 0xbc6;
1752 - $sub[1] = 0xbd7;
1753 - return $sub;
1754 - /* telugu */
1755 - case 0xc48 : $sub[0] = 0xc46;
1756 - $sub[1] = 0xc56;
1757 - return $sub;
1758 - /* kannada */
1759 - case 0xcc0 : $sub[0] = 0xcbf;
1760 - $sub[1] = 0xcd5;
1761 - return $sub;
1762 - case 0xcc7 : $sub[0] = 0xcc6;
1763 - $sub[1] = 0xcd5;
1764 - return $sub;
1765 - case 0xcc8 : $sub[0] = 0xcc6;
1766 - $sub[1] = 0xcd6;
1767 - return $sub;
1768 - case 0xcca : $sub[0] = 0xcc6;
1769 - $sub[1] = 0xcc2;
1770 - return $sub;
1771 - case 0xccb : $sub[0] = 0xcc6;
1772 - $sub[1] = 0xcc2;
1773 - $sub[2] = 0xcd5;
1774 - return $sub;
1775 - /* malayalam */
1776 - case 0xd4a : $sub[0] = 0xd46;
1777 - $sub[1] = 0xd3e;
1778 - return $sub;
1779 - case 0xd4b : $sub[0] = 0xd47;
1780 - $sub[1] = 0xd3e;
1781 - return $sub;
1782 - case 0xd4c : $sub[0] = 0xd46;
1783 - $sub[1] = 0xd57;
1784 - return $sub;
1785 - /* sinhala */
1786 - // NB Some fonts break with these Sinhala decomps (although this is Uniscribe spec)
1787 - // Can check if character would be substituted by pstf and only decompose if true
1788 - // e.g. if (isset($GSUBdata['pstf'][$ab])) - would need to pass $GSUBdata as parameter to this function
1789 - case 0xdda : $sub[0] = 0xdd9;
1790 - $sub[1] = 0xdca;
1791 - return $sub;
1792 - case 0xddc : $sub[0] = 0xdd9;
1793 - $sub[1] = 0xdcf;
1794 - return $sub;
1795 - case 0xddd : $sub[0] = 0xdd9;
1796 - $sub[1] = 0xdcf;
1797 - $sub[2] = 0xdca;
1798 - return $sub;
1799 - case 0xdde : $sub[0] = 0xdd9;
1800 - $sub[1] = 0xddf;
1801 - return $sub;
1802 - /* khmer */
1803 - case 0x17be : $sub[0] = 0x17c1;
1804 - $sub[1] = 0x17be;
1805 - return $sub;
1806 - case 0x17bf : $sub[0] = 0x17c1;
1807 - $sub[1] = 0x17bf;
1808 - return $sub;
1809 - case 0x17c0 : $sub[0] = 0x17c1;
1810 - $sub[1] = 0x17c0;
1811 - return $sub;
1812 -
1813 - case 0x17c4 : $sub[0] = 0x17c1;
1814 - $sub[1] = 0x17c4;
1815 - return $sub;
1816 - case 0x17c5 : $sub[0] = 0x17c1;
1817 - $sub[1] = 0x17c5;
1818 - return $sub;
1819 - /* tibetan - included here although does not use Inidc shaper in other ways */
1820 - case 0xf73 : $sub[0] = 0xf71;
1821 - $sub[1] = 0xf72;
1822 - return $sub;
1823 - case 0xf75 : $sub[0] = 0xf71;
1824 - $sub[1] = 0xf74;
1825 - return $sub;
1826 - case 0xf76 : $sub[0] = 0xfb2;
1827 - $sub[1] = 0xf80;
1828 - return $sub;
1829 - case 0xf77 : $sub[0] = 0xfb2;
1830 - $sub[1] = 0xf81;
1831 - return $sub;
1832 - case 0xf78 : $sub[0] = 0xfb3;
1833 - $sub[1] = 0xf80;
1834 - return $sub;
1835 - case 0xf79 : $sub[0] = 0xfb3;
1836 - $sub[1] = 0xf71;
1837 - $sub[2] = 0xf80;
1838 - return $sub;
1839 - case 0xf81 : $sub[0] = 0xf71;
1840 - $sub[1] = 0xf80;
1841 - return $sub;
1842 - }
1843 - return false;
1844 - }
1845 -
1846 - public static function bubble_sort(&$arr, $start, $len)
1847 - {
1848 - if ($len < 2) {
1849 - return;
1850 - }
1851 - $k = $start + $len - 2;
1852 - while ($k >= $start) {
1853 - for ($j = $start; $j <= $k; $j++) {
1854 - if ($arr[$j]['indic_position'] > $arr[$j + 1]['indic_position']) {
1855 - $t = $arr[$j];
1856 - $arr[$j] = $arr[$j + 1];
1857 - $arr[$j + 1] = $t;
1858 - }
1859 - }
1860 - $k--;
1861 - }
1862 - }
1863 -
1864 -}
1 +<?php
2 +
3 +class indic {
4 +
5 +function indic() {
6 +
7 +}
8 +
9 +
10 +function substituteIndic($earr, $lang, $font) {
11 + global $voltdata;
12 +
13 + if (!isset($voltdata[$font])) {
14 + include_once(_MPDF_PATH.'includes/'.$font.'.volt.php');
15 + $voltdata[$font] = $volt;
16 + }
17 +
18 + foreach($earr as $eid=>$char) {
19 + $earr[$eid] = sprintf("%04s", strtoupper(dechex($char)));
20 + }
21 + $vstr = "0020 ".implode(" ",$earr)." 0020";
22 + //============================
23 + // Common Indic Punctuation marks
24 + // If NOT devanagari
25 + if ($lang!='hi') {
26 + $vstr = str_replace('0964','007C', $vstr); // U+0964 replace with "|"
27 + $vstr = str_replace('0965','007C 007C', $vstr); // U+0964 replace with "|"
28 + }
29 + //============================
30 + // Tamil numeral for Zero missing Added mPDF 4.2
31 + if ($lang=='ta') {
32 + $vstr = str_replace('0BE6','0030', $vstr); // U+0BEB replace with "0"
33 + }
34 +
35 + //============================
36 + // Re-order vowels
37 +
38 + // DEVANAGARI vowel sign matraI[093F] before consonant
39 + if ($lang=='hi') {
40 + $prebasedvowels = "(093F)";
41 + $nukta = "093C";
42 + $halant = "094D";
43 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.'/','\\2 \\1', $vstr); // vowel sign pre-based shift left
44 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.' '.$nukta.'/','\\2 \\1 '.$nukta, $vstr); // before NUKTA
45 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$prebasedvowels.'/','\\2 \\1 '.$halant, $vstr); // before CHAR HALANT == VIRAMA
46 + }
47 +
48 + // BENGALI vowels [09BF 09C7 09C8]
49 + else if ($lang=='bn') {
50 +
51 + // Khanda Ta 09CE not in font -> replace with 09A4|09CD
52 + $vstr = preg_replace('/09CE/','09A4 09CD 200D', $vstr); // mPDF 5.3.09
53 +
54 + // BENGALI double-part vowels [09CB 09C7 09BE][09CC 09C7 09D7]
55 + $vstr = str_replace('09CB','09C7 09BE', $vstr); // convert to 2 parts
56 + $vstr = str_replace('09CC','09C7 09D7', $vstr); // 09C7 pre-based is then shifted below
57 + $prebasedvowels = "(09BF|09C7|09C8)";
58 + $nukta = "09BC";
59 + $halant = "09CD";
60 + // mPDF 5.0.044
61 + $bnfullcons = "0995|0996|0997|0998|0999|099A|099B|099C|099D|099F|09A0|09A1|09A2|09A3|09A4|09A5|09A6|09A7|09A8|09AA|09AB|09AC|09AD|09AE|09AF|09B0|09B2|09B6|09B7|09B8|09B9|09DC|09DD|09DF";
62 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.'/','\\2 \\1', $vstr); // vowel sign pre-based shift left
63 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.' '.$nukta.'/','\\2 \\1 '.$nukta, $vstr); // before NUKTA
64 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$prebasedvowels.'/','\\2 \\1 '.$halant, $vstr); // before CHAR HALANT
65 + // mPDF 5.0.044
66 + // .. and shifting left-based vowel further to the left in case 3 consonants together.
67 + $vstr = preg_replace('/('.$bnfullcons.') '.$halant.' '.$prebasedvowels.'/','\\2 \\1 '.$halant, $vstr);
68 +
69 + // mPDF 5.0.044
70 + // If left-based vowel has now been shifted to left of RA/Halant (09B0/09CD)
71 + // Convert here to above-line form (E068) as it would get missed later
72 + // e.g. 09B0 09CD 09AD 09C7 would be changed above =>
73 + // e.g. 09C7 09B0 09CD 09AD. The 09B0 09CD should => E068
74 + // ??? need to add 09BF as well (09BF|09C7|09C8)
75 + $vstr = preg_replace('/(09C7|09C8) 09B0 09CD/', '\\1 E068', $vstr);
76 +
77 + }
78 +
79 + // GUJARATI pre-based vowel [0ABF]
80 + else if ($lang=='gu') {
81 + $prebasedvowels = "(0ABF)";
82 + $nukta = "0ABC";
83 + $halant = "0ACD";
84 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.'/','\\2 \\1', $vstr); // vowel sign pre-based shift left
85 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.' '.$nukta.'/','\\2 \\1 '.$nukta, $vstr); // before NUKTA
86 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$prebasedvowels.'/','\\2 \\1 '.$halant, $vstr); // before CHAR HALANT
87 + }
88 +
89 + // GURMUKHI/PUNJABI pre-based vowel [0ABF]
90 + else if ($lang=='pa') {
91 + $prebasedvowels = "(0A3F)";
92 + $nukta = "0A3C";
93 + $halant = "0A4D";
94 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.'/','\\2 \\1', $vstr); // vowel sign pre-based shift left
95 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.' '.$nukta.'/','\\2 \\1 '.$nukta, $vstr); // before NUKTA
96 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$prebasedvowels.'/','\\2 \\1 '.$halant, $vstr); // before CHAR HALANT
97 + }
98 +
99 + // TAMIL pre-based vowel [0ABF]
100 + else if ($lang=='ta') {
101 + // Shrii (Shree)
102 + $vstr = preg_replace('/0BB6 0BCD 0BB0 0BC0/','E04B', $vstr);
103 +
104 + // TAMIL double-part vowels [0BCA 0BC6 0BBE][0BCB 0BC7 0BBE][0BCC 0BC6 0BD7]
105 + $vstr = preg_replace('/0BCA/','0BC6 0BBE', $vstr); // convert to 2 parts
106 + $vstr = preg_replace('/0BCB/','0BC7 0BBE', $vstr); // pre-based is then shifted below
107 + $vstr = preg_replace('/0BCC/','0BC6 0BD7', $vstr);
108 + $prebasedvowels = "(0BC6|0BC7|0BC8)";
109 + // No nukta
110 + $halant = "0BCD"; // Doesn't seem to move most in front of halanted consonants
111 + $vstr = preg_replace('/([A-F0-9]{4}) '.$prebasedvowels.'/','\\2 \\1', $vstr); // vowel sign pre-based shift left
112 + // ? Only for special case KSS (already moved to left of 0BB7)
113 + $vstr = preg_replace('/0B95 '.$halant.' '.$prebasedvowels.' 0BB7/','\\1 0B95 '.$halant.' 0BB7', $vstr);
114 + }
115 +
116 + // ORIYA
117 + else if ($lang=='or') {
118 + // ORIYA double-part vowels []
119 + $vstr = str_replace('0B48','0B47 0B56', $vstr); // 2-part Vowel
120 + $vstr = str_replace('0B4B','0B47 0B3E', $vstr); // 2-part Vowel
121 + $vstr = str_replace('0B4C','0B47 0B57', $vstr); // 2-part Vowel
122 + $orprebasedvowels = "(0B47)";
123 + // No nukta
124 + $halant = "0B4D";
125 + $vstr = preg_replace('/([A-F0-9]{4}) '.$orprebasedvowels.'/','\\2 \\1', $vstr); // vowel sign pre-based shift left
126 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$orprebasedvowels.'/','\\2 \\1 '.$halant, $vstr); // before CHAR HALANT
127 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$orprebasedvowels.'/','\\2 \\1 '.$halant, $vstr); // before CHAR HALANT
128 + }
129 +
130 + // MALAYALAM
131 + else if ($lang=='ml') {
132 + // Chillus - old forms - remove ZWNJ after
133 + // This font Volt rules recognises e.g. "Na Halant(Virama)" as ChilluN
134 + $vstr = preg_replace('/(0D23 0D4D|0D28 0D4D|0D30 0D4D|0D32 0D4D|0D33 0D4D) 200D/','\\1', $vstr);
135 + // See Chillus in Unicode [http://en.wikipedia.org/wiki/Malayalam_script]
136 + $vstr = str_replace('0D7A','0D23 0D4D', $vstr); // [mlymChilluNn]
137 + $vstr = str_replace('0D7B','0D28 0D4D', $vstr); // [mlymChilluN]
138 + $vstr = str_replace('0D7C','0D30 0D4D', $vstr); // [mlymChilluR]
139 + $vstr = str_replace('0D7D','0D32 0D4D', $vstr); // [mlymChilluL]
140 + $vstr = str_replace('0D7E','0D33 0D4D', $vstr); // [mlymChilluLl]
141 +/*
142 + // Chillus - 0D7A-0D7E not in font directly, but as E005-E009
143 + $vstr = preg_replace('/0D23 0D4D 200D/','0D7A', $vstr);
144 + $vstr = preg_replace('/0D28 0D4D 200D/','0D7B', $vstr);
145 + $vstr = preg_replace('/0D30 0D4D 200D/','0D7C', $vstr);
146 + $vstr = preg_replace('/0D32 0D4D 200D/','0D7D', $vstr);
147 + $vstr = preg_replace('/0D33 0D4D 200D/','0D7E', $vstr);
148 +
149 + $vstr = preg_replace('/0D7F/','E004', $vstr); // [mlymChilluK]
150 + $vstr = preg_replace('/0D7A/','E005', $vstr); // [mlymChilluNn]
151 + $vstr = preg_replace('/0D7B/','E006', $vstr); // [mlymChilluN]
152 + $vstr = preg_replace('/0D7C/','E007', $vstr); // [mlymChilluR]
153 + $vstr = preg_replace('/0D7D/','E008', $vstr); // [mlymChilluL]
154 + $vstr = preg_replace('/0D7E/','E009', $vstr); // [mlymChilluLl]
155 +*/
156 +
157 + // MALAYALAM double-part vowels []
158 + $vstr = str_replace('0D4A','0D46 0D3E', $vstr); // 2-part Vowel
159 + $vstr = str_replace('0D4B','0D47 0D3E', $vstr); // 2-part Vowel
160 + $vstr = str_replace('0D4C','0D46 0D57', $vstr); // 2-part Vowel
161 + $mlprebasedvowels = "(0D46|0D47|0D48)";
162 + // No nukta
163 + $halant = "0D4D";
164 + $vstr = preg_replace('/([A-F0-9]{4}) '.$mlprebasedvowels.'/','\\2 \\1', $vstr); // vowel sign pre-based shift left
165 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$mlprebasedvowels.'/','\\2 \\1 '.$halant, $vstr); // before CHAR HALANT
166 + }
167 +
168 + // TELUGU
169 + else if ($lang=='te') {
170 + // TELUGU double-part vowels [0C48 -> 0C46 0C56]
171 + $vstr = str_replace('0C48','0C46 0C56', $vstr); // 2-part Vowel
172 + $prebasedvowels = "(0C46)";
173 + $abvvowels = "(0C3E|0C3F|0C40|0C46|0C47|0C4A|0C4B|0C4C|0C55)";
174 + // No nukta
175 + $halant = "0C4D";
176 + $tefullforms = "0C15|0C17|0C18|0C1A|0C1B|0C1C|0C1D|0C20|0C21|0C22|0C24|0C25|0C26|0C27|0C28|0C2A|0C2B|0C2D|0C2E|0C2F|0C30|0C33|0C35|0C36|0C37|0C38|0C39|E028|E029|E02A|E02B|E078|E07A|E07B";
177 + $vstr = preg_replace('/('.$tefullforms .') '.$halant.' ('.$tefullforms .') '.$abvvowels .'/','\\1 \\3 '.$halant.' \\2', $vstr); // before HALANT
178 + }
179 +
180 +
181 + // KANNADA
182 + else if ($lang=='kn') {
183 + // KANNADA double-part vowels [0CC8 -> 0CC6 0CD6]
184 + $vstr = str_replace('0CC0','0CBF 0CD5', $vstr); // 2-part Vowel
185 + $vstr = str_replace('0CC7','0CC6 0CD5', $vstr); // 2-part Vowel
186 + $vstr = str_replace('0CC8','0CC6 0CD6', $vstr); // 2-part Vowel AI - no glyph for single
187 + $vstr = str_replace('0CCA','0CC6 0CC2', $vstr); // 2-part Vowel
188 + $vstr = str_replace('0CCB','0CC6 0CC2 0CD5', $vstr); // 2-part Vowel
189 + $prebasedvowels = "(0CBF|0CC6)";
190 + $halant = "0CCD";
191 + }
192 +
193 +
194 + //============================
195 +
196 + // SPECIALS
197 +
198 + // DEVANAGARI Ra Halant Ra
199 + if ($lang=='hi') {
200 + $vstr = str_replace('0930 094D 0930','E05D 0930', $vstr); // Ra Halant Ra => halfRa FullRa
201 + }
202 +
203 + // GUJARATI
204 + if ($lang=='gu') {
205 + $vstr = str_replace('0AB0 0AC2','E02E', $vstr); // Ra VowelUu => SpecialForm RaUu
206 + }
207 +
208 + // TELUGU Ra Halant <Consonant> Halant => halfRa Halant<Consonant> Halant
209 + if ($lang=='te') {
210 + $vstr = preg_replace('/0C30 0C4D ([A-F0-9]{4}) 0C4D/','E021 0C4D \\1 0C4D', $vstr);
211 + }
212 +
213 + // KANNADA
214 + // Reph at end of word becomes E0CC instead of E00B
215 + if ($lang=='kn') {
216 + $vstr = str_replace('0CB0 0CCD 0020','E0CC 0020', $vstr); // mPDF 5.3.87
217 + }
218 +
219 +
220 + //============================
221 + // MAIN BIT FROM VOLT RULES
222 + foreach($voltdata[$font] AS $rid=>$reps) {
223 +//echo $rid . ': ' . $vstr.'<br />';
224 + $vstr = preg_replace('/'.$reps['match'].'/',$reps['replace'], $vstr);
225 + }
226 +//echo $vstr.'<br />'; exit;
227 +
228 +
229 + //============================
230 +
231 + // SPECIALS
232 +
233 + // KANNADA
234 + // <Base> <BelowBase1> [<BelowBase2> ] MatraI -> <Base/MatraI ligature> <Belowbase1> etc
235 + if ($lang=='kn') {
236 + $matraI = "0CBF";
237 + $knbase = preg_split('/\|/', "0C95|0C96|0C97|0C98|0C9A|0C9B|0C9C|0C9D|0CA0|0CA1|0CA2|0CA3|0CA4|0CA5|0CA6|0CA7|0CA8|0CAA|0CAB|0CAC|0CAD|0CAE|0CAF|0CB0|0CB2|0CB3|0CB5|0CB6|0CB7|0CB8|0CB9|E0A3|E07D|E07E");
238 + $knmatraIligature = preg_split('/\|/', "E082|E083|E084|E085|E086|E087|E088|E089|E08A|E08B|E08C|E08D|E08E|E08F|E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A4|E0A1|E0A2");
239 + $belowbase1 = "E02E|E02F|E030|E031|E032|E033|E034|E035|E036|E037|E038|E039|E03A|E03B|E03C|E03D|E03E|E03F|E040|E041|E042|E043|E044|E045|E046|E047|E048|E049|E04A|E04B|E04C|E04D|E04E|E04F|E050|E081";
240 + $belowbase2 = "E052|E053|E054|E055|E056|E057|E058|E059|E05A|E05B|E05C|E05D|E05E|E05F|E060|E061|E062|E063|E064|E065|E066|E067|E068|E069|E06A|E06B|E06C|E06D|E06E|E06F|E070|E071|E072|E073|E074|E081";
241 + for ($i=0; $i<count($knbase);$i++) {
242 + $vstr = preg_replace('/'.$knbase[$i].' ('.$belowbase1.') ('.$belowbase2.') '.$matraI.'/', $knmatraIligature[$i].' \\1 \\2', $vstr);
243 + $vstr = preg_replace('/'.$knbase[$i].' ('.$belowbase1.') '.$matraI.'/', $knmatraIligature[$i].' \\1', $vstr);
244 + }
245 + }
246 +
247 + // KANNADA
248 + // [KanTtaFull] [matraI] => [KanTtaPartial] [matraI]
249 + if ($lang=='kn') {
250 + $vstr = preg_replace('/0C9F '.$matraI.'/', 'E015 '.$matraI, $vstr);
251 + }
252 +
253 + // ORIYA
254 + if ($lang=='or') {
255 + // SpecialCase Ra[0B30] Halant still left before [oryaFullNnNna] => E00F
256 + $vstr = preg_replace('/0B30 '.$halant.' E00F/','E00F E069', $vstr); // convert to Reph
257 + }
258 +
259 + //============================
260 + // SHIFT REPH
261 +
262 + // DEVANAGARI Shift Reph [E015]
263 + if ($lang=='hi') {
264 + // FIRSTLY - halfRa = E05D - Change this to Reph [E015]
265 + $himatchhalfforms = "E043|E044|E045|E046|E047|E048|E049|E04A|E04B|E04C|E04D|E04E|E04F|E050|E051|E052|E053|E054|E055|E056|E057|E058|E059|E05A|E05B|E05C|E05D|E05E|E05F|E060|E061|E062|E063|E064|E065|E066|E067|E068|E069|E06A|E06B|E06C|E06D|E06E|E06F|E070|E071|E072|E073|E074|E075|E076|E077|E078|E079|E07A|E07B|E07C|E07D|E07E|E07F|E080|E081|E082|E083|E084|E085|E086|E087|E088|E089|E08A|E0D3|E0D4|E0D5|E0D6|E0D7|E0D8|E0D9|E0DA|E0DB|E0DC|E0DD|E0DE|E0DF|E0E0|E0E1|E0E2|E0E3|E0E4|E0E5|E0E6|E0E7|E0E8|E0E9|E0EA|E0EB|E0EC|E0ED|E0EE|E0EF|E0F0|E0F1|E0F2|E0F3|E0F4|E0F5|E0F6|E0F7|E0F8|E0F9|E0FA|E0FB|E0FC|E0FD|E0FE|E0FF|E100|E101|E102|E103|E104|E105|E106|E107|E108|E109|E10A|E10B|E10C|E10D|E10E|E10F|E110|E111|E112|E113|E114|E115|E116|E117|E118|E119|E11A|E13D|E13E|E13F|E140|E141|E142|E143|E144|E145";
266 + $himatchfullforms = "0915|0916|0917|0918|0919|091A|091B|091C|091D|091E|091F|0920|0921|0922|0923|0924|0925|0926|0927|0928|092A|092B|092C|092D|092E|092F|0930|0932|0933|0935|0936|0937|0938|0939|E028|E029|0958|0959|095A|E02A|E02B|E02C|E02D|095B|E02E|E02F|E030|E031|095C|095D|E032|E033|E034|E035|E036|0929|E037|095E|E038|E039|E03A|095F|0931|E03B|0934|E03C|E03D|E03E|E03F|E040|E041|E042|E08B|E08C|E08D|E08E|E08F|E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A1|E0A2|E0A3|E0A4|E0A5|E0A6|E0A7|E0A8|E0A9|E0AA|E0AB|E0AC|E0AD|E0AE|E0AF|E0B0|E0B1|E0B2|E0B3|E0B4|E0B5|E0B6|E0B7|E0B8|E0B9|E0BA|E0BB|E0BC|E0BD|E0BE|E0BF|E0C0|E0C1|E0C2|E0C3|E0C4|E0C5|E0C6|E0C7|E0C8|E0C9|E0CA|E0CB|E0CC|E0CD|E0CE|E0CF|E0D0|E0D1|E0D2|E11E|E11F|E120|E121|E122|E123|E124|E125|E126|E127|E128|E129|E12A|E12B|E12C|E12D|E12E|E12F|E130|E131|E132|E133";
267 + $vstr = preg_replace('/E05D ('.$himatchhalfforms.'|'.$himatchfullforms.')/', 'E015 \\1', $vstr);
268 +
269 + // Reph = E015 - Shift Right to just after end of syllable
270 + // FullAllForms + HalfAllForms + 093E matraA
271 + while(preg_match('/E015 ('.$himatchhalfforms.')/', $vstr)) {
272 + $vstr = preg_replace('/E015 ('.$himatchhalfforms.')/', '\\1 E015', $vstr);
273 + }
274 + $vstr = preg_replace('/E015 ('.$himatchfullforms.')/', '\\1 E015', $vstr);
275 +
276 + // Now shift it beyond post-based vowels // ??? Need to add others e.g. 0949,094A,094B,094C + presentation forms like E198
277 + $vstr = str_replace('E015 093E', '093E E015', $vstr);
278 + $vstr = preg_replace('/E015 (0940|E194|E195|E196|E197|E198)/', '\\1 E014', $vstr); // (Small) reph [E014] to Right of matraI
279 + $vstr = str_replace('E015 0947', '0947 E014', $vstr); // (Small) reph [E014] to Right of matraI
280 + }
281 +
282 + // BENGALI Shift Reph [E068]
283 + else if ($lang=='bn') {
284 + $bnfullconjuncts = "E002|E003|E004|E041|E042|E043|E044|E045|E046|E047|E048|E049|E04A|E04B|E04C|E04D|E04E|E04F|E050|E051|E052|E053|E054|E055|E056|E057|E058|E059|E05A|E05B|E05C|E05D|E05E|E05F|E060|E061|E062|E063|E064|E065|E06A|E06B|E06C|E06D|E06E|E06F|E070|E071|E072|E073|E074|E075|E076|E077|E078|E079|E07A|E07B|E07C|E07D|E07E|E07F|E080|E081|E082|E083|E084|E085|E086|E087|E088|E089|E08A|E08B|E08C|E08D|E08E|E08F|E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A1|E0A2|E0A3|E0A4|E0A5|E0A6|E0A7|E0A8|E0A9|E0AA|E0AB|E0AC|E0AD|E0AE|E0AF|E0B0|E0B1|E0B2|E0B3|E0B4|E0B5|E0B6|E0B7|E0B8|E0B9|E0BA|E0BB|E0BC|E0BD|E0BE|E0BF|E0C0|E0C1|E0C2|E0C3|E0C4|E0C5|E0C6|E0C7|E0C8|E0C9|E0CA|E0CB|E0CC|E0CD|E0CE|E0CF|E0D0|E0D1|E0D2|E0D3|E0D4|E0D5|E0D6|E0D7|E0D8|E0D9|E0DA|E0DB|E0DC|E0DD|E0DE|E0DF|E0E0|E0E1|E0E2|E0E3|E0E4|E0E5|E0E6|E0E7|E0E8|E0E9|E0EA|E0EB|E0EC|E0ED|E0EE|E0EF|E0F0|E0F1|E0F2|E0F3|E0F4|E0F5|E0F6|E0F7|E0F8|E0F9|E0FA|E0FB|E0FC|E0FD|E0FE|E0FF|E100|E101|E102|E103|E104|E105|E106|E107|E108|E109|E10A|E10B|E10C|E10D|E10E|E10F|E110|E111|E112|E113|E114|E115|E116|E117|E118|E119|E11A|E11B|E11C|E11D|E11E|E11F|E120|E121|E122|E123|E124|E125|E126|E127|E128|E129|E12A|E12B|E12C|E12D|E12E|E12F|E130|E131|E132|E133|E134|E135|E136|E137|E138|E139|E13A|E13B|E13C|E13D|E13E|E13F|E140|E141|E142|E143|E144|E145|E146|E147|E148|E149|E14A|E14B|E14C|E14D|E14E|E14F|E150|E151|E152|E153|E154|E155|E156|E157|E158|E159|E15A|E15B|E15C|E15D|E15E|E15F|E160|E161|E162|E163|E164|E165|E166|E167|E168|E169|E16A|E16B|E16C|E16D|E16E|E16F|E170|E171|E172|E173|E174|E175|E176|E177|E178|E179|E17A|E17B|E17C|E17D|E17E|E17F|E180|E181|E182|E183|E184|E185|E186|E187|E188|E189|E18A|E18B|E18C|E18D|E18E|E18F|E190|E191|E192|E193|E194|E195|E196|E197|E198|E199|E19A";
285 + // $bnfullcons - set above;
286 + $vstr = preg_replace('/E068 ('.$bnfullconjuncts.'|'.$bnfullcons.')/', '\\1 E068', $vstr);
287 + // ? Need to shift it beyond post-base vowels 09BE, 09C0, 09D7 haven't found so can't test??
288 + $vstr = preg_replace('/E068 (09BE|09C0|09D7)/', '\\1 E068', $vstr);
289 + }
290 +
291 + // GUJARATI Shift Reph [E032]
292 + else if ($lang=='gu') {
293 + $gufullforms = "0A95|0A96|0A97|0A98|0A99|0A9A|0A9B|0A9C|0A9D|0A9E|0A9F|0AA0|0AA1|0AA2|0AA3|0AA4|0AA5|0AA6|0AA7|0AA8|0AAA|0AAB|0AAC|0AAD|0AAE|0AAF|0AB0|0AB2|0AB3|0AB5|0AB6|0AB7|0AB8|0AB9|E002|E003|E004|E005|E006|E007|E008|E009|E00A|E00B|E00C|E00D|E00E|E00F|E010|E011|E012|E013|E014|E015|E016|E017|E018|E019|E01A|E01B|E01C|E01D|E01E|E01F|E020|E021|E022|E023|E024|E025|E026|E027|E05E|E05F|E060|E061|E062|E063|E064|E065|E066|E067|E068|E069|E06A|E06B|E06C|E06D|E06E|E06F|E070|E071|E072|E073|E074|E075|E076|E077|E078|E079|E07A|E07B|E07C|E07D|E07E|E07F|E080|E081|E082|E083|E084|E085|E086|E087|E088|E089|E08A|E08B|E08C|E08D|E08E|E08F|E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A1|E0A2|E0A3|E0A4|E0A5";
294 + $vstr = preg_replace('/E032 ('.$gufullforms.')/', '\\1 E032', $vstr);
295 + // Now shift it beyond post-based vowels // ??? Need to add others e.g. 0949,094A,094B,094C + presentation forms like E198
296 + // ? Need to shift it beyond post-base vowels 0ABE, 0AC0 haven't found so can't test??
297 + $vstr = preg_replace('/E032 (0ABE|0AC0)/', '\\1 E032', $vstr);
298 + }
299 +
300 +
301 + // TELUGU Shift Reph to LEFT [E046|E069|E077] [TelRaSmallOne] => E046 [TelRaSmallTwo] => E069 [TelRaSmallThree] => E077
302 + else if ($lang=='te') {
303 + // tefullforms defined earlier
304 + $tepartialforms = "E00D|E00E|E00F|E010|E011|E012|E013|E014|E015|E016|E017|E018|E019|E01A|E01B|E01C|E01D|E01E|E01F|E020|E021|E022|E023|E024|E025|E026|E027|E07C|E07D|E07E";
305 + $matraligs = "E07F|E080|E081|E082|E083|E084|E085|E086|E087|E088|E089|E08A|E08B|E08C|E08D|E08E|E08F|E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A1|E0A2|E0A3|E0A4|E0A5|E0A6|E0A7|E0A8|E0A9|E0AA|E0AB|E0AC|E0AD|E0AE|E0AF";
306 + $tevowels = "0C3E|0C3F|0C40|0C46|0C47|0C56|0C4A|0C4B|0C4C"
307 + ."|0C41|0C42|0C43|0C44"; // post matras
308 + $vstr = preg_replace('/('.$tevowels.') (E046|E069|E077)/', '\\2 \\1', $vstr);
309 + while(preg_match('/('.$tepartialforms.') (E046|E069|E077)/', $vstr)) {
310 + $vstr = preg_replace('/('.$tepartialforms.') (E046|E069|E077)/', '\\2 \\1', $vstr);
311 + }
312 + $vstr = preg_replace('/('.$tefullforms .'|'.$matraligs.') (E046|E069|E077)/', '\\2 \\1', $vstr);
313 + }
314 +
315 +
316 + // KANNADA Shift Reph to RIGHT [E00B]
317 + else if ($lang=='kn') {
318 + $knfullforms = "0C95|0C96|0C97|0C98|0C99|0C9A|0C9B|0C9C|0C9D|0C9E|0C9F|0CA0|0CA1|0CA2|0CA3|0CA4|0CA5|0CA6|0CA7|0CA8|0CAA|0CAB|0CAC|0CAD|0CAE|0CAF|0CB0|0CB1|0CB2|0CB3|0CB5|0CB6|0CB7|0CB8|0CB9|E07D|E07E|E0A3";
319 + $knpartialforms = "E00C|E00D|E00E|E00F|E010|E011|E012|E013|E014|0C9E|E015|E016|E017|E018|E019|E01A|E01B|E01C|E01D|E01E|E01F|E020|E021|E022|E023|E024|E025|E026|E027|E028|E029|E02A|E02B|E02C|E02D|E07F";
320 + while(preg_match('/E00B ('.$knpartialforms.')/', $vstr)) {
321 + $vstr = preg_replace('/E00B ('.$knpartialforms.')/', '\\1 E00B', $vstr);
322 + }
323 + // mPDF 5.3.47 Also move Reph to right of matraIligatures
324 + $knfullforms .= "|E082|E083|E084|E085|E086|E087|E088|E089|E08A|E08B|E08C|E08D|E08E|E08F|E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A4|E0A1|E0A2";
325 + $vstr = preg_replace('/E00B ('.$knfullforms.')/', '\\1 E00B', $vstr);
326 +
327 + // ? Need to shift it beyond base or below-base forms - haven't found so can't test??
328 + // mPDF 5.3.87
329 + // E004 added to list (which is a transformed version of 0CBE)
330 + $knvowels = "0CBE|0CC0|0CC1|0CC2|0CC3|0CC4|0CC7|0CC8|0CCA|0CCB|0CD5|0CD6|E004";
331 + $vstr = preg_replace('/E00B ('.$knvowels.')/', '\\1 E00B', $vstr);
332 + }
333 +
334 +
335 + // ORIYA Shift Reph to RIGHT [E069|E06A|E06B|E06C]
336 + else if ($lang=='or') {
337 + $orrephs = "E069|E06A|E06B|E06C";
338 + $orfullforms = "0B15|0B16|0B17|0B18|0B19|0B1A|0B1B|0B1C|0B1D|0B1E|0B1F|0B20|0B21|0B22|0B23|0B24|0B25|0B26|0B27|0B28|0B29|0B2A|0B2B|0B2C|0B2D|0B2E|0B2F|0B30|0B31|0B32|0B33|0B34|0B35|0B36|0B37|0B38|E003|E004|E005|E006|E007|E008|E009|E00A|E00B|E00C|E00D|E00E|E00F|E010|E011|E012|E013|E014|E015|E016|E017|E018|E019|E01A|E01B|E01C|E01D|E01E|E01F|E020|E021|E022|E023|E024|E025|E026|E027|E028|E029|E02A|E02B|E02C|E02D|E02E|E02F|E030|E031|E032|E033|E034|E035|E036|E037";
339 + // E123 - E147 FullHalant forms ? add to FullForms
340 + $orpartialforms = "E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A1|E0A2|E0A3|E0A4|E0A5|E0A6|E0A7|E0A8|E0A9|E0AA|E0AB|E0AC|E0AD|E0AE|E0AF|E0B0|E0B1|E0B2|E0B3|E0B4|E0B5|E0B6|E0B7|E0B8|E0B9|E0BA|E0BB|E0BC|E0BD|E0BE|E0BF|E0C0|E0C1|E0C2|E0C3|E0C4|E0C5|E0C6|E0C7|E0C8|E0C9|E0CA|E0CB|E0CC|E0CD|E0CE|E0CF|E0D0|E0D1|E0D2|E0D3|E0D4|E0D5|E0D6|E0D7|E0D8|E0D9|E0DA|E0DB|E0DC|E0DD|E0DE|E0DF|E0E0|E0E1|E0E2|E0E3|E0E4|E0E5|E0E6|E0E7|E0E8|E0E9|E0EA|E0EB|E0EC|E0ED|E0EE|E0EF|E0F0|E0F1|E0F2|E0F3|E0F4|E0F5";
341 +
342 + // Combined MatraIReph[E06D] split [0B3F & E069] to allow reph to be shifted forwards
343 + $vstr = preg_replace('/('.$orfullforms.') E06D ('.$orfullforms.') 0B3E/', '\\1 0B3F E069 \\2 0B3E', $vstr);
344 +
345 +
346 + while(preg_match('/('.$orrephs.') ('.$orpartialforms.')/', $vstr)) {
347 + $vstr = preg_replace('/('.$orrephs.') ('.$orpartialforms.')/', '\\2 \\1', $vstr);
348 + }
349 + $vstr = preg_replace('/('.$orrephs.') ('.$orfullforms.')/', '\\2 \\1', $vstr);
350 +
351 +
352 + // Combine Reph and MatraI
353 + $vstr = str_replace('E069 0B3F', 'E06D', $vstr); // Reph and MatraI -> MatraIReph
354 + $vstr = str_replace('E06A 0B3F', 'E06E', $vstr); // Reph and MatraI -> MatraIReph
355 + $vstr = str_replace('E06B 0B3F', 'E06F', $vstr); // Reph and MatraI -> MatraIReph
356 + }
357 +
358 +
359 + // MALAYALAM Shift Reph to LEFT [E00E] (mlylmRaVattu)
360 + else if ($lang=='ml') {
361 + $halant = "0D4D";
362 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' 0D30/','E00E \\1', $vstr); // 0D30 = Ra
363 + $vstr = preg_replace('/([A-F0-9]{4}) '.$halant.' '.$mlprebasedvowels .' 0D30/','\\2 E00E \\1', $vstr); // 0D30 = Ra
364 +
365 + $mlfullforms = "0D15|0D16|0D17|0D18|0D19|0D1A|0D1B|0D1C|0D1D|0D1E|0D1F|0D20|0D21|0D22|0D23|0D24|0D25|0D26|0D27|0D28|0D2A|0D2B|0D2C|0D2D|0D2E|0D2F|0D30|0D31|0D32|0D33|0D34|0D35|0D36|0D37|0D38|0D39"
366 + ."|E010|E011|E012|E013|E014|E015|E016|E017|E018|E019|E01A|E01B|E01C|E01D|E01E|E01F|E020|E021|E022|E023|E024|E025|E026|E027|E028|E029|E02A|E02B|E02C|E02D|E02E|E02F|E030|E031|E032|E033|E034|E035|E036|E037|E038|E039|E03A|E03B|E03C|E03D|E03E|E03F|E040|E041|E042|E043|E044|E045|E046|E047|E048|E049|E04A|E04B|E04C|E04D|E04E|E04F|E050|E051|E052|E053|E054|E055|E056|E057|E058|E059|E05A|E05B|E05C|E05D|E05E|E05F|E060|E061|E062|E063|E064|E065|E066|E067|E068|E069|E06A|E06B|E06C|E06D|E06E|E06F|E070|E071|E072|E073|E074|E075|E076|E077|E078|E079|E07A|E07B|E07C|E07D";
367 + // = FullConsonants + FullConjuncts
368 +
369 + // = Add Chillu characters // mPDF 5.0.024
370 + $mlfullforms .= "|E004|E005|E006|E007|E008|E009";
371 + while(preg_match('/('.$mlfullforms.') E00E/', $vstr))
372 + $vstr = preg_replace('/('.$mlfullforms.') E00E/', 'E00E \\1', $vstr);
373 + }
374 +
375 + //============================
376 +
377 + // SHIFT post-based vowels to Left of SmallForms (NOT to left of full forms)
378 +
379 + // TELUGU Shift
380 + if ($lang=='te') {
381 + // NB $tevowels defined above
382 + // NB $tefullforms defined above
383 + $tebelowbase1 = "E02C|E02D|E02E|E02F|E030|E031|E032|E033|E034|E035|E036|E037|E038|E039|E03A|E03B|E03C|E03D|E03E|E03F|E040|E041|E042|E043|E044|E045|E046|E047|E048|E049|E04A|E04B|E04C|E04D|E04E"; //'Small1KaToHa'
384 + $tebelowbase2 = "E04F|E050|E051|E052|E053|E054|E055|E056|E057|E058|E059|E05A|E05B|E05C|E05D|E05E|E05F|E060|E061|E062|E063|E064|E065|E066|E067|E068|E069|E06A|E06B|E06C|E06D|E06E|E06F|E070|E071"; // 'Small2KaToHa'
385 + $vstr = preg_replace('/('.$tebelowbase2.') ('.$tevowels.')/', '\\2 \\1', $vstr);
386 + $vstr = preg_replace('/('.$tebelowbase1.') ('.$tevowels.')/', '\\2 \\1', $vstr);
387 + }
388 +
389 +
390 + // KANNADA Shift
391 + else if ($lang=='kn') {
392 + $knvowels = "0CBE|0CC0|0CC1|0CC2|0CC3|0CC4|0CC7|0CC8|0CCA|0CCB|0CD5|0CD6"
393 + // mPDF 5.3.87 Shouldn't swop E082 and E047 (belowbase1) below
394 + // E082 is a matraIligature
395 + // ."|E082|E083|E084|E085|E086|E087|E088|E089|E08A|E08B|E08C|E08D|E08E|E08F|E090|E091|E092|E093|E094|E095|E096|E097|E098|E099|E09A|E09B|E09C|E09D|E09E|E09F|E0A0|E0A1|E0A2|E0A3|E0A4|E0A5|E0A6|E0A7|E0A8|E0A9|E0AA|E0AB"
396 + ."|E004|E007|E008|E009|E00A";
397 +
398 +
399 + // NB $knvowels defined above
400 + // NB $fullforms defined above
401 + // $belowbase1/2 defined above
402 + $vstr = preg_replace('/('.$belowbase2.') ('.$knvowels.')/', '\\2 \\1', $vstr);
403 + // mPDF 5.3.87
404 + $vstr = preg_replace('/('.$belowbase1.') ('.$knvowels.')/', '\\2 \\1', $vstr);
405 +
406 + //$vstr = preg_replace('/('.$fullforms.') ('.$knvowels.')/', '\\2 \\1', $vstr);
407 + }
408 +
409 + //============================
410 + // Clear unwanted ZWJ, ZWNJ
411 + // MALAYALAM
412 + if ($lang=='ml') {
413 + $vstr = preg_replace('/(200C|200D) /','', $vstr);
414 + }
415 +
416 + //============================
417 + // END & PUT IT BACK TOGETHER
418 + $vstr = preg_replace('/^0020 (.*) 0020$/', '\\1', $vstr);
419 +
420 + $varr = explode(" ",$vstr);
421 + $e = '';
422 + foreach($varr AS $v) {
423 + $e.=code2utf(hexdec($v));
424 + }
425 + //============================
426 +
427 + return $e;
428 +}
429 +
430 +
431 +}
432 +
433 +?>