summaryrefslogtreecommitdiff
path: root/source/fitz/bidi.c
blob: 74548d4d34b0d3677220ca358909a2197888438a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
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
164
165
166
167
168
169
170
171
172
173
174
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
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
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
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
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
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
/*
 * Bidirectional text processing.
 *
 * Processes unicode text by arranging the characters into an order suitable
 * for display. E.g. Hebrew text will be arranged from right-to-left and
 * any English within the text will remain in the left-to-right order.
 * Characters such as parenthesis will be substituted for their mirrored
 * equivalents if they are part of text which must be reversed.
 *
 * This is an implementation of the unicode Bidirectional Algorithm which
 * can be found here: http://www.unicode.org/reports/tr9/ and is based
 * on the reference implementation of the algorithm found on that page.
 *
 * For a nice overview of how it works, read this...
 * http://www.w3.org/TR/REC-html40/struct/dirlang.html
 *
 * Extracted from the SmartOffice code, where it was modified by Ian
 * Beveridge.
 *
 * Copyright (C) Picsel, 2004. All Rights Reserved.
 */

/*
 * Original copyright notice from unicode reference implementation.
 * ----------------------------------------------------------------
 * Written by: Asmus Freytag
 *	C++ and Windows dependencies removed, and
 *	command line interface added by: Rick McGowan
 *
 *	Copyright (C) 1999, ASMUS, Inc. All Rights Reserved
 */

/*
 * Includes...
 */

#include "mupdf/fitz.h"
#include "bidi-impl.h" /* standard bidi code interface */

/*
 * Macros...
 */

#define ODD(x) ((x) & 1)

#define REPLACEABLE_TYPE(t) ( \
		((t)==BDI_ES) || ((t)==BDI_ET) || ((t)==BDI_CS) || \
		((t)==BDI_NSM) || ((t)==BDI_PDF) || ((t)==BDI_BN) || \
		((t)==BDI_S) || ((t)==BDI_WS) || ((t)==BDI_N) )

#ifdef DEBUG_BIDI_VERBOSE
#define DBUGVF(params) do { fz_warn params; } while (0)
#else
#define DBUGVF(params) do {} while (0)
#endif

#ifdef DEBUG_BIDI_OUTLINE
#define DBUGH(params) do { fz_warn params; } while (0)
#else
#define DBUGH(params) do {} while (0)
#endif

#define UNICODE_EOS					0
#define UNICODE_DIGIT_ZERO				0x0030
#define UNICODE_DIGIT_NINE				0x0039
#define UNICODE_SUPERSCRIPT_TWO				0x00B2
#define UNICODE_SUPERSCRIPT_THREE			0x00B3
#define UNICODE_SUPERSCRIPT_ONE				0x00B9
#define UNICODE_RTL_START				0x0590
#define UNICODE_RTL_END					0x07BF
#define UNICODE_ARABIC_INDIC_DIGIT_ZERO			0x0660
#define UNICODE_ARABIC_INDIC_DIGIT_NINE			0x0669
#define UNICODE_EXTENDED_ARABIC_INDIC_DIGIT_ZERO	0x06F0
#define UNICODE_EXTENDED_ARABIC_INDIC_DIGIT_NINE	0x06F9
#define UNICODE_ZERO_WIDTH_NON_JOINER			0x200C
#define UNICODE_SUPERSCRIPT_ZERO			0x2070
#define UNICODE_SUPERSCRIPT_FOUR			0x2074
#define UNICODE_SUPERSCRIPT_NINE			0x2079
#define UNICODE_SUBSCRIPT_ZERO				0x2080
#define UNICODE_SUBSCRIPT_NINE				0x2089
#define UNICODE_CIRCLED_DIGIT_ONE			0x2460
#define UNICODE_NUMBER_TWENTY_FULL_STOP			0x249B
#define UNICODE_CIRCLED_DIGIT_ZERO			0x24EA
#define UNICODE_FULLWIDTH_DIGIT_ZERO			0xFF10
#define UNICODE_FULLWIDTH_DIGIT_NINE			0xFF19

#ifndef TRUE
#define TRUE (1)
#endif
#ifndef FALSE
#define FALSE (0)
#endif

/*
 * Enumerations...
 */

#ifdef DEBUG_BIDI_VERBOSE
/* display support: */
static const char char_from_types[] =
{
	' ',	/* ON */
	'>',	/* L */
	'<',	/* R */
	'9',	/* AN */
	'1',	/* EN */
	'a',	/* AL */
	'@',	/* NSM */
	'.',	/* CS */
	',',	/* ES */
	'$',	/* ET */
	':',	/* BN */
	'X',	/* S */
	'_',	/* WS */
	'B',	/* B */
	'+',	/* RLO */
	'+',	/* RLE */
	'+',	/* LRO */
	'+',	/* LRE */
	'-',	/* PDF */
	'='	/* LS */
};
#endif

/*
 * Functions and static functions...
 */

/* UCDN uses a different ordering than Bidi does. We cannot
 * change to the UCDN ordering, as the bidi-std.c code relies
 * on the exact ordering (at least that N = ON = 0). We
 * therefore map between the two using this small table. It
 * also takes care of fudging LRI, RLI, FSI and PDI, that this
 * code does not currently support. */
static const uint8_t ucdn_to_bidi[] =
{
	BDI_L,		/* UCDN_BIDI_CLASS_L = 0 */
	BDI_LRE,	/* UCDN_BIDI_CLASS_LRE = 1 */
	BDI_LRO,	/* UCDN_BIDI_CLASS_LRO = 2 */
	BDI_R,		/* UCDN_BIDI_CLASS_R = 3 */
	BDI_AL,		/* UCDN_BIDI_CLASS_AL = 4 */
	BDI_RLE,	/* UCDN_BIDI_CLASS_RLE = 5 */
	BDI_RLO,	/* UCDN_BIDI_CLASS_RLO = 6 */
	BDI_PDF,	/* UCDN_BIDI_CLASS_PDF = 7 */
	BDI_EN,		/* UCDN_BIDI_CLASS_EN = 8 */
	BDI_ES,		/* UCDN_BIDI_CLASS_ES = 9 */
	BDI_ET,		/* UCDN_BIDI_CLASS_ET = 10 */
	BDI_AN,		/* UCDN_BIDI_CLASS_AN = 11 */
	BDI_CS,		/* UCDN_BIDI_CLASS_CS = 12 */
	BDI_NSM,	/* UCDN_BIDI_CLASS_NSM = 13 */
	BDI_BN,		/* UCDN_BIDI_CLASS_BN = 14 */
	BDI_B,		/* UCDN_BIDI_CLASS_B = 15 */
	BDI_S,		/* UCDN_BIDI_CLASS_S = 16 */
	BDI_WS,		/* UCDN_BIDI_CLASS_WS = 17 */
	BDI_ON,		/* UCDN_BIDI_CLASS_ON = 18 */
	BDI_LRE,	/* UCDN_BIDI_CLASS_LRI = 19 */
	BDI_RLE,	/* UCDN_BIDI_CLASS_RLI = 20 */
	BDI_N,		/* UCDN_BIDI_CLASS_FSI = 21 */
	BDI_N,		/* UCDN_BIDI_CLASS_PDI = 22 */
};

#define class_from_ch_ws(ch) (ucdn_to_bidi[ucdn_get_bidi_class(ch)])

/* Return a direction for white-space on the second pass of the algorithm. */
static fz_bidi_chartype class_from_ch_n(uint32_t ch)
{
	fz_bidi_chartype from_ch_ws = class_from_ch_ws(ch);
	if (from_ch_ws == BDI_S || from_ch_ws == BDI_WS)
		return BDI_N;
	return from_ch_ws;
}

static int
is_european_number(const uint32_t *str, unsigned int len)
{
	const uint32_t *end = str + len;

	for ( ; str != end; str++)
	{
		const uint32_t u = *str;
		if ((u >= UNICODE_RTL_START && u < UNICODE_ARABIC_INDIC_DIGIT_ZERO) ||
			(u > UNICODE_ARABIC_INDIC_DIGIT_NINE && u < UNICODE_EXTENDED_ARABIC_INDIC_DIGIT_ZERO) ||
			(u > UNICODE_EXTENDED_ARABIC_INDIC_DIGIT_NINE && u <= UNICODE_RTL_END))
		{
			/* This is just a normal RTL character or accent */
			return FALSE;
		}
		else if (!((u >= UNICODE_DIGIT_ZERO && u <= UNICODE_DIGIT_NINE) ||
			(u == UNICODE_SUPERSCRIPT_TWO) ||
			(u == UNICODE_SUPERSCRIPT_THREE) ||
			(u == UNICODE_SUPERSCRIPT_ONE) ||
			(u >= UNICODE_ARABIC_INDIC_DIGIT_ZERO && u <= UNICODE_ARABIC_INDIC_DIGIT_NINE) ||
			(u >= UNICODE_EXTENDED_ARABIC_INDIC_DIGIT_ZERO && u <= UNICODE_EXTENDED_ARABIC_INDIC_DIGIT_NINE) ||
			(u == UNICODE_SUPERSCRIPT_ZERO) ||
			(u >= UNICODE_SUPERSCRIPT_FOUR && u <= UNICODE_SUPERSCRIPT_NINE) ||
			(u >= UNICODE_SUBSCRIPT_ZERO && u <= UNICODE_SUBSCRIPT_NINE) ||
			(u >= UNICODE_CIRCLED_DIGIT_ONE && u <= UNICODE_NUMBER_TWENTY_FULL_STOP) ||
			(u == UNICODE_CIRCLED_DIGIT_ZERO) ||
			(u >= UNICODE_FULLWIDTH_DIGIT_ZERO && u <= UNICODE_FULLWIDTH_DIGIT_NINE) ||
			(u == UNICODE_ZERO_WIDTH_NON_JOINER)))
		{
			return FALSE;
		}
	}
	return TRUE;
}

/* Split fragments into single scripts (or punctation + single script) */
static void
split_at_script(const uint32_t *fragment,
		size_t fragment_len,
		int block_r2l,
		int char_r2l,
		void *arg,
		fz_bidi_fragment_callback *callback)
{
	int script = UCDN_SCRIPT_COMMON;
	size_t script_start, i;

	script_start = 0;
	for (i = 0; i < fragment_len; i++)
	{
		int s = ucdn_get_script(fragment[i]);
		if (s == UCDN_SCRIPT_COMMON || s == UCDN_SCRIPT_INHERITED)
		{
			/* Punctuation etc. This is fine. */
		}
		else if (s == script)
		{
			/* Same script. Still fine. */
		}
		else if (script == UCDN_SCRIPT_COMMON || script == UCDN_SCRIPT_INHERITED)
		{
			/* First non punctuation thing. Set the script. */
			script = s;
		}
		else
		{
			/* Change of script. Break the fragment. */
			(*callback)(&fragment[script_start], i - script_start, block_r2l, char_r2l, script, arg);
			script_start = i+1;
			script = s;
		}
	}
	if (script_start != fragment_len)
	{
		(*callback)(&fragment[script_start], fragment_len - script_start, block_r2l, char_r2l, script, arg);
	}
}

static void
detect_numbers(const uint32_t *fragment,
		size_t fragment_len,
		size_t start,
		size_t end,
		const fz_bidi_level *levels,
		void *arg,
		fz_bidi_fragment_callback *callback)
{
	int block_r2l = ODD(levels[start]);
	int char_r2l = block_r2l;

	/* Check to see if we've got a number. Numbers should
	 * never be block_r2l, so we can avoid the test. */
	if (block_r2l || !is_european_number(&fragment[start], end-start))
	{
		/* No number, just split as normal */
		split_at_script(&fragment[start],
				end-start,
				block_r2l,
				char_r2l,
				arg,
				callback);
		return;
	}

	/* We have a number. We have to check to see whether this
	 * should be handled as a block_r2l thing. */
	if (start != 0)
		block_r2l = ODD(levels[start-1]);
	if (block_r2l && end != fragment_len)
		block_r2l = ODD(levels[end]);

	split_at_script(&fragment[start], end-start, block_r2l, char_r2l, arg, callback);
}

/* Determines the character classes for all following
 * passes of the algorithm. A character class is basically the type of Bidi
 * behaviour that the character exhibits.
 */
static void
classify_characters(const uint32_t *text,
		fz_bidi_chartype *types,
		int len,
		fz_bidi_flags flags)
{
	int i;

	if ((flags & BIDI_CLASSIFY_WHITE_SPACE)!=0)
	{
		for (i = 0; i < len; i++)
		{
			types[i] = class_from_ch_ws(text[i]);
		}
	}
	else
	{
#ifdef DEBUG_BIDI_VERBOSE
		fprintf(stderr, "Text:  ");
		for (i = 0; i < len; i++)
		{
			/* So that we can actually sort of read the debug string, any
			 * non-ascii characters are replaced with a 1-digit hash
			 * value from 0-9, making non-english characters appear
			 * as numbers
			 */
			fprintf(stderr, "%c", (text[i] <= 127 && text[i] >= 32) ?
					text[i] : text[i] % 9 + '0');
		}
		fprintf(stderr, "\nTypes: ");
#endif
		for (i = 0; i < len; i++)
		{
			types[i] = class_from_ch_n(text[i]);
#ifdef DEBUG_BIDI_VERBOSE
			fprintf(stderr, "%c", char_from_types[(int)types[i]]);
#endif
		}
#ifdef DEBUG_BIDI_VERBOSE
		fprintf(stderr, "\n");
#endif
	}
}

/* Determines the base level of the text.
 * Implements rule P2 of the Unicode Bidi Algorithm.
 * Note: Ignores explicit embeddings
 */
static fz_bidi_level base_level_from_text(fz_bidi_chartype *types, int len)
{
	int i;

	for (i = 0; i < len; i++)
	{
		switch (types[i])
		{
		/* strong left */
		case BDI_L:
			return BIDI_LEFT_TO_RIGHT;

		/* strong right */
		case BDI_R:
		case BDI_AL:
			return BIDI_RIGHT_TO_LEFT;
		}
	}
	return BIDI_LEFT_TO_RIGHT;
}

static fz_bidi_direction direction_from_type(fz_bidi_chartype type)
{
	switch (type)
	{
	case BDI_L:
	case BDI_EN:
		return BIDI_LEFT_TO_RIGHT;

	case BDI_R:
	case BDI_AL:
		return BIDI_RIGHT_TO_LEFT;

	default:
		return BIDI_NEUTRAL;
	}
}

static void
classify_quoted_blocks(const uint32_t *text,
		fz_bidi_chartype *types,
		size_t len)
{
	size_t i;
	int inQuote = FALSE;
	int pdfNeeded = FALSE;
	int ltrFound = FALSE;
	int rtlFound = FALSE;

	/* Only do anything special here if there is mixed content
	 * (LTR *and* RTL) in the text.
	 */
	for (i = 0; i < len; i++)
	{
		switch (direction_from_type(types[i]))
		{
		case BIDI_LEFT_TO_RIGHT:
			ltrFound = TRUE;
			break;

		case BIDI_RIGHT_TO_LEFT:
			rtlFound = TRUE;
			break;

		default:
			break;
		}
	}

	/* Only make any changes if *both* LTR and RTL characters exist
	 * in this text.
	 */
	if (!ltrFound || !rtlFound)
	{
		return;
	}

	for (i = 0; i < len; i++)
	{
		if (text[i]=='"')
		{
			/* If we're already in a quote then terminate it,
			 * else start a new block.
			 */
			if (inQuote)
			{
				inQuote = FALSE;
				if (pdfNeeded)
				{
					pdfNeeded = FALSE;
					types[i] = BDI_PDF;
				}
			}
			else
			{
				size_t j;
				int done = FALSE;

				inQuote = TRUE;

				/* Find the first strong right or left type and
				 * use that to determine whether we should classify
				 * the quote as LRE or RLE. Or neither, if we
				 * hit another quote before any strongly-directional
				 * character.
				 */
				for (j = i + 1; !done && (j < len) && text[j] != '"'; ++j)
				{
					switch(types[j])
					{
					case BDI_RLE:
					case BDI_LRE:
						done = TRUE;
						break;

					case BDI_L:
					case BDI_EN:
						types[i] = BDI_LRE;
						pdfNeeded = TRUE;
						done = TRUE;
						break;

					case BDI_R:
					case BDI_AL:
						types[i] = BDI_RLE;
						pdfNeeded = TRUE;
						done = TRUE;
						break;

					default:
						break;
					}
				}
			}
		}
	}
}

/* Creates a buffer with an embedding level for every character in the
 * given text. Also determines the base level and returns it in
 * *baseDir if *baseDir does not initially contain a valid direction.
 */
static fz_bidi_level *
create_levels(fz_context *ctx,
		const uint32_t *text,
		size_t len,
		fz_bidi_direction *baseDir,
		int resolveWhiteSpace,
		int flags)
{
	fz_bidi_level *levels;
	fz_bidi_chartype *types = NULL;
	fz_bidi_level baseLevel;

	levels = fz_malloc(ctx, len * sizeof(*levels));

	fz_var(types);

	fz_try(ctx)
	{
		types = fz_malloc(ctx, len * sizeof(fz_bidi_chartype));

		classify_characters(text, types, len, flags);

		if (*baseDir != BIDI_LEFT_TO_RIGHT && *baseDir != BIDI_RIGHT_TO_LEFT)
		{
			/* Derive the base level from the text and
			 * update *baseDir in case the caller wants to know.
			 */
			baseLevel = base_level_from_text(types, len);
			*baseDir = ODD(baseLevel)==1 ? BIDI_RIGHT_TO_LEFT : BIDI_LEFT_TO_RIGHT;
		}
		else
		{
			baseLevel = (fz_bidi_level)*baseDir;
		}

		{
			/* Replace tab with base direction, i.e. make tab appear as
			 * 'strong left' if the base direction is left-to-right and
			 * 'strong right' if base direction is right-to-left. This
			 * allows Layout to implicitly treat tabs as 'segment separators'.
			 */
			size_t i;

			for (i = 0u; i < len; i++)
			{
				if (text[i]=='\t')
				{
					types[i] = (*baseDir == BIDI_RIGHT_TO_LEFT) ? BDI_R : BDI_L;
				}
			}
		}

		/* Look for quotation marks. Classify them as RLE or LRE
		 * or leave them alone, depending on what follows them.
		 */
		classify_quoted_blocks(text, types, len);

		/* Work out the levels and character types... */
		(void)fz_bidi_resolve_explicit(baseLevel, BDI_N, types, levels, len, 0);
		fz_bidi_resolve_weak(ctx, baseLevel, types, levels, len);
		fz_bidi_resolve_neutrals(baseLevel,types, levels, len);
		fz_bidi_resolve_implicit(types, levels, len);

		classify_characters(text, types, len, BIDI_CLASSIFY_WHITE_SPACE);

		if (resolveWhiteSpace)
		{
			/* resolve whitespace */
			fz_bidi_resolve_whitespace(baseLevel, types, levels, len);
		}

		/* The levels buffer now has odd and even numbers indicating
		 * rtl or ltr characters, respectively.
		 */
#ifdef DEBUG_BIDI_VERBOSE
		fprintf(stderr, "Levels: ");
		{
			size_t i;
			for (i = 0; i < len; i++)
			{
				fprintf(stderr, "%d", levels[i]>9?0:levels[i]);
			}
			fprintf(stderr, "\n");
		}
#endif
	}
	fz_always(ctx)
	{
		fz_free(ctx, types);
	}
	fz_catch(ctx)
	{
		fz_free(ctx, levels);
		fz_rethrow(ctx);
	}
	return levels;
}

/* Partitions the given character sequence into one or more unidirectional
 * fragments and invokes the given callback function for each fragment.
 */
void fz_bidi_fragment_text(fz_context *ctx,
		const uint32_t *text,
		size_t textlen,
		fz_bidi_direction *baseDir,
		fz_bidi_fragment_callback *callback,
		void *arg,
		int flags)
{
	size_t startOfFragment;
	size_t i;
	fz_bidi_level *levels;

	if (text == NULL || callback == NULL || textlen == 0)
		return;

	DBUGH(("fz_bidi_fragment_text('%S', len = %d)\n", text, textlen));

	levels = create_levels(ctx, text, textlen, baseDir, FALSE, flags);

	/* We now have an array with an embedding level
	 * for each character in text.
	 */
	assert(levels != NULL);

	fz_try(ctx)
	{
		startOfFragment = 0;
		for (i = 1; i < textlen; i++)
		{
			if (levels[i] != levels[i-1])
			{
				/* We've gone past the end of the fragment.
				 * Create a text object for it, then start
				 * a new fragment.
				 */
				detect_numbers(text,
						textlen,
						startOfFragment,
						i,
						levels,
						arg,
						callback);
				startOfFragment = i;
			}
		}
		/* Now i == textlen. Deal with the final (or maybe only) fragment. */
		/* otherwise create 1 fragment */
		detect_numbers(text,
				textlen,
				startOfFragment,
				i,
				levels,
				arg,
				callback);
	}
	fz_always(ctx)
	{
		fz_free(ctx, levels);
	}
	fz_catch(ctx)
	{
		fz_rethrow(ctx);
	}
}