bidi.c 68.3 KB
Newer Older
1
/* Low-level bidirectional buffer/string-scanning functions for GNU Emacs.
2
   Copyright (C) 2000-2001, 2004-2005, 2009-2011
3
   Free Software Foundation, Inc.
Eli Zaretskii's avatar
Eli Zaretskii committed
4 5 6

This file is part of GNU Emacs.

Eli Zaretskii's avatar
Eli Zaretskii committed
7
GNU Emacs is free software: you can redistribute it and/or modify
Eli Zaretskii's avatar
Eli Zaretskii committed
8
it under the terms of the GNU General Public License as published by
Eli Zaretskii's avatar
Eli Zaretskii committed
9 10
the Free Software Foundation, either version 3 of the License, or
(at your option) any later version.
Eli Zaretskii's avatar
Eli Zaretskii committed
11 12 13 14 15 16 17

GNU Emacs is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
GNU General Public License for more details.

You should have received a copy of the GNU General Public License
Eli Zaretskii's avatar
Eli Zaretskii committed
18
along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.  */
Eli Zaretskii's avatar
Eli Zaretskii committed
19

20 21 22
/* Written by Eli Zaretskii <eliz@gnu.org>.

   A sequential implementation of the Unicode Bidirectional algorithm,
23
   (UBA) as per UAX#9, a part of the Unicode Standard.
Eli Zaretskii's avatar
Eli Zaretskii committed
24 25

   Unlike the reference and most other implementations, this one is
26 27
   designed to be called once for every character in the buffer or
   string.
Eli Zaretskii's avatar
Eli Zaretskii committed
28

29
   The main entry point is bidi_move_to_visually_next.  Each time it
Eli Zaretskii's avatar
Eli Zaretskii committed
30 31 32
   is called, it finds the next character in the visual order, and
   returns its information in a special structure.  The caller is then
   expected to process this character for display or any other
33 34 35
   purposes, and call bidi_move_to_visually_next for the next
   character.  See the comments in bidi_move_to_visually_next for more
   details about its algorithm that finds the next visual-order
Eli Zaretskii's avatar
Eli Zaretskii committed
36 37
   character by resolving their levels on the fly.

38 39 40 41 42
   The two other entry points are bidi_paragraph_init and
   bidi_mirror_char.  The first determines the base direction of a
   paragraph, while the second returns the mirrored version of its
   argument character.

43 44 45 46
   If you want to understand the code, you will have to read it
   together with the relevant portions of UAX#9.  The comments include
   references to UAX#9 rules, for that very reason.

Eli Zaretskii's avatar
Eli Zaretskii committed
47 48 49 50 51 52 53
   A note about references to UAX#9 rules: if the reference says
   something like "X9/Retaining", it means that you need to refer to
   rule X9 and to its modifications decribed in the "Implementation
   Notes" section of UAX#9, under "Retaining Format Codes".  */

#include <config.h>
#include <stdio.h>
54 55
#include <setjmp.h>

Eli Zaretskii's avatar
Eli Zaretskii committed
56 57 58 59 60 61 62
#include "lisp.h"
#include "buffer.h"
#include "character.h"
#include "dispextern.h"

static int bidi_initialized = 0;

63
static Lisp_Object bidi_type_table, bidi_mirror_table;
Eli Zaretskii's avatar
Eli Zaretskii committed
64 65 66 67 68 69 70 71 72 73 74 75 76

#define LRM_CHAR   0x200E
#define RLM_CHAR   0x200F
#define BIDI_EOB   -1

/* Data type for describing the bidirectional character categories.  */
typedef enum {
  UNKNOWN_BC,
  NEUTRAL,
  WEAK,
  STRONG
} bidi_category_t;

77
extern int bidi_ignore_explicit_marks_for_paragraph_level EXTERNALLY_VISIBLE;
Eli Zaretskii's avatar
Eli Zaretskii committed
78 79
int bidi_ignore_explicit_marks_for_paragraph_level = 1;

Eli Zaretskii's avatar
Eli Zaretskii committed
80
static Lisp_Object paragraph_start_re, paragraph_separate_re;
81
static Lisp_Object Qparagraph_start, Qparagraph_separate;
Eli Zaretskii's avatar
Eli Zaretskii committed
82

83 84 85 86

/***********************************************************************
			Utilities
 ***********************************************************************/
Eli Zaretskii's avatar
Eli Zaretskii committed
87

88 89
/* Return the bidi type of a character CH, subject to the current
   directional OVERRIDE.  */
90
static INLINE bidi_type_t
91
bidi_get_type (int ch, bidi_dir_t override)
Eli Zaretskii's avatar
Eli Zaretskii committed
92
{
93 94
  bidi_type_t default_type;

95 96
  if (ch == BIDI_EOB)
    return NEUTRAL_B;
97 98
  if (ch < 0 || ch > MAX_CHAR)
    abort ();
99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130

  default_type = (bidi_type_t) XINT (CHAR_TABLE_REF (bidi_type_table, ch));

  if (override == NEUTRAL_DIR)
    return default_type;

  switch (default_type)
    {
      /* Although UAX#9 does not tell, it doesn't make sense to
	 override NEUTRAL_B and LRM/RLM characters.  */
      case NEUTRAL_B:
      case LRE:
      case LRO:
      case RLE:
      case RLO:
      case PDF:
	return default_type;
      default:
	switch (ch)
	  {
	    case LRM_CHAR:
	    case RLM_CHAR:
	      return default_type;
	    default:
	      if (override == L2R) /* X6 */
		return STRONG_L;
	      else if (override == R2L)
		return STRONG_R;
	      else
		abort ();	/* can't happen: handled above */
	  }
    }
Eli Zaretskii's avatar
Eli Zaretskii committed
131 132
}

133
static void
134 135 136 137 138 139
bidi_check_type (bidi_type_t type)
{
  if (type < UNKNOWN_BT || type > NEUTRAL_ON)
    abort ();
}

Eli Zaretskii's avatar
Eli Zaretskii committed
140
/* Given a bidi TYPE of a character, return its category.  */
141
static INLINE bidi_category_t
Eli Zaretskii's avatar
Eli Zaretskii committed
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
bidi_get_category (bidi_type_t type)
{
  switch (type)
    {
      case UNKNOWN_BT:
	return UNKNOWN_BC;
      case STRONG_L:
      case STRONG_R:
      case STRONG_AL:
      case LRE:
      case LRO:
      case RLE:
      case RLO:
	return STRONG;
      case PDF:		/* ??? really?? */
      case WEAK_EN:
      case WEAK_ES:
      case WEAK_ET:
      case WEAK_AN:
      case WEAK_CS:
      case WEAK_NSM:
      case WEAK_BN:
	return WEAK;
      case NEUTRAL_B:
      case NEUTRAL_S:
      case NEUTRAL_WS:
      case NEUTRAL_ON:
	return NEUTRAL;
      default:
	abort ();
    }
}

175 176 177 178
/* Return the mirrored character of C, if it has one.  If C has no
   mirrored counterpart, return C.
   Note: The conditions in UAX#9 clause L4 regarding the surrounding
   context must be tested by the caller.  */
Eli Zaretskii's avatar
Eli Zaretskii committed
179 180 181
int
bidi_mirror_char (int c)
{
182 183 184 185 186 187
  Lisp_Object val;

  if (c == BIDI_EOB)
    return c;
  if (c < 0 || c > MAX_CHAR)
    abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
188

189 190
  val = CHAR_TABLE_REF (bidi_mirror_table, c);
  if (INTEGERP (val))
Eli Zaretskii's avatar
Eli Zaretskii committed
191
    {
192
      int v = XINT (val);
Eli Zaretskii's avatar
Eli Zaretskii committed
193

194 195 196 197
      if (v < 0 || v > MAX_CHAR)
	abort ();

      return v;
Eli Zaretskii's avatar
Eli Zaretskii committed
198
    }
199

Eli Zaretskii's avatar
Eli Zaretskii committed
200 201 202
  return c;
}

203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274
/* Determine the start-of-run (sor) directional type given the two
   embedding levels on either side of the run boundary.  Also, update
   the saved info about previously seen characters, since that info is
   generally valid for a single level run.  */
static INLINE void
bidi_set_sor_type (struct bidi_it *bidi_it, int level_before, int level_after)
{
  int higher_level = level_before > level_after ? level_before : level_after;

  /* The prev_was_pdf gork is required for when we have several PDFs
     in a row.  In that case, we want to compute the sor type for the
     next level run only once: when we see the first PDF.  That's
     because the sor type depends only on the higher of the two levels
     that we find on the two sides of the level boundary (see UAX#9,
     clause X10), and so we don't need to know the final embedding
     level to which we descend after processing all the PDFs.  */
  if (!bidi_it->prev_was_pdf || level_before < level_after)
    /* FIXME: should the default sor direction be user selectable?  */
    bidi_it->sor = (higher_level & 1) != 0 ? R2L : L2R;
  if (level_before > level_after)
    bidi_it->prev_was_pdf = 1;

  bidi_it->prev.type = UNKNOWN_BT;
  bidi_it->last_strong.type = bidi_it->last_strong.type_after_w1 =
    bidi_it->last_strong.orig_type = UNKNOWN_BT;
  bidi_it->prev_for_neutral.type = bidi_it->sor == R2L ? STRONG_R : STRONG_L;
  bidi_it->prev_for_neutral.charpos = bidi_it->charpos;
  bidi_it->prev_for_neutral.bytepos = bidi_it->bytepos;
  bidi_it->next_for_neutral.type = bidi_it->next_for_neutral.type_after_w1 =
    bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
  bidi_it->ignore_bn_limit = -1; /* meaning it's unknown */
}

/* Push the current embedding level and override status; reset the
   current level to LEVEL and the current override status to OVERRIDE.  */
static INLINE void
bidi_push_embedding_level (struct bidi_it *bidi_it,
			   int level, bidi_dir_t override)
{
  bidi_it->stack_idx++;
  if (bidi_it->stack_idx >= BIDI_MAXLEVEL)
    abort ();
  bidi_it->level_stack[bidi_it->stack_idx].level = level;
  bidi_it->level_stack[bidi_it->stack_idx].override = override;
}

/* Pop the embedding level and directional override status from the
   stack, and return the new level.  */
static INLINE int
bidi_pop_embedding_level (struct bidi_it *bidi_it)
{
  /* UAX#9 says to ignore invalid PDFs.  */
  if (bidi_it->stack_idx > 0)
    bidi_it->stack_idx--;
  return bidi_it->level_stack[bidi_it->stack_idx].level;
}

/* Record in SAVED_INFO the information about the current character.  */
static INLINE void
bidi_remember_char (struct bidi_saved_info *saved_info,
		    struct bidi_it *bidi_it)
{
  saved_info->charpos = bidi_it->charpos;
  saved_info->bytepos = bidi_it->bytepos;
  saved_info->type = bidi_it->type;
  bidi_check_type (bidi_it->type);
  saved_info->type_after_w1 = bidi_it->type_after_w1;
  bidi_check_type (bidi_it->type_after_w1);
  saved_info->orig_type = bidi_it->orig_type;
  bidi_check_type (bidi_it->orig_type);
}

Eli Zaretskii's avatar
Eli Zaretskii committed
275 276
/* Copy the bidi iterator from FROM to TO.  To save cycles, this only
   copies the part of the level stack that is actually in use.  */
277
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
278 279 280 281
bidi_copy_it (struct bidi_it *to, struct bidi_it *from)
{
  int i;

282
  /* Copy everything except the level stack and beyond.  */
283
  memcpy (to, from, offsetof (struct bidi_it, level_stack[0]));
Eli Zaretskii's avatar
Eli Zaretskii committed
284 285 286 287 288 289 290

  /* Copy the active part of the level stack.  */
  to->level_stack[0] = from->level_stack[0]; /* level zero is always in use */
  for (i = 1; i <= from->stack_idx; i++)
    to->level_stack[i] = from->level_stack[i];
}

291 292 293 294

/***********************************************************************
			Caching the bidi iterator states
 ***********************************************************************/
Eli Zaretskii's avatar
Eli Zaretskii committed
295

296 297 298
#define BIDI_CACHE_CHUNK 200
static struct bidi_it *bidi_cache;
static size_t bidi_cache_size = 0;
299
static size_t elsz = sizeof (struct bidi_it);
300 301 302 303 304 305 306 307 308 309 310
static EMACS_INT bidi_cache_idx;	/* next unused cache slot */
static EMACS_INT bidi_cache_last_idx;	/* slot of last cache hit */
static EMACS_INT bidi_cache_start = 0;	/* start of cache for this
					   "stack" level */

/* Reset the cache state to the empty state.  We only reset the part
   of the cache relevant to iteration of the current object.  Previous
   objects, which are pushed on the display iterator's stack, are left
   intact.  This is called when the cached information is no more
   useful for the current iteration, e.g. when we were reseated to a
   new position on the same object.  */
311
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
312 313
bidi_cache_reset (void)
{
314
  bidi_cache_idx = bidi_cache_start;
Eli Zaretskii's avatar
Eli Zaretskii committed
315 316 317
  bidi_cache_last_idx = -1;
}

318 319 320 321
/* Shrink the cache to its minimal size.  Called when we init the bidi
   iterator for reordering a buffer or a string that does not come
   from display properties, because that means all the previously
   cached info is of no further use.  */
322 323 324 325 326
static INLINE void
bidi_cache_shrink (void)
{
  if (bidi_cache_size > BIDI_CACHE_CHUNK)
    {
327 328 329
      bidi_cache_size = BIDI_CACHE_CHUNK;
      bidi_cache =
	(struct bidi_it *) xrealloc (bidi_cache, bidi_cache_size * elsz);
330 331 332 333
    }
  bidi_cache_reset ();
}

334
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
335 336 337 338
bidi_cache_fetch_state (int idx, struct bidi_it *bidi_it)
{
  int current_scan_dir = bidi_it->scan_dir;

339
  if (idx < bidi_cache_start || idx >= bidi_cache_idx)
Eli Zaretskii's avatar
Eli Zaretskii committed
340 341 342 343 344 345 346 347 348 349 350
    abort ();

  bidi_copy_it (bidi_it, &bidi_cache[idx]);
  bidi_it->scan_dir = current_scan_dir;
  bidi_cache_last_idx = idx;
}

/* Find a cached state with a given CHARPOS and resolved embedding
   level less or equal to LEVEL.  if LEVEL is -1, disregard the
   resolved levels in cached states.  DIR, if non-zero, means search
   in that direction from the last cache hit.  */
351
static INLINE int
352
bidi_cache_search (EMACS_INT charpos, int level, int dir)
Eli Zaretskii's avatar
Eli Zaretskii committed
353 354 355 356 357 358
{
  int i, i_start;

  if (bidi_cache_idx)
    {
      if (charpos < bidi_cache[bidi_cache_last_idx].charpos)
359 360 361 362 363 364 365 366 367 368 369
	{
	  dir = -1;
	  i_start = bidi_cache_last_idx - 1;
	}
      else if (charpos > (bidi_cache[bidi_cache_last_idx].charpos
			  + bidi_cache[bidi_cache_last_idx].nchars - 1))
	{
	  dir = 1;
	  i_start = bidi_cache_last_idx + 1;
	}
      else if (dir)
Eli Zaretskii's avatar
Eli Zaretskii committed
370 371 372 373 374 375 376 377 378 379
	i_start = bidi_cache_last_idx;
      else
	{
	  dir = -1;
	  i_start = bidi_cache_idx - 1;
	}

      if (dir < 0)
	{
	  /* Linear search for now; FIXME!  */
380
	  for (i = i_start; i >= bidi_cache_start; i--)
381 382
	    if (bidi_cache[i].charpos <= charpos
		&& charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
Eli Zaretskii's avatar
Eli Zaretskii committed
383 384 385 386 387 388
		&& (level == -1 || bidi_cache[i].resolved_level <= level))
	      return i;
	}
      else
	{
	  for (i = i_start; i < bidi_cache_idx; i++)
389 390
	    if (bidi_cache[i].charpos <= charpos
		&& charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
Eli Zaretskii's avatar
Eli Zaretskii committed
391 392 393 394 395 396 397 398 399 400 401
		&& (level == -1 || bidi_cache[i].resolved_level <= level))
	      return i;
	}
    }

  return -1;
}

/* Find a cached state where the resolved level changes to a value
   that is lower than LEVEL, and return its cache slot index.  DIR is
   the direction to search, starting with the last used cache slot.
402 403 404
   If DIR is zero, we search backwards from the last occupied cache
   slot.  BEFORE, if non-zero, means return the index of the slot that
   is ``before'' the level change in the search direction.  That is,
Eli Zaretskii's avatar
Eli Zaretskii committed
405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428
   given the cached levels like this:

	 1122333442211
	  AB        C

   and assuming we are at the position cached at the slot marked with
   C, searching backwards (DIR = -1) for LEVEL = 2 will return the
   index of slot B or A, depending whether BEFORE is, respectively,
   non-zero or zero.  */
static int
bidi_cache_find_level_change (int level, int dir, int before)
{
  if (bidi_cache_idx)
    {
      int i = dir ? bidi_cache_last_idx : bidi_cache_idx - 1;
      int incr = before ? 1 : 0;

      if (!dir)
	dir = -1;
      else if (!incr)
	i += dir;

      if (dir < 0)
	{
429
	  while (i >= bidi_cache_start + incr)
Eli Zaretskii's avatar
Eli Zaretskii committed
430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451
	    {
	      if (bidi_cache[i - incr].resolved_level >= 0
		  && bidi_cache[i - incr].resolved_level < level)
		return i;
	      i--;
	    }
	}
      else
	{
	  while (i < bidi_cache_idx - incr)
	    {
	      if (bidi_cache[i + incr].resolved_level >= 0
		  && bidi_cache[i + incr].resolved_level < level)
		return i;
	      i++;
	    }
	}
    }

  return -1;
}

452
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
453 454 455 456 457 458 459 460 461 462 463 464
bidi_cache_iterator_state (struct bidi_it *bidi_it, int resolved)
{
  int idx;

  /* We should never cache on backward scans.  */
  if (bidi_it->scan_dir == -1)
    abort ();
  idx = bidi_cache_search (bidi_it->charpos, -1, 1);

  if (idx < 0)
    {
      idx = bidi_cache_idx;
465 466 467
      /* Enlarge the cache as needed.  */
      if (idx >= bidi_cache_size)
	{
468
	  bidi_cache_size += BIDI_CACHE_CHUNK;
469
	  bidi_cache =
470
	    (struct bidi_it *) xrealloc (bidi_cache, bidi_cache_size * elsz);
471
	}
472 473 474
      /* Character positions should correspond to cache positions 1:1.
	 If we are outside the range of cached positions, the cache is
	 useless and must be reset.  */
475
      if (idx > bidi_cache_start &&
476 477
	  (bidi_it->charpos > (bidi_cache[idx - 1].charpos
			       + bidi_cache[idx - 1].nchars)
478 479 480
	   || bidi_it->charpos < bidi_cache[0].charpos))
	{
	  bidi_cache_reset ();
481
	  idx = bidi_cache_start;
482
	}
483 484
      if (bidi_it->nchars <= 0)
	abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
485 486 487 488 489 490 491 492 493
      bidi_copy_it (&bidi_cache[idx], bidi_it);
      if (!resolved)
	bidi_cache[idx].resolved_level = -1;
    }
  else
    {
      /* Copy only the members which could have changed, to avoid
	 costly copying of the entire struct.  */
      bidi_cache[idx].type = bidi_it->type;
494
      bidi_check_type (bidi_it->type);
495 496
      bidi_cache[idx].type_after_w1 = bidi_it->type_after_w1;
      bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512
      if (resolved)
	bidi_cache[idx].resolved_level = bidi_it->resolved_level;
      else
	bidi_cache[idx].resolved_level = -1;
      bidi_cache[idx].invalid_levels = bidi_it->invalid_levels;
      bidi_cache[idx].invalid_rl_levels = bidi_it->invalid_rl_levels;
      bidi_cache[idx].next_for_neutral = bidi_it->next_for_neutral;
      bidi_cache[idx].next_for_ws = bidi_it->next_for_ws;
      bidi_cache[idx].ignore_bn_limit = bidi_it->ignore_bn_limit;
    }

  bidi_cache_last_idx = idx;
  if (idx >= bidi_cache_idx)
    bidi_cache_idx = idx + 1;
}

513
static INLINE bidi_type_t
514
bidi_cache_find (EMACS_INT charpos, int level, struct bidi_it *bidi_it)
Eli Zaretskii's avatar
Eli Zaretskii committed
515 516 517
{
  int i = bidi_cache_search (charpos, level, bidi_it->scan_dir);

518
  if (i >= bidi_cache_start)
Eli Zaretskii's avatar
Eli Zaretskii committed
519 520 521
    {
      bidi_dir_t current_scan_dir = bidi_it->scan_dir;

522
      bidi_copy_it (bidi_it, &bidi_cache[i]);
Eli Zaretskii's avatar
Eli Zaretskii committed
523 524 525 526 527 528 529 530 531 532
      bidi_cache_last_idx = i;
      /* Don't let scan direction from from the cached state override
	 the current scan direction.  */
      bidi_it->scan_dir = current_scan_dir;
      return bidi_it->type;
    }

  return UNKNOWN_BT;
}

533
static INLINE int
Eli Zaretskii's avatar
Eli Zaretskii committed
534 535
bidi_peek_at_next_level (struct bidi_it *bidi_it)
{
536
  if (bidi_cache_idx == bidi_cache_start || bidi_cache_last_idx == -1)
Eli Zaretskii's avatar
Eli Zaretskii committed
537 538 539 540
    abort ();
  return bidi_cache[bidi_cache_last_idx + bidi_it->scan_dir].resolved_level;
}

541 542 543 544 545 546

/***********************************************************************
			Initialization
 ***********************************************************************/
static void
bidi_initialize (void)
Eli Zaretskii's avatar
Eli Zaretskii committed
547
{
548

549 550
#include "biditype.h"
#include "bidimirror.h"
551

552
  int i;
Eli Zaretskii's avatar
Eli Zaretskii committed
553

554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580
  bidi_type_table = Fmake_char_table (Qnil, make_number (STRONG_L));
  staticpro (&bidi_type_table);

  for (i = 0; i < sizeof bidi_type / sizeof bidi_type[0]; i++)
    char_table_set_range (bidi_type_table, bidi_type[i].from, bidi_type[i].to,
			  make_number (bidi_type[i].type));

  bidi_mirror_table = Fmake_char_table (Qnil, Qnil);
  staticpro (&bidi_mirror_table);

  for (i = 0; i < sizeof bidi_mirror / sizeof bidi_mirror[0]; i++)
    char_table_set (bidi_mirror_table, bidi_mirror[i].from,
		    make_number (bidi_mirror[i].to));

  Qparagraph_start = intern ("paragraph-start");
  staticpro (&Qparagraph_start);
  paragraph_start_re = Fsymbol_value (Qparagraph_start);
  if (!STRINGP (paragraph_start_re))
    paragraph_start_re = build_string ("\f\\|[ \t]*$");
  staticpro (&paragraph_start_re);
  Qparagraph_separate = intern ("paragraph-separate");
  staticpro (&Qparagraph_separate);
  paragraph_separate_re = Fsymbol_value (Qparagraph_separate);
  if (!STRINGP (paragraph_separate_re))
    paragraph_separate_re = build_string ("[ \t\f]*$");
  staticpro (&paragraph_separate_re);
  bidi_initialized = 1;
Eli Zaretskii's avatar
Eli Zaretskii committed
581 582
}

583 584
/* Do whatever UAX#9 clause X8 says should be done at paragraph's
   end.  */
585
static INLINE void
586
bidi_set_paragraph_end (struct bidi_it *bidi_it)
Eli Zaretskii's avatar
Eli Zaretskii committed
587
{
588 589 590 591 592
  bidi_it->invalid_levels = 0;
  bidi_it->invalid_rl_levels = -1;
  bidi_it->stack_idx = 0;
  bidi_it->resolved_level = bidi_it->level_stack[0].level;
}
Eli Zaretskii's avatar
Eli Zaretskii committed
593

594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616
/* Initialize the bidi iterator from buffer/string position CHARPOS.  */
void
bidi_init_it (EMACS_INT charpos, EMACS_INT bytepos, int frame_window_p,
	      struct bidi_it *bidi_it)
{
  if (! bidi_initialized)
    bidi_initialize ();
  if (charpos >= 0)
    bidi_it->charpos = charpos;
  if (bytepos >= 0)
    bidi_it->bytepos = bytepos;
  bidi_it->frame_window_p = frame_window_p;
  bidi_it->nchars = -1;	/* to be computed in bidi_resolve_explicit_1 */
  bidi_it->first_elt = 1;
  bidi_set_paragraph_end (bidi_it);
  bidi_it->new_paragraph = 1;
  bidi_it->separator_limit = -1;
  bidi_it->type = NEUTRAL_B;
  bidi_it->type_after_w1 = NEUTRAL_B;
  bidi_it->orig_type = NEUTRAL_B;
  bidi_it->prev_was_pdf = 0;
  bidi_it->prev.type = bidi_it->prev.type_after_w1 =
    bidi_it->prev.orig_type = UNKNOWN_BT;
617 618
  bidi_it->last_strong.type = bidi_it->last_strong.type_after_w1 =
    bidi_it->last_strong.orig_type = UNKNOWN_BT;
619 620 621
  bidi_it->next_for_neutral.charpos = -1;
  bidi_it->next_for_neutral.type =
    bidi_it->next_for_neutral.type_after_w1 =
622
    bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
623 624 625 626 627 628 629 630 631 632
  bidi_it->prev_for_neutral.charpos = -1;
  bidi_it->prev_for_neutral.type =
    bidi_it->prev_for_neutral.type_after_w1 =
    bidi_it->prev_for_neutral.orig_type = UNKNOWN_BT;
  bidi_it->sor = L2R;	 /* FIXME: should it be user-selectable? */
  bidi_it->disp_pos = -1;	/* invalid/unknown */
  /* We can only shrink the cache if we are at the bottom level of its
     "stack".  */
  if (bidi_cache_start == 0)
    bidi_cache_shrink ();
Eli Zaretskii's avatar
Eli Zaretskii committed
633 634
}

635
/* Perform initializations for reordering a new line of bidi text.  */
636
static void
637 638 639 640 641 642 643 644 645
bidi_line_init (struct bidi_it *bidi_it)
{
  bidi_it->scan_dir = 1; /* FIXME: do we need to have control on this? */
  bidi_it->resolved_level = bidi_it->level_stack[0].level;
  bidi_it->level_stack[0].override = NEUTRAL_DIR; /* X1 */
  bidi_it->invalid_levels = 0;
  bidi_it->invalid_rl_levels = -1;
  bidi_it->next_en_pos = -1;
  bidi_it->next_for_ws.type = UNKNOWN_BT;
646 647
  bidi_set_sor_type (bidi_it,
		     bidi_it->paragraph_dir == R2L ? 1 : 0,
648 649 650 651 652
		     bidi_it->level_stack[0].level); /* X10 */

  bidi_cache_reset ();
}

653 654 655 656 657

/***********************************************************************
			Fetching characters
 ***********************************************************************/

658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691
/* Count bytes in multibyte string S between BEG/BEGBYTE and END.  BEG
   and END are zero-based character positions in S, BEGBYTE is byte
   position corresponding to BEG.  */
static inline EMACS_INT
bidi_count_bytes (const unsigned char *s, const EMACS_INT beg,
		  const EMACS_INT begbyte, const EMACS_INT end)
{
  EMACS_INT pos = beg;
  const unsigned char *p = s + begbyte, *start = p;

  if (!CHAR_HEAD_P (*p))
    abort ();

  while (pos < end)
    {
      p += BYTES_BY_CHAR_HEAD (*p);
      pos++;
    }

  return p - start;
}

/* Fetch and returns the character at byte position BYTEPOS.  If S is
   non-NULL, fetch the character from string S; otherwise fetch the
   character from the current buffer.  */
static inline int
bidi_char_at_pos (EMACS_INT bytepos, const unsigned char *s)
{
  if (s)
    return STRING_CHAR (s + bytepos);
  else
    return FETCH_MULTIBYTE_CHAR (bytepos);
}

692 693 694 695 696 697 698
/* Fetch and return the character at BYTEPOS/CHARPOS.  If that
   character is covered by a display string, treat the entire run of
   covered characters as a single character u+FFFC, and return their
   combined length in CH_LEN and NCHARS.  DISP_POS specifies the
   character position of the next display string, or -1 if not yet
   computed.  When the next character is at or beyond that position,
   the function updates DISP_POS with the position of the next display
699 700 701
   string.  STRING->s is the C string to iterate, or NULL if iterating
   over a buffer or a Lisp string; in the latter case, STRING->lstring
   is the Lisp string.  */
702
static inline int
703
bidi_fetch_char (EMACS_INT bytepos, EMACS_INT charpos, EMACS_INT *disp_pos,
704
		 struct bidi_string_data *string,
705
		 int frame_window_p, EMACS_INT *ch_len, EMACS_INT *nchars)
706 707
{
  int ch;
708 709
  EMACS_INT endpos =
    (string->s || STRINGP (string->lstring)) ? string->schars : ZV;
710
  struct text_pos pos;
711 712

  /* If we got past the last known position of display string, compute
713 714
     the position of the next one.  That position could be at CHARPOS.  */
  if (charpos < endpos && charpos > *disp_pos)
715 716 717 718
    {
      SET_TEXT_POS (pos, charpos, bytepos);
      *disp_pos = compute_display_string_pos (&pos, string, frame_window_p);
    }
719 720

  /* Fetch the character at BYTEPOS.  */
721
  if (charpos >= endpos)
722 723 724 725
    {
      ch = BIDI_EOB;
      *ch_len = 1;
      *nchars = 1;
726
      *disp_pos = endpos;
727
    }
728
  else if (charpos >= *disp_pos)
729
    {
730 731 732 733 734 735 736
      EMACS_INT disp_end_pos;

      /* We don't expect to find ourselves in the middle of a display
	 property.  Hopefully, it will never be needed.  */
      if (charpos > *disp_pos)
	abort ();
      /* Return the Unicode Object Replacement Character to represent
737
	 the entire run of characters covered by the display string.  */
738
      ch = 0xFFFC;
739
      disp_end_pos = compute_display_string_end (*disp_pos, string);
740
      *nchars = disp_end_pos - *disp_pos;
741 742 743
      if (string->s)
	*ch_len = bidi_count_bytes (string->s, *disp_pos, bytepos,
				    disp_end_pos);
744 745 746
      else if (STRINGP (string->lstring))
	*ch_len = bidi_count_bytes (SDATA (string->lstring), *disp_pos,
				    bytepos, disp_end_pos);
747 748
      else
	*ch_len = CHAR_TO_BYTE (disp_end_pos) - bytepos;
749 750 751
    }
  else
    {
752 753 754 755 756 757 758
      if (string->s)
	{
	  EMACS_INT len;

	  ch = STRING_CHAR_AND_LENGTH (string->s + bytepos, len);
	  *ch_len = len;
	}
759 760 761 762 763 764 765
      else if (STRINGP (string->lstring))
	{
	  EMACS_INT len;

	  ch = STRING_CHAR_AND_LENGTH (SDATA (string->lstring) + bytepos, len);
	  *ch_len = len;
	}
766 767 768 769 770
      else
	{
	  ch = FETCH_MULTIBYTE_CHAR (bytepos);
	  *ch_len = CHAR_BYTES (ch);
	}
771 772 773 774 775
      *nchars = 1;
    }

  /* If we just entered a run of characters covered by a display
     string, compute the position of the next display string.  */
776
  if (charpos + *nchars <= endpos && charpos + *nchars > *disp_pos)
777 778 779 780
    {
      SET_TEXT_POS (pos, charpos + *nchars, bytepos + *ch_len);
      *disp_pos = compute_display_string_pos (&pos, string, frame_window_p);
    }
781 782 783 784

  return ch;
}

785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816

/***********************************************************************
			Determining paragraph direction
 ***********************************************************************/

/* Check if buffer position CHARPOS/BYTEPOS is the end of a paragraph.
   Value is the non-negative length of the paragraph separator
   following the buffer position, -1 if position is at the beginning
   of a new paragraph, or -2 if position is neither at beginning nor
   at end of a paragraph.  */
static EMACS_INT
bidi_at_paragraph_end (EMACS_INT charpos, EMACS_INT bytepos)
{
  Lisp_Object sep_re;
  Lisp_Object start_re;
  EMACS_INT val;

  sep_re = paragraph_separate_re;
  start_re = paragraph_start_re;

  val = fast_looking_at (sep_re, charpos, bytepos, ZV, ZV_BYTE, Qnil);
  if (val < 0)
    {
      if (fast_looking_at (start_re, charpos, bytepos, ZV, ZV_BYTE, Qnil) >= 0)
	val = -1;
      else
	val = -2;
    }

  return val;
}

817 818 819
/* Find the beginning of this paragraph by looking back in the buffer.
   Value is the byte position of the paragraph's beginning.  */
static EMACS_INT
820
bidi_find_paragraph_start (EMACS_INT pos, EMACS_INT pos_byte)
821
{
Eli Zaretskii's avatar
Eli Zaretskii committed
822
  Lisp_Object re = paragraph_start_re;
823 824 825 826 827
  EMACS_INT limit = ZV, limit_byte = ZV_BYTE;

  while (pos_byte > BEGV_BYTE
	 && fast_looking_at (re, pos, pos_byte, limit, limit_byte, Qnil) < 0)
    {
828 829 830 831
      /* FIXME: What if the paragraph beginning is covered by a
	 display string?  And what if a display string covering some
	 of the text over which we scan back includes
	 paragraph_start_re?  */
832 833
      pos = find_next_newline_no_quit (pos - 1, -1);
      pos_byte = CHAR_TO_BYTE (pos);
834
    }
835
  return pos_byte;
836 837
}

838
/* Determine the base direction, a.k.a. base embedding level, of the
839 840
   paragraph we are about to iterate through.  If DIR is either L2R or
   R2L, just use that.  Otherwise, determine the paragraph direction
841 842
   from the first strong directional character of the paragraph.

843
   NO_DEFAULT_P non-zero means don't default to L2R if the paragraph
844 845 846 847 848 849 850 851 852
   has no strong directional characters and both DIR and
   bidi_it->paragraph_dir are NEUTRAL_DIR.  In that case, search back
   in the buffer until a paragraph is found with a strong character,
   or until hitting BEGV.  In the latter case, fall back to L2R.  This
   flag is used in current-bidi-paragraph-direction.

   Note that this function gives the paragraph separator the same
   direction as the preceding paragraph, even though Emacs generally
   views the separartor as not belonging to any paragraph.  */
Eli Zaretskii's avatar
Eli Zaretskii committed
853
void
854
bidi_paragraph_init (bidi_dir_t dir, struct bidi_it *bidi_it, int no_default_p)
Eli Zaretskii's avatar
Eli Zaretskii committed
855
{
856
  EMACS_INT bytepos = bidi_it->bytepos;
857
  int string_p = bidi_it->string.s != NULL || STRINGP (bidi_it->string.lstring);
858
  EMACS_INT pstartbyte;
859 860 861 862 863
  /* Note that begbyte is a byte position, while end is a character
     position.  Yes, this is ugly, but we are trying to avoid costly
     calls to BYTE_TO_CHAR and its ilk.  */
  EMACS_INT begbyte = string_p ? 0 : BEGV_BYTE;
  EMACS_INT end = string_p ? bidi_it->string.schars : ZV;
864

865
  /* Special case for an empty buffer. */
866
  if (bytepos == begbyte && bidi_it->charpos == end)
867
    dir = L2R;
868
  /* We should never be called at EOB or before BEGV.  */
869
  else if (bidi_it->charpos >= end || bytepos < begbyte)
870 871
    abort ();

872 873 874 875 876 877 878 879 880 881
  if (dir == L2R)
    {
      bidi_it->paragraph_dir = L2R;
      bidi_it->new_paragraph = 0;
    }
  else if (dir == R2L)
    {
      bidi_it->paragraph_dir = R2L;
      bidi_it->new_paragraph = 0;
    }
Eli Zaretskii's avatar
Eli Zaretskii committed
882 883
  else if (dir == NEUTRAL_DIR)	/* P2 */
    {
884 885 886
      int ch;
      EMACS_INT ch_len, nchars;
      EMACS_INT pos, disp_pos = -1;
887
      bidi_type_t type;
888
      const unsigned char *s;
889

890 891 892
      if (!bidi_initialized)
	bidi_initialize ();

893 894
      /* If we are inside a paragraph separator, we are just waiting
	 for the separator to be exhausted; use the previous paragraph
895 896 897 898
	 direction.  But don't do that if we have been just reseated,
	 because we need to reinitialize below in that case.  */
      if (!bidi_it->first_elt
	  && bidi_it->charpos < bidi_it->separator_limit)
899 900
	return;

901
      /* If we are on a newline, get past it to where the next
902 903 904
	 paragraph might start.  But don't do that at BEGV since then
	 we are potentially in a new paragraph that doesn't yet
	 exist.  */
905
      pos = bidi_it->charpos;
906 907 908
      s = STRINGP (bidi_it->string.lstring) ?
	SDATA (bidi_it->string.lstring) : bidi_it->string.s;
      if (bytepos > begbyte && bidi_char_at_pos (bytepos, s) == '\n')
909
	{
910
	  bytepos++;
911
	  pos++;
912
	}
913 914 915

      /* We are either at the beginning of a paragraph or in the
	 middle of it.  Find where this paragraph starts.  */
916 917 918 919 920 921 922 923
      if (string_p)
	{
	  /* We don't support changes of paragraph direction inside a
	     string.  It is treated as a single paragraph.  */
	  pstartbyte = 0;
	}
      else
	pstartbyte = bidi_find_paragraph_start (pos, bytepos);
924 925
      bidi_it->separator_limit = -1;
      bidi_it->new_paragraph = 0;
926 927

      /* The following loop is run more than once only if NO_DEFAULT_P
928
	 is non-zero, and only if we are iterating on a buffer.  */
929 930
      do {
	bytepos = pstartbyte;
931 932 933 934
	if (!string_p)
	  pos = BYTE_TO_CHAR (bytepos);
	ch = bidi_fetch_char (bytepos, pos, &disp_pos, &bidi_it->string,
			      bidi_it->frame_window_p, &ch_len, &nchars);
935 936
	type = bidi_get_type (ch, NEUTRAL_DIR);

937
	for (pos += nchars, bytepos += ch_len;
938 939 940 941 942 943 944 945 946 947
	     /* NOTE: UAX#9 says to search only for L, AL, or R types
		of characters, and ignore RLE, RLO, LRE, and LRO.
		However, I'm not sure it makes sense to omit those 4;
		should try with and without that to see the effect.  */
	     (bidi_get_category (type) != STRONG)
	       || (bidi_ignore_explicit_marks_for_paragraph_level
		   && (type == RLE || type == RLO
		       || type == LRE || type == LRO));
	     type = bidi_get_type (ch, NEUTRAL_DIR))
	  {
948 949 950
	    if (!string_p
		&& type == NEUTRAL_B
		&& bidi_at_paragraph_end (pos, bytepos) >= -1)
951
	      break;
952
	    if (pos >= end)
953 954
	      {
		/* Pretend there's a paragraph separator at end of
955
		   buffer/string.  */
956 957 958
		type = NEUTRAL_B;
		break;
	      }
959
	    /* Fetch next character and advance to get past it.  */
960
	    ch = bidi_fetch_char (bytepos, pos, &disp_pos, &bidi_it->string,
961
				  bidi_it->frame_window_p, &ch_len, &nchars);
962 963
	    pos += nchars;
	    bytepos += ch_len;
964 965 966 967 968
	  }
	if (type == STRONG_R || type == STRONG_AL) /* P3 */
	  bidi_it->paragraph_dir = R2L;
	else if (type == STRONG_L)
	  bidi_it->paragraph_dir = L2R;
969 970
	if (!string_p
	    && no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR)
971 972 973 974 975 976 977 978 979 980 981 982
	  {
	    /* If this paragraph is at BEGV, default to L2R.  */
	    if (pstartbyte == BEGV_BYTE)
	      bidi_it->paragraph_dir = L2R; /* P3 and HL1 */
	    else
	      {
		EMACS_INT prevpbyte = pstartbyte;
		EMACS_INT p = BYTE_TO_CHAR (pstartbyte), pbyte = pstartbyte;

		/* Find the beginning of the previous paragraph, if any.  */
		while (pbyte > BEGV_BYTE && prevpbyte >= pstartbyte)
		  {
983 984 985
		    /* FXIME: What if p is covered by a display
		       string?  See also a FIXME inside
		       bidi_find_paragraph_start.  */
986 987 988 989 990 991 992
		    p--;
		    pbyte = CHAR_TO_BYTE (p);
		    prevpbyte = bidi_find_paragraph_start (p, pbyte);
		  }
		pstartbyte = prevpbyte;
	      }
	  }
993 994
      } while (!string_p
	       && no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR);
Eli Zaretskii's avatar
Eli Zaretskii committed
995
    }
996 997
  else
    abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
998

999 1000
  /* Contrary to UAX#9 clause P3, we only default the paragraph
     direction to L2R if we have no previous usable paragraph
1001
     direction.  This is allowed by the HL1 clause.  */
1002
  if (bidi_it->paragraph_dir != L2R && bidi_it->paragraph_dir != R2L)
1003
    bidi_it->paragraph_dir = L2R; /* P3 and HL1 ``higher-level protocols'' */
1004
  if (bidi_it->paragraph_dir == R2L)
1005
    bidi_it->level_stack[0].level = 1;
1006
  else
1007
    bidi_it->level_stack[0].level = 0;
1008 1009

  bidi_line_init (bidi_it);
Eli Zaretskii's avatar
Eli Zaretskii committed
1010 1011
}

1012 1013 1014 1015 1016 1017

/***********************************************************************
		 Resolving explicit and implicit levels.
		 The rest of the file constitutes the core
		 of the UBA implementation.
 ***********************************************************************/
Eli Zaretskii's avatar
Eli Zaretskii committed
1018

1019
static INLINE int
1020
bidi_explicit_dir_char (int ch)
Eli Zaretskii's avatar
Eli Zaretskii committed
1021
{
1022 1023 1024 1025 1026 1027 1028 1029
  bidi_type_t ch_type;

  if (!bidi_initialized)
    abort ();
  ch_type = (bidi_type_t) XINT (CHAR_TABLE_REF (bidi_type_table, ch));
  return (ch_type == LRE || ch_type == LRO
	  || ch_type == RLE || ch_type == RLO
	  || ch_type == PDF);
Eli Zaretskii's avatar
Eli Zaretskii committed
1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
}

/* A helper function for bidi_resolve_explicit.  It advances to the
   next character in logical order and determines the new embedding
   level and directional override, but does not take into account
   empty embeddings.  */
static int
bidi_resolve_explicit_1 (struct bidi_it *bidi_it)
{
  int curchar;
  bidi_type_t type;
  int current_level;
  int new_level;
  bidi_dir_t override;
1044
  int string_p = bidi_it->string.s != NULL || STRINGP (bidi_it->string.lstring);
Eli Zaretskii's avatar
Eli Zaretskii committed
1045

1046 1047 1048
  /* If reseat()'ed, don't advance, so as to start iteration from the
     position where we were reseated.  bidi_it->bytepos can be less
     than BEGV_BYTE after reseat to BEGV.  */
1049
  if (bidi_it->bytepos < (string_p ? 0 : BEGV_BYTE)
1050
      || bidi_it->first_elt)
1051
    {
1052
      bidi_it->first_elt = 0;
1053 1054
      if (string_p)
	{
1055 1056 1057 1058
	  const unsigned char *p =
	    STRINGP (bidi_it->string.lstring)
	    ? SDATA (bidi_it->string.lstring) : bidi_it->string.s;

1059 1060
	  if (bidi_it->charpos < 0)
	    bidi_it->charpos = 0;
1061
	  bidi_it->bytepos = bidi_count_bytes (p, 0, 0, bidi_it->charpos);
1062 1063 1064 1065 1066 1067 1068
	}
      else
	{
	  if (bidi_it->charpos < BEGV)
	    bidi_it->charpos = BEGV;
	  bidi_it->bytepos = CHAR_TO_BYTE (bidi_it->charpos);
	}
1069
    }
1070 1071
  /* Don't move at end of buffer/string.  */
  else if (bidi_it->charpos < (string_p ? bidi_it->string.schars : ZV))
Eli Zaretskii's avatar
Eli Zaretskii committed
1072
    {
1073 1074
      /* Advance to the next character, skipping characters covered by
	 display strings (nchars > 1).  */
1075 1076
      if (bidi_it->nchars <= 0)
	abort ();
1077
      bidi_it->charpos += bidi_it->nchars;
1078 1079
      if (bidi_it->ch_len == 0)
	abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
1080 1081 1082 1083 1084 1085 1086
      bidi_it->bytepos += bidi_it->ch_len;
    }

  current_level = bidi_it->level_stack[bidi_it->stack_idx].level; /* X1 */
  override = bidi_it->level_stack[bidi_it->stack_idx].override;
  new_level = current_level;

1087
  if (bidi_it->charpos >= (string_p ? bidi_it->string.schars : ZV))
1088 1089 1090
    {
      curchar = BIDI_EOB;
      bidi_it->ch_len = 1;
1091
      bidi_it->nchars = 1;
1092
      bidi_it->disp_pos = (string_p ? bidi_it->string.schars : ZV);
1093 1094 1095
    }
  else
    {
1096 1097 1098
      /* Fetch the character at BYTEPOS.  If it is covered by a
	 display string, treat the entire run of covered characters as
	 a single character u+FFFC.  */
1099
      curchar = bidi_fetch_char (bidi_it->bytepos, bidi_it->charpos,
1100 1101
				 &bidi_it->disp_pos, &bidi_it->string,
				 bidi_it->frame_window_p,
1102
				 &bidi_it->ch_len, &bidi_it->nchars);
1103
    }
Eli Zaretskii's avatar
Eli Zaretskii committed
1104 1105
  bidi_it->ch = curchar;

1106 1107 1108 1109 1110
  /* Don't apply directional override here, as all the types we handle
     below will not be affected by the override anyway, and we need
     the original type unaltered.  The override will be applied in
     bidi_resolve_weak.  */
  type = bidi_get_type (curchar, NEUTRAL_DIR);
1111 1112
  bidi_it->orig_type = type;
  bidi_check_type (bidi_it->orig_type);
Eli Zaretskii's avatar
Eli Zaretskii committed
1113 1114 1115 1116

  if (type != PDF)
    bidi_it->prev_was_pdf = 0;

1117
  bidi_it->type_after_w1 = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
1118 1119 1120 1121 1122

  switch (type)
    {
      case RLE:	/* X2 */
      case RLO:	/* X4 */
1123 1124
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1125
	type = WEAK_BN; /* X9/Retaining */
1126
	if (bidi_it->ignore_bn_limit <= -1)
Eli Zaretskii's avatar
Eli Zaretskii committed
1127 1128 1129 1130 1131 1132
	  {
	    if (current_level <= BIDI_MAXLEVEL - 4)
	      {
		/* Compute the least odd embedding level greater than
		   the current level.  */
		new_level = ((current_level + 1) & ~1) + 1;
1133
		if (bidi_it->type_after_w1 == RLE)
Eli Zaretskii's avatar
Eli Zaretskii committed
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
		  override = NEUTRAL_DIR;
		else
		  override = R2L;
		if (current_level == BIDI_MAXLEVEL - 4)
		  bidi_it->invalid_rl_levels = 0;
		bidi_push_embedding_level (bidi_it, new_level, override);
	      }
	    else
	      {
		bidi_it->invalid_levels++;
		/* See the commentary about invalid_rl_levels below.  */
		if (bidi_it->invalid_rl_levels < 0)
		  bidi_it->invalid_rl_levels = 0;
		bidi_it->invalid_rl_levels++;
	      }
	  }
1150
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1151 1152 1153 1154 1155
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      case LRE:	/* X3 */
      case LRO:	/* X5 */
1156 1157
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1158
	type = WEAK_BN; /* X9/Retaining */
1159
	if (bidi_it->ignore_bn_limit <= -1)
Eli Zaretskii's avatar
Eli Zaretskii committed
1160 1161 1162 1163 1164 1165
	  {
	    if (current_level <= BIDI_MAXLEVEL - 5)
	      {
		/* Compute the least even embedding level greater than
		   the current level.  */
		new_level = ((current_level + 2) & ~1);
1166
		if (bidi_it->type_after_w1 == LRE)
Eli Zaretskii's avatar
Eli Zaretskii committed
1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185
		  override = NEUTRAL_DIR;
		else
		  override = L2R;
		bidi_push_embedding_level (bidi_it, new_level, override);
	      }
	    else
	      {
		bidi_it->invalid_levels++;
		/* invalid_rl_levels counts invalid levels encountered
		   while the embedding level was already too high for
		   LRE/LRO, but not for RLE/RLO.  That is because
		   there may be exactly one PDF which we should not
		   ignore even though invalid_levels is non-zero.
		   invalid_rl_levels helps to know what PDF is
		   that.  */
		if (bidi_it->invalid_rl_levels >= 0)
		  bidi_it->invalid_rl_levels++;
	      }
	  }
1186
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1187 1188 1189 1190
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      case PDF:	/* X7 */
1191 1192
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1193
	type = WEAK_BN; /* X9/Retaining */
1194
	if (bidi_it->ignore_bn_limit <= -1)
Eli Zaretskii's avatar
Eli Zaretskii committed
1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211
	  {
	    if (!bidi_it->invalid_rl_levels)
	      {
		new_level = bidi_pop_embedding_level (bidi_it);
		bidi_it->invalid_rl_levels = -1;
		if (bidi_it->invalid_levels)
		  bidi_it->invalid_levels--;
		/* else nothing: UAX#9 says to ignore invalid PDFs */
	      }
	    if (!bidi_it->invalid_levels)
	      new_level = bidi_pop_embedding_level (bidi_it);
	    else
	      {
		bidi_it->invalid_levels--;
		bidi_it->invalid_rl_levels--;
	      }
	  }
1212
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1213 1214 1215 1216 1217 1218 1219 1220 1221
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      default:
	/* Nothing.  */
	break;
    }

  bidi_it->type = type;
1222
  bidi_check_type (bidi_it->type);
Eli Zaretskii's avatar
Eli Zaretskii committed
1223 1224 1225 1226 1227

  return new_level;
}

/* Given an iterator state in BIDI_IT, advance one character position
1228 1229 1230 1231
   in the buffer/string to the next character (in the logical order),
   resolve any explicit embeddings and directional overrides, and
   return the embedding level of the character after resolving
   explicit directives and ignoring empty embeddings.  */
Eli Zaretskii's avatar
Eli Zaretskii committed
1232 1233 1234 1235 1236
static int
bidi_resolve_explicit (struct bidi_it *bidi_it)
{
  int prev_level = bidi_it->level_stack[bidi_it->stack_idx].level;
  int new_level  = bidi_resolve_explicit_1 (bidi_it);
1237
  EMACS_INT eob = bidi_it->string.s ? bidi_it->string.schars : ZV;
1238 1239
  const unsigned char *s = STRINGP (bidi_it->string.lstring)
    ? SDATA (bidi_it->string.