blob: cb43b802e88c257522c0ae60c69b2140a445836f [file] [log] [blame]
Bram Moolenaaredf3f972016-08-29 22:49:24 +02001/* vi:set ts=8 sts=4 sw=4 noet:
Bram Moolenaar071d4272004-06-13 20:20:40 +00002 *
3 * VIM - Vi IMproved by Bram Moolenaar
4 *
5 * Do ":help uganda" in Vim to read copying and usage conditions.
6 * Do ":help credits" in Vim to see a list of people who contributed.
7 * See README.txt for an overview of the Vim source code.
8 */
9
10/*
11 * arabic.c: functions for Arabic language
12 *
Bram Moolenaar071d4272004-06-13 20:20:40 +000013 * Author: Nadim Shaikli & Isam Bayazidi
Bram Moolenaar071d4272004-06-13 20:20:40 +000014 */
15
Bram Moolenaar75464dc2016-07-02 20:27:50 +020016#include "vim.h"
17
18#if defined(FEAT_ARABIC) || defined(PROTO)
19
Bram Moolenaarf28dbce2016-01-29 22:03:47 +010020static int A_is_a(int cur_c);
21static int A_is_s(int cur_c);
22static int A_is_f(int cur_c);
23static int chg_c_a2s(int cur_c);
24static int chg_c_a2i(int cur_c);
25static int chg_c_a2m(int cur_c);
26static int chg_c_a2f(int cur_c);
27static int chg_c_i2m(int cur_c);
28static int chg_c_f2m(int cur_c);
29static int chg_c_laa2i(int hid_c);
30static int chg_c_laa2f(int hid_c);
31static int half_shape(int c);
32static int A_firstc_laa(int c1, int c);
33static int A_is_harakat(int c);
34static int A_is_iso(int c);
35static int A_is_formb(int c);
36static int A_is_ok(int c);
37static int A_is_valid(int c);
38static int A_is_special(int c);
Bram Moolenaar071d4272004-06-13 20:20:40 +000039
40
41/*
42 * Returns True if c is an ISO-8859-6 shaped ARABIC letter (user entered)
43 */
44 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +010045A_is_a(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +000046{
47 switch (cur_c)
48 {
49 case a_HAMZA:
50 case a_ALEF_MADDA:
51 case a_ALEF_HAMZA_ABOVE:
52 case a_WAW_HAMZA:
53 case a_ALEF_HAMZA_BELOW:
54 case a_YEH_HAMZA:
55 case a_ALEF:
56 case a_BEH:
57 case a_TEH_MARBUTA:
58 case a_TEH:
59 case a_THEH:
60 case a_JEEM:
61 case a_HAH:
62 case a_KHAH:
63 case a_DAL:
64 case a_THAL:
65 case a_REH:
66 case a_ZAIN:
67 case a_SEEN:
68 case a_SHEEN:
69 case a_SAD:
70 case a_DAD:
71 case a_TAH:
72 case a_ZAH:
73 case a_AIN:
74 case a_GHAIN:
75 case a_TATWEEL:
76 case a_FEH:
77 case a_QAF:
78 case a_KAF:
79 case a_LAM:
80 case a_MEEM:
81 case a_NOON:
82 case a_HEH:
83 case a_WAW:
84 case a_ALEF_MAKSURA:
85 case a_YEH:
86 return TRUE;
87 }
88
89 return FALSE;
90}
91
92
93/*
94 * Returns True if c is an Isolated Form-B ARABIC letter
95 */
96 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +010097A_is_s(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +000098{
99 switch (cur_c)
100 {
101 case a_s_HAMZA:
102 case a_s_ALEF_MADDA:
103 case a_s_ALEF_HAMZA_ABOVE:
104 case a_s_WAW_HAMZA:
105 case a_s_ALEF_HAMZA_BELOW:
106 case a_s_YEH_HAMZA:
107 case a_s_ALEF:
108 case a_s_BEH:
109 case a_s_TEH_MARBUTA:
110 case a_s_TEH:
111 case a_s_THEH:
112 case a_s_JEEM:
113 case a_s_HAH:
114 case a_s_KHAH:
115 case a_s_DAL:
116 case a_s_THAL:
117 case a_s_REH:
118 case a_s_ZAIN:
119 case a_s_SEEN:
120 case a_s_SHEEN:
121 case a_s_SAD:
122 case a_s_DAD:
123 case a_s_TAH:
124 case a_s_ZAH:
125 case a_s_AIN:
126 case a_s_GHAIN:
127 case a_s_FEH:
128 case a_s_QAF:
129 case a_s_KAF:
130 case a_s_LAM:
131 case a_s_MEEM:
132 case a_s_NOON:
133 case a_s_HEH:
134 case a_s_WAW:
135 case a_s_ALEF_MAKSURA:
136 case a_s_YEH:
137 return TRUE;
138 }
139
140 return FALSE;
141}
142
143
144/*
145 * Returns True if c is a Final shape of an ARABIC letter
146 */
147 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100148A_is_f(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000149{
150 switch (cur_c)
151 {
152 case a_f_ALEF_MADDA:
153 case a_f_ALEF_HAMZA_ABOVE:
154 case a_f_WAW_HAMZA:
155 case a_f_ALEF_HAMZA_BELOW:
156 case a_f_YEH_HAMZA:
157 case a_f_ALEF:
158 case a_f_BEH:
159 case a_f_TEH_MARBUTA:
160 case a_f_TEH:
161 case a_f_THEH:
162 case a_f_JEEM:
163 case a_f_HAH:
164 case a_f_KHAH:
165 case a_f_DAL:
166 case a_f_THAL:
167 case a_f_REH:
168 case a_f_ZAIN:
169 case a_f_SEEN:
170 case a_f_SHEEN:
171 case a_f_SAD:
172 case a_f_DAD:
173 case a_f_TAH:
174 case a_f_ZAH:
175 case a_f_AIN:
176 case a_f_GHAIN:
177 case a_f_FEH:
178 case a_f_QAF:
179 case a_f_KAF:
180 case a_f_LAM:
181 case a_f_MEEM:
182 case a_f_NOON:
183 case a_f_HEH:
184 case a_f_WAW:
185 case a_f_ALEF_MAKSURA:
186 case a_f_YEH:
187 case a_f_LAM_ALEF_MADDA_ABOVE:
188 case a_f_LAM_ALEF_HAMZA_ABOVE:
189 case a_f_LAM_ALEF_HAMZA_BELOW:
190 case a_f_LAM_ALEF:
191 return TRUE;
192 }
193 return FALSE;
194}
195
196
197/*
198 * Change shape - from ISO-8859-6/Isolated to Form-B Isolated
199 */
200 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100201chg_c_a2s(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000202{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000203 switch (cur_c)
204 {
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100205 case a_HAMZA: return a_s_HAMZA;
206 case a_ALEF_MADDA: return a_s_ALEF_MADDA;
207 case a_ALEF_HAMZA_ABOVE: return a_s_ALEF_HAMZA_ABOVE;
208 case a_WAW_HAMZA: return a_s_WAW_HAMZA;
209 case a_ALEF_HAMZA_BELOW: return a_s_ALEF_HAMZA_BELOW;
210 case a_YEH_HAMZA: return a_s_YEH_HAMZA;
211 case a_ALEF: return a_s_ALEF;
212 case a_TEH_MARBUTA: return a_s_TEH_MARBUTA;
213 case a_DAL: return a_s_DAL;
214 case a_THAL: return a_s_THAL;
215 case a_REH: return a_s_REH;
216 case a_ZAIN: return a_s_ZAIN;
217 case a_TATWEEL: return cur_c; /* exceptions */
218 case a_WAW: return a_s_WAW;
219 case a_ALEF_MAKSURA: return a_s_ALEF_MAKSURA;
220 case a_BEH: return a_s_BEH;
221 case a_TEH: return a_s_TEH;
222 case a_THEH: return a_s_THEH;
223 case a_JEEM: return a_s_JEEM;
224 case a_HAH: return a_s_HAH;
225 case a_KHAH: return a_s_KHAH;
226 case a_SEEN: return a_s_SEEN;
227 case a_SHEEN: return a_s_SHEEN;
228 case a_SAD: return a_s_SAD;
229 case a_DAD: return a_s_DAD;
230 case a_TAH: return a_s_TAH;
231 case a_ZAH: return a_s_ZAH;
232 case a_AIN: return a_s_AIN;
233 case a_GHAIN: return a_s_GHAIN;
234 case a_FEH: return a_s_FEH;
235 case a_QAF: return a_s_QAF;
236 case a_KAF: return a_s_KAF;
237 case a_LAM: return a_s_LAM;
238 case a_MEEM: return a_s_MEEM;
239 case a_NOON: return a_s_NOON;
240 case a_HEH: return a_s_HEH;
241 case a_YEH: return a_s_YEH;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000242 }
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100243 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000244}
245
246
247/*
248 * Change shape - from ISO-8859-6/Isolated to Initial
249 */
250 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100251chg_c_a2i(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000252{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000253 switch (cur_c)
254 {
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100255 case a_YEH_HAMZA: return a_i_YEH_HAMZA;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000256 case a_HAMZA: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100257 return a_s_HAMZA;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000258 case a_ALEF_MADDA: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100259 return a_s_ALEF_MADDA;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000260 case a_ALEF_HAMZA_ABOVE: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100261 return a_s_ALEF_HAMZA_ABOVE;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000262 case a_WAW_HAMZA: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100263 return a_s_WAW_HAMZA;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000264 case a_ALEF_HAMZA_BELOW: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100265 return a_s_ALEF_HAMZA_BELOW;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000266 case a_ALEF: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100267 return a_s_ALEF;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000268 case a_TEH_MARBUTA: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100269 return a_s_TEH_MARBUTA;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000270 case a_DAL: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100271 return a_s_DAL;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000272 case a_THAL: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100273 return a_s_THAL;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000274 case a_REH: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100275 return a_s_REH;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000276 case a_ZAIN: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100277 return a_s_ZAIN;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000278 case a_TATWEEL: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100279 return cur_c;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000280 case a_WAW: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100281 return a_s_WAW;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000282 case a_ALEF_MAKSURA: /* exceptions */
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100283 return a_s_ALEF_MAKSURA;
284 case a_BEH: return a_i_BEH;
285 case a_TEH: return a_i_TEH;
286 case a_THEH: return a_i_THEH;
287 case a_JEEM: return a_i_JEEM;
288 case a_HAH: return a_i_HAH;
289 case a_KHAH: return a_i_KHAH;
290 case a_SEEN: return a_i_SEEN;
291 case a_SHEEN: return a_i_SHEEN;
292 case a_SAD: return a_i_SAD;
293 case a_DAD: return a_i_DAD;
294 case a_TAH: return a_i_TAH;
295 case a_ZAH: return a_i_ZAH;
296 case a_AIN: return a_i_AIN;
297 case a_GHAIN: return a_i_GHAIN;
298 case a_FEH: return a_i_FEH;
299 case a_QAF: return a_i_QAF;
300 case a_KAF: return a_i_KAF;
301 case a_LAM: return a_i_LAM;
302 case a_MEEM: return a_i_MEEM;
303 case a_NOON: return a_i_NOON;
304 case a_HEH: return a_i_HEH;
305 case a_YEH: return a_i_YEH;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000306 }
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100307 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000308}
309
310
311/*
312 * Change shape - from ISO-8859-6/Isolated to Medial
313 */
314 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100315chg_c_a2m(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000316{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000317 switch (cur_c)
318 {
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100319 case a_HAMZA: return a_s_HAMZA; /* exception */
320 case a_ALEF_MADDA: return a_f_ALEF_MADDA; /* exception */
321 case a_ALEF_HAMZA_ABOVE: return a_f_ALEF_HAMZA_ABOVE; /* exception */
322 case a_WAW_HAMZA: return a_f_WAW_HAMZA; /* exception */
323 case a_ALEF_HAMZA_BELOW: return a_f_ALEF_HAMZA_BELOW; /* exception */
324 case a_YEH_HAMZA: return a_m_YEH_HAMZA;
325 case a_ALEF: return a_f_ALEF; /* exception */
326 case a_BEH: return a_m_BEH;
327 case a_TEH_MARBUTA: return a_f_TEH_MARBUTA; /* exception */
328 case a_TEH: return a_m_TEH;
329 case a_THEH: return a_m_THEH;
330 case a_JEEM: return a_m_JEEM;
331 case a_HAH: return a_m_HAH;
332 case a_KHAH: return a_m_KHAH;
333 case a_DAL: return a_f_DAL; /* exception */
334 case a_THAL: return a_f_THAL; /* exception */
335 case a_REH: return a_f_REH; /* exception */
336 case a_ZAIN: return a_f_ZAIN; /* exception */
337 case a_SEEN: return a_m_SEEN;
338 case a_SHEEN: return a_m_SHEEN;
339 case a_SAD: return a_m_SAD;
340 case a_DAD: return a_m_DAD;
341 case a_TAH: return a_m_TAH;
342 case a_ZAH: return a_m_ZAH;
343 case a_AIN: return a_m_AIN;
344 case a_GHAIN: return a_m_GHAIN;
345 case a_TATWEEL: return cur_c; /* exception */
346 case a_FEH: return a_m_FEH;
347 case a_QAF: return a_m_QAF;
348 case a_KAF: return a_m_KAF;
349 case a_LAM: return a_m_LAM;
350 case a_MEEM: return a_m_MEEM;
351 case a_NOON: return a_m_NOON;
352 case a_HEH: return a_m_HEH;
353 case a_WAW: return a_f_WAW; /* exception */
354 case a_ALEF_MAKSURA: return a_f_ALEF_MAKSURA; /* exception */
355 case a_YEH: return a_m_YEH;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000356 }
Bram Moolenaar5f53dd32017-03-01 14:02:30 +0100357 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000358}
359
360
361/*
362 * Change shape - from ISO-8859-6/Isolated to final
363 */
364 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100365chg_c_a2f(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000366{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000367 /* NOTE: these encodings need to be accounted for
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100368 * a_f_ALEF_MADDA;
369 * a_f_ALEF_HAMZA_ABOVE;
370 * a_f_ALEF_HAMZA_BELOW;
371 * a_f_LAM_ALEF_MADDA_ABOVE;
372 * a_f_LAM_ALEF_HAMZA_ABOVE;
373 * a_f_LAM_ALEF_HAMZA_BELOW;
374 */
Bram Moolenaar071d4272004-06-13 20:20:40 +0000375 switch (cur_c)
376 {
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100377 case a_HAMZA: return a_s_HAMZA; /* exception */
378 case a_ALEF_MADDA: return a_f_ALEF_MADDA;
379 case a_ALEF_HAMZA_ABOVE: return a_f_ALEF_HAMZA_ABOVE;
380 case a_WAW_HAMZA: return a_f_WAW_HAMZA;
381 case a_ALEF_HAMZA_BELOW: return a_f_ALEF_HAMZA_BELOW;
382 case a_YEH_HAMZA: return a_f_YEH_HAMZA;
383 case a_ALEF: return a_f_ALEF;
384 case a_BEH: return a_f_BEH;
385 case a_TEH_MARBUTA: return a_f_TEH_MARBUTA;
386 case a_TEH: return a_f_TEH;
387 case a_THEH: return a_f_THEH;
388 case a_JEEM: return a_f_JEEM;
389 case a_HAH: return a_f_HAH;
390 case a_KHAH: return a_f_KHAH;
391 case a_DAL: return a_f_DAL;
392 case a_THAL: return a_f_THAL;
393 case a_REH: return a_f_REH;
394 case a_ZAIN: return a_f_ZAIN;
395 case a_SEEN: return a_f_SEEN;
396 case a_SHEEN: return a_f_SHEEN;
397 case a_SAD: return a_f_SAD;
398 case a_DAD: return a_f_DAD;
399 case a_TAH: return a_f_TAH;
400 case a_ZAH: return a_f_ZAH;
401 case a_AIN: return a_f_AIN;
402 case a_GHAIN: return a_f_GHAIN;
403 case a_TATWEEL: return cur_c; /* exception */
404 case a_FEH: return a_f_FEH;
405 case a_QAF: return a_f_QAF;
406 case a_KAF: return a_f_KAF;
407 case a_LAM: return a_f_LAM;
408 case a_MEEM: return a_f_MEEM;
409 case a_NOON: return a_f_NOON;
410 case a_HEH: return a_f_HEH;
411 case a_WAW: return a_f_WAW;
412 case a_ALEF_MAKSURA: return a_f_ALEF_MAKSURA;
413 case a_YEH: return a_f_YEH;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000414 }
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100415 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000416}
417
418
419/*
420 * Change shape - from Initial to Medial
421 */
422 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100423chg_c_i2m(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000424{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000425 switch (cur_c)
426 {
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100427 case a_i_YEH_HAMZA: return a_m_YEH_HAMZA;
428 case a_i_BEH: return a_m_BEH;
429 case a_i_TEH: return a_m_TEH;
430 case a_i_THEH: return a_m_THEH;
431 case a_i_JEEM: return a_m_JEEM;
432 case a_i_HAH: return a_m_HAH;
433 case a_i_KHAH: return a_m_KHAH;
434 case a_i_SEEN: return a_m_SEEN;
435 case a_i_SHEEN: return a_m_SHEEN;
436 case a_i_SAD: return a_m_SAD;
437 case a_i_DAD: return a_m_DAD;
438 case a_i_TAH: return a_m_TAH;
439 case a_i_ZAH: return a_m_ZAH;
440 case a_i_AIN: return a_m_AIN;
441 case a_i_GHAIN: return a_m_GHAIN;
442 case a_i_FEH: return a_m_FEH;
443 case a_i_QAF: return a_m_QAF;
444 case a_i_KAF: return a_m_KAF;
445 case a_i_LAM: return a_m_LAM;
446 case a_i_MEEM: return a_m_MEEM;
447 case a_i_NOON: return a_m_NOON;
448 case a_i_HEH: return a_m_HEH;
449 case a_i_YEH: return a_m_YEH;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000450 }
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100451 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000452}
453
454
455/*
456 * Change shape - from Final to Medial
457 */
458 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100459chg_c_f2m(int cur_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000460{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000461 switch (cur_c)
462 {
463 /* NOTE: these encodings are multi-positional, no ?
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100464 * case a_f_ALEF_MADDA:
465 * case a_f_ALEF_HAMZA_ABOVE:
466 * case a_f_ALEF_HAMZA_BELOW:
467 */
468 case a_f_YEH_HAMZA: return a_m_YEH_HAMZA;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000469 case a_f_WAW_HAMZA: /* exceptions */
470 case a_f_ALEF:
471 case a_f_TEH_MARBUTA:
472 case a_f_DAL:
473 case a_f_THAL:
474 case a_f_REH:
475 case a_f_ZAIN:
476 case a_f_WAW:
477 case a_f_ALEF_MAKSURA:
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100478 return cur_c;
479 case a_f_BEH: return a_m_BEH;
480 case a_f_TEH: return a_m_TEH;
481 case a_f_THEH: return a_m_THEH;
482 case a_f_JEEM: return a_m_JEEM;
483 case a_f_HAH: return a_m_HAH;
484 case a_f_KHAH: return a_m_KHAH;
485 case a_f_SEEN: return a_m_SEEN;
486 case a_f_SHEEN: return a_m_SHEEN;
487 case a_f_SAD: return a_m_SAD;
488 case a_f_DAD: return a_m_DAD;
489 case a_f_TAH: return a_m_TAH;
490 case a_f_ZAH: return a_m_ZAH;
491 case a_f_AIN: return a_m_AIN;
492 case a_f_GHAIN: return a_m_GHAIN;
493 case a_f_FEH: return a_m_FEH;
494 case a_f_QAF: return a_m_QAF;
495 case a_f_KAF: return a_m_KAF;
496 case a_f_LAM: return a_m_LAM;
497 case a_f_MEEM: return a_m_MEEM;
498 case a_f_NOON: return a_m_NOON;
499 case a_f_HEH: return a_m_HEH;
500 case a_f_YEH: return a_m_YEH;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000501
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100502 /* NOTE: these encodings are multi-positional, no ?
503 * case a_f_LAM_ALEF_MADDA_ABOVE:
504 * case a_f_LAM_ALEF_HAMZA_ABOVE:
505 * case a_f_LAM_ALEF_HAMZA_BELOW:
506 * case a_f_LAM_ALEF:
507 */
508 }
509 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000510}
511
512
513/*
514 * Change shape - from Combination (2 char) to an Isolated
515 */
516 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100517chg_c_laa2i(int hid_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000518{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000519 switch (hid_c)
520 {
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100521 case a_ALEF_MADDA: return a_s_LAM_ALEF_MADDA_ABOVE;
522 case a_ALEF_HAMZA_ABOVE: return a_s_LAM_ALEF_HAMZA_ABOVE;
523 case a_ALEF_HAMZA_BELOW: return a_s_LAM_ALEF_HAMZA_BELOW;
524 case a_ALEF: return a_s_LAM_ALEF;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000525 }
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100526 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000527}
528
529
530/*
531 * Change shape - from Combination-Isolated to Final
532 */
533 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100534chg_c_laa2f(int hid_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000535{
Bram Moolenaar071d4272004-06-13 20:20:40 +0000536 switch (hid_c)
537 {
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100538 case a_ALEF_MADDA: return a_f_LAM_ALEF_MADDA_ABOVE;
539 case a_ALEF_HAMZA_ABOVE: return a_f_LAM_ALEF_HAMZA_ABOVE;
540 case a_ALEF_HAMZA_BELOW: return a_f_LAM_ALEF_HAMZA_BELOW;
541 case a_ALEF: return a_f_LAM_ALEF;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000542 }
Bram Moolenaar7f73b542017-03-04 14:50:19 +0100543 return 0;
Bram Moolenaar071d4272004-06-13 20:20:40 +0000544}
545
546/*
547 * Do "half-shaping" on character "c". Return zero if no shaping.
548 */
549 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100550half_shape(int c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000551{
552 if (A_is_a(c))
553 return chg_c_a2i(c);
554 if (A_is_valid(c) && A_is_f(c))
555 return chg_c_f2m(c);
556 return 0;
557}
558
559/*
560 * Do Arabic shaping on character "c". Returns the shaped character.
561 * out: "ccp" points to the first byte of the character to be shaped.
562 * in/out: "c1p" points to the first composing char for "c".
563 * in: "prev_c" is the previous character (not shaped)
564 * in: "prev_c1" is the first composing char for the previous char
565 * (not shaped)
566 * in: "next_c" is the next character (not shaped).
567 */
568 int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100569arabic_shape(
570 int c,
571 int *ccp,
572 int *c1p,
573 int prev_c,
574 int prev_c1,
575 int next_c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000576{
577 int curr_c;
578 int shape_c;
579 int curr_laa;
580 int prev_laa;
581
582 /* Deal only with Arabic character, pass back all others */
583 if (!A_is_ok(c))
584 return c;
585
586 /* half-shape current and previous character */
587 shape_c = half_shape(prev_c);
588
589 /* Save away current character */
590 curr_c = c;
591
592 curr_laa = A_firstc_laa(c, *c1p);
593 prev_laa = A_firstc_laa(prev_c, prev_c1);
594
595 if (curr_laa)
596 {
597 if (A_is_valid(prev_c) && !A_is_f(shape_c)
598 && !A_is_s(shape_c) && !prev_laa)
599 curr_c = chg_c_laa2f(curr_laa);
600 else
601 curr_c = chg_c_laa2i(curr_laa);
602
603 /* Remove the composing character */
604 *c1p = 0;
605 }
606 else if (!A_is_valid(prev_c) && A_is_valid(next_c))
607 curr_c = chg_c_a2i(c);
608 else if (!shape_c || A_is_f(shape_c) || A_is_s(shape_c) || prev_laa)
609 curr_c = A_is_valid(next_c) ? chg_c_a2i(c) : chg_c_a2s(c);
610 else if (A_is_valid(next_c))
611 curr_c = A_is_iso(c) ? chg_c_a2m(c) : chg_c_i2m(c);
612 else if (A_is_valid(prev_c))
613 curr_c = chg_c_a2f(c);
614 else
615 curr_c = chg_c_a2s(c);
616
617 /* Sanity check -- curr_c should, in the future, never be 0.
618 * We should, in the future, insert a fatal error here. */
619 if (curr_c == NUL)
620 curr_c = c;
621
622 if (curr_c != c && ccp != NULL)
623 {
Bram Moolenaar9a920d82012-06-01 15:21:02 +0200624 char_u buf[MB_MAXBYTES + 1];
Bram Moolenaar071d4272004-06-13 20:20:40 +0000625
626 /* Update the first byte of the character. */
627 (*mb_char2bytes)(curr_c, buf);
628 *ccp = buf[0];
629 }
630
631 /* Return the shaped character */
632 return curr_c;
633}
634
635
636/*
637 * A_firstc_laa returns first character of LAA combination if it exists
638 */
639 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100640A_firstc_laa(
641 int c, /* base character */
642 int c1) /* first composing character */
Bram Moolenaar071d4272004-06-13 20:20:40 +0000643{
644 if (c1 != NUL && c == a_LAM && !A_is_harakat(c1))
645 return c1;
646 return 0;
647}
648
649
650/*
651 * A_is_harakat returns TRUE if 'c' is an Arabic Harakat character
652 * (harakat/tanween)
653 */
654 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100655A_is_harakat(int c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000656{
657 return (c >= a_FATHATAN && c <= a_SUKUN);
658}
659
660
661/*
662 * A_is_iso returns TRUE if 'c' is an Arabic ISO-8859-6 character
663 * (alphabet/number/punctuation)
664 */
665 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100666A_is_iso(int c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000667{
668 return ((c >= a_HAMZA && c <= a_GHAIN)
669 || (c >= a_TATWEEL && c <= a_HAMZA_BELOW)
670 || c == a_MINI_ALEF);
671}
672
673
674/*
675 * A_is_formb returns TRUE if 'c' is an Arabic 10646-1 FormB character
676 * (alphabet/number/punctuation)
677 */
678 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100679A_is_formb(int c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000680{
681 return ((c >= a_s_FATHATAN && c <= a_s_DAMMATAN)
682 || c == a_s_KASRATAN
683 || (c >= a_s_FATHA && c <= a_f_LAM_ALEF)
684 || c == a_BYTE_ORDER_MARK);
685}
686
687
688/*
689 * A_is_ok returns TRUE if 'c' is an Arabic 10646 (8859-6 or Form-B)
690 */
691 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100692A_is_ok(int c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000693{
694 return (A_is_iso(c) || A_is_formb(c));
695}
696
697
698/*
699 * A_is_valid returns TRUE if 'c' is an Arabic 10646 (8859-6 or Form-B)
700 * with some exceptions/exclusions
701 */
702 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100703A_is_valid(int c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000704{
705 return (A_is_ok(c) && !A_is_special(c));
706}
707
708
709/*
710 * A_is_special returns TRUE if 'c' is not a special Arabic character.
711 * Specials don't adhere to most of the rules.
712 */
713 static int
Bram Moolenaar7454a062016-01-30 15:14:10 +0100714A_is_special(int c)
Bram Moolenaar071d4272004-06-13 20:20:40 +0000715{
716 return (c == a_HAMZA || c == a_s_HAMZA);
717}
Bram Moolenaar75464dc2016-07-02 20:27:50 +0200718
719#endif /* FEAT_ARABIC */