src/arabic.c

   1 /* vi:set ts=8 sts=4 sw=4:
   2  *
   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  *
  13  * Included by main.c, when FEAT_ARABIC & FEAT_GUI is defined.
  14  *
  15  * --
  16  *
  17  * Author: Nadim Shaikli & Isam Bayazidi
  18  *
  19  */
  20
  21 static int  A_is_a __ARGS((int cur_c));
  22 static int  A_is_s __ARGS((int cur_c));
  23 static int  A_is_f __ARGS((int cur_c));
  24 static int  chg_c_a2s __ARGS((int cur_c));
  25 static int  chg_c_a2i __ARGS((int cur_c));
  26 static int  chg_c_a2m __ARGS((int cur_c));
  27 static int  chg_c_a2f __ARGS((int cur_c));
  28 static int  chg_c_i2m __ARGS((int cur_c));
  29 static int  chg_c_f2m __ARGS((int cur_c));
  30 static int  chg_c_laa2i __ARGS((int hid_c));
  31 static int  chg_c_laa2f __ARGS((int hid_c));
  32 static int  half_shape __ARGS((int c));
  33 static int  A_firstc_laa __ARGS((int c1, int c));
  34 static int  A_is_harakat __ARGS((int c));
  35 static int  A_is_iso __ARGS((int c));
  36 static int  A_is_formb __ARGS((int c));
  37 static int  A_is_ok __ARGS((int c));
  38 static int  A_is_valid __ARGS((int c));
  39 static int  A_is_special __ARGS((int c));
  40
  41
  42 /*
  43  * Returns True if c is an ISO-8859-6 shaped ARABIC letter (user entered)
  44  */
  45     static int
  46 A_is_a(cur_c)
  47     int cur_c;
  48 {
  49     switch (cur_c)
  50     {
  51         case a_HAMZA:
  52         case a_ALEF_MADDA:
  53         case a_ALEF_HAMZA_ABOVE:
  54         case a_WAW_HAMZA:
  55         case a_ALEF_HAMZA_BELOW:
  56         case a_YEH_HAMZA:
  57         case a_ALEF:
  58         case a_BEH:
  59         case a_TEH_MARBUTA:
  60         case a_TEH:
  61         case a_THEH:
  62         case a_JEEM:
  63         case a_HAH:
  64         case a_KHAH:
  65         case a_DAL:
  66         case a_THAL:
  67         case a_REH:
  68         case a_ZAIN:
  69         case a_SEEN:
  70         case a_SHEEN:
  71         case a_SAD:
  72         case a_DAD:
  73         case a_TAH:
  74         case a_ZAH:
  75         case a_AIN:
  76         case a_GHAIN:
  77         case a_TATWEEL:
  78         case a_FEH:
  79         case a_QAF:
  80         case a_KAF:
  81         case a_LAM:
  82         case a_MEEM:
  83         case a_NOON:
  84         case a_HEH:
  85         case a_WAW:
  86         case a_ALEF_MAKSURA:
  87         case a_YEH:
  88             return TRUE;
  89     }
  90
  91     return FALSE;
  92 }
  93
  94
  95 /*
  96  * Returns True if c is an Isolated Form-B ARABIC letter
  97  */
  98     static int
  99 A_is_s(cur_c)
 100     int cur_c;
 101 {
 102     switch (cur_c)
 103     {
 104         case a_s_HAMZA:
 105         case a_s_ALEF_MADDA:
 106         case a_s_ALEF_HAMZA_ABOVE:
 107         case a_s_WAW_HAMZA:
 108         case a_s_ALEF_HAMZA_BELOW:
 109         case a_s_YEH_HAMZA:
 110         case a_s_ALEF:
 111         case a_s_BEH:
 112         case a_s_TEH_MARBUTA:
 113         case a_s_TEH:
 114         case a_s_THEH:
 115         case a_s_JEEM:
 116         case a_s_HAH:
 117         case a_s_KHAH:
 118         case a_s_DAL:
 119         case a_s_THAL:
 120         case a_s_REH:
 121         case a_s_ZAIN:
 122         case a_s_SEEN:
 123         case a_s_SHEEN:
 124         case a_s_SAD:
 125         case a_s_DAD:
 126         case a_s_TAH:
 127         case a_s_ZAH:
 128         case a_s_AIN:
 129         case a_s_GHAIN:
 130         case a_s_FEH:
 131         case a_s_QAF:
 132         case a_s_KAF:
 133         case a_s_LAM:
 134         case a_s_MEEM:
 135         case a_s_NOON:
 136         case a_s_HEH:
 137         case a_s_WAW:
 138         case a_s_ALEF_MAKSURA:
 139         case a_s_YEH:
 140             return TRUE;
 141     }
 142
 143     return FALSE;
 144 }
 145
 146
 147 /*
 148  * Returns True if c is a Final shape of an ARABIC letter
 149  */
 150     static int
 151 A_is_f(cur_c)
 152     int cur_c;
 153 {
 154     switch (cur_c)
 155     {
 156         case a_f_ALEF_MADDA:
 157         case a_f_ALEF_HAMZA_ABOVE:
 158         case a_f_WAW_HAMZA:
 159         case a_f_ALEF_HAMZA_BELOW:
 160         case a_f_YEH_HAMZA:
 161         case a_f_ALEF:
 162         case a_f_BEH:
 163         case a_f_TEH_MARBUTA:
 164         case a_f_TEH:
 165         case a_f_THEH:
 166         case a_f_JEEM:
 167         case a_f_HAH:
 168         case a_f_KHAH:
 169         case a_f_DAL:
 170         case a_f_THAL:
 171         case a_f_REH:
 172         case a_f_ZAIN:
 173         case a_f_SEEN:
 174         case a_f_SHEEN:
 175         case a_f_SAD:
 176         case a_f_DAD:
 177         case a_f_TAH:
 178         case a_f_ZAH:
 179         case a_f_AIN:
 180         case a_f_GHAIN:
 181         case a_f_FEH:
 182         case a_f_QAF:
 183         case a_f_KAF:
 184         case a_f_LAM:
 185         case a_f_MEEM:
 186         case a_f_NOON:
 187         case a_f_HEH:
 188         case a_f_WAW:
 189         case a_f_ALEF_MAKSURA:
 190         case a_f_YEH:
 191         case a_f_LAM_ALEF_MADDA_ABOVE:
 192         case a_f_LAM_ALEF_HAMZA_ABOVE:
 193         case a_f_LAM_ALEF_HAMZA_BELOW:
 194         case a_f_LAM_ALEF:
 195             return TRUE;
 196     }
 197     return FALSE;
 198 }
 199
 200
 201 /*
 202  * Change shape - from ISO-8859-6/Isolated to Form-B Isolated
 203  */
 204     static int
 205 chg_c_a2s(cur_c)
 206     int cur_c;
 207 {
 208     int tempc;
 209
 210     switch (cur_c)
 211     {
 212         case a_HAMZA:
 213             tempc = a_s_HAMZA;
 214             break;
 215         case a_ALEF_MADDA:
 216             tempc = a_s_ALEF_MADDA;
 217             break;
 218         case a_ALEF_HAMZA_ABOVE:
 219             tempc = a_s_ALEF_HAMZA_ABOVE;
 220             break;
 221         case a_WAW_HAMZA:
 222             tempc = a_s_WAW_HAMZA;
 223             break;
 224         case a_ALEF_HAMZA_BELOW:
 225             tempc = a_s_ALEF_HAMZA_BELOW;
 226             break;
 227         case a_YEH_HAMZA:
 228             tempc = a_s_YEH_HAMZA;
 229             break;
 230         case a_ALEF:
 231             tempc = a_s_ALEF;
 232             break;
 233         case a_TEH_MARBUTA:
 234             tempc = a_s_TEH_MARBUTA;
 235             break;
 236         case a_DAL:
 237             tempc = a_s_DAL;
 238             break;
 239         case a_THAL:
 240             tempc = a_s_THAL;
 241             break;
 242         case a_REH:
 243             tempc = a_s_REH;
 244             break;
 245         case a_ZAIN:
 246             tempc = a_s_ZAIN;
 247             break;
 248         case a_TATWEEL:                 /* exceptions */
 249             tempc = cur_c;
 250             break;
 251         case a_WAW:
 252             tempc = a_s_WAW;
 253             break;
 254         case a_ALEF_MAKSURA:
 255             tempc = a_s_ALEF_MAKSURA;
 256             break;
 257         case a_BEH:
 258             tempc = a_s_BEH;
 259             break;
 260         case a_TEH:
 261             tempc = a_s_TEH;
 262             break;
 263         case a_THEH:
 264             tempc = a_s_THEH;
 265             break;
 266         case a_JEEM:
 267             tempc = a_s_JEEM;
 268             break;
 269         case a_HAH:
 270             tempc = a_s_HAH;
 271             break;
 272         case a_KHAH:
 273             tempc = a_s_KHAH;
 274             break;
 275         case a_SEEN:
 276             tempc = a_s_SEEN;
 277             break;
 278         case a_SHEEN:
 279             tempc = a_s_SHEEN;
 280             break;
 281         case a_SAD:
 282             tempc = a_s_SAD;
 283             break;
 284         case a_DAD:
 285             tempc = a_s_DAD;
 286             break;
 287         case a_TAH:
 288             tempc = a_s_TAH;
 289             break;
 290         case a_ZAH:
 291             tempc = a_s_ZAH;
 292             break;
 293         case a_AIN:
 294             tempc = a_s_AIN;
 295             break;
 296         case a_GHAIN:
 297             tempc = a_s_GHAIN;
 298             break;
 299         case a_FEH:
 300             tempc = a_s_FEH;
 301             break;
 302         case a_QAF:
 303             tempc = a_s_QAF;
 304             break;
 305         case a_KAF:
 306             tempc = a_s_KAF;
 307             break;
 308         case a_LAM:
 309             tempc = a_s_LAM;
 310             break;
 311         case a_MEEM:
 312             tempc = a_s_MEEM;
 313             break;
 314         case a_NOON:
 315             tempc = a_s_NOON;
 316             break;
 317         case a_HEH:
 318             tempc = a_s_HEH;
 319             break;
 320         case a_YEH:
 321             tempc = a_s_YEH;
 322             break;
 323         default:
 324             tempc = 0;
 325     }
 326
 327     return tempc;
 328 }
 329
 330
 331 /*
 332  * Change shape - from ISO-8859-6/Isolated to Initial
 333  */
 334     static int
 335 chg_c_a2i(cur_c)
 336     int cur_c;
 337 {
 338     int tempc;
 339
 340     switch (cur_c)
 341     {
 342         case a_YEH_HAMZA:
 343             tempc = a_i_YEH_HAMZA;
 344             break;
 345         case a_HAMZA:                   /* exceptions */
 346             tempc = a_s_HAMZA;
 347             break;
 348         case a_ALEF_MADDA:              /* exceptions */
 349             tempc = a_s_ALEF_MADDA;
 350             break;
 351         case a_ALEF_HAMZA_ABOVE:        /* exceptions */
 352             tempc = a_s_ALEF_HAMZA_ABOVE;
 353             break;
 354         case a_WAW_HAMZA:               /* exceptions */
 355             tempc = a_s_WAW_HAMZA;
 356             break;
 357         case a_ALEF_HAMZA_BELOW:        /* exceptions */
 358             tempc = a_s_ALEF_HAMZA_BELOW;
 359             break;
 360         case a_ALEF:                    /* exceptions */
 361             tempc = a_s_ALEF;
 362             break;
 363         case a_TEH_MARBUTA:             /* exceptions */
 364             tempc = a_s_TEH_MARBUTA;
 365             break;
 366         case a_DAL:                     /* exceptions */
 367             tempc = a_s_DAL;
 368             break;
 369         case a_THAL:                    /* exceptions */
 370             tempc = a_s_THAL;
 371             break;
 372         case a_REH:                     /* exceptions */
 373             tempc = a_s_REH;
 374             break;
 375         case a_ZAIN:                    /* exceptions */
 376             tempc = a_s_ZAIN;
 377             break;
 378         case a_TATWEEL:                 /* exceptions */
 379             tempc = cur_c;
 380             break;
 381         case a_WAW:                     /* exceptions */
 382             tempc = a_s_WAW;
 383             break;
 384         case a_ALEF_MAKSURA:            /* exceptions */
 385             tempc = a_s_ALEF_MAKSURA;
 386             break;
 387         case a_BEH:
 388             tempc = a_i_BEH;
 389             break;
 390         case a_TEH:
 391             tempc = a_i_TEH;
 392             break;
 393         case a_THEH:
 394             tempc = a_i_THEH;
 395             break;
 396         case a_JEEM:
 397             tempc = a_i_JEEM;
 398             break;
 399         case a_HAH:
 400             tempc = a_i_HAH;
 401             break;
 402         case a_KHAH:
 403             tempc = a_i_KHAH;
 404             break;
 405         case a_SEEN:
 406             tempc = a_i_SEEN;
 407             break;
 408         case a_SHEEN:
 409             tempc = a_i_SHEEN;
 410             break;
 411         case a_SAD:
 412             tempc = a_i_SAD;
 413             break;
 414         case a_DAD:
 415             tempc = a_i_DAD;
 416             break;
 417         case a_TAH:
 418             tempc = a_i_TAH;
 419             break;
 420         case a_ZAH:
 421             tempc = a_i_ZAH;
 422             break;
 423         case a_AIN:
 424             tempc = a_i_AIN;
 425             break;
 426         case a_GHAIN:
 427             tempc = a_i_GHAIN;
 428             break;
 429         case a_FEH:
 430             tempc = a_i_FEH;
 431             break;
 432         case a_QAF:
 433             tempc = a_i_QAF;
 434             break;
 435         case a_KAF:
 436             tempc = a_i_KAF;
 437             break;
 438         case a_LAM:
 439             tempc = a_i_LAM;
 440             break;
 441         case a_MEEM:
 442             tempc = a_i_MEEM;
 443             break;
 444         case a_NOON:
 445             tempc = a_i_NOON;
 446             break;
 447         case a_HEH:
 448             tempc = a_i_HEH;
 449             break;
 450         case a_YEH:
 451             tempc = a_i_YEH;
 452             break;
 453         default:
 454             tempc = 0;
 455     }
 456
 457     return tempc;
 458 }
 459
 460
 461 /*
 462  * Change shape - from ISO-8859-6/Isolated to Medial
 463  */
 464     static int
 465 chg_c_a2m(cur_c)
 466     int cur_c;
 467 {
 468     int tempc;
 469
 470     switch (cur_c)
 471     {
 472         case a_HAMZA:                   /* exception */
 473             tempc = a_s_HAMZA;
 474             break;
 475         case a_ALEF_MADDA:              /* exception */
 476             tempc = a_f_ALEF_MADDA;
 477             break;
 478         case a_ALEF_HAMZA_ABOVE:        /* exception */
 479             tempc = a_f_ALEF_HAMZA_ABOVE;
 480             break;
 481         case a_WAW_HAMZA:               /* exception */
 482             tempc = a_f_WAW_HAMZA;
 483             break;
 484         case a_ALEF_HAMZA_BELOW:        /* exception */
 485             tempc = a_f_ALEF_HAMZA_BELOW;
 486             break;
 487         case a_YEH_HAMZA:
 488             tempc = a_m_YEH_HAMZA;
 489             break;
 490         case a_ALEF:                    /* exception */
 491             tempc = a_f_ALEF;
 492             break;
 493         case a_BEH:
 494             tempc = a_m_BEH;
 495             break;
 496         case a_TEH_MARBUTA:             /* exception */
 497             tempc = a_f_TEH_MARBUTA;
 498             break;
 499         case a_TEH:
 500             tempc = a_m_TEH;
 501             break;
 502         case a_THEH:
 503             tempc = a_m_THEH;
 504             break;
 505         case a_JEEM:
 506             tempc = a_m_JEEM;
 507             break;
 508         case a_HAH:
 509             tempc = a_m_HAH;
 510             break;
 511         case a_KHAH:
 512             tempc = a_m_KHAH;
 513             break;
 514         case a_DAL:                     /* exception */
 515             tempc = a_f_DAL;
 516             break;
 517         case a_THAL:                    /* exception */
 518             tempc = a_f_THAL;
 519             break;
 520         case a_REH:                     /* exception */
 521             tempc = a_f_REH;
 522             break;
 523         case a_ZAIN:                    /* exception */
 524             tempc = a_f_ZAIN;
 525             break;
 526         case a_SEEN:
 527             tempc = a_m_SEEN;
 528             break;
 529         case a_SHEEN:
 530             tempc = a_m_SHEEN;
 531             break;
 532         case a_SAD:
 533             tempc = a_m_SAD;
 534             break;
 535         case a_DAD:
 536             tempc = a_m_DAD;
 537             break;
 538         case a_TAH:
 539             tempc = a_m_TAH;
 540             break;
 541         case a_ZAH:
 542             tempc = a_m_ZAH;
 543             break;
 544         case a_AIN:
 545             tempc = a_m_AIN;
 546             break;
 547         case a_GHAIN:
 548             tempc = a_m_GHAIN;
 549             break;
 550         case a_TATWEEL:                 /* exception */
 551             tempc = cur_c;
 552             break;
 553         case a_FEH:
 554             tempc = a_m_FEH;
 555             break;
 556         case a_QAF:
 557             tempc = a_m_QAF;
 558             break;
 559         case a_KAF:
 560             tempc = a_m_KAF;
 561             break;
 562         case a_LAM:
 563             tempc = a_m_LAM;
 564             break;
 565         case a_MEEM:
 566             tempc = a_m_MEEM;
 567             break;
 568         case a_NOON:
 569             tempc = a_m_NOON;
 570             break;
 571         case a_HEH:
 572             tempc = a_m_HEH;
 573             break;
 574         case a_WAW:                     /* exception */
 575             tempc = a_f_WAW;
 576             break;
 577         case a_ALEF_MAKSURA:            /* exception */
 578             tempc = a_f_ALEF_MAKSURA;
 579             break;
 580         case a_YEH:
 581             tempc = a_m_YEH;
 582             break;
 583         default:
 584             tempc = 0;
 585     }
 586
 587     return tempc;
 588 }
 589
 590
 591 /*
 592  * Change shape - from ISO-8859-6/Isolated to final
 593  */
 594     static int
 595 chg_c_a2f(cur_c)
 596     int cur_c;
 597 {
 598     int tempc;
 599
 600     /* NOTE: these encodings need to be accounted for
 601
 602         a_f_ALEF_MADDA;
 603         a_f_ALEF_HAMZA_ABOVE;
 604         a_f_ALEF_HAMZA_BELOW;
 605         a_f_LAM_ALEF_MADDA_ABOVE;
 606         a_f_LAM_ALEF_HAMZA_ABOVE;
 607         a_f_LAM_ALEF_HAMZA_BELOW;
 608         */
 609
 610     switch (cur_c)
 611     {
 612         case a_HAMZA:                   /* exception */
 613             tempc = a_s_HAMZA;
 614             break;
 615         case a_ALEF_MADDA:
 616             tempc = a_f_ALEF_MADDA;
 617             break;
 618         case a_ALEF_HAMZA_ABOVE:
 619             tempc = a_f_ALEF_HAMZA_ABOVE;
 620             break;
 621         case a_WAW_HAMZA:
 622             tempc = a_f_WAW_HAMZA;
 623             break;
 624         case a_ALEF_HAMZA_BELOW:
 625             tempc = a_f_ALEF_HAMZA_BELOW;
 626             break;
 627         case a_YEH_HAMZA:
 628             tempc = a_f_YEH_HAMZA;
 629             break;
 630         case a_ALEF:
 631             tempc = a_f_ALEF;
 632             break;
 633         case a_BEH:
 634             tempc = a_f_BEH;
 635             break;
 636         case a_TEH_MARBUTA:
 637             tempc = a_f_TEH_MARBUTA;
 638             break;
 639         case a_TEH:
 640             tempc = a_f_TEH;
 641             break;
 642         case a_THEH:
 643             tempc = a_f_THEH;
 644             break;
 645         case a_JEEM:
 646             tempc = a_f_JEEM;
 647             break;
 648         case a_HAH:
 649             tempc = a_f_HAH;
 650             break;
 651         case a_KHAH:
 652             tempc = a_f_KHAH;
 653             break;
 654         case a_DAL:
 655             tempc = a_f_DAL;
 656             break;
 657         case a_THAL:
 658             tempc = a_f_THAL;
 659             break;
 660         case a_REH:
 661             tempc = a_f_REH;
 662             break;
 663         case a_ZAIN:
 664             tempc = a_f_ZAIN;
 665             break;
 666         case a_SEEN:
 667             tempc = a_f_SEEN;
 668             break;
 669         case a_SHEEN:
 670             tempc = a_f_SHEEN;
 671             break;
 672         case a_SAD:
 673             tempc = a_f_SAD;
 674             break;
 675         case a_DAD:
 676             tempc = a_f_DAD;
 677             break;
 678         case a_TAH:
 679             tempc = a_f_TAH;
 680             break;
 681         case a_ZAH:
 682             tempc = a_f_ZAH;
 683             break;
 684         case a_AIN:
 685             tempc = a_f_AIN;
 686             break;
 687         case a_GHAIN:
 688             tempc = a_f_GHAIN;
 689             break;
 690         case a_TATWEEL:                 /* exception */
 691             tempc = cur_c;
 692             break;
 693         case a_FEH:
 694             tempc = a_f_FEH;
 695             break;
 696         case a_QAF:
 697             tempc = a_f_QAF;
 698             break;
 699         case a_KAF:
 700             tempc = a_f_KAF;
 701             break;
 702         case a_LAM:
 703             tempc = a_f_LAM;
 704             break;
 705         case a_MEEM:
 706             tempc = a_f_MEEM;
 707             break;
 708         case a_NOON:
 709             tempc = a_f_NOON;
 710             break;
 711         case a_HEH:
 712             tempc = a_f_HEH;
 713             break;
 714         case a_WAW:
 715             tempc = a_f_WAW;
 716             break;
 717         case a_ALEF_MAKSURA:
 718             tempc = a_f_ALEF_MAKSURA;
 719             break;
 720         case a_YEH:
 721             tempc = a_f_YEH;
 722             break;
 723         default:
 724             tempc = 0;
 725     }
 726
 727     return tempc;
 728 }
 729
 730
 731 /*
 732  * Change shape - from Initial to Medial
 733  */
 734     static int
 735 chg_c_i2m(cur_c)
 736     int cur_c;
 737 {
 738     int tempc;
 739
 740     switch (cur_c)
 741     {
 742         case a_i_YEH_HAMZA:
 743             tempc = a_m_YEH_HAMZA;
 744             break;
 745         case a_i_BEH:
 746             tempc = a_m_BEH;
 747             break;
 748         case a_i_TEH:
 749             tempc = a_m_TEH;
 750             break;
 751         case a_i_THEH:
 752             tempc = a_m_THEH;
 753             break;
 754         case a_i_JEEM:
 755             tempc = a_m_JEEM;
 756             break;
 757         case a_i_HAH:
 758             tempc = a_m_HAH;
 759             break;
 760         case a_i_KHAH:
 761             tempc = a_m_KHAH;
 762             break;
 763         case a_i_SEEN:
 764             tempc = a_m_SEEN;
 765             break;
 766         case a_i_SHEEN:
 767             tempc = a_m_SHEEN;
 768             break;
 769         case a_i_SAD:
 770             tempc = a_m_SAD;
 771             break;
 772         case a_i_DAD:
 773             tempc = a_m_DAD;
 774             break;
 775         case a_i_TAH:
 776             tempc = a_m_TAH;
 777             break;
 778         case a_i_ZAH:
 779             tempc = a_m_ZAH;
 780             break;
 781         case a_i_AIN:
 782             tempc = a_m_AIN;
 783             break;
 784         case a_i_GHAIN:
 785             tempc = a_m_GHAIN;
 786             break;
 787         case a_i_FEH:
 788             tempc = a_m_FEH;
 789             break;
 790         case a_i_QAF:
 791             tempc = a_m_QAF;
 792             break;
 793         case a_i_KAF:
 794             tempc = a_m_KAF;
 795             break;
 796         case a_i_LAM:
 797             tempc = a_m_LAM;
 798             break;
 799         case a_i_MEEM:
 800             tempc = a_m_MEEM;
 801             break;
 802         case a_i_NOON:
 803             tempc = a_m_NOON;
 804             break;
 805         case a_i_HEH:
 806             tempc = a_m_HEH;
 807             break;
 808         case a_i_YEH:
 809             tempc = a_m_YEH;
 810             break;
 811         default:
 812             tempc = 0;
 813     }
 814
 815     return tempc;
 816 }
 817
 818
 819 /*
 820  * Change shape - from Final to Medial
 821  */
 822     static int
 823 chg_c_f2m(cur_c)
 824     int cur_c;
 825 {
 826     int tempc;
 827
 828     switch (cur_c)
 829     {
 830         /* NOTE: these encodings are multi-positional, no ?
 831            case a_f_ALEF_MADDA:
 832            case a_f_ALEF_HAMZA_ABOVE:
 833            case a_f_ALEF_HAMZA_BELOW:
 834            */
 835         case a_f_YEH_HAMZA:
 836             tempc = a_m_YEH_HAMZA;
 837             break;
 838         case a_f_WAW_HAMZA:             /* exceptions */
 839         case a_f_ALEF:
 840         case a_f_TEH_MARBUTA:
 841         case a_f_DAL:
 842         case a_f_THAL:
 843         case a_f_REH:
 844         case a_f_ZAIN:
 845         case a_f_WAW:
 846         case a_f_ALEF_MAKSURA:
 847             tempc = cur_c;
 848             break;
 849         case a_f_BEH:
 850             tempc = a_m_BEH;
 851             break;
 852         case a_f_TEH:
 853             tempc = a_m_TEH;
 854             break;
 855         case a_f_THEH:
 856             tempc = a_m_THEH;
 857             break;
 858         case a_f_JEEM:
 859             tempc = a_m_JEEM;
 860             break;
 861         case a_f_HAH:
 862             tempc = a_m_HAH;
 863             break;
 864         case a_f_KHAH:
 865             tempc = a_m_KHAH;
 866             break;
 867         case a_f_SEEN:
 868             tempc = a_m_SEEN;
 869             break;
 870         case a_f_SHEEN:
 871             tempc = a_m_SHEEN;
 872             break;
 873         case a_f_SAD:
 874             tempc = a_m_SAD;
 875             break;
 876         case a_f_DAD:
 877             tempc = a_m_DAD;
 878             break;
 879         case a_f_TAH:
 880             tempc = a_m_TAH;
 881             break;
 882         case a_f_ZAH:
 883             tempc = a_m_ZAH;
 884             break;
 885         case a_f_AIN:
 886             tempc = a_m_AIN;
 887             break;
 888         case a_f_GHAIN:
 889             tempc = a_m_GHAIN;
 890             break;
 891         case a_f_FEH:
 892             tempc = a_m_FEH;
 893             break;
 894         case a_f_QAF:
 895             tempc = a_m_QAF;
 896             break;
 897         case a_f_KAF:
 898             tempc = a_m_KAF;
 899             break;
 900         case a_f_LAM:
 901             tempc = a_m_LAM;
 902             break;
 903         case a_f_MEEM:
 904             tempc = a_m_MEEM;
 905             break;
 906         case a_f_NOON:
 907             tempc = a_m_NOON;
 908             break;
 909         case a_f_HEH:
 910             tempc = a_m_HEH;
 911             break;
 912         case a_f_YEH:
 913             tempc = a_m_YEH;
 914             break;
 915             /* NOTE: these encodings are multi-positional, no ?
 916                 case a_f_LAM_ALEF_MADDA_ABOVE:
 917                 case a_f_LAM_ALEF_HAMZA_ABOVE:
 918                 case a_f_LAM_ALEF_HAMZA_BELOW:
 919                 case a_f_LAM_ALEF:
 920                 */
 921         default:
 922             tempc = 0;
 923     }
 924
 925     return tempc;
 926 }
 927
 928
 929 /*
 930  * Change shape - from Combination (2 char) to an Isolated
 931  */
 932     static int
 933 chg_c_laa2i(hid_c)
 934     int hid_c;
 935 {
 936     int tempc;
 937
 938     switch (hid_c)
 939     {
 940         case a_ALEF_MADDA:
 941             tempc = a_s_LAM_ALEF_MADDA_ABOVE;
 942             break;
 943         case a_ALEF_HAMZA_ABOVE:
 944             tempc = a_s_LAM_ALEF_HAMZA_ABOVE;
 945             break;
 946         case a_ALEF_HAMZA_BELOW:
 947             tempc = a_s_LAM_ALEF_HAMZA_BELOW;
 948             break;
 949         case a_ALEF:
 950             tempc = a_s_LAM_ALEF;
 951             break;
 952         default:
 953             tempc = 0;
 954     }
 955
 956     return tempc;
 957 }
 958
 959
 960 /*
 961  * Change shape - from Combination-Isolated to Final
 962  */
 963     static int
 964 chg_c_laa2f(hid_c)
 965     int hid_c;
 966 {
 967     int tempc;
 968
 969     switch (hid_c)
 970     {
 971         case a_ALEF_MADDA:
 972             tempc = a_f_LAM_ALEF_MADDA_ABOVE;
 973             break;
 974         case a_ALEF_HAMZA_ABOVE:
 975             tempc = a_f_LAM_ALEF_HAMZA_ABOVE;
 976             break;
 977         case a_ALEF_HAMZA_BELOW:
 978             tempc = a_f_LAM_ALEF_HAMZA_BELOW;
 979             break;
 980         case a_ALEF:
 981             tempc = a_f_LAM_ALEF;
 982             break;
 983         default:
 984             tempc = 0;
 985     }
 986
 987     return tempc;
 988 }
 989
 990 /*
 991  * Do "half-shaping" on character "c".  Return zero if no shaping.
 992  */
 993     static int
 994 half_shape(c)
 995     int         c;
 996 {
 997     if (A_is_a(c))
 998         return chg_c_a2i(c);
 999     if (A_is_valid(c) && A_is_f(c))
1000         return chg_c_f2m(c);
1001     return 0;
1002 }
1003
1004 /*
1005  * Do Arabic shaping on character "c".  Returns the shaped character.
1006  * out:    "ccp" points to the first byte of the character to be shaped.
1007  * in/out: "c1p" points to the first composing char for "c".
1008  * in:     "prev_c"  is the previous character (not shaped)
1009  * in:     "prev_c1" is the first composing char for the previous char
1010  *                   (not shaped)
1011  * in:     "next_c"  is the next character (not shaped).
1012  */
1013     int
1014 arabic_shape(c, ccp, c1p, prev_c, prev_c1, next_c)
1015     int         c;
1016     int         *ccp;
1017     int         *c1p;
1018     int         prev_c;
1019     int         prev_c1;
1020     int         next_c;
1021 {
1022     int         curr_c;
1023     int         shape_c;
1024     int         curr_laa;
1025     int         prev_laa;
1026
1027     /* Deal only with Arabic character, pass back all others */
1028     if (!A_is_ok(c))
1029         return c;
1030
1031     /* half-shape current and previous character */
1032     shape_c = half_shape(prev_c);
1033
1034     /* Save away current character */
1035     curr_c = c;
1036
1037     curr_laa = A_firstc_laa(c, *c1p);
1038     prev_laa = A_firstc_laa(prev_c, prev_c1);
1039
1040     if (curr_laa)
1041     {
1042         if (A_is_valid(prev_c) && !A_is_f(shape_c)
1043                                          && !A_is_s(shape_c) && !prev_laa)
1044             curr_c = chg_c_laa2f(curr_laa);
1045         else
1046             curr_c = chg_c_laa2i(curr_laa);
1047
1048         /* Remove the composing character */
1049         *c1p = 0;
1050     }
1051     else if (!A_is_valid(prev_c) && A_is_valid(next_c))
1052         curr_c = chg_c_a2i(c);
1053     else if (!shape_c || A_is_f(shape_c) || A_is_s(shape_c) || prev_laa)
1054         curr_c = A_is_valid(next_c) ? chg_c_a2i(c) : chg_c_a2s(c);
1055     else if (A_is_valid(next_c))
1056         curr_c = A_is_iso(c) ? chg_c_a2m(c) : chg_c_i2m(c);
1057     else if (A_is_valid(prev_c))
1058         curr_c = chg_c_a2f(c);
1059     else
1060         curr_c = chg_c_a2s(c);
1061
1062     /* Sanity check -- curr_c should, in the future, never be 0.
1063      * We should, in the future, insert a fatal error here. */
1064     if (curr_c == NUL)
1065         curr_c = c;
1066
1067     if (curr_c != c && ccp != NULL)
1068     {
1069         char_u buf[MB_MAXBYTES];
1070
1071         /* Update the first byte of the character. */
1072         (*mb_char2bytes)(curr_c, buf);
1073         *ccp = buf[0];
1074     }
1075
1076     /* Return the shaped character */
1077     return curr_c;
1078 }
1079
1080
1081 /*
1082  * A_firstc_laa returns first character of LAA combination if it exists
1083  */
1084     static int
1085 A_firstc_laa(c, c1)
1086     int c;      /* base character */
1087     int c1;     /* first composing character */
1088 {
1089     if (c1 != NUL && c == a_LAM && !A_is_harakat(c1))
1090         return c1;
1091     return 0;
1092 }
1093
1094
1095 /*
1096  * A_is_harakat returns TRUE if 'c' is an Arabic Harakat character
1097  *              (harakat/tanween)
1098  */
1099     static int
1100 A_is_harakat(c)
1101     int c;
1102 {
1103     return (c >= a_FATHATAN && c <= a_SUKUN);
1104 }
1105
1106
1107 /*
1108  * A_is_iso returns TRUE if 'c' is an Arabic ISO-8859-6 character
1109  *              (alphabet/number/punctuation)
1110  */
1111     static int
1112 A_is_iso(c)
1113     int c;
1114 {
1115     return ((c >= a_HAMZA && c <= a_GHAIN)
1116             || (c >= a_TATWEEL && c <= a_HAMZA_BELOW)
1117             || c == a_MINI_ALEF);
1118 }
1119
1120
1121 /*
1122  * A_is_formb returns TRUE if 'c' is an Arabic 10646-1 FormB character
1123  *              (alphabet/number/punctuation)
1124  */
1125     static int
1126 A_is_formb(c)
1127     int c;
1128 {
1129     return ((c >= a_s_FATHATAN && c <= a_s_DAMMATAN)
1130             || c == a_s_KASRATAN
1131             || (c >= a_s_FATHA && c <= a_f_LAM_ALEF)
1132             || c == a_BYTE_ORDER_MARK);
1133 }
1134
1135
1136 /*
1137  * A_is_ok returns TRUE if 'c' is an Arabic 10646 (8859-6 or Form-B)
1138  */
1139     static int
1140 A_is_ok(c)
1141     int c;
1142 {
1143     return (A_is_iso(c) || A_is_formb(c));
1144 }
1145
1146
1147 /*
1148  * A_is_valid returns TRUE if 'c' is an Arabic 10646 (8859-6 or Form-B)
1149  *              with some exceptions/exclusions
1150  */
1151     static int
1152 A_is_valid(c)
1153     int c;
1154 {
1155     return (A_is_ok(c) && !A_is_special(c));
1156 }
1157
1158
1159 /*
1160  * A_is_special returns TRUE if 'c' is not a special Arabic character.
1161  *              Specials don't adhere to most of the rules.
1162  */
1163     static int
1164 A_is_special(c)
1165     int c;
1166 {
1167     return (c == a_HAMZA || c == a_s_HAMZA);
1168 }