bidi.c 61.6 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

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

/* Local data structures.  (Look in dispextern.h for the rest.)  */

/* What we need to know about the current paragraph.  */
struct bidi_paragraph_info {
73 74 75 76
  EMACS_INT start_bytepos;	/* byte position where it begins */
  EMACS_INT end_bytepos;	/* byte position where it ends */
  int	    embedding_level;	/* its basic embedding level */
  bidi_dir_t base_dir;		/* its base direction */
Eli Zaretskii's avatar
Eli Zaretskii committed
77 78 79 80 81 82 83 84 85 86
};

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

87
extern int bidi_ignore_explicit_marks_for_paragraph_level EXTERNALLY_VISIBLE;
Eli Zaretskii's avatar
Eli Zaretskii committed
88 89
int bidi_ignore_explicit_marks_for_paragraph_level = 1;

Eli Zaretskii's avatar
Eli Zaretskii committed
90
static Lisp_Object paragraph_start_re, paragraph_separate_re;
91
static Lisp_Object Qparagraph_start, Qparagraph_separate;
Eli Zaretskii's avatar
Eli Zaretskii committed
92 93

static void
94
bidi_initialize (void)
Eli Zaretskii's avatar
Eli Zaretskii committed
95
{
96 97

#include "biditype.h"
98
#include "bidimirror.h"
99

Eli Zaretskii's avatar
Eli Zaretskii committed
100 101 102
  int i;

  bidi_type_table = Fmake_char_table (Qnil, make_number (STRONG_L));
103
  staticpro (&bidi_type_table);
Eli Zaretskii's avatar
Eli Zaretskii committed
104 105

  for (i = 0; i < sizeof bidi_type / sizeof bidi_type[0]; i++)
106
    char_table_set_range (bidi_type_table, bidi_type[i].from, bidi_type[i].to,
Eli Zaretskii's avatar
Eli Zaretskii committed
107
			  make_number (bidi_type[i].type));
108

109 110 111 112 113 114 115
  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));

116 117
  Qparagraph_start = intern ("paragraph-start");
  staticpro (&Qparagraph_start);
Eli Zaretskii's avatar
Eli Zaretskii committed
118 119 120 121
  paragraph_start_re = Fsymbol_value (Qparagraph_start);
  if (!STRINGP (paragraph_start_re))
    paragraph_start_re = build_string ("\f\\|[ \t]*$");
  staticpro (&paragraph_start_re);
122 123
  Qparagraph_separate = intern ("paragraph-separate");
  staticpro (&Qparagraph_separate);
Eli Zaretskii's avatar
Eli Zaretskii committed
124 125 126 127
  paragraph_separate_re = Fsymbol_value (Qparagraph_separate);
  if (!STRINGP (paragraph_separate_re))
    paragraph_separate_re = build_string ("[ \t\f]*$");
  staticpro (&paragraph_separate_re);
Eli Zaretskii's avatar
Eli Zaretskii committed
128 129 130
  bidi_initialized = 1;
}

131 132
/* Return the bidi type of a character CH, subject to the current
   directional OVERRIDE.  */
133
static INLINE bidi_type_t
134
bidi_get_type (int ch, bidi_dir_t override)
Eli Zaretskii's avatar
Eli Zaretskii committed
135
{
136 137
  bidi_type_t default_type;

138 139
  if (ch == BIDI_EOB)
    return NEUTRAL_B;
140 141
  if (ch < 0 || ch > MAX_CHAR)
    abort ();
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

  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
174 175
}

176
static void
177 178 179 180 181 182
bidi_check_type (bidi_type_t type)
{
  if (type < UNKNOWN_BT || type > NEUTRAL_ON)
    abort ();
}

Eli Zaretskii's avatar
Eli Zaretskii committed
183
/* Given a bidi TYPE of a character, return its category.  */
184
static INLINE bidi_category_t
Eli Zaretskii's avatar
Eli Zaretskii committed
185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217
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 ();
    }
}

218 219 220 221
/* 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
222 223 224
int
bidi_mirror_char (int c)
{
225 226 227 228 229 230
  Lisp_Object val;

  if (c == BIDI_EOB)
    return c;
  if (c < 0 || c > MAX_CHAR)
    abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
231

232 233
  val = CHAR_TABLE_REF (bidi_mirror_table, c);
  if (INTEGERP (val))
Eli Zaretskii's avatar
Eli Zaretskii committed
234
    {
235
      int v = XINT (val);
Eli Zaretskii's avatar
Eli Zaretskii committed
236

237 238 239 240
      if (v < 0 || v > MAX_CHAR)
	abort ();

      return v;
Eli Zaretskii's avatar
Eli Zaretskii committed
241
    }
242

Eli Zaretskii's avatar
Eli Zaretskii committed
243 244 245 246 247
  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.  */
248
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
249 250 251 252
bidi_copy_it (struct bidi_it *to, struct bidi_it *from)
{
  int i;

253
  /* Copy everything except the level stack and beyond.  */
254
  memcpy (to, from, offsetof (struct bidi_it, level_stack[0]));
Eli Zaretskii's avatar
Eli Zaretskii committed
255 256 257 258 259 260 261 262 263

  /* 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.  */

264 265 266
#define BIDI_CACHE_CHUNK 200
static struct bidi_it *bidi_cache;
static size_t bidi_cache_size = 0;
267
static size_t elsz = sizeof (struct bidi_it);
268 269
static int bidi_cache_idx;	/* next unused cache slot */
static int bidi_cache_last_idx;	/* slot of last cache hit */
Eli Zaretskii's avatar
Eli Zaretskii committed
270

271
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
272 273 274 275 276 277
bidi_cache_reset (void)
{
  bidi_cache_idx = 0;
  bidi_cache_last_idx = -1;
}

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

290
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
bidi_cache_fetch_state (int idx, struct bidi_it *bidi_it)
{
  int current_scan_dir = bidi_it->scan_dir;

  if (idx < 0 || idx >= bidi_cache_idx)
    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.  */
307
static INLINE int
308
bidi_cache_search (EMACS_INT charpos, int level, int dir)
Eli Zaretskii's avatar
Eli Zaretskii committed
309 310 311 312 313 314
{
  int i, i_start;

  if (bidi_cache_idx)
    {
      if (charpos < bidi_cache[bidi_cache_last_idx].charpos)
315 316 317 318 319 320 321 322 323 324 325
	{
	  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
326 327 328 329 330 331 332 333 334 335 336
	i_start = bidi_cache_last_idx;
      else
	{
	  dir = -1;
	  i_start = bidi_cache_idx - 1;
	}

      if (dir < 0)
	{
	  /* Linear search for now; FIXME!  */
	  for (i = i_start; i >= 0; i--)
337 338
	    if (bidi_cache[i].charpos <= charpos
		&& charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
Eli Zaretskii's avatar
Eli Zaretskii committed
339 340 341 342 343 344
		&& (level == -1 || bidi_cache[i].resolved_level <= level))
	      return i;
	}
      else
	{
	  for (i = i_start; i < bidi_cache_idx; i++)
345 346
	    if (bidi_cache[i].charpos <= charpos
		&& charpos < bidi_cache[i].charpos + bidi_cache[i].nchars
Eli Zaretskii's avatar
Eli Zaretskii committed
347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
		&& (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.
   BEFORE, if non-zero, means return the index of the slot that is
   ``before'' the level change in the search direction.  That is,
   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)
	{
	  while (i >= incr)
	    {
	      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;
}

407
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
408 409 410 411 412 413 414 415 416 417 418 419
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;
420 421 422
      /* Enlarge the cache as needed.  */
      if (idx >= bidi_cache_size)
	{
423
	  bidi_cache_size += BIDI_CACHE_CHUNK;
424
	  bidi_cache =
425
	    (struct bidi_it *) xrealloc (bidi_cache, bidi_cache_size * elsz);
426
	}
427 428 429 430
      /* 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.  */
      if (idx > 0 &&
431 432
	  (bidi_it->charpos > (bidi_cache[idx - 1].charpos
			       + bidi_cache[idx - 1].nchars)
433 434 435 436 437
	   || bidi_it->charpos < bidi_cache[0].charpos))
	{
	  bidi_cache_reset ();
	  idx = 0;
	}
438 439
      if (bidi_it->nchars <= 0)
	abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
440 441 442 443 444 445 446 447 448
      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;
449
      bidi_check_type (bidi_it->type);
450 451
      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
452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467
      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;
}

468
static INLINE bidi_type_t
469
bidi_cache_find (EMACS_INT charpos, int level, struct bidi_it *bidi_it)
Eli Zaretskii's avatar
Eli Zaretskii committed
470 471 472 473 474 475 476
{
  int i = bidi_cache_search (charpos, level, bidi_it->scan_dir);

  if (i >= 0)
    {
      bidi_dir_t current_scan_dir = bidi_it->scan_dir;

477
      bidi_copy_it (bidi_it, &bidi_cache[i]);
Eli Zaretskii's avatar
Eli Zaretskii committed
478 479 480 481 482 483 484 485 486 487
      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;
}

488
static INLINE int
Eli Zaretskii's avatar
Eli Zaretskii committed
489 490 491 492 493 494 495
bidi_peek_at_next_level (struct bidi_it *bidi_it)
{
  if (bidi_cache_idx == 0 || bidi_cache_last_idx == -1)
    abort ();
  return bidi_cache[bidi_cache_last_idx + bidi_it->scan_dir].resolved_level;
}

496 497 498 499 500
/* 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.  */
501
static EMACS_INT
502
bidi_at_paragraph_end (EMACS_INT charpos, EMACS_INT bytepos)
Eli Zaretskii's avatar
Eli Zaretskii committed
503
{
Eli Zaretskii's avatar
Eli Zaretskii committed
504 505
  Lisp_Object sep_re;
  Lisp_Object start_re;
506 507
  EMACS_INT val;

Eli Zaretskii's avatar
Eli Zaretskii committed
508 509
  sep_re = paragraph_separate_re;
  start_re = paragraph_start_re;
510 511 512 513 514 515 516 517 518

  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
519

520
  return val;
Eli Zaretskii's avatar
Eli Zaretskii committed
521 522 523 524 525 526
}

/* 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.  */
527
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
528 529 530 531 532 533 534 535 536 537 538
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.  */
539
  if (!bidi_it->prev_was_pdf || level_before < level_after)
Eli Zaretskii's avatar
Eli Zaretskii committed
540 541 542 543 544 545
    /* 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;
546 547
  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
548 549 550
  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;
551 552
  bidi_it->next_for_neutral.type = bidi_it->next_for_neutral.type_after_w1 =
    bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
553 554 555
  bidi_it->ignore_bn_limit = 0; /* meaning it's unknown */
}

556
/* Perform initializations for reordering a new line of bidi text.  */
557
static void
558 559 560 561 562 563 564 565 566
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;
567 568
  bidi_set_sor_type (bidi_it,
		     bidi_it->paragraph_dir == R2L ? 1 : 0,
569 570 571 572 573
		     bidi_it->level_stack[0].level); /* X10 */

  bidi_cache_reset ();
}

574 575 576 577 578 579 580 581
/* 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
   string.  */
582
static INLINE int
583
bidi_fetch_char (EMACS_INT bytepos, EMACS_INT charpos, EMACS_INT *disp_pos,
584 585 586 587 588 589 590
		 EMACS_INT *ch_len, EMACS_INT *nchars)
{
  int ch;

  /* FIXME: Support strings in addition to buffers.  */
  /* If we got past the last known position of display string, compute
     the position of the next one.  That position could be at BYTEPOS.  */
591 592 593 594
  if (charpos < ZV && charpos > *disp_pos)
    *disp_pos = compute_display_string_pos (charpos);

  /* Fetch the character at BYTEPOS.  */
595 596 597 598 599
  if (bytepos >= ZV_BYTE)
    {
      ch = BIDI_EOB;
      *ch_len = 1;
      *nchars = 1;
600
      *disp_pos = ZV;
601
    }
602
  else if (charpos >= *disp_pos)
603
    {
604 605 606 607 608 609 610 611 612 613 614 615 616
      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
	 the entire run of characters covered by the display
	 string.  */
      ch = 0xFFFC;
      disp_end_pos = compute_display_string_end (*disp_pos);
      *nchars = disp_end_pos - *disp_pos;
      *ch_len = CHAR_TO_BYTE (disp_end_pos) - bytepos;
617 618 619 620 621
    }
  else
    {
      ch = FETCH_MULTIBYTE_CHAR (bytepos);
      *nchars = 1;
622
      *ch_len = CHAR_BYTES (ch);
623 624 625 626
    }

  /* If we just entered a run of characters covered by a display
     string, compute the position of the next display string.  */
627 628
  if (charpos + *nchars <= ZV && charpos + *nchars > *disp_pos)
    *disp_pos = compute_display_string_pos (charpos + *nchars);
629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657

  return ch;
}

/* Looks like we won't need this one.  */
#if 0
/* Fetch character at CHARPOS/BYTEPOS.  Return the character, and
   advance CHARPOS and BYTEPOS to the next character in logical
   order.  */
static INLINE int
bidi_fetch_char_advance (EMACS_INT *charpos, EMACS_INT *bytepos)
{
  int ch;

  /* FIXME: Support strings in addition to buffers.  */
  FETCH_CHAR_ADVANCE_NO_CHECK (ch, charpos, bytepos);

#if 0
  if (...)
    {
      /* FIXME: Support characters covered by display strings.  */
      ch = 0xFFFC;
    }
#endif

  return ch;
}
#endif

658 659 660
/* 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
661
bidi_find_paragraph_start (EMACS_INT pos, EMACS_INT pos_byte)
662
{
Eli Zaretskii's avatar
Eli Zaretskii committed
663
  Lisp_Object re = paragraph_start_re;
664 665 666 667 668
  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)
    {
669 670 671 672
      /* 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?  */
673 674
      pos = find_next_newline_no_quit (pos - 1, -1);
      pos_byte = CHAR_TO_BYTE (pos);
675
    }
676
  return pos_byte;
677 678
}

679
/* Determine the base direction, a.k.a. base embedding level, of the
680 681
   paragraph we are about to iterate through.  If DIR is either L2R or
   R2L, just use that.  Otherwise, determine the paragraph direction
682 683
   from the first strong directional character of the paragraph.

684
   NO_DEFAULT_P non-zero means don't default to L2R if the paragraph
685 686 687 688 689 690 691 692 693
   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
694
void
695
bidi_paragraph_init (bidi_dir_t dir, struct bidi_it *bidi_it, int no_default_p)
Eli Zaretskii's avatar
Eli Zaretskii committed
696
{
697
  EMACS_INT bytepos = bidi_it->bytepos;
698
  EMACS_INT pstartbyte;
699

700 701 702
  /* Special case for an empty buffer. */
  if (bytepos == BEGV_BYTE && bytepos == ZV_BYTE)
    dir = L2R;
703
  /* We should never be called at EOB or before BEGV.  */
704
  else if (bytepos >= ZV_BYTE || bytepos < BEGV_BYTE)
705 706
    abort ();

707 708 709 710 711 712 713 714 715 716
  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
717 718
  else if (dir == NEUTRAL_DIR)	/* P2 */
    {
719 720 721
      int ch;
      EMACS_INT ch_len, nchars;
      EMACS_INT pos, disp_pos = -1;
722
      bidi_type_t type;
723

724 725 726
      if (!bidi_initialized)
	bidi_initialize ();

727 728
      /* If we are inside a paragraph separator, we are just waiting
	 for the separator to be exhausted; use the previous paragraph
729 730 731 732
	 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)
733 734
	return;

735
      /* If we are on a newline, get past it to where the next
736 737 738
	 paragraph might start.  But don't do that at BEGV since then
	 we are potentially in a new paragraph that doesn't yet
	 exist.  */
739
      pos = bidi_it->charpos;
740
      if (bytepos > BEGV_BYTE && FETCH_CHAR (bytepos) == '\n')
741
	{
742
	  bytepos++;
743
	  pos++;
744
	}
745 746 747

      /* We are either at the beginning of a paragraph or in the
	 middle of it.  Find where this paragraph starts.  */
748
      pstartbyte = bidi_find_paragraph_start (pos, bytepos);
749 750
      bidi_it->separator_limit = -1;
      bidi_it->new_paragraph = 0;
751 752 753 754 755 756

      /* The following loop is run more than once only if NO_DEFAULT_P
	 is non-zero.  */
      do {
	bytepos = pstartbyte;
	pos = BYTE_TO_CHAR (bytepos);
757
	ch = bidi_fetch_char (bytepos, pos, &disp_pos, &ch_len, &nchars);
758 759
	type = bidi_get_type (ch, NEUTRAL_DIR);

760
	for (pos += nchars, bytepos += ch_len;
761 762 763 764 765 766 767 768 769 770 771
	     /* 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))
	  {
	    if (type == NEUTRAL_B && bidi_at_paragraph_end (pos, bytepos) >= -1)
772
	      break;
773 774 775 776 777 778 779
	    if (bytepos >= ZV_BYTE)
	      {
		/* Pretend there's a paragraph separator at end of
		   buffer.  */
		type = NEUTRAL_B;
		break;
	      }
780 781
	    /* Fetch next character and advance to get past it.  */
	    ch = bidi_fetch_char (bytepos, pos, &disp_pos, &ch_len, &nchars);
782 783
	    pos += nchars;
	    bytepos += ch_len;
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801
	  }
	if (type == STRONG_R || type == STRONG_AL) /* P3 */
	  bidi_it->paragraph_dir = R2L;
	else if (type == STRONG_L)
	  bidi_it->paragraph_dir = L2R;
	if (no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR)
	  {
	    /* 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)
		  {
802 803 804
		    /* FXIME: What if p is covered by a display
		       string?  See also a FIXME inside
		       bidi_find_paragraph_start.  */
805 806 807 808 809 810 811 812
		    p--;
		    pbyte = CHAR_TO_BYTE (p);
		    prevpbyte = bidi_find_paragraph_start (p, pbyte);
		  }
		pstartbyte = prevpbyte;
	      }
	  }
      } while (no_default_p && bidi_it->paragraph_dir == NEUTRAL_DIR);
Eli Zaretskii's avatar
Eli Zaretskii committed
813
    }
814 815
  else
    abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
816

817 818
  /* Contrary to UAX#9 clause P3, we only default the paragraph
     direction to L2R if we have no previous usable paragraph
819
     direction.  This is allowed by the HL1 clause.  */
820
  if (bidi_it->paragraph_dir != L2R && bidi_it->paragraph_dir != R2L)
821
    bidi_it->paragraph_dir = L2R; /* P3 and HL1 ``higher-level protocols'' */
822
  if (bidi_it->paragraph_dir == R2L)
823
    bidi_it->level_stack[0].level = 1;
824
  else
825
    bidi_it->level_stack[0].level = 0;
826 827

  bidi_line_init (bidi_it);
Eli Zaretskii's avatar
Eli Zaretskii committed
828 829
}

830 831
/* Do whatever UAX#9 clause X8 says should be done at paragraph's
   end.  */
832
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
833 834 835 836 837 838 839 840
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;
}

841
/* Initialize the bidi iterator from buffer/string position CHARPOS.  */
Eli Zaretskii's avatar
Eli Zaretskii committed
842
void
843
bidi_init_it (EMACS_INT charpos, EMACS_INT bytepos, struct bidi_it *bidi_it)
Eli Zaretskii's avatar
Eli Zaretskii committed
844 845 846
{
  if (! bidi_initialized)
    bidi_initialize ();
847 848
  bidi_it->charpos = charpos;
  bidi_it->bytepos = bytepos;
849
  bidi_it->nchars = -1;	/* to be computed in bidi_resolve_explicit_1 */
850 851 852
  bidi_it->first_elt = 1;
  bidi_set_paragraph_end (bidi_it);
  bidi_it->new_paragraph = 1;
853
  bidi_it->separator_limit = -1;
Eli Zaretskii's avatar
Eli Zaretskii committed
854
  bidi_it->type = NEUTRAL_B;
855 856
  bidi_it->type_after_w1 = NEUTRAL_B;
  bidi_it->orig_type = NEUTRAL_B;
Eli Zaretskii's avatar
Eli Zaretskii committed
857
  bidi_it->prev_was_pdf = 0;
858 859
  bidi_it->prev.type = bidi_it->prev.type_after_w1 =
    bidi_it->prev.orig_type = UNKNOWN_BT;
860 861
  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
862 863
  bidi_it->next_for_neutral.charpos = -1;
  bidi_it->next_for_neutral.type =
864 865
    bidi_it->next_for_neutral.type_after_w1 =
    bidi_it->next_for_neutral.orig_type = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
866 867
  bidi_it->prev_for_neutral.charpos = -1;
  bidi_it->prev_for_neutral.type =
868 869
    bidi_it->prev_for_neutral.type_after_w1 =
    bidi_it->prev_for_neutral.orig_type = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
870
  bidi_it->sor = L2R;	 /* FIXME: should it be user-selectable? */
871
  bidi_it->disp_pos = -1;	/* invalid/unknown */
872
  bidi_cache_shrink ();
Eli Zaretskii's avatar
Eli Zaretskii committed
873 874 875 876
}

/* Push the current embedding level and override status; reset the
   current level to LEVEL and the current override status to OVERRIDE.  */
877
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
878 879 880 881 882 883 884 885 886 887 888 889
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.  */
890
static INLINE int
Eli Zaretskii's avatar
Eli Zaretskii committed
891 892 893 894 895 896 897 898 899
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.  */
900
static INLINE void
Eli Zaretskii's avatar
Eli Zaretskii committed
901 902 903 904 905 906
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;
907
  bidi_check_type (bidi_it->type);
908 909
  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
910
  saved_info->orig_type = bidi_it->orig_type;
911
  bidi_check_type (bidi_it->orig_type);
Eli Zaretskii's avatar
Eli Zaretskii committed
912 913 914 915
}

/* Resolve the type of a neutral character according to the type of
   surrounding strong text and the current embedding level.  */
916
static INLINE bidi_type_t
Eli Zaretskii's avatar
Eli Zaretskii committed
917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932
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;
}

933
static INLINE int
934
bidi_explicit_dir_char (int ch)
Eli Zaretskii's avatar
Eli Zaretskii committed
935
{
936 937 938 939 940 941 942 943
  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
944 945 946 947 948 949 950 951 952 953 954 955 956 957 958
}

/* 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;

959 960 961 962
  /* 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.  */
  if (bidi_it->bytepos < BEGV_BYTE
963
      || bidi_it->first_elt)
964
    {
965 966 967 968
      bidi_it->first_elt = 0;
      if (bidi_it->charpos < BEGV)
	bidi_it->charpos = BEGV;
      bidi_it->bytepos = CHAR_TO_BYTE (bidi_it->charpos);
969
    }
970
  else if (bidi_it->bytepos < ZV_BYTE)	/* don't move at ZV */
Eli Zaretskii's avatar
Eli Zaretskii committed
971
    {
972 973
      /* Advance to the next character, skipping characters covered by
	 display strings (nchars > 1).  */
974 975
      if (bidi_it->nchars <= 0)
	abort ();
976
      bidi_it->charpos += bidi_it->nchars;
977 978
      if (bidi_it->ch_len == 0)
	abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
979 980 981 982 983 984 985
      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;

986 987 988 989
  if (bidi_it->bytepos >= ZV_BYTE)
    {
      curchar = BIDI_EOB;
      bidi_it->ch_len = 1;
990
      bidi_it->nchars = 1;
991
      bidi_it->disp_pos = ZV;
992 993 994
    }
  else
    {
995 996 997
      /* 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.  */
998 999 1000
      curchar = bidi_fetch_char (bidi_it->bytepos, bidi_it->charpos,
				 &bidi_it->disp_pos,
				 &bidi_it->ch_len, &bidi_it->nchars);
1001
    }
Eli Zaretskii's avatar
Eli Zaretskii committed
1002 1003
  bidi_it->ch = curchar;

1004 1005 1006 1007 1008
  /* 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);
1009 1010
  bidi_it->orig_type = type;
  bidi_check_type (bidi_it->orig_type);
Eli Zaretskii's avatar
Eli Zaretskii committed
1011 1012 1013 1014

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

1015
  bidi_it->type_after_w1 = UNKNOWN_BT;
Eli Zaretskii's avatar
Eli Zaretskii committed
1016 1017 1018 1019 1020

  switch (type)
    {
      case RLE:	/* X2 */
      case RLO:	/* X4 */
1021 1022
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1023 1024 1025 1026 1027 1028 1029 1030
	type = WEAK_BN; /* X9/Retaining */
	if (bidi_it->ignore_bn_limit <= 0)
	  {
	    if (current_level <= BIDI_MAXLEVEL - 4)
	      {
		/* Compute the least odd embedding level greater than
		   the current level.  */
		new_level = ((current_level + 1) & ~1) + 1;
1031
		if (bidi_it->type_after_w1 == RLE)
Eli Zaretskii's avatar
Eli Zaretskii committed
1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047
		  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++;
	      }
	  }
1048
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1049 1050 1051 1052 1053
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      case LRE:	/* X3 */
      case LRO:	/* X5 */
1054 1055
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1056 1057 1058 1059 1060 1061 1062 1063
	type = WEAK_BN; /* X9/Retaining */
	if (bidi_it->ignore_bn_limit <= 0)
	  {
	    if (current_level <= BIDI_MAXLEVEL - 5)
	      {
		/* Compute the least even embedding level greater than
		   the current level.  */
		new_level = ((current_level + 2) & ~1);
1064
		if (bidi_it->type_after_w1 == LRE)
Eli Zaretskii's avatar
Eli Zaretskii committed
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083
		  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++;
	      }
	  }
1084
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1085 1086 1087 1088
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      case PDF:	/* X7 */
1089 1090
	bidi_it->type_after_w1 = type;
	bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109
	type = WEAK_BN; /* X9/Retaining */
	if (bidi_it->ignore_bn_limit <= 0)
	  {
	    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--;
	      }
	  }
1110
	else if (bidi_it->prev.type_after_w1 == WEAK_EN /* W5/Retaining */
Eli Zaretskii's avatar
Eli Zaretskii committed
1111 1112 1113 1114 1115 1116 1117 1118 1119
		 || bidi_it->next_en_pos > bidi_it->charpos)
	  type = WEAK_EN;
	break;
      default:
	/* Nothing.  */
	break;
    }

  bidi_it->type = type;
1120
  bidi_check_type (bidi_it->type);
Eli Zaretskii's avatar
Eli Zaretskii committed
1121 1122 1123 1124 1125

  return new_level;
}

/* Given an iterator state in BIDI_IT, advance one character position
1126 1127 1128 1129
   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
1130 1131 1132 1133 1134 1135 1136 1137 1138
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);

  if (prev_level < new_level
      && bidi_it->type == WEAK_BN
      && bidi_it->ignore_bn_limit == 0 /* only if not already known */
1139
      && bidi_it->bytepos < ZV_BYTE    /* not already at EOB */
1140 1141
      && bidi_explicit_dir_char (FETCH_MULTIBYTE_CHAR (bidi_it->bytepos
						       + bidi_it->ch_len)))
Eli Zaretskii's avatar
Eli Zaretskii committed
1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
    {
      /* Avoid pushing and popping embedding levels if the level run
	 is empty, as this breaks level runs where it shouldn't.
	 UAX#9 removes all the explicit embedding and override codes,
	 so empty embeddings disappear without a trace.  We need to
	 behave as if we did the same.  */
      struct bidi_it saved_it;
      int level = prev_level;

      bidi_copy_it (&saved_it, bidi_it);

1153 1154
      while (bidi_explicit_dir_char (FETCH_MULTIBYTE_CHAR (bidi_it->bytepos
							   + bidi_it->ch_len)))
Eli Zaretskii's avatar
Eli Zaretskii committed
1155
	{
1156 1157
	  /* This advances to the next character, skipping any
	     characters covered by display strings.  */
Eli Zaretskii's avatar
Eli Zaretskii committed
1158 1159 1160
	  level = bidi_resolve_explicit_1 (bidi_it);
	}

1161 1162
      if (bidi_it->nchars <= 0)
	abort ();
Eli Zaretskii's avatar
Eli Zaretskii committed
1163
      if (level == prev_level)	/* empty embedding */
1164
	saved_it.ignore_bn_limit = bidi_it->charpos + bidi_it->nchars;
Eli Zaretskii's avatar
Eli Zaretskii committed
1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190
      else			/* this embedding is non-empty */
	saved_it.ignore_bn_limit = -1;

      bidi_copy_it (bidi_it, &saved_it);
      if (bidi_it->ignore_bn_limit > 0)
	{
	  /* We pushed a level, but we shouldn't have.  Undo that. */
	  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--;
	    }
	  if (!bidi_it->invalid_levels)
	    new_level = bidi_pop_embedding_level (bidi_it);
	  else
	    {
	      bidi_it->invalid_levels--;
	      bidi_it->invalid_rl_levels--;
	    }
	}
    }

  if (bidi_it->type == NEUTRAL_B)	/* X8 */
    {
1191
      bidi_set_paragraph_end (bidi_it);
1192 1193
      /* This is needed by bidi_resolve_weak below, and in L1.  */
      bidi_it->type_after_w1 = bidi_it->type;
1194
      bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1195 1196 1197 1198 1199
    }

  return new_level;
}

1200 1201
/* Advance in the buffer/string, resolve weak types and return the
   type of the next character after weak type resolution.  */
1202
static bidi_type_t
Eli Zaretskii's avatar
Eli Zaretskii committed
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233
bidi_resolve_weak (struct bidi_it *bidi_it)
{
  bidi_type_t type;
  bidi_dir_t override;
  int prev_level = bidi_it->level_stack[bidi_it->stack_idx].level;
  int new_level  = bidi_resolve_explicit (bidi_it);
  int next_char;
  bidi_type_t type_of_next;
  struct bidi_it saved_it;

  type = bidi_it->type;
  override = bidi_it->level_stack[bidi_it->stack_idx].override;

  if (type == UNKNOWN_BT
      || type == LRE
      || type == LRO
      || type == RLE
      || type == RLO
      || type == PDF)
    abort ();

  if (new_level != prev_level
      || bidi_it->type == NEUTRAL_B)
    {
      /* We've got a new embedding level run, compute the directional
         type of sor and initialize per-run variables (UAX#9, clause
         X10).  */
      bidi_set_sor_type (bidi_it, prev_level, new_level);
    }
  else if (type == NEUTRAL_S || type == NEUTRAL_WS
	   || type == WEAK_BN || type == STRONG_AL)
1234 1235
    bidi_it->type_after_w1 = type;	/* needed in L1 */
  bidi_check_type (bidi_it->type_after_w1);
Eli Zaretskii's avatar
Eli Zaretskii committed
1236 1237 1238 1239 1240 1241 1242

  /* Level and directional override status are already recorded in
     bidi_it, and do not need any change; see X6.  */
  if (override == R2L)		/* X6 */
    type = STRONG_R;
  else if (override == L2R)
    type = STRONG_L;
1243
  else
Eli Zaretskii's avatar
Eli Zaretskii committed
1244
    {
1245
      if (type == WEAK_NSM)	/* W1 */
Eli Zaretskii's avatar
Eli Zaretskii committed
1246
	{
1247
	  /* Note that we don't need to consider the case where the
1248 1249 1250 1251 1252 1253
	     prev character has its type overridden by an RLO or LRO,
	     because then either the type of this NSM would have been
	     also overridden, or the previous character is outside the
	     current level run, and thus not relevant to this NSM.
	     This is why NSM gets the type_after_w1 of the previous
	     character.  */
1254 1255 1256
	  if (bidi_it->prev.type_after_w1 != UNKNOWN_BT
	      /* if type_after_w1 is NEUTRAL_B, this NSM is at sor */
	      && bidi_it->prev.type_after_w1 != NEUTRAL_B)
1257
	    type = bidi_it->prev.type_after_w1;