bidi.c 65.4 KB
Newer Older
Eli Zaretskii's avatar
Eli Zaretskii committed
1
/* Low-level bidirectional buffer-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,
Eli Zaretskii's avatar
Eli Zaretskii committed
23 24 25
   as per UAX#9, a part of the Unicode Standard.

   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

static void
84
bidi_initialize (void)
Eli Zaretskii's avatar
Eli Zaretskii committed
85
{
86 87

#include "biditype.h"
88
#include "bidimirror.h"
89

Eli Zaretskii's avatar
Eli Zaretskii committed
90 91 92
  int i;

  bidi_type_table = Fmake_char_table (Qnil, make_number (STRONG_L));
93
  staticpro (&bidi_type_table);
Eli Zaretskii's avatar
Eli Zaretskii committed
94 95

  for (i = 0; i < sizeof bidi_type / sizeof bidi_type[0]; i++)
96
    char_table_set_range (bidi_type_table, bidi_type[i].from, bidi_type[i].to,
Eli Zaretskii's avatar
Eli Zaretskii committed
97
			  make_number (bidi_type[i].type));
98

99 100 101 102 103 104 105
  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));

106 107
  Qparagraph_start = intern ("paragraph-start");
  staticpro (&Qparagraph_start);
Eli Zaretskii's avatar
Eli Zaretskii committed
108 109 110 111
  paragraph_start_re = Fsymbol_value (Qparagraph_start);
  if (!STRINGP (paragraph_start_re))
    paragraph_start_re = build_string ("\f\\|[ \t]*$");
  staticpro (&paragraph_start_re);
112 113
  Qparagraph_separate = intern ("paragraph-separate");
  staticpro (&Qparagraph_separate);
Eli Zaretskii's avatar
Eli Zaretskii committed
114 115 116 117
  paragraph_separate_re = Fsymbol_value (Qparagraph_separate);
  if (!STRINGP (paragraph_separate_re))
    paragraph_separate_re = build_string ("[ \t\f]*$");
  staticpro (&paragraph_separate_re);
Eli Zaretskii's avatar
Eli Zaretskii committed
118 119 120
  bidi_initialized = 1;
}

121 122
/* Return the bidi type of a character CH, subject to the current
   directional OVERRIDE.  */
123
static INLINE bidi_type_t
124
bidi_get_type (int ch, bidi_dir_t override)
Eli Zaretskii's avatar
Eli Zaretskii committed
125
{
126 127
  bidi_type_t default_type;

128 129
  if (ch == BIDI_EOB)
    return NEUTRAL_B;
130 131
  if (ch < 0 || ch > MAX_CHAR)
    abort ();
132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163

  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
164 165
}

166
static void
167 168 169 170 171 172
bidi_check_type (bidi_type_t type)
{
  if (type < UNKNOWN_BT || type > NEUTRAL_ON)
    abort ();
}

Eli Zaretskii's avatar
Eli Zaretskii committed
173
/* Given a bidi TYPE of a character, return its category.  */
174
static INLINE bidi_category_t
Eli Zaretskii's avatar
Eli Zaretskii committed
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207
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 ();
    }
}

208 209 210 211
/* 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
212 213 214
int
bidi_mirror_char (int c)
{
215 216 217 218 219 220
  Lisp_Object val;

  if (c == BIDI_EOB)
    return c;
  if (c < 0 || c > MAX_CHAR)
    abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
221

222 223
  val = CHAR_TABLE_REF (bidi_mirror_table, c);
  if (INTEGERP (val))
Eli Zaretskii's avatar
Eli Zaretskii committed
224
    {
225
      int v = XINT (val);
Eli Zaretskii's avatar
Eli Zaretskii committed
226

227 228 229 230
      if (v < 0 || v > MAX_CHAR)
	abort ();

      return v;
Eli Zaretskii's avatar
Eli Zaretskii committed
231
    }
232

Eli Zaretskii's avatar
Eli Zaretskii committed
233 234 235 236 237
  return c;
}

/* 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.  */
238
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
239 240 241 242
bidi_copy_it (struct bidi_it *to, struct bidi_it *from)
{
  int i;

243
  /* Copy everything except the level stack and beyond.  */
244
  memcpy (to, from, offsetof (struct bidi_it, level_stack[0]));
Eli Zaretskii's avatar
Eli Zaretskii committed
245 246 247 248 249 250 251 252 253

  /* 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];
}

/* Caching the bidi iterator states.  */

254 255 256
#define BIDI_CACHE_CHUNK 200
static struct bidi_it *bidi_cache;
static size_t bidi_cache_size = 0;
257
static size_t elsz = sizeof (struct bidi_it);
258 259 260 261 262 263 264 265 266 267 268
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.  */
269
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
270 271
bidi_cache_reset (void)
{
272
  bidi_cache_idx = bidi_cache_start;
Eli Zaretskii's avatar
Eli Zaretskii committed
273 274 275
  bidi_cache_last_idx = -1;
}

276 277 278 279
/* 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.  */
280 281 282 283 284
static INLINE void
bidi_cache_shrink (void)
{
  if (bidi_cache_size > BIDI_CACHE_CHUNK)
    {
285 286 287
      bidi_cache_size = BIDI_CACHE_CHUNK;
      bidi_cache =
	(struct bidi_it *) xrealloc (bidi_cache, bidi_cache_size * elsz);
288 289 290 291
    }
  bidi_cache_reset ();
}

292
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
293 294 295 296
bidi_cache_fetch_state (int idx, struct bidi_it *bidi_it)
{
  int current_scan_dir = bidi_it->scan_dir;

297
  if (idx < bidi_cache_start || idx >= bidi_cache_idx)
Eli Zaretskii's avatar
Eli Zaretskii committed
298 299 300 301 302 303 304 305 306 307 308
    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.  */
309
static INLINE int
310
bidi_cache_search (EMACS_INT charpos, int level, int dir)
Eli Zaretskii's avatar
Eli Zaretskii committed
311 312 313 314 315 316
{
  int i, i_start;

  if (bidi_cache_idx)
    {
      if (charpos < bidi_cache[bidi_cache_last_idx].charpos)
317 318 319 320 321 322 323 324 325 326 327
	{
	  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
328 329 330 331 332 333 334 335 336 337
	i_start = bidi_cache_last_idx;
      else
	{
	  dir = -1;
	  i_start = bidi_cache_idx - 1;
	}

      if (dir < 0)
	{
	  /* Linear search for now; FIXME!  */
338
	  for (i = i_start; i >= bidi_cache_start; i--)
339 340
	    if (bidi_cache[i].charpos <= charpos
		&& charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
Eli Zaretskii's avatar
Eli Zaretskii committed
341 342 343 344 345 346
		&& (level == -1 || bidi_cache[i].resolved_level <= level))
	      return i;
	}
      else
	{
	  for (i = i_start; i < bidi_cache_idx; i++)
347 348
	    if (bidi_cache[i].charpos <= charpos
		&& charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
Eli Zaretskii's avatar
Eli Zaretskii committed
349 350 351 352 353 354 355 356 357 358 359
		&& (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.
360 361 362
   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
363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386
   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)
	{
387
	  while (i >= bidi_cache_start + incr)
Eli Zaretskii's avatar
Eli Zaretskii committed
388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409
	    {
	      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;
}

410
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
411 412 413 414 415 416 417 418 419 420 421 422
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;
423 424 425
      /* Enlarge the cache as needed.  */
      if (idx >= bidi_cache_size)
	{
426
	  bidi_cache_size += BIDI_CACHE_CHUNK;
427
	  bidi_cache =
428
	    (struct bidi_it *) xrealloc (bidi_cache, bidi_cache_size * elsz);
429
	}
430 431 432
      /* 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.  */
433
      if (idx > bidi_cache_start &&
434 435
	  (bidi_it->charpos > (bidi_cache[idx - 1].charpos
			       + bidi_cache[idx - 1].nchars)
436 437 438
	   || bidi_it->charpos < bidi_cache[0].charpos))
	{
	  bidi_cache_reset ();
439
	  idx = bidi_cache_start;
440
	}
441 442
      if (bidi_it->nchars <= 0)
	abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
443 444 445 446 447 448 449 450 451
      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;
452
      bidi_check_type (bidi_it->type);
453 454
      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
455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470
      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;
}

471
static INLINE bidi_type_t
472
bidi_cache_find (EMACS_INT charpos, int level, struct bidi_it *bidi_it)
Eli Zaretskii's avatar
Eli Zaretskii committed
473 474 475
{
  int i = bidi_cache_search (charpos, level, bidi_it->scan_dir);

476
  if (i >= bidi_cache_start)
Eli Zaretskii's avatar
Eli Zaretskii committed
477 478 479
    {
      bidi_dir_t current_scan_dir = bidi_it->scan_dir;

480
      bidi_copy_it (bidi_it, &bidi_cache[i]);
Eli Zaretskii's avatar
Eli Zaretskii committed
481 482 483 484 485 486 487 488 489 490
      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;
}

491
static INLINE int
Eli Zaretskii's avatar
Eli Zaretskii committed
492 493
bidi_peek_at_next_level (struct bidi_it *bidi_it)
{
494
  if (bidi_cache_idx == bidi_cache_start || bidi_cache_last_idx == -1)
Eli Zaretskii's avatar
Eli Zaretskii committed
495 496 497 498
    abort ();
  return bidi_cache[bidi_cache_last_idx + bidi_it->scan_dir].resolved_level;
}

499 500 501 502 503
/* 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.  */
504
static EMACS_INT
505
bidi_at_paragraph_end (EMACS_INT charpos, EMACS_INT bytepos)
Eli Zaretskii's avatar
Eli Zaretskii committed
506
{
Eli Zaretskii's avatar
Eli Zaretskii committed
507 508
  Lisp_Object sep_re;
  Lisp_Object start_re;
509 510
  EMACS_INT val;

Eli Zaretskii's avatar
Eli Zaretskii committed
511 512
  sep_re = paragraph_separate_re;
  start_re = paragraph_start_re;
513 514 515 516 517 518 519 520 521

  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;
    }
Eli Zaretskii's avatar
Eli Zaretskii committed
522

523
  return val;
Eli Zaretskii's avatar
Eli Zaretskii committed
524 525 526 527 528 529
}

/* 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.  */
530
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
531 532 533 534 535 536 537 538 539 540 541
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.  */
542
  if (!bidi_it->prev_was_pdf || level_before < level_after)
Eli Zaretskii's avatar
Eli Zaretskii committed
543 544 545 546 547 548
    /* 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;
549 550
  bidi_it->last_strong.type = bidi_it->last_strong.type_after_w1 =
    bidi_it->last_strong.orig_type = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
551 552 553
  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;
554 555
  bidi_it->next_for_neutral.type = bidi_it->next_for_neutral.type_after_w1 =
    bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
556
  bidi_it->ignore_bn_limit = -1; /* meaning it's unknown */
Eli Zaretskii's avatar
Eli Zaretskii committed
557 558
}

559
/* Perform initializations for reordering a new line of bidi text.  */
560
static void
561 562 563 564 565 566 567 568 569
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;
570 571
  bidi_set_sor_type (bidi_it,
		     bidi_it->paragraph_dir == R2L ? 1 : 0,
572 573 574 575 576
		     bidi_it->level_stack[0].level); /* X10 */

  bidi_cache_reset ();
}

577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610
/* 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);
}

611 612 613 614 615 616 617
/* 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
618 619 620
   string.  STRING->s is the string to iterate, or NULL if iterating over
   a buffer.  */
static inline int
621
bidi_fetch_char (EMACS_INT bytepos, EMACS_INT charpos, EMACS_INT *disp_pos,
622
		 struct bidi_string_data *string,
623
		 int frame_window_p, EMACS_INT *ch_len, EMACS_INT *nchars)
624 625
{
  int ch;
626
  EMACS_INT endpos = string->s ? string->schars : ZV;
627 628

  /* If we got past the last known position of display string, compute
629 630 631
     the position of the next one.  That position could be at CHARPOS.  */
  if (charpos < endpos && charpos > *disp_pos)
    *disp_pos = compute_display_string_pos (charpos, string, frame_window_p);
632 633

  /* Fetch the character at BYTEPOS.  */
634
  if (charpos >= endpos)
635 636 637 638
    {
      ch = BIDI_EOB;
      *ch_len = 1;
      *nchars = 1;
639
      *disp_pos = endpos;
640
    }
641
  else if (charpos >= *disp_pos)
642
    {
643 644 645 646 647 648 649
      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
650
	 the entire run of characters covered by the display string.  */
651
      ch = 0xFFFC;
652
      disp_end_pos = compute_display_string_end (*disp_pos, string);
653
      *nchars = disp_end_pos - *disp_pos;
654 655 656 657 658
      if (string->s)
	*ch_len = bidi_count_bytes (string->s, *disp_pos, bytepos,
				    disp_end_pos);
      else
	*ch_len = CHAR_TO_BYTE (disp_end_pos) - bytepos;
659 660 661
    }
  else
    {
662 663 664 665 666 667 668 669 670 671 672 673
      if (string->s)
	{
	  EMACS_INT len;

	  ch = STRING_CHAR_AND_LENGTH (string->s + bytepos, len);
	  *ch_len = len;
	}
      else
	{
	  ch = FETCH_MULTIBYTE_CHAR (bytepos);
	  *ch_len = CHAR_BYTES (ch);
	}
674 675 676 677 678
      *nchars = 1;
    }

  /* If we just entered a run of characters covered by a display
     string, compute the position of the next display string.  */
679 680 681
  if (charpos + *nchars <= endpos && charpos + *nchars > *disp_pos)
    *disp_pos = compute_display_string_pos (charpos + *nchars, string,
					    frame_window_p);
682 683 684 685

  return ch;
}

686 687 688
/* 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
689
bidi_find_paragraph_start (EMACS_INT pos, EMACS_INT pos_byte)
690
{
Eli Zaretskii's avatar
Eli Zaretskii committed
691
  Lisp_Object re = paragraph_start_re;
692 693 694 695 696
  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)
    {
697 698 699 700
      /* 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?  */
701 702
      pos = find_next_newline_no_quit (pos - 1, -1);
      pos_byte = CHAR_TO_BYTE (pos);
703
    }
704
  return pos_byte;
705 706
}

707
/* Determine the base direction, a.k.a. base embedding level, of the
708 709
   paragraph we are about to iterate through.  If DIR is either L2R or
   R2L, just use that.  Otherwise, determine the paragraph direction
710 711
   from the first strong directional character of the paragraph.

712
   NO_DEFAULT_P non-zero means don't default to L2R if the paragraph
713 714 715 716 717 718 719 720 721
   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
722
void
723
bidi_paragraph_init (bidi_dir_t dir, struct bidi_it *bidi_it, int no_default_p)
Eli Zaretskii's avatar
Eli Zaretskii committed
724
{
725
  EMACS_INT bytepos = bidi_it->bytepos;
726
  int string_p = bidi_it->string.s != NULL;
727
  EMACS_INT pstartbyte;
728 729 730 731 732
  /* 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;
733

734
  /* Special case for an empty buffer. */
735
  if (bytepos == begbyte && bidi_it->charpos == end)
736
    dir = L2R;
737
  /* We should never be called at EOB or before BEGV.  */
738
  else if (bidi_it->charpos >= end || bytepos < begbyte)
739 740
    abort ();

741 742 743 744 745 746 747 748 749 750
  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
751 752
  else if (dir == NEUTRAL_DIR)	/* P2 */
    {
753 754 755
      int ch;
      EMACS_INT ch_len, nchars;
      EMACS_INT pos, disp_pos = -1;
756
      bidi_type_t type;
757

758 759 760
      if (!bidi_initialized)
	bidi_initialize ();

761 762
      /* If we are inside a paragraph separator, we are just waiting
	 for the separator to be exhausted; use the previous paragraph
763 764 765 766
	 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)
767 768
	return;

769
      /* If we are on a newline, get past it to where the next
770 771 772
	 paragraph might start.  But don't do that at BEGV since then
	 we are potentially in a new paragraph that doesn't yet
	 exist.  */
773
      pos = bidi_it->charpos;
774 775
      if (bytepos > begbyte
	  && bidi_char_at_pos (bytepos, bidi_it->string.s) == '\n')
776
	{
777
	  bytepos++;
778
	  pos++;
779
	}
780 781 782

      /* We are either at the beginning of a paragraph or in the
	 middle of it.  Find where this paragraph starts.  */
783 784 785 786 787 788 789 790
      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);
791 792
      bidi_it->separator_limit = -1;
      bidi_it->new_paragraph = 0;
793 794

      /* The following loop is run more than once only if NO_DEFAULT_P
795
	 is non-zero, and only if we are iterating on a buffer.  */
796 797
      do {
	bytepos = pstartbyte;
798 799 800 801
	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);
802 803
	type = bidi_get_type (ch, NEUTRAL_DIR);

804
	for (pos += nchars, bytepos += ch_len;
805 806 807 808 809 810 811 812 813 814
	     /* 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))
	  {
815 816 817
	    if (!string_p
		&& type == NEUTRAL_B
		&& bidi_at_paragraph_end (pos, bytepos) >= -1)
818
	      break;
819
	    if (pos >= end)
820 821
	      {
		/* Pretend there's a paragraph separator at end of
822
		   buffer/string.  */
823 824 825
		type = NEUTRAL_B;
		break;
	      }
826
	    /* Fetch next character and advance to get past it.  */
827
	    ch = bidi_fetch_char (bytepos, pos, &disp_pos, &bidi_it->string,
828
				  bidi_it->frame_window_p, &ch_len, &nchars);
829 830
	    pos += nchars;
	    bytepos += ch_len;
831 832 833 834 835
	  }
	if (type == STRONG_R || type == STRONG_AL) /* P3 */
	  bidi_it->paragraph_dir = R2L;
	else if (type == STRONG_L)
	  bidi_it->paragraph_dir = L2R;
836 837
	if (!string_p
	    && no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR)
838 839 840 841 842 843 844 845 846 847 848 849
	  {
	    /* 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)
		  {
850 851 852
		    /* FXIME: What if p is covered by a display
		       string?  See also a FIXME inside
		       bidi_find_paragraph_start.  */
853 854 855 856 857 858 859
		    p--;
		    pbyte = CHAR_TO_BYTE (p);
		    prevpbyte = bidi_find_paragraph_start (p, pbyte);
		  }
		pstartbyte = prevpbyte;
	      }
	  }
860 861
      } while (!string_p
	       && no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR);
Eli Zaretskii's avatar
Eli Zaretskii committed
862
    }
863 864
  else
    abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
865

866 867
  /* Contrary to UAX#9 clause P3, we only default the paragraph
     direction to L2R if we have no previous usable paragraph
868
     direction.  This is allowed by the HL1 clause.  */
869
  if (bidi_it->paragraph_dir != L2R && bidi_it->paragraph_dir != R2L)
870
    bidi_it->paragraph_dir = L2R; /* P3 and HL1 ``higher-level protocols'' */
871
  if (bidi_it->paragraph_dir == R2L)
872
    bidi_it->level_stack[0].level = 1;
873
  else
874
    bidi_it->level_stack[0].level = 0;
875 876

  bidi_line_init (bidi_it);
Eli Zaretskii's avatar
Eli Zaretskii committed
877 878
}

879 880
/* Do whatever UAX#9 clause X8 says should be done at paragraph's
   end.  */
881
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
882 883 884 885 886 887 888 889
bidi_set_paragraph_end (struct bidi_it *bidi_it)
{
  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;
}

890
/* Initialize the bidi iterator from buffer/string position CHARPOS.  */
Eli Zaretskii's avatar
Eli Zaretskii committed
891
void
892 893
bidi_init_it (EMACS_INT charpos, EMACS_INT bytepos, int frame_window_p,
	      struct bidi_it *bidi_it)
Eli Zaretskii's avatar
Eli Zaretskii committed
894 895 896
{
  if (! bidi_initialized)
    bidi_initialize ();
897 898 899 900
  if (charpos >= 0)
    bidi_it->charpos = charpos;
  if (bytepos >= 0)
    bidi_it->bytepos = bytepos;
901
  bidi_it->frame_window_p = frame_window_p;
902
  bidi_it->nchars = -1;	/* to be computed in bidi_resolve_explicit_1 */
903 904 905
  bidi_it->first_elt = 1;
  bidi_set_paragraph_end (bidi_it);
  bidi_it->new_paragraph = 1;
906
  bidi_it->separator_limit = -1;
Eli Zaretskii's avatar
Eli Zaretskii committed
907
  bidi_it->type = NEUTRAL_B;
908 909
  bidi_it->type_after_w1 = NEUTRAL_B;
  bidi_it->orig_type = NEUTRAL_B;
Eli Zaretskii's avatar
Eli Zaretskii committed
910
  bidi_it->prev_was_pdf = 0;
911 912
  bidi_it->prev.type = bidi_it->prev.type_after_w1 =
    bidi_it->prev.orig_type = UNKNOWN_BT;
913 914
  bidi_it->last_strong.type = bidi_it->last_strong.type_after_w1 =
    bidi_it->last_strong.orig_type = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
915 916
  bidi_it->next_for_neutral.charpos = -1;
  bidi_it->next_for_neutral.type =
917 918
    bidi_it->next_for_neutral.type_after_w1 =
    bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
919 920
  bidi_it->prev_for_neutral.charpos = -1;
  bidi_it->prev_for_neutral.type =
921 922
    bidi_it->prev_for_neutral.type_after_w1 =
    bidi_it->prev_for_neutral.orig_type = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
923
  bidi_it->sor = L2R;	 /* FIXME: should it be user-selectable? */
924
  bidi_it->disp_pos = -1;	/* invalid/unknown */
925 926 927 928
  /* 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
929 930 931 932
}

/* Push the current embedding level and override status; reset the
   current level to LEVEL and the current override status to OVERRIDE.  */
933
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
934 935 936 937 938 939 940 941 942 943 944 945
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.  */
946
static INLINE int
Eli Zaretskii's avatar
Eli Zaretskii committed
947 948 949 950 951 952 953 954 955
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.  */
956
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
957 958 959 960 961 962
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;
963
  bidi_check_type (bidi_it->type);
964 965
  saved_info->type_after_w1 = bidi_it->type_after_w1;
  bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
966
  saved_info->orig_type = bidi_it->orig_type;
967
  bidi_check_type (bidi_it->orig_type);
Eli Zaretskii's avatar
Eli Zaretskii committed
968 969 970 971
}

/* Resolve the type of a neutral character according to the type of
   surrounding strong text and the current embedding level.  */
972
static INLINE bidi_type_t
Eli Zaretskii's avatar
Eli Zaretskii committed
973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988
bidi_resolve_neutral_1 (bidi_type_t prev_type, bidi_type_t next_type, int lev)
{
  /* N1: European and Arabic numbers are treated as though they were R.  */
  if (next_type == WEAK_EN || next_type == WEAK_AN)
    next_type = STRONG_R;
  if (prev_type == WEAK_EN || prev_type == WEAK_AN)
    prev_type = STRONG_R;

  if (next_type == prev_type)	/* N1 */
    return next_type;
  else if ((lev & 1) == 0)	/* N2 */
    return STRONG_L;
  else
    return STRONG_R;
}

989
static INLINE int
990
bidi_explicit_dir_char (int ch)
Eli Zaretskii's avatar
Eli Zaretskii committed
991
{
992 993 994 995 996 997 998 999
  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
1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013
}

/* 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;
1014
  int string_p = bidi_it->string.s != NULL;
Eli Zaretskii's avatar
Eli Zaretskii committed
1015

1016 1017 1018
  /* 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.  */
1019
  if (bidi_it->bytepos < (string_p ? 0 : BEGV_BYTE)
1020
      || bidi_it->first_elt)
1021
    {
1022
      bidi_it->first_elt = 0;
1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
      if (string_p)
	{
	  if (bidi_it->charpos < 0)
	    bidi_it->charpos = 0;
	  bidi_it->bytepos = bidi_count_bytes (bidi_it->string.s, 0, 0,
					       bidi_it->charpos);
	}
      else
	{
	  if (bidi_it->charpos < BEGV)
	    bidi_it->charpos = BEGV;
	  bidi_it->bytepos = CHAR_TO_BYTE (bidi_it->charpos);
	}
1036
    }
1037 1038
  /* 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
1039
    {
1040 1041
      /* Advance to the next character, skipping characters covered by
	 display strings (nchars > 1).  */
1042 1043
      if (bidi_it->nchars <= 0)
	abort ();
1044
      bidi_it->charpos += bidi_it->nchars;
1045 1046
      if (bidi_it->ch_len == 0)
	abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
1047 1048 1049 1050 1051 1052 1053
      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;

1054
  if (bidi_it->charpos >= (string_p ? bidi_it->string.schars : ZV))
1055 1056 1057
    {
      curchar = BIDI_EOB;
      bidi_it->ch_len = 1;
1058
      bidi_it->nchars = 1;
1059
      bidi_it->disp_pos = (string_p ? bidi_it->string.schars : ZV);
1060 1061 1062
    }
  else
    {
1063 1064 1065
      /* 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.  */
1066
      curchar = bidi_fetch_char (bidi_it->bytepos, bidi_it->charpos,
1067 1068
				 &bidi_it->disp_pos, &bidi_it->string,
				 bidi_it->frame_window_p,
1069
				 &bidi_it->ch_len, &bidi_it->nchars);
1070
    }
Eli Zaretskii's avatar
Eli Zaretskii committed
1071 1072
  bidi_it->ch = curchar;

1073 1074 1075 1076 1077
  /* 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);
1078 1079
  bidi_it->orig_type = type;
  bidi_check_type (bidi_it->orig_type);
Eli Zaretskii's avatar
Eli Zaretskii committed
1080 1081 1082 1083

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

1084
  bidi_it->type_after_w1 = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
1085 1086 1087 1088 1089

  switch (type)
    {
      case RLE:	/* X2 */
      case RLO:	/* X4 */
1090 1091
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1092
	type = WEAK_BN; /* X9/Retaining */
1093
	if (bidi_it->ignore_bn_limit <= -1)
Eli Zaretskii's avatar
Eli Zaretskii committed
1094 1095 1096 1097 1098 1099
	  {
	    if (current_level <= BIDI_MAXLEVEL - 4)
	      {
		/* Compute the least odd embedding level greater than
		   the current level.  */
		new_level = ((current_level + 1) & ~1) + 1;
1100
		if (bidi_it->type_after_w1 == RLE)
Eli Zaretskii's avatar
Eli Zaretskii committed
1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116
		  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++;
	      }
	  }
1117
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1118 1119 1120 1121 1122
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      case LRE:	/* X3 */
      case LRO:	/* X5 */
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 - 5)
	      {
		/* Compute the least even embedding level greater than
		   the current level.  */
		new_level = ((current_level + 2) & ~1);
1133
		if (bidi_it->type_after_w1 == LRE)
Eli Zaretskii's avatar
Eli Zaretskii committed
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
		  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++;
	      }
	  }
1153
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1154 1155 1156 1157
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      case PDF:	/* X7 */
1158 1159
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1160
	type = WEAK_BN; /* X9/Retaining */
1161
	if (bidi_it->ignore_bn_limit <= -1)
Eli Zaretskii's avatar
Eli Zaretskii committed
1162 1163 1164 1165 1166 1167 1168 1169