src/editor/syntax.c

   1 /*
   2    Editor syntax highlighting.
   3
   4    Copyright (C) 1996-2016
   5    Free Software Foundation, Inc.
   6
   7    Written by:
   8    Paul Sheer, 1998
   9    Egmont Koblinger <egmont@gmail.com>, 2010
  10    Slava Zanko <slavazanko@gmail.com>, 2013
  11    Andrew Borodin <aborodin@vmail.ru>, 2013, 2014
  12
  13    This file is part of the Midnight Commander.
  14
  15    The Midnight Commander is free software: you can redistribute it
  16    and/or modify it under the terms of the GNU General Public License as
  17    published by the Free Software Foundation, either version 3 of the License,
  18    or (at your option) any later version.
  19
  20    The Midnight Commander is distributed in the hope that it will be useful,
  21    but WITHOUT ANY WARRANTY; without even the implied warranty of
  22    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  23    GNU General Public License for more details.
  24
  25    You should have received a copy of the GNU General Public License
  26    along with this program.  If not, see <http://www.gnu.org/licenses/>.
  27  */
  28
  29 /** \file
  30  *  \brief Source: editor syntax highlighting
  31  *  \author Paul Sheer
  32  *  \date 1996, 1997
  33  *  \author Mikhail Pobolovets
  34  *  \date 2010
  35  *
  36  *  Mispelled words are flushed from the syntax highlighting rules
  37  *  when they have been around longer than
  38  *  TRANSIENT_WORD_TIME_OUT seconds. At a cursor rate of 30
  39  *  chars per second and say 3 chars + a space per word, we can
  40  *  accumulate 450 words absolute max with a value of 60. This is
  41  *  below this limit of 1024 words in a context.
  42  */
  43
  44 #include <config.h>
  45
  46 #include <stdio.h>
  47 #include <stdarg.h>
  48 #include <sys/types.h>
  49 #include <unistd.h>
  50 #include <string.h>
  51 #include <ctype.h>
  52 #include <errno.h>
  53 #include <sys/stat.h>
  54 #include <stdlib.h>
  55
  56 #include "lib/global.h"
  57 #include "lib/search.h"         /* search engine */
  58 #include "lib/skin.h"
  59 #include "lib/fileloc.h"        /* EDIT_DIR, EDIT_SYNTAX_FILE */
  60 #include "lib/strutil.h"        /* utf string functions */
  61 #include "lib/util.h"
  62 #include "lib/widget.h"         /* message() */
  63
  64 #include "edit-impl.h"
  65 #include "editwidget.h"
  66
  67 /*** global variables ****************************************************************************/
  68
  69 int option_syntax_highlighting = 1;
  70 int option_auto_syntax = 1;
  71
  72 /*** file scope macro definitions ****************************************************************/
  73
  74 /* bytes */
  75 #define SYNTAX_MARKER_DENSITY 512
  76
  77 #define RULE_ON_LEFT_BORDER 1
  78 #define RULE_ON_RIGHT_BORDER 2
  79
  80 #define SYNTAX_TOKEN_STAR       '\001'
  81 #define SYNTAX_TOKEN_PLUS       '\002'
  82 #define SYNTAX_TOKEN_BRACKET    '\003'
  83 #define SYNTAX_TOKEN_BRACE      '\004'
  84
  85 #define whiteness(x) ((x) == '\t' || (x) == '\n' || (x) == ' ')
  86
  87 #define free_args(x)
  88 #define break_a {result=line;break;}
  89 #define check_a {if(!*a){result=line;break;}}
  90 #define check_not_a {if(*a){result=line;break;}}
  91
  92 #define SYNTAX_KEYWORD(x) ((syntax_keyword_t *) (x))
  93 #define CONTEXT_RULE(x) ((context_rule_t *) (x))
  94
  95 /*** file scope type declarations ****************************************************************/
  96
  97 typedef struct
  98 {
  99     char *keyword;
 100     char *whole_word_chars_left;
 101     char *whole_word_chars_right;
 102     long line_start;
 103     int color;
 104 } syntax_keyword_t;
 105
 106 typedef struct
 107 {
 108     char *left;
 109     unsigned char first_left;
 110     char *right;
 111     unsigned char first_right;
 112     char line_start_left;
 113     char line_start_right;
 114     int between_delimiters;
 115     char *whole_word_chars_left;
 116     char *whole_word_chars_right;
 117     char *keyword_first_chars;
 118     gboolean spelling;
 119     /* first word is word[1] */
 120     GPtrArray *keyword;
 121 } context_rule_t;
 122
 123 typedef struct
 124 {
 125     off_t offset;
 126     edit_syntax_rule_t rule;
 127 } syntax_marker_t;
 128
 129 /*** file scope variables ************************************************************************/
 130
 131 static char *error_file_name = NULL;
 132
 133 /* --------------------------------------------------------------------------------------------- */
 134 /*** file scope functions ************************************************************************/
 135 /* --------------------------------------------------------------------------------------------- */
 136
 137 static void
 138 syntax_keyword_free (gpointer keyword)
 139 {
 140     syntax_keyword_t *k = SYNTAX_KEYWORD (keyword);
 141
 142     g_free (k->keyword);
 143     g_free (k->whole_word_chars_left);
 144     g_free (k->whole_word_chars_right);
 145     g_free (k);
 146 }
 147
 148 /* --------------------------------------------------------------------------------------------- */
 149
 150 static void
 151 context_rule_free (gpointer rule)
 152 {
 153     context_rule_t *r = CONTEXT_RULE (rule);
 154
 155     g_free (r->left);
 156     g_free (r->right);
 157     g_free (r->whole_word_chars_left);
 158     g_free (r->whole_word_chars_right);
 159     g_free (r->keyword_first_chars);
 160
 161     if (r->keyword != NULL)
 162     {
 163         g_ptr_array_foreach (r->keyword, (GFunc) syntax_keyword_free, NULL);
 164         g_ptr_array_free (r->keyword, TRUE);
 165     }
 166
 167     g_free (r);
 168 }
 169
 170 /* --------------------------------------------------------------------------------------------- */
 171
 172 static gint
 173 mc_defines_destroy (gpointer key, gpointer value, gpointer data)
 174 {
 175     (void) data;
 176
 177     g_free (key);
 178     g_strfreev ((char **) value);
 179
 180     return FALSE;
 181 }
 182
 183 /* --------------------------------------------------------------------------------------------- */
 184 /** Completely destroys the defines tree */
 185
 186 static void
 187 destroy_defines (GTree ** defines)
 188 {
 189     g_tree_foreach (*defines, mc_defines_destroy, NULL);
 190     g_tree_destroy (*defines);
 191     *defines = NULL;
 192 }
 193
 194 /* --------------------------------------------------------------------------------------------- */
 195
 196 /** Wrapper for case insensitive mode */
 197 inline static int
 198 xx_tolower (const WEdit * edit, int c)
 199 {
 200     return edit->is_case_insensitive ? tolower (c) : c;
 201 }
 202
 203 /* --------------------------------------------------------------------------------------------- */
 204
 205 static void
 206 subst_defines (GTree * defines, char **argv, char **argv_end)
 207 {
 208     char **t, **p;
 209     int argc;
 210
 211     while (*argv != NULL && argv < argv_end)
 212     {
 213         t = g_tree_lookup (defines, *argv);
 214         if (t != NULL)
 215         {
 216             int count = 0;
 217
 218             /* Count argv array members */
 219             argc = 0;
 220             for (p = &argv[1]; *p != NULL; p++)
 221                 argc++;
 222
 223             /* Count members of definition array */
 224             for (p = t; *p != NULL; p++)
 225                 count++;
 226             p = &argv[count + argc];
 227
 228             /* Buffer overflow or infinitive loop in define */
 229             if (p >= argv_end)
 230                 break;
 231
 232             /* Move rest of argv after definition members */
 233             while (argc >= 0)
 234                 *p-- = argv[argc-- + 1];
 235
 236             /* Copy definition members to argv */
 237             for (p = argv; *t != NULL; *p++ = *t++)
 238                 ;
 239         }
 240         argv++;
 241     }
 242 }
 243
 244 /* --------------------------------------------------------------------------------------------- */
 245
 246 static off_t
 247 compare_word_to_right (const WEdit * edit, off_t i, const char *text,
 248                        const char *whole_left, const char *whole_right, long line_start)
 249 {
 250     const unsigned char *p, *q;
 251     int c, d, j;
 252
 253     if (*text == '\0')
 254         return -1;
 255
 256     c = xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i - 1));
 257     if ((line_start != 0 && c != '\n') || (whole_left != NULL && strchr (whole_left, c) != NULL))
 258         return -1;
 259
 260     for (p = (const unsigned char *) text, q = p + strlen ((const char *) p); p < q; p++, i++)
 261     {
 262         switch (*p)
 263         {
 264         case SYNTAX_TOKEN_STAR:
 265             if (++p > q)
 266                 return -1;
 267             while (TRUE)
 268             {
 269                 c = xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i));
 270                 if (*p == '\0' && whole_right != NULL && strchr (whole_right, c) == NULL)
 271                     break;
 272                 if (c == *p)
 273                     break;
 274                 if (c == '\n')
 275                     return -1;
 276                 i++;
 277             }
 278             break;
 279         case SYNTAX_TOKEN_PLUS:
 280             if (++p > q)
 281                 return -1;
 282             j = 0;
 283             while (TRUE)
 284             {
 285                 c = xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i));
 286                 if (c == *p)
 287                 {
 288                     j = i;
 289                     if (*p == *text && p[1] == '\0')    /* handle eg '+' and @+@ keywords properly */
 290                         break;
 291                 }
 292                 if (j != 0 && strchr ((const char *) p + 1, c) != NULL) /* c exists further down, so it will get matched later */
 293                     break;
 294                 if (c == '\n' || c == '\t' || c == ' ' ||
 295                     (whole_right != NULL && strchr (whole_right, c) == NULL))
 296                 {
 297                     if (*p == '\0')
 298                     {
 299                         i--;
 300                         break;
 301                     }
 302                     if (j == 0)
 303                         return -1;
 304                     i = j;
 305                     break;
 306                 }
 307                 i++;
 308             }
 309             break;
 310         case SYNTAX_TOKEN_BRACKET:
 311             if (++p > q)
 312                 return -1;
 313             c = -1;
 314             while (TRUE)
 315             {
 316                 d = c;
 317                 c = xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i));
 318                 for (j = 0; p[j] != SYNTAX_TOKEN_BRACKET && p[j]; j++)
 319                     if (c == p[j])
 320                         goto found_char2;
 321                 break;
 322               found_char2:
 323                 i++;
 324             }
 325             i--;
 326             while (*p != SYNTAX_TOKEN_BRACKET && p <= q)
 327                 p++;
 328             if (p > q)
 329                 return -1;
 330             if (p[1] == d)
 331                 i--;
 332             break;
 333         case SYNTAX_TOKEN_BRACE:
 334             if (++p > q)
 335                 return -1;
 336             c = xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i));
 337             for (; *p != SYNTAX_TOKEN_BRACE && *p; p++)
 338                 if (c == *p)
 339                     goto found_char3;
 340             return -1;
 341           found_char3:
 342             while (*p != SYNTAX_TOKEN_BRACE && p < q)
 343                 p++;
 344             break;
 345         default:
 346             if (*p != xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i)))
 347                 return -1;
 348         }
 349     }
 350     return (whole_right != NULL &&
 351             strchr (whole_right,
 352                     xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i))) != NULL) ? -1 : i;
 353 }
 354
 355 /* --------------------------------------------------------------------------------------------- */
 356
 357 static const char *
 358 xx_strchr (const WEdit * edit, const unsigned char *s, int char_byte)
 359 {
 360     while (*s >= '\005' && xx_tolower (edit, *s) != char_byte)
 361         s++;
 362
 363     return (const char *) s;
 364 }
 365
 366 /* --------------------------------------------------------------------------------------------- */
 367
 368 static void
 369 apply_rules_going_right (WEdit * edit, off_t i)
 370 {
 371     context_rule_t *r;
 372     int c;
 373     gboolean contextchanged = FALSE;
 374     gboolean found_left = FALSE, found_right = FALSE;
 375     gboolean keyword_foundleft = FALSE, keyword_foundright = FALSE;
 376     gboolean is_end;
 377     off_t end = 0;
 378     edit_syntax_rule_t _rule = edit->rule;
 379
 380     c = xx_tolower (edit, edit_buffer_get_byte (&edit->buffer, i));
 381     if (c == 0)
 382         return;
 383
 384     is_end = (edit->rule.end == i);
 385
 386     /* check to turn off a keyword */
 387     if (_rule.keyword != 0)
 388     {
 389         if (edit_buffer_get_byte (&edit->buffer, i - 1) == '\n')
 390             _rule.keyword = 0;
 391         if (is_end)
 392         {
 393             _rule.keyword = 0;
 394             keyword_foundleft = TRUE;
 395         }
 396     }
 397
 398     /* check to turn off a context */
 399     if (_rule.context != 0 && _rule.keyword == 0)
 400     {
 401         off_t e;
 402
 403         r = CONTEXT_RULE (g_ptr_array_index (edit->rules, _rule.context));
 404         if (r->first_right == c && (edit->rule.border & RULE_ON_RIGHT_BORDER) == 0
 405             && (e =
 406                 compare_word_to_right (edit, i, r->right, r->whole_word_chars_left,
 407                                        r->whole_word_chars_right, r->line_start_right)) > 0)
 408         {
 409             _rule.end = e;
 410             found_right = TRUE;
 411             _rule.border = RULE_ON_RIGHT_BORDER;
 412             if (r->between_delimiters)
 413                 _rule.context = 0;
 414         }
 415         else if (is_end && (edit->rule.border & RULE_ON_RIGHT_BORDER) != 0)
 416         {
 417             /* always turn off a context at 4 */
 418             found_left = TRUE;
 419             _rule.border = 0;
 420             if (!keyword_foundleft)
 421                 _rule.context = 0;
 422         }
 423         else if (is_end && (edit->rule.border & RULE_ON_LEFT_BORDER) != 0)
 424         {
 425             /* never turn off a context at 2 */
 426             found_left = TRUE;
 427             _rule.border = 0;
 428         }
 429     }
 430
 431     /* check to turn on a keyword */
 432     if (_rule.keyword == 0)
 433     {
 434         const char *p;
 435
 436         r = CONTEXT_RULE (g_ptr_array_index (edit->rules, _rule.context));
 437         p = r->keyword_first_chars;
 438
 439         if (p != NULL)
 440             while (*(p = xx_strchr (edit, (const unsigned char *) p + 1, c)) != '\0')
 441             {
 442                 syntax_keyword_t *k;
 443                 int count;
 444                 off_t e;
 445
 446                 count = p - r->keyword_first_chars;
 447                 k = SYNTAX_KEYWORD (g_ptr_array_index (r->keyword, count));
 448                 e = compare_word_to_right (edit, i, k->keyword, k->whole_word_chars_left,
 449                                            k->whole_word_chars_right, k->line_start);
 450                 if (e > 0)
 451                 {
 452                     end = e;
 453                     _rule.end = e;
 454                     _rule.keyword = count;
 455                     keyword_foundright = TRUE;
 456                     break;
 457                 }
 458             }
 459     }
 460
 461     /* check to turn on a context */
 462     if (_rule.context == 0)
 463     {
 464         if (!found_left && is_end)
 465         {
 466             if ((edit->rule.border & RULE_ON_RIGHT_BORDER) != 0)
 467             {
 468                 _rule.border = 0;
 469                 _rule.context = 0;
 470                 contextchanged = TRUE;
 471                 _rule.keyword = 0;
 472
 473             }
 474             else if ((edit->rule.border & RULE_ON_LEFT_BORDER) != 0)
 475             {
 476                 r = CONTEXT_RULE (g_ptr_array_index (edit->rules, _rule._context));
 477                 _rule.border = 0;
 478                 if (r->between_delimiters)
 479                 {
 480                     _rule.context = _rule._context;
 481                     contextchanged = TRUE;
 482                     _rule.keyword = 0;
 483
 484                     if (r->first_right == c)
 485                     {
 486                         off_t e;
 487
 488                         e = compare_word_to_right (edit, i, r->right, r->whole_word_chars_left,
 489                                                    r->whole_word_chars_right, r->line_start_right);
 490                         if (e >= end)
 491                         {
 492                             _rule.end = e;
 493                             found_right = TRUE;
 494                             _rule.border = RULE_ON_RIGHT_BORDER;
 495                             _rule.context = 0;
 496                         }
 497                     }
 498                 }
 499             }
 500         }
 501
 502         if (!found_right)
 503         {
 504             size_t count;
 505
 506             for (count = 1; count < edit->rules->len; count++)
 507             {
 508                 r = CONTEXT_RULE (g_ptr_array_index (edit->rules, count));
 509                 if (r->first_left == c)
 510                 {
 511                     off_t e;
 512
 513                     e = compare_word_to_right (edit, i, r->left, r->whole_word_chars_left,
 514                                                r->whole_word_chars_right, r->line_start_left);
 515                     if (e >= end && (_rule.keyword == 0 || keyword_foundright))
 516                     {
 517                         _rule.end = e;
 518                         found_right = TRUE;
 519                         _rule.border = RULE_ON_LEFT_BORDER;
 520                         _rule._context = count;
 521                         if (!r->between_delimiters && _rule.keyword == 0)
 522                         {
 523                             _rule.context = count;
 524                             contextchanged = TRUE;
 525                         }
 526                         break;
 527                     }
 528                 }
 529             }
 530         }
 531     }
 532
 533     /* check again to turn on a keyword if the context switched */
 534     if (contextchanged && _rule.keyword == 0)
 535     {
 536         const char *p;
 537
 538         r = CONTEXT_RULE (g_ptr_array_index (edit->rules, _rule.context));
 539         p = r->keyword_first_chars;
 540
 541         while (*(p = xx_strchr (edit, (const unsigned char *) p + 1, c)) != '\0')
 542         {
 543             syntax_keyword_t *k;
 544             int count;
 545             off_t e;
 546
 547             count = p - r->keyword_first_chars;
 548             k = SYNTAX_KEYWORD (g_ptr_array_index (r->keyword, count));
 549             e = compare_word_to_right (edit, i, k->keyword, k->whole_word_chars_left,
 550                                        k->whole_word_chars_right, k->line_start);
 551             if (e > 0)
 552             {
 553                 _rule.end = e;
 554                 _rule.keyword = count;
 555                 break;
 556             }
 557         }
 558     }
 559
 560     edit->rule = _rule;
 561 }
 562
 563 /* --------------------------------------------------------------------------------------------- */
 564
 565 static void
 566 edit_get_rule (WEdit * edit, off_t byte_index)
 567 {
 568     off_t i;
 569
 570     if (byte_index > edit->last_get_rule)
 571     {
 572         for (i = edit->last_get_rule + 1; i <= byte_index; i++)
 573         {
 574             off_t d = SYNTAX_MARKER_DENSITY;
 575
 576             apply_rules_going_right (edit, i);
 577
 578             if (edit->syntax_marker != NULL)
 579                 d += ((syntax_marker_t *) edit->syntax_marker->data)->offset;
 580
 581             if (i > d)
 582             {
 583                 syntax_marker_t *s;
 584
 585                 s = g_new (syntax_marker_t, 1);
 586                 s->offset = i;
 587                 s->rule = edit->rule;
 588                 edit->syntax_marker = g_slist_prepend (edit->syntax_marker, s);
 589             }
 590         }
 591     }
 592     else if (byte_index < edit->last_get_rule)
 593     {
 594         while (TRUE)
 595         {
 596             syntax_marker_t *s;
 597
 598             if (edit->syntax_marker == NULL)
 599             {
 600                 memset (&edit->rule, 0, sizeof (edit->rule));
 601                 for (i = -1; i <= byte_index; i++)
 602                     apply_rules_going_right (edit, i);
 603                 break;
 604             }
 605
 606             s = (syntax_marker_t *) edit->syntax_marker->data;
 607
 608             if (byte_index >= s->offset)
 609             {
 610                 edit->rule = s->rule;
 611                 for (i = s->offset + 1; i <= byte_index; i++)
 612                     apply_rules_going_right (edit, i);
 613                 break;
 614             }
 615
 616             g_free (s);
 617             edit->syntax_marker = g_slist_delete_link (edit->syntax_marker, edit->syntax_marker);
 618         }
 619     }
 620     edit->last_get_rule = byte_index;
 621 }
 622
 623 /* --------------------------------------------------------------------------------------------- */
 624
 625 static int
 626 translate_rule_to_color (const WEdit * edit, const edit_syntax_rule_t * rule)
 627 {
 628     syntax_keyword_t *k;
 629     context_rule_t *r;
 630
 631     r = CONTEXT_RULE (g_ptr_array_index (edit->rules, rule->context));
 632     k = SYNTAX_KEYWORD (g_ptr_array_index (r->keyword, rule->keyword));
 633
 634     return k->color;
 635 }
 636
 637 /* --------------------------------------------------------------------------------------------- */
 638 /**
 639    Returns 0 on error/eof or a count of the number of bytes read
 640    including the newline. Result must be free'd.
 641    In case of an error, *line will not be modified.
 642  */
 643
 644 static size_t
 645 read_one_line (char **line, FILE * f)
 646 {
 647     GString *p;
 648     size_t r = 0;
 649
 650     /* not reallocate string too often */
 651     p = g_string_sized_new (64);
 652
 653     while (TRUE)
 654     {
 655         int c;
 656
 657         c = fgetc (f);
 658         if (c == EOF)
 659         {
 660             if (ferror (f))
 661             {
 662                 if (errno == EINTR)
 663                     continue;
 664                 r = 0;
 665             }
 666             break;
 667         }
 668         r++;
 669
 670         /* handle all of \r\n, \r, \n correctly. */
 671         if (c == '\n')
 672             break;
 673         if (c == '\r')
 674         {
 675             c = fgetc (f);
 676             if (c == '\n')
 677                 r++;
 678             else
 679                 ungetc (c, f);
 680             break;
 681         }
 682
 683         g_string_append_c (p, c);
 684     }
 685     if (r != 0)
 686         *line = g_string_free (p, FALSE);
 687     else
 688         g_string_free (p, TRUE);
 689
 690     return r;
 691 }
 692
 693 /* --------------------------------------------------------------------------------------------- */
 694
 695 static char *
 696 convert (char *s)
 697 {
 698     char *r, *p;
 699
 700     p = r = s;
 701     while (*s)
 702     {
 703         switch (*s)
 704         {
 705         case '\\':
 706             s++;
 707             switch (*s)
 708             {
 709             case ' ':
 710                 *p = ' ';
 711                 s--;
 712                 break;
 713             case 'n':
 714                 *p = '\n';
 715                 break;
 716             case 'r':
 717                 *p = '\r';
 718                 break;
 719             case 't':
 720                 *p = '\t';
 721                 break;
 722             case 's':
 723                 *p = ' ';
 724                 break;
 725             case '*':
 726                 *p = '*';
 727                 break;
 728             case '\\':
 729                 *p = '\\';
 730                 break;
 731             case '[':
 732             case ']':
 733                 *p = SYNTAX_TOKEN_BRACKET;
 734                 break;
 735             case '{':
 736             case '}':
 737                 *p = SYNTAX_TOKEN_BRACE;
 738                 break;
 739             case 0:
 740                 *p = *s;
 741                 return r;
 742             default:
 743                 *p = *s;
 744                 break;
 745             }
 746             break;
 747         case '*':
 748             *p = SYNTAX_TOKEN_STAR;
 749             break;
 750         case '+':
 751             *p = SYNTAX_TOKEN_PLUS;
 752             break;
 753         default:
 754             *p = *s;
 755             break;
 756         }
 757         s++;
 758         p++;
 759     }
 760     *p = '\0';
 761     return r;
 762 }
 763
 764 /* --------------------------------------------------------------------------------------------- */
 765
 766 static int
 767 get_args (char *l, char **args, int args_size)
 768 {
 769     int argc = 0;
 770
 771     while (argc < args_size)
 772     {
 773         char *p = l;
 774         while (*p != '\0' && whiteness (*p))
 775             p++;
 776         if (*p == '\0')
 777             break;
 778         for (l = p + 1; *l != '\0' && !whiteness (*l); l++)
 779             ;
 780         if (*l != '\0')
 781             *l++ = '\0';
 782         args[argc++] = convert (p);
 783     }
 784     args[argc] = (char *) NULL;
 785     return argc;
 786 }
 787
 788 /* --------------------------------------------------------------------------------------------- */
 789
 790 static int
 791 this_try_alloc_color_pair (const char *fg, const char *bg, const char *attrs)
 792 {
 793     char f[80], b[80], a[80], *p;
 794
 795     if (bg != NULL && *bg == '\0')
 796         bg = NULL;
 797     if (fg != NULL && *fg == '\0')
 798         fg = NULL;
 799     if (attrs != NULL && *attrs == '\0')
 800         attrs = NULL;
 801
 802     if ((fg == NULL) && (bg == NULL))
 803         return EDITOR_NORMAL_COLOR;
 804
 805     if (fg != NULL)
 806     {
 807         g_strlcpy (f, fg, sizeof (f));
 808         p = strchr (f, '/');
 809         if (p != NULL)
 810             *p = '\0';
 811         fg = f;
 812     }
 813     if (bg != NULL)
 814     {
 815         g_strlcpy (b, bg, sizeof (b));
 816         p = strchr (b, '/');
 817         if (p != NULL)
 818             *p = '\0';
 819         bg = b;
 820     }
 821     if ((fg == NULL) || (bg == NULL))
 822     {
 823         /* get colors from skin */
 824         char *editnormal;
 825
 826         editnormal = mc_skin_get ("editor", "_default_", "default;default");
 827
 828         if (fg == NULL)
 829         {
 830             g_strlcpy (f, editnormal, sizeof (f));
 831             p = strchr (f, ';');
 832             if (p != NULL)
 833                 *p = '\0';
 834             if (f[0] == '\0')
 835                 g_strlcpy (f, "default", sizeof (f));
 836             fg = f;
 837         }
 838         if (bg == NULL)
 839         {
 840             p = strchr (editnormal, ';');
 841             if ((p != NULL) && (*(++p) != '\0'))
 842                 g_strlcpy (b, p, sizeof (b));
 843             else
 844                 g_strlcpy (b, "default", sizeof (b));
 845             bg = b;
 846         }
 847
 848         g_free (editnormal);
 849     }
 850
 851     if (attrs != NULL)
 852     {
 853         g_strlcpy (a, attrs, sizeof (a));
 854         p = strchr (a, '/');
 855         if (p != NULL)
 856             *p = '\0';
 857         /* get_args() mangles the + signs, unmangle 'em */
 858         p = a;
 859         while ((p = strchr (p, SYNTAX_TOKEN_PLUS)) != NULL)
 860             *p++ = '+';
 861         attrs = a;
 862     }
 863     return tty_try_alloc_color_pair (fg, bg, attrs);
 864 }
 865
 866 /* --------------------------------------------------------------------------------------------- */
 867
 868 static FILE *
 869 open_include_file (const char *filename)
 870 {
 871     FILE *f;
 872
 873     MC_PTR_FREE (error_file_name);
 874     error_file_name = g_strdup (filename);
 875     if (g_path_is_absolute (filename))
 876         return fopen (filename, "r");
 877
 878     g_free (error_file_name);
 879     error_file_name =
 880         g_build_filename (mc_config_get_data_path (), EDIT_DIR, filename, (char *) NULL);
 881     f = fopen (error_file_name, "r");
 882     if (f != NULL)
 883         return f;
 884
 885     g_free (error_file_name);
 886     error_file_name = g_build_filename (mc_global.sysconfig_dir, "syntax", filename, (char *) NULL);
 887     f = fopen (error_file_name, "r");
 888     if (f != NULL)
 889         return f;
 890
 891     g_free (error_file_name);
 892     error_file_name =
 893         g_build_filename (mc_global.share_data_dir, "syntax", filename, (char *) NULL);
 894
 895     return fopen (error_file_name, "r");
 896 }
 897
 898 /* --------------------------------------------------------------------------------------------- */
 899
 900 inline static void
 901 xx_lowerize_line (WEdit * edit, char *line, size_t len)
 902 {
 903     if (edit->is_case_insensitive)
 904     {
 905         size_t i;
 906         for (i = 0; i < len; ++i)
 907             line[i] = tolower (line[i]);
 908     }
 909 }
 910
 911 /* --------------------------------------------------------------------------------------------- */
 912 /** returns line number on error */
 913
 914 static int
 915 edit_read_syntax_rules (WEdit * edit, FILE * f, char **args, int args_size)
 916 {
 917     FILE *g = NULL;
 918     char *fg, *bg, *attrs;
 919     char last_fg[32] = "", last_bg[32] = "", last_attrs[64] = "";
 920     char whole_right[512];
 921     char whole_left[512];
 922     char *l = 0;
 923     int save_line = 0, line = 0;
 924     context_rule_t *c = NULL;
 925     gboolean no_words = TRUE;
 926     int result = 0;
 927
 928     args[0] = NULL;
 929     edit->is_case_insensitive = FALSE;
 930
 931     strcpy (whole_left, "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_01234567890");
 932     strcpy (whole_right, "abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ_01234567890");
 933
 934     edit->rules = g_ptr_array_new ();
 935
 936     if (!edit->defines)
 937         edit->defines = g_tree_new ((GCompareFunc) strcmp);
 938
 939     while (TRUE)
 940     {
 941         char **a;
 942         size_t len;
 943         int argc;
 944
 945         line++;
 946         l = 0;
 947
 948         len = read_one_line (&l, f);
 949         if (len != 0)
 950             xx_lowerize_line (edit, l, len);
 951         else
 952         {
 953             if (g == NULL)
 954                 break;
 955
 956             fclose (f);
 957             f = g;
 958             g = NULL;
 959             line = save_line + 1;
 960             MC_PTR_FREE (error_file_name);
 961             MC_PTR_FREE (l);
 962             len = read_one_line (&l, f);
 963             if (len == 0)
 964                 break;
 965             xx_lowerize_line (edit, l, len);
 966         }
 967
 968         argc = get_args (l, args, args_size);
 969         a = args + 1;
 970         if (args[0] == NULL)
 971         {
 972             /* do nothing */
 973         }
 974         else if (strcmp (args[0], "include") == 0)
 975         {
 976             if (g != NULL || argc != 2)
 977             {
 978                 result = line;
 979                 break;
 980             }
 981             g = f;
 982             f = open_include_file (args[1]);
 983             if (f == NULL)
 984             {
 985                 MC_PTR_FREE (error_file_name);
 986                 result = line;
 987                 break;
 988             }
 989             save_line = line;
 990             line = 0;
 991         }
 992         else if (strcmp (args[0], "caseinsensitive") == 0)
 993         {
 994             edit->is_case_insensitive = TRUE;
 995         }
 996         else if (strcmp (args[0], "wholechars") == 0)
 997         {
 998             check_a;
 999             if (strcmp (*a, "left") == 0)
1000             {
1001                 a++;
1002                 g_strlcpy (whole_left, *a, sizeof (whole_left));
1003             }
1004             else if (strcmp (*a, "right") == 0)
1005             {
1006                 a++;
1007                 g_strlcpy (whole_right, *a, sizeof (whole_right));
1008             }
1009             else
1010             {
1011                 g_strlcpy (whole_left, *a, sizeof (whole_left));
1012                 g_strlcpy (whole_right, *a, sizeof (whole_right));
1013             }
1014             a++;
1015             check_not_a;
1016         }
1017         else if (strcmp (args[0], "context") == 0)
1018         {
1019             syntax_keyword_t *k;
1020
1021             check_a;
1022             if (edit->rules->len == 0)
1023             {
1024                 /* first context is the default */
1025                 if (strcmp (*a, "default") != 0)
1026                     break_a;
1027
1028                 a++;
1029                 c = g_new0 (context_rule_t, 1);
1030                 g_ptr_array_add (edit->rules, c);
1031                 c->left = g_strdup (" ");
1032                 c->right = g_strdup (" ");
1033             }
1034             else
1035             {
1036                 /* Start new context.  */
1037                 c = g_new0 (context_rule_t, 1);
1038                 g_ptr_array_add (edit->rules, c);
1039                 if (strcmp (*a, "exclusive") == 0)
1040                 {
1041                     a++;
1042                     c->between_delimiters = 1;
1043                 }
1044                 check_a;
1045                 if (strcmp (*a, "whole") == 0)
1046                 {
1047                     a++;
1048                     c->whole_word_chars_left = g_strdup (whole_left);
1049                     c->whole_word_chars_right = g_strdup (whole_right);
1050                 }
1051                 else if (strcmp (*a, "wholeleft") == 0)
1052                 {
1053                     a++;
1054                     c->whole_word_chars_left = g_strdup (whole_left);
1055                 }
1056                 else if (strcmp (*a, "wholeright") == 0)
1057                 {
1058                     a++;
1059                     c->whole_word_chars_right = g_strdup (whole_right);
1060                 }
1061                 check_a;
1062                 if (strcmp (*a, "linestart") == 0)
1063                 {
1064                     a++;
1065                     c->line_start_left = 1;
1066                 }
1067                 check_a;
1068                 c->left = g_strdup (*a++);
1069                 check_a;
1070                 if (strcmp (*a, "linestart") == 0)
1071                 {
1072                     a++;
1073                     c->line_start_right = 1;
1074                 }
1075                 check_a;
1076                 c->right = g_strdup (*a++);
1077                 c->first_left = *c->left;
1078                 c->first_right = *c->right;
1079             }
1080             c->keyword = g_ptr_array_new ();
1081             k = g_new0 (syntax_keyword_t, 1);
1082             g_ptr_array_add (c->keyword, k);
1083             no_words = FALSE;
1084             subst_defines (edit->defines, a, &args[1024]);
1085             fg = *a;
1086             if (*a != NULL)
1087                 a++;
1088             bg = *a;
1089             if (*a != NULL)
1090                 a++;
1091             attrs = *a;
1092             if (*a != NULL)
1093                 a++;
1094             g_strlcpy (last_fg, fg != NULL ? fg : "", sizeof (last_fg));
1095             g_strlcpy (last_bg, bg != NULL ? bg : "", sizeof (last_bg));
1096             g_strlcpy (last_attrs, attrs != NULL ? attrs : "", sizeof (last_attrs));
1097             k->color = this_try_alloc_color_pair (fg, bg, attrs);
1098             k->keyword = g_strdup (" ");
1099             check_not_a;
1100         }
1101         else if (strcmp (args[0], "spellcheck") == 0)
1102         {
1103             if (c == NULL)
1104             {
1105                 result = line;
1106                 break;
1107             }
1108             c->spelling = TRUE;
1109         }
1110         else if (strcmp (args[0], "keyword") == 0)
1111         {
1112             context_rule_t *last_rule;
1113             syntax_keyword_t *k;
1114
1115             if (no_words)
1116                 break_a;
1117             check_a;
1118             last_rule = CONTEXT_RULE (g_ptr_array_index (edit->rules, edit->rules->len - 1));
1119             k = g_new0 (syntax_keyword_t, 1);
1120             g_ptr_array_add (last_rule->keyword, k);
1121             if (strcmp (*a, "whole") == 0)
1122             {
1123                 a++;
1124                 k->whole_word_chars_left = g_strdup (whole_left);
1125                 k->whole_word_chars_right = g_strdup (whole_right);
1126             }
1127             else if (strcmp (*a, "wholeleft") == 0)
1128             {
1129                 a++;
1130                 k->whole_word_chars_left = g_strdup (whole_left);
1131             }
1132             else if (strcmp (*a, "wholeright") == 0)
1133             {
1134                 a++;
1135                 k->whole_word_chars_right = g_strdup (whole_right);
1136             }
1137             check_a;
1138             if (strcmp (*a, "linestart") == 0)
1139             {
1140                 a++;
1141                 k->line_start = 1;
1142             }
1143             check_a;
1144             if (strcmp (*a, "whole") == 0)
1145             {
1146                 break_a;
1147             }
1148             k->keyword = g_strdup (*a++);
1149             subst_defines (edit->defines, a, &args[1024]);
1150             fg = *a;
1151             if (*a != NULL)
1152                 a++;
1153             bg = *a;
1154             if (*a != NULL)
1155                 a++;
1156             attrs = *a;
1157             if (*a != NULL)
1158                 a++;
1159             if (fg == NULL)
1160                 fg = last_fg;
1161             if (bg == NULL)
1162                 bg = last_bg;
1163             if (attrs == NULL)
1164                 attrs = last_attrs;
1165             k->color = this_try_alloc_color_pair (fg, bg, attrs);
1166             check_not_a;
1167         }
1168         else if (*(args[0]) == '#')
1169         {
1170             /* do nothing for comment */
1171         }
1172         else if (strcmp (args[0], "file") == 0)
1173         {
1174             break;
1175         }
1176         else if (strcmp (args[0], "define") == 0)
1177         {
1178             char *key = *a++;
1179             char **argv;
1180
1181             if (argc < 3)
1182                 break_a;
1183             argv = g_tree_lookup (edit->defines, key);
1184             if (argv != NULL)
1185                 mc_defines_destroy (NULL, argv, NULL);
1186             else
1187                 key = g_strdup (key);
1188
1189             argv = g_new (char *, argc - 1);
1190             g_tree_insert (edit->defines, key, argv);
1191             while (*a != NULL)
1192                 *argv++ = g_strdup (*a++);
1193             *argv = NULL;
1194         }
1195         else
1196         {                       /* anything else is an error */
1197             break_a;
1198         }
1199         free_args (args);
1200         MC_PTR_FREE (l);
1201     }
1202     free_args (args);
1203     MC_PTR_FREE (l);
1204
1205     if (edit->rules->len == 0)
1206     {
1207         g_ptr_array_free (edit->rules, TRUE);
1208         edit->rules = NULL;
1209     }
1210
1211     if (result == 0)
1212     {
1213         size_t i;
1214         GString *first_chars;
1215
1216         if (edit->rules == NULL)
1217             return line;
1218
1219         first_chars = g_string_sized_new (32);
1220
1221         /* collect first character of keywords */
1222         for (i = 0; i < edit->rules->len; i++)
1223         {
1224             size_t j;
1225
1226             g_string_set_size (first_chars, 0);
1227             c = CONTEXT_RULE (g_ptr_array_index (edit->rules, i));
1228
1229             g_string_append_c (first_chars, (char) 1);
1230             for (j = 1; j < c->keyword->len; j++)
1231             {
1232                 syntax_keyword_t *k;
1233
1234                 k = SYNTAX_KEYWORD (g_ptr_array_index (c->keyword, j));
1235                 g_string_append_c (first_chars, k->keyword[0]);
1236             }
1237
1238             c->keyword_first_chars = g_strndup (first_chars->str, first_chars->len);
1239         }
1240
1241         g_string_free (first_chars, TRUE);
1242     }
1243
1244     return result;
1245 }
1246
1247 /* --------------------------------------------------------------------------------------------- */
1248
1249 /* returns -1 on file error, line number on error in file syntax */
1250 static int
1251 edit_read_syntax_file (WEdit * edit, GPtrArray * pnames, const char *syntax_file,
1252                        const char *editor_file, const char *first_line, const char *type)
1253 {
1254     FILE *f, *g = NULL;
1255     char *args[1024], *l = NULL;
1256     long line = 0;
1257     int result = 0;
1258     char *lib_file;
1259     gboolean found = FALSE;
1260
1261     f = fopen (syntax_file, "r");
1262     if (f == NULL)
1263     {
1264         lib_file = g_build_filename (mc_global.share_data_dir, "syntax", "Syntax", (char *) NULL);
1265         f = fopen (lib_file, "r");
1266         g_free (lib_file);
1267         if (f == NULL)
1268             return -1;
1269     }
1270
1271     args[0] = NULL;
1272     while (TRUE)
1273     {
1274         line++;
1275         MC_PTR_FREE (l);
1276         if (read_one_line (&l, f) == 0)
1277             break;
1278         (void) get_args (l, args, 1023);        /* Final NULL */
1279         if (args[0] == NULL)
1280             continue;
1281
1282         /* Looking for 'include ...' lines before first 'file ...' ones */
1283         if (!found && strcmp (args[0], "include") == 0)
1284         {
1285             if (g != NULL)
1286                 continue;
1287
1288             if (!args[1] || !(g = open_include_file (args[1])))
1289             {
1290                 result = line;
1291                 break;
1292             }
1293             goto found_type;
1294         }
1295
1296         /* looking for 'file ...' lines only */
1297         if (strcmp (args[0], "file") != 0)
1298             continue;
1299
1300         found = TRUE;
1301
1302         /* must have two args or report error */
1303         if (!args[1] || !args[2])
1304         {
1305             result = line;
1306             break;
1307         }
1308
1309         if (pnames != NULL)
1310         {
1311             /* 1: just collecting a list of names of rule sets */
1312             g_ptr_array_add (pnames, g_strdup (args[2]));
1313         }
1314         else if (type)
1315         {
1316             /* 2: rule set was explicitly specified by the caller */
1317             if (strcmp (type, args[2]) == 0)
1318                 goto found_type;
1319         }
1320         else if (editor_file && edit)
1321         {
1322             /* 3: auto-detect rule set from regular expressions */
1323             int q;
1324
1325             q = mc_search (args[1], DEFAULT_CHARSET, editor_file, MC_SEARCH_T_REGEX);
1326             /* does filename match arg 1 ? */
1327             if (!q && args[3])
1328             {
1329                 /* does first line match arg 3 ? */
1330                 q = mc_search (args[3], DEFAULT_CHARSET, first_line, MC_SEARCH_T_REGEX);
1331             }
1332             if (q)
1333             {
1334                 int line_error;
1335                 char *syntax_type;
1336               found_type:
1337                 syntax_type = args[2];
1338                 line_error = edit_read_syntax_rules (edit, g ? g : f, args, 1023);
1339                 if (line_error)
1340                 {
1341                     if (!error_file_name)       /* an included file */
1342                         result = line + line_error;
1343                     else
1344                         result = line_error;
1345                 }
1346                 else
1347                 {
1348                     g_free (edit->syntax_type);
1349                     edit->syntax_type = g_strdup (syntax_type);
1350                     /* if there are no rules then turn off syntax highlighting for speed */
1351                     if (g == NULL && edit->rules->len == 1)
1352                     {
1353                         context_rule_t *r0;
1354
1355                         r0 = CONTEXT_RULE (g_ptr_array_index (edit->rules, 0));
1356                         if (r0->keyword->len == 1 && !r0->spelling)
1357                         {
1358                             edit_free_syntax_rules (edit);
1359                             break;
1360                         }
1361                     }
1362                 }
1363
1364                 if (g == NULL)
1365                     break;
1366
1367                 fclose (g);
1368                 g = NULL;
1369             }
1370         }
1371     }
1372     g_free (l);
1373     fclose (f);
1374     return result;
1375 }
1376
1377 /* --------------------------------------------------------------------------------------------- */
1378
1379 static const char *
1380 get_first_editor_line (WEdit * edit)
1381 {
1382     static char s[256];
1383
1384     s[0] = '\0';
1385
1386     if (edit != NULL)
1387     {
1388         size_t i;
1389
1390         for (i = 0; i < sizeof (s) - 1; i++)
1391         {
1392             s[i] = edit_buffer_get_byte (&edit->buffer, i);
1393             if (s[i] == '\n')
1394             {
1395                 s[i] = '\0';
1396                 break;
1397             }
1398         }
1399
1400         s[sizeof (s) - 1] = '\0';
1401     }
1402
1403     return s;
1404 }
1405
1406 /* --------------------------------------------------------------------------------------------- */
1407 /*** public functions ****************************************************************************/
1408 /* --------------------------------------------------------------------------------------------- */
1409
1410 int
1411 edit_get_syntax_color (WEdit * edit, off_t byte_index)
1412 {
1413     if (!tty_use_colors ())
1414         return 0;
1415
1416     if (edit->rules != NULL && byte_index < edit->buffer.size && option_syntax_highlighting)
1417     {
1418         edit_get_rule (edit, byte_index);
1419         return translate_rule_to_color (edit, &edit->rule);
1420     }
1421
1422     return EDITOR_NORMAL_COLOR;
1423 }
1424
1425 /* --------------------------------------------------------------------------------------------- */
1426
1427 void
1428 edit_free_syntax_rules (WEdit * edit)
1429 {
1430     if (edit == NULL)
1431         return;
1432
1433     if (edit->defines != NULL)
1434         destroy_defines (&edit->defines);
1435
1436     if (edit->rules == NULL)
1437         return;
1438
1439     edit_get_rule (edit, -1);
1440     MC_PTR_FREE (edit->syntax_type);
1441
1442     g_ptr_array_foreach (edit->rules, (GFunc) context_rule_free, NULL);
1443     g_ptr_array_free (edit->rules, TRUE);
1444     edit->rules = NULL;
1445     g_slist_free_full (edit->syntax_marker, g_free);
1446     edit->syntax_marker = NULL;
1447     tty_color_free_all_tmp ();
1448 }
1449
1450 /* --------------------------------------------------------------------------------------------- */
1451 /**
1452  * Load rules into edit struct.  Either edit or *pnames must be NULL.  If
1453  * edit is NULL, a list of types will be stored into names.  If type is
1454  * NULL, then the type will be selected according to the filename.
1455  * type must be edit->syntax_type or NULL
1456  */
1457 void
1458 edit_load_syntax (WEdit * edit, GPtrArray * pnames, const char *type)
1459 {
1460     int r;
1461     char *f = NULL;
1462
1463     if (option_auto_syntax)
1464         type = NULL;
1465
1466     if (edit != NULL)
1467     {
1468         char *saved_type;
1469
1470         saved_type = g_strdup (type);   /* save edit->syntax_type */
1471         edit_free_syntax_rules (edit);
1472         edit->syntax_type = saved_type; /* restore edit->syntax_type */
1473     }
1474
1475     if (!tty_use_colors ())
1476         return;
1477
1478     if (!option_syntax_highlighting && (pnames == NULL || pnames->len == 0))
1479         return;
1480
1481     if (edit != NULL && edit->filename_vpath == NULL)
1482         return;
1483
1484     f = mc_config_get_full_path (EDIT_SYNTAX_FILE);
1485     if (edit != NULL)
1486         r = edit_read_syntax_file (edit, pnames, f, vfs_path_as_str (edit->filename_vpath),
1487                                    get_first_editor_line (edit),
1488                                    option_auto_syntax ? NULL : edit->syntax_type);
1489     else
1490         r = edit_read_syntax_file (NULL, pnames, f, NULL, "", NULL);
1491     if (r == -1)
1492     {
1493         edit_free_syntax_rules (edit);
1494         message (D_ERROR, _("Load syntax file"),
1495                  _("Cannot open file %s\n%s"), f, unix_error_string (errno));
1496     }
1497     else if (r != 0)
1498     {
1499         edit_free_syntax_rules (edit);
1500         message (D_ERROR, _("Load syntax file"),
1501                  _("Error in file %s on line %d"), error_file_name ? error_file_name : f, r);
1502         MC_PTR_FREE (error_file_name);
1503     }
1504
1505     g_free (f);
1506 }
1507
1508 /* --------------------------------------------------------------------------------------------- */
1509
1510 const char *
1511 edit_get_syntax_type (const WEdit * edit)
1512 {
1513     return edit->syntax_type;
1514 }
1515
1516 /* --------------------------------------------------------------------------------------------- */