combine-diff.c 40.3 KB
Newer Older
1 2
#include "cache.h"
#include "commit.h"
3
#include "blob.h"
4 5 6
#include "diff.h"
#include "diffcore.h"
#include "quote.h"
7
#include "xdiff-interface.h"
8
#include "xdiff/xmacros.h"
9
#include "log-tree.h"
10
#include "refs.h"
11
#include "userdiff.h"
12
#include "sha1-array.h"
13
#include "revision.h"
14

15 16 17 18 19 20 21 22 23 24
static int compare_paths(const struct combine_diff_path *one,
			  const struct diff_filespec *two)
{
	if (!S_ISDIR(one->mode) && !S_ISDIR(two->mode))
		return strcmp(one->path, two->path);

	return base_name_compare(one->path, strlen(one->path), one->mode,
				 two->path, strlen(two->path), two->mode);
}

25
static struct combine_diff_path *intersect_paths(struct combine_diff_path *curr, int n, int num_parent)
26 27
{
	struct diff_queue_struct *q = &diff_queued_diff;
28
	struct combine_diff_path *p, **tail = &curr;
29
	int i, cmp;
30 31 32 33 34

	if (!n) {
		for (i = 0; i < q->nr; i++) {
			int len;
			const char *path;
35
			if (diff_unmodified_pair(q->queue[i]))
36 37 38
				continue;
			path = q->queue[i]->two->path;
			len = strlen(path);
39
			p = xmalloc(combine_diff_path_size(num_parent, len));
40
			p->path = (char *) &(p->parent[num_parent]);
41 42 43
			memcpy(p->path, path, len);
			p->path[len] = 0;
			p->next = NULL;
44 45 46
			memset(p->parent, 0,
			       sizeof(p->parent[0]) * num_parent);

47
			oidcpy(&p->oid, &q->queue[i]->two->oid);
48
			p->mode = q->queue[i]->two->mode;
49
			oidcpy(&p->parent[n].oid, &q->queue[i]->one->oid);
50
			p->parent[n].mode = q->queue[i]->one->mode;
51
			p->parent[n].status = q->queue[i]->status;
52 53
			*tail = p;
			tail = &p->next;
54
		}
55
		return curr;
56 57
	}

58
	/*
59 60
	 * paths in curr (linked list) and q->queue[] (array) are
	 * both sorted in the tree order.
61 62
	 */
	i = 0;
63 64
	while ((p = *tail) != NULL) {
		cmp = ((i >= q->nr)
65
		       ? -1 : compare_paths(p, q->queue[i]->two));
66 67

		if (cmp < 0) {
68 69 70
			/* p->path not in q->queue[]; drop it */
			*tail = p->next;
			free(p);
71
			continue;
72
		}
73

74
		if (cmp > 0) {
75
			/* q->queue[i] not in p->path; skip it */
76 77
			i++;
			continue;
78
		}
79

80
		oidcpy(&p->parent[n].oid, &q->queue[i]->one->oid);
81 82 83
		p->parent[n].mode = q->queue[i]->one->mode;
		p->parent[n].status = q->queue[i]->status;

84
		tail = &p->next;
85
		i++;
86 87 88 89
	}
	return curr;
}

90
/* Lines lost from parent */
91
struct lline {
92
	struct lline *next, *prev;
93 94 95 96 97
	int len;
	unsigned long parent_map;
	char line[FLEX_ARRAY];
};

98 99 100 101 102 103
/* Lines lost from current parent (before coalescing) */
struct plost {
	struct lline *lost_head, *lost_tail;
	int len;
};

104
/* Lines surviving in the merge result */
105
struct sline {
106 107 108 109
	/* Accumulated and coalesced lost lines */
	struct lline *lost;
	int lenlost;
	struct plost plost;
110 111
	char *bol;
	int len;
Junio C Hamano's avatar
Junio C Hamano committed
112 113
	/* bit 0 up to (N-1) are on if the parent has this line (i.e.
	 * we did not change it).
114
	 * bit N is used for "interesting" lines, including context.
115
	 * bit (N+1) is used for "do not show deletion before this".
116
	 */
117
	unsigned long flag;
118
	unsigned long *p_lno;
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
static int match_string_spaces(const char *line1, int len1,
			       const char *line2, int len2,
			       long flags)
{
	if (flags & XDF_WHITESPACE_FLAGS) {
		for (; len1 > 0 && XDL_ISSPACE(line1[len1 - 1]); len1--);
		for (; len2 > 0 && XDL_ISSPACE(line2[len2 - 1]); len2--);
	}

	if (!(flags & (XDF_IGNORE_WHITESPACE | XDF_IGNORE_WHITESPACE_CHANGE)))
		return (len1 == len2 && !memcmp(line1, line2, len1));

	while (len1 > 0 && len2 > 0) {
		len1--;
		len2--;
		if (XDL_ISSPACE(line1[len1]) || XDL_ISSPACE(line2[len2])) {
			if ((flags & XDF_IGNORE_WHITESPACE_CHANGE) &&
			    (!XDL_ISSPACE(line1[len1]) || !XDL_ISSPACE(line2[len2])))
				return 0;

			for (; len1 > 0 && XDL_ISSPACE(line1[len1]); len1--);
			for (; len2 > 0 && XDL_ISSPACE(line2[len2]); len2--);
		}
		if (line1[len1] != line2[len2])
			return 0;
	}

	if (flags & XDF_IGNORE_WHITESPACE) {
		/* Consume remaining spaces */
		for (; len1 > 0 && XDL_ISSPACE(line1[len1 - 1]); len1--);
		for (; len2 > 0 && XDL_ISSPACE(line2[len2 - 1]); len2--);
	}

	/* We matched full line1 and line2 */
	if (!len1 && !len2)
		return 1;

	return 0;
}

161 162 163 164
enum coalesce_direction { MATCH, BASE, NEW };

/* Coalesce new lines into base by finding LCS */
static struct lline *coalesce_lines(struct lline *base, int *lenbase,
165
				    struct lline *newline, int lennew,
166
				    unsigned long parent, long flags)
167
{
168 169 170 171
	int **lcs;
	enum coalesce_direction **directions;
	struct lline *baseend, *newend = NULL;
	int i, j, origbaselen = *lenbase;
172

173
	if (newline == NULL)
174 175 176 177
		return base;

	if (base == NULL) {
		*lenbase = lennew;
178
		return newline;
179 180 181 182
	}

	/*
	 * Coalesce new lines into base by finding the LCS
Ondřej Bílka's avatar
Ondřej Bílka committed
183
	 * - Create the table to run dynamic programming
184 185 186 187 188 189 190 191
	 * - Compute the LCS
	 * - Then reverse read the direction structure:
	 *   - If we have MATCH, assign parent to base flag, and consume
	 *   both baseend and newend
	 *   - Else if we have BASE, consume baseend
	 *   - Else if we have NEW, insert newend lline into base and
	 *   consume newend
	 */
192 193
	lcs = xcalloc(st_add(origbaselen, 1), sizeof(int*));
	directions = xcalloc(st_add(origbaselen, 1), sizeof(enum coalesce_direction*));
194
	for (i = 0; i < origbaselen + 1; i++) {
195 196
		lcs[i] = xcalloc(st_add(lennew, 1), sizeof(int));
		directions[i] = xcalloc(st_add(lennew, 1), sizeof(enum coalesce_direction));
197 198 199 200 201 202
		directions[i][0] = BASE;
	}
	for (j = 1; j < lennew + 1; j++)
		directions[0][j] = NEW;

	for (i = 1, baseend = base; i < origbaselen + 1; i++) {
203
		for (j = 1, newend = newline; j < lennew + 1; j++) {
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
			if (match_string_spaces(baseend->line, baseend->len,
						newend->line, newend->len, flags)) {
				lcs[i][j] = lcs[i - 1][j - 1] + 1;
				directions[i][j] = MATCH;
			} else if (lcs[i][j - 1] >= lcs[i - 1][j]) {
				lcs[i][j] = lcs[i][j - 1];
				directions[i][j] = NEW;
			} else {
				lcs[i][j] = lcs[i - 1][j];
				directions[i][j] = BASE;
			}
			if (newend->next)
				newend = newend->next;
		}
		if (baseend->next)
			baseend = baseend->next;
	}

	for (i = 0; i < origbaselen + 1; i++)
		free(lcs[i]);
	free(lcs);

	/* At this point, baseend and newend point to the end of each lists */
	i--;
	j--;
	while (i != 0 || j != 0) {
		if (directions[i][j] == MATCH) {
			baseend->parent_map |= 1<<parent;
			baseend = baseend->prev;
			newend = newend->prev;
			i--;
			j--;
		} else if (directions[i][j] == NEW) {
			struct lline *lline;

			lline = newend;
			/* Remove lline from new list and update newend */
			if (lline->prev)
				lline->prev->next = lline->next;
			else
244
				newline = lline->next;
245 246 247 248 249 250 251 252 253 254 255 256
			if (lline->next)
				lline->next->prev = lline->prev;

			newend = lline->prev;
			j--;

			/* Add lline to base list */
			if (baseend) {
				lline->next = baseend->next;
				lline->prev = baseend;
				if (lline->prev)
					lline->prev->next = lline;
257
			}
258 259 260 261 262 263 264 265 266 267 268 269
			else {
				lline->next = base;
				base = lline;
			}
			(*lenbase)++;

			if (lline->next)
				lline->next->prev = lline;

		} else {
			baseend = baseend->prev;
			i--;
270 271 272
		}
	}

273
	newend = newline;
274 275 276 277 278 279 280 281 282 283 284 285 286
	while (newend) {
		struct lline *lline = newend;
		newend = newend->next;
		free(lline);
	}

	for (i = 0; i < origbaselen + 1; i++)
		free(directions[i]);
	free(directions);

	return base;
}

287
static char *grab_blob(const struct object_id *oid, unsigned int mode,
288 289 290 291 292 293 294
		       unsigned long *size, struct userdiff_driver *textconv,
		       const char *path)
{
	char *blob;
	enum object_type type;

	if (S_ISGITLINK(mode)) {
295 296 297 298
		struct strbuf buf = STRBUF_INIT;
		strbuf_addf(&buf, "Subproject commit %s\n", oid_to_hex(oid));
		*size = buf.len;
		blob = strbuf_detach(&buf, NULL);
299
	} else if (is_null_oid(oid)) {
300 301 302 303 304
		/* deleted blob */
		*size = 0;
		return xcalloc(1, 1);
	} else if (textconv) {
		struct diff_filespec *df = alloc_filespec(path);
305
		fill_filespec(df, oid, 1, mode);
306 307 308
		*size = fill_textconv(textconv, df, &blob);
		free_filespec(df);
	} else {
309
		blob = read_object_file(oid, &type, size);
310
		if (type != OBJ_BLOB)
311
			die("object '%s' is not a blob!", oid_to_hex(oid));
312 313 314 315 316 317 318 319 320 321 322
	}
	return blob;
}

static void append_lost(struct sline *sline, int n, const char *line, int len)
{
	struct lline *lline;
	unsigned long this_mask = (1UL<<n);
	if (line[len-1] == '\n')
		len--;

323
	FLEX_ALLOC_MEM(lline, line, line, len);
324 325
	lline->len = len;
	lline->next = NULL;
326 327 328 329 330 331 332
	lline->prev = sline->plost.lost_tail;
	if (lline->prev)
		lline->prev->next = lline;
	else
		sline->plost.lost_head = lline;
	sline->plost.lost_tail = lline;
	sline->plost.len++;
333 334 335
	lline->parent_map = this_mask;
}

336
struct combine_diff_state {
337 338
	unsigned int lno;
	int ob, on, nb, nn;
339 340 341 342 343 344 345
	unsigned long nmask;
	int num_parent;
	int n;
	struct sline *sline;
	struct sline *lost_bucket;
};

346
static void consume_line(void *state_, char *line, unsigned long len)
347
{
348
	struct combine_diff_state *state = state_;
349 350 351 352 353 354
	if (5 < len && !memcmp("@@ -", line, 4)) {
		if (parse_hunk_header(line, len,
				      &state->ob, &state->on,
				      &state->nb, &state->nn))
			return;
		state->lno = state->nb;
355
		if (state->nn == 0) {
356 357 358 359
			/* @@ -X,Y +N,0 @@ removed Y lines
			 * that would have come *after* line N
			 * in the result.  Our lost buckets hang
			 * to the line after the removed lines,
360 361 362 363
			 *
			 * Note that this is correct even when N == 0,
			 * in which case the hunk removes the first
			 * line in the file.
364 365
			 */
			state->lost_bucket = &state->sline[state->nb];
366 367 368
			if (!state->nb)
				state->nb = 1;
		} else {
369
			state->lost_bucket = &state->sline[state->nb-1];
370
		}
371 372 373 374 375 376 377 378 379 380 381
		if (!state->sline[state->nb-1].p_lno)
			state->sline[state->nb-1].p_lno =
				xcalloc(state->num_parent,
					sizeof(unsigned long));
		state->sline[state->nb-1].p_lno[state->n] = state->ob;
		return;
	}
	if (!state->lost_bucket)
		return; /* not in any hunk yet */
	switch (line[0]) {
	case '-':
382
		append_lost(state->lost_bucket, state->n, line+1, len-1);
383 384 385 386 387 388 389 390
		break;
	case '+':
		state->sline[state->lno-1].flag |= state->nmask;
		state->lno++;
		break;
	}
}

391
static void combine_diff(const struct object_id *parent, unsigned int mode,
392
			 mmfile_t *result_file,
Junio C Hamano's avatar
Junio C Hamano committed
393
			 struct sline *sline, unsigned int cnt, int n,
394 395
			 int num_parent, int result_deleted,
			 struct userdiff_driver *textconv,
396
			 const char *path, long flags)
397
{
398
	unsigned int p_lno, lno;
399
	unsigned long nmask = (1UL << n);
400 401 402 403 404
	xpparam_t xpp;
	xdemitconf_t xecfg;
	mmfile_t parent_file;
	struct combine_diff_state state;
	unsigned long sz;
405

406
	if (result_deleted)
407 408
		return; /* result deleted */

409
	parent_file.ptr = grab_blob(parent, mode, &sz, textconv, path);
410
	parent_file.size = sz;
411
	memset(&xpp, 0, sizeof(xpp));
412
	xpp.flags = flags;
413
	memset(&xecfg, 0, sizeof(xecfg));
414 415 416 417 418 419 420
	memset(&state, 0, sizeof(state));
	state.nmask = nmask;
	state.sline = sline;
	state.lno = 1;
	state.num_parent = num_parent;
	state.n = n;

Jeff King's avatar
Jeff King committed
421 422 423
	if (xdi_diff_outf(&parent_file, result_file, consume_line, &state,
			  &xpp, &xecfg))
		die("unable to generate combined diff for %s",
Junio C Hamano's avatar
Junio C Hamano committed
424
		    oid_to_hex(parent));
425
	free(parent_file.ptr);
426 427 428 429 430 431 432 433

	/* Assign line numbers for this parent.
	 *
	 * sline[lno].p_lno[n] records the first line number
	 * (counting from 1) for parent N if the final hunk display
	 * started by showing sline[lno] (possibly showing the lost
	 * lines attached to it first).
	 */
434
	for (lno = 0,  p_lno = 1; lno <= cnt; lno++) {
435 436 437
		struct lline *ll;
		sline[lno].p_lno[n] = p_lno;

438 439 440 441 442 443 444 445 446 447
		/* Coalesce new lines */
		if (sline[lno].plost.lost_head) {
			struct sline *sl = &sline[lno];
			sl->lost = coalesce_lines(sl->lost, &sl->lenlost,
						  sl->plost.lost_head,
						  sl->plost.len, n, flags);
			sl->plost.lost_head = sl->plost.lost_tail = NULL;
			sl->plost.len = 0;
		}

448
		/* How many lines would this sline advance the p_lno? */
449
		ll = sline[lno].lost;
450 451 452 453 454
		while (ll) {
			if (ll->parent_map & nmask)
				p_lno++; /* '-' means parent had it */
			ll = ll->next;
		}
455
		if (lno < cnt && !(sline[lno].flag & nmask))
456 457 458
			p_lno++; /* no '+' means parent had it */
	}
	sline[lno].p_lno[n] = p_lno; /* trailer */
459 460 461 462 463 464 465
}

static unsigned long context = 3;
static char combine_marker = '@';

static int interesting(struct sline *sline, unsigned long all_mask)
{
Junio C Hamano's avatar
Junio C Hamano committed
466 467 468
	/* If some parents lost lines here, or if we have added to
	 * some parent, it is interesting.
	 */
469
	return ((sline->flag & all_mask) || sline->lost);
470 471
}

472 473 474 475 476
static unsigned long adjust_hunk_tail(struct sline *sline,
				      unsigned long all_mask,
				      unsigned long hunk_begin,
				      unsigned long i)
{
Junio C Hamano's avatar
Junio C Hamano committed
477 478 479 480 481 482
	/* i points at the first uninteresting line.  If the last line
	 * of the hunk was interesting only because it has some
	 * deletion, then it is not all that interesting for the
	 * purpose of giving trailing context lines.  This is because
	 * we output '-' line and then unmodified sline[i-1] itself in
	 * that case which gives us one extra context line.
483
	 */
Junio C Hamano's avatar
Junio C Hamano committed
484
	if ((hunk_begin + 1 <= i) && !(sline[i-1].flag & all_mask))
485 486 487 488
		i--;
	return i;
}

Junio C Hamano's avatar
Junio C Hamano committed
489 490 491 492
static unsigned long find_next(struct sline *sline,
			       unsigned long mark,
			       unsigned long i,
			       unsigned long cnt,
Junio C Hamano's avatar
Junio C Hamano committed
493
			       int look_for_uninteresting)
494
{
Junio C Hamano's avatar
Junio C Hamano committed
495 496 497 498 499 500 501
	/* We have examined up to i-1 and are about to look at i.
	 * Find next interesting or uninteresting line.  Here,
	 * "interesting" does not mean interesting(), but marked by
	 * the give_context() function below (i.e. it includes context
	 * lines that are not interesting to interesting() function
	 * that are surrounded by interesting() ones.
	 */
502
	while (i <= cnt)
Junio C Hamano's avatar
Junio C Hamano committed
503
		if (look_for_uninteresting
Junio C Hamano's avatar
Junio C Hamano committed
504 505
		    ? !(sline[i].flag & mark)
		    : (sline[i].flag & mark))
506 507 508
			return i;
		else
			i++;
509
	return i;
510 511 512
}

static int give_context(struct sline *sline, unsigned long cnt, int num_parent)
513 514 515
{
	unsigned long all_mask = (1UL<<num_parent) - 1;
	unsigned long mark = (1UL<<num_parent);
516
	unsigned long no_pre_delete = (2UL<<num_parent);
517 518
	unsigned long i;

Junio C Hamano's avatar
Junio C Hamano committed
519
	/* Two groups of interesting lines may have a short gap of
520
	 * uninteresting lines.  Connect such groups to give them a
Junio C Hamano's avatar
Junio C Hamano committed
521 522 523 524 525 526 527 528
	 * bit of context.
	 *
	 * We first start from what the interesting() function says,
	 * and mark them with "mark", and paint context lines with the
	 * mark.  So interesting() would still say false for such context
	 * lines but they are treated as "interesting" in the end.
	 */
	i = find_next(sline, mark, 0, cnt, 0);
529
	if (cnt < i)
530 531
		return 0;

532
	while (i <= cnt) {
533 534
		unsigned long j = (context < i) ? (i - context) : 0;
		unsigned long k;
Junio C Hamano's avatar
Junio C Hamano committed
535 536

		/* Paint a few lines before the first interesting line. */
537 538 539 540 541
		while (j < i) {
			if (!(sline[j].flag & mark))
				sline[j].flag |= no_pre_delete;
			sline[j++].flag |= mark;
		}
542 543

	again:
Junio C Hamano's avatar
Junio C Hamano committed
544 545 546 547
		/* we know up to i is to be included.  where does the
		 * next uninteresting one start?
		 */
		j = find_next(sline, mark, i, cnt, 1);
548
		if (cnt < j)
549 550 551
			break; /* the rest are all interesting */

		/* lookahead context lines */
Junio C Hamano's avatar
Junio C Hamano committed
552
		k = find_next(sline, mark, j, cnt, 0);
553 554 555 556 557 558 559
		j = adjust_hunk_tail(sline, all_mask, i, j);

		if (k < j + context) {
			/* k is interesting and [j,k) are not, but
			 * paint them interesting because the gap is small.
			 */
			while (j < k)
560
				sline[j++].flag |= mark;
561 562
			i = k;
			goto again;
563
		}
564 565

		/* j is the first uninteresting line and there is
Junio C Hamano's avatar
Junio C Hamano committed
566 567
		 * no overlap beyond it within context lines.  Paint
		 * the trailing edge a bit.
568 569
		 */
		i = k;
570
		k = (j + context < cnt+1) ? j + context : cnt+1;
571 572 573 574 575 576 577 578 579 580 581 582 583 584
		while (j < k)
			sline[j++].flag |= mark;
	}
	return 1;
}

static int make_hunks(struct sline *sline, unsigned long cnt,
		       int num_parent, int dense)
{
	unsigned long all_mask = (1UL<<num_parent) - 1;
	unsigned long mark = (1UL<<num_parent);
	unsigned long i;
	int has_interesting = 0;

585
	for (i = 0; i <= cnt; i++) {
586 587 588 589
		if (interesting(&sline[i], all_mask))
			sline[i].flag |= mark;
		else
			sline[i].flag &= ~mark;
590
	}
591
	if (!dense)
592
		return give_context(sline, cnt, num_parent);
593

594 595 596
	/* Look at each hunk, and if we have changes from only one
	 * parent, or the changes are the same from all but one
	 * parent, mark that uninteresting.
597 598
	 */
	i = 0;
599
	while (i <= cnt) {
600
		unsigned long j, hunk_begin, hunk_end;
601
		unsigned long same_diff;
602
		while (i <= cnt && !(sline[i].flag & mark))
603
			i++;
604
		if (cnt < i)
605
			break; /* No more interesting hunks */
606
		hunk_begin = i;
607
		for (j = i + 1; j <= cnt; j++) {
608 609 610 611 612 613 614 615 616
			if (!(sline[j].flag & mark)) {
				/* Look beyond the end to see if there
				 * is an interesting line after this
				 * hunk within context span.
				 */
				unsigned long la; /* lookahead */
				int contin = 0;
				la = adjust_hunk_tail(sline, all_mask,
						     hunk_begin, j);
617 618
				la = (la + context < cnt + 1) ?
					(la + context) : cnt + 1;
619
				while (la && j <= --la) {
620 621 622 623 624 625 626 627 628 629 630 631
					if (sline[la].flag & mark) {
						contin = 1;
						break;
					}
				}
				if (!contin)
					break;
				j = la;
			}
		}
		hunk_end = j;

632
		/* [i..hunk_end) are interesting.  Now is it really
633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649
		 * interesting?  We check if there are only two versions
		 * and the result matches one of them.  That is, we look
		 * at:
		 *   (+) line, which records lines added to which parents;
		 *       this line appears in the result.
		 *   (-) line, which records from what parents the line
		 *       was removed; this line does not appear in the result.
		 * then check the set of parents the result has difference
		 * from, from all lines.  If there are lines that has
		 * different set of parents that the result has differences
		 * from, that means we have more than two versions.
		 *
		 * Even when we have only two versions, if the result does
		 * not match any of the parents, the it should be considered
		 * interesting.  In such a case, we would have all '+' line.
		 * After passing the above "two versions" test, that would
		 * appear as "the same set of parents" to be "all parents".
650
		 */
651 652 653
		same_diff = 0;
		has_interesting = 0;
		for (j = i; j < hunk_end && !has_interesting; j++) {
Junio C Hamano's avatar
Junio C Hamano committed
654
			unsigned long this_diff = sline[j].flag & all_mask;
655
			struct lline *ll = sline[j].lost;
656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678
			if (this_diff) {
				/* This has some changes.  Is it the
				 * same as others?
				 */
				if (!same_diff)
					same_diff = this_diff;
				else if (same_diff != this_diff) {
					has_interesting = 1;
					break;
				}
			}
			while (ll && !has_interesting) {
				/* Lost this line from these parents;
				 * who are they?  Are they the same?
				 */
				this_diff = ll->parent_map;
				if (!same_diff)
					same_diff = this_diff;
				else if (same_diff != this_diff) {
					has_interesting = 1;
				}
				ll = ll->next;
			}
679
		}
680

681
		if (!has_interesting && same_diff != all_mask) {
682
			/* This hunk is not that interesting after all */
683
			for (j = hunk_begin; j < hunk_end; j++)
684 685 686 687
				sline[j].flag &= ~mark;
		}
		i = hunk_end;
	}
688 689

	has_interesting = give_context(sline, cnt, num_parent);
690
	return has_interesting;
691 692
}

693
static void show_parent_lno(struct sline *sline, unsigned long l0, unsigned long l1, int n, unsigned long null_context)
694 695 696
{
	l0 = sline[l0].p_lno[n];
	l1 = sline[l1].p_lno[n];
697
	printf(" -%lu,%lu", l0, l1-l0-null_context);
698 699
}

700 701
static int hunk_comment_line(const char *bol)
{
702 703 704 705 706
	int ch;

	if (!bol)
		return 0;
	ch = *bol & 0xff;
707 708 709
	return (isalpha(ch) || ch == '_' || ch == '$');
}

710 711 712 713 714 715 716 717 718 719 720 721
static void show_line_to_eol(const char *line, int len, const char *reset)
{
	int saw_cr_at_eol = 0;
	if (len < 0)
		len = strlen(line);
	saw_cr_at_eol = (len && line[len-1] == '\r');

	printf("%.*s%s%s\n", len - saw_cr_at_eol, line,
	       reset,
	       saw_cr_at_eol ? "\r" : "");
}

722 723
static void dump_sline(struct sline *sline, const char *line_prefix,
		       unsigned long cnt, int num_parent,
724
		       int use_color, int result_deleted)
725 726
{
	unsigned long mark = (1UL<<num_parent);
727
	unsigned long no_pre_delete = (2UL<<num_parent);
728
	int i;
729
	unsigned long lno = 0;
Junio C Hamano's avatar
Junio C Hamano committed
730
	const char *c_frag = diff_get_color(use_color, DIFF_FRAGINFO);
731
	const char *c_func = diff_get_color(use_color, DIFF_FUNCINFO);
Junio C Hamano's avatar
Junio C Hamano committed
732 733
	const char *c_new = diff_get_color(use_color, DIFF_FILE_NEW);
	const char *c_old = diff_get_color(use_color, DIFF_FILE_OLD);
734
	const char *c_context = diff_get_color(use_color, DIFF_CONTEXT);
Junio C Hamano's avatar
Junio C Hamano committed
735
	const char *c_reset = diff_get_color(use_color, DIFF_RESET);
736

737
	if (result_deleted)
738 739
		return; /* result deleted */

740
	while (1) {
Junio C Hamano's avatar
Junio C Hamano committed
741 742
		unsigned long hunk_end;
		unsigned long rlines;
743
		const char *hunk_comment = NULL;
744
		unsigned long null_context = 0;
745 746 747 748

		while (lno <= cnt && !(sline[lno].flag & mark)) {
			if (hunk_comment_line(sline[lno].bol))
				hunk_comment = sline[lno].bol;
749
			lno++;
750
		}
751
		if (cnt < lno)
752
			break;
753
		else {
754
			for (hunk_end = lno + 1; hunk_end <= cnt; hunk_end++)
755 756 757
				if (!(sline[hunk_end].flag & mark))
					break;
		}
758 759 760
		rlines = hunk_end - lno;
		if (cnt < hunk_end)
			rlines--; /* pointing at the last delete hunk */
761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778

		if (!context) {
			/*
			 * Even when running with --unified=0, all
			 * lines in the hunk needs to be processed in
			 * the loop below in order to show the
			 * deletion recorded in lost_head.  However,
			 * we do not want to show the resulting line
			 * with all blank context markers in such a
			 * case.  Compensate.
			 */
			unsigned long j;
			for (j = lno; j < hunk_end; j++)
				if (!(sline[j].flag & (mark-1)))
					null_context++;
			rlines -= null_context;
		}

779
		printf("%s%s", line_prefix, c_frag);
780
		for (i = 0; i <= num_parent; i++) putchar(combine_marker);
781
		for (i = 0; i < num_parent; i++)
782
			show_parent_lno(sline, lno, hunk_end, i, null_context);
783
		printf(" +%lu,%lu ", lno+1, rlines);
784
		for (i = 0; i <= num_parent; i++) putchar(combine_marker);
785 786 787 788 789 790 791 792 793 794 795

		if (hunk_comment) {
			int comment_end = 0;
			for (i = 0; i < 40; i++) {
				int ch = hunk_comment[i] & 0xff;
				if (!ch || ch == '\n')
					break;
				if (!isspace(ch))
				    comment_end = i;
			}
			if (comment_end)
796
				printf("%s%s %s%s", c_reset,
797
						    c_context, c_reset,
798
						    c_func);
799 800 801 802
			for (i = 0; i < comment_end; i++)
				putchar(hunk_comment[i]);
		}

Junio C Hamano's avatar
Junio C Hamano committed
803
		printf("%s\n", c_reset);
804 805 806
		while (lno < hunk_end) {
			struct lline *ll;
			int j;
Junio C Hamano's avatar
Junio C Hamano committed
807
			unsigned long p_mask;
808
			struct sline *sl = &sline[lno++];
809
			ll = (sl->flag & no_pre_delete) ? NULL : sl->lost;
810
			while (ll) {
811
				printf("%s%s", line_prefix, c_old);
812 813 814 815 816 817
				for (j = 0; j < num_parent; j++) {
					if (ll->parent_map & (1UL<<j))
						putchar('-');
					else
						putchar(' ');
				}
818
				show_line_to_eol(ll->line, -1, c_reset);
819 820
				ll = ll->next;
			}
821
			if (cnt < lno)
822
				break;
Junio C Hamano's avatar
Junio C Hamano committed
823
			p_mask = 1;
824
			fputs(line_prefix, stdout);
825 826 827 828 829 830 831
			if (!(sl->flag & (mark-1))) {
				/*
				 * This sline was here to hang the
				 * lost lines in front of it.
				 */
				if (!context)
					continue;
832
				fputs(c_context, stdout);
833
			}
Junio C Hamano's avatar
Junio C Hamano committed
834 835
			else
				fputs(c_new, stdout);
836
			for (j = 0; j < num_parent; j++) {
Junio C Hamano's avatar
Junio C Hamano committed
837
				if (p_mask & sl->flag)
838
					putchar('+');
Junio C Hamano's avatar
Junio C Hamano committed
839 840 841
				else
					putchar(' ');
				p_mask <<= 1;
842
			}
843
			show_line_to_eol(sl->bol, sl->len, c_reset);
844 845 846 847
		}
	}
}

848 849 850 851 852 853 854 855 856 857 858
static void reuse_combine_diff(struct sline *sline, unsigned long cnt,
			       int i, int j)
{
	/* We have already examined parent j and we know parent i
	 * and parent j are the same, so reuse the combined result
	 * of parent j for parent i.
	 */
	unsigned long lno, imask, jmask;
	imask = (1UL<<i);
	jmask = (1UL<<j);

859
	for (lno = 0; lno <= cnt; lno++) {
860
		struct lline *ll = sline->lost;
861
		sline->p_lno[i] = sline->p_lno[j];
862 863 864 865 866
		while (ll) {
			if (ll->parent_map & jmask)
				ll->parent_map |= imask;
			ll = ll->next;
		}
Junio C Hamano's avatar
Junio C Hamano committed
867 868
		if (sline->flag & jmask)
			sline->flag |= imask;
869 870
		sline++;
	}
871 872
	/* the overall size of the file (sline[cnt]) */
	sline->p_lno[i] = sline->p_lno[j];
873 874
}

875 876 877
static void dump_quoted_path(const char *head,
			     const char *prefix,
			     const char *path,
878
			     const char *line_prefix,
Junio C Hamano's avatar
Junio C Hamano committed
879
			     const char *c_meta, const char *c_reset)
880
{
881 882 883
	static struct strbuf buf = STRBUF_INIT;

	strbuf_reset(&buf);
884
	strbuf_addstr(&buf, line_prefix);
885 886
	strbuf_addstr(&buf, c_meta);
	strbuf_addstr(&buf, head);
887
	quote_two_c_style(&buf, prefix, path, 0);
888 889
	strbuf_addstr(&buf, c_reset);
	puts(buf.buf);
890 891
}

892 893 894 895
static void show_combined_header(struct combine_diff_path *elem,
				 int num_parent,
				 int dense,
				 struct rev_info *rev,
896
				 const char *line_prefix,
897 898
				 int mode_differs,
				 int show_file_header)
899 900
{
	struct diff_options *opt = &rev->diffopt;
901
	int abbrev = opt->flags.full_index ? GIT_SHA1_HEXSZ : DEFAULT_ABBREV;
902 903
	const char *a_prefix = opt->a_prefix ? opt->a_prefix : "a/";
	const char *b_prefix = opt->b_prefix ? opt->b_prefix : "b/";
904 905
	const char *c_meta = diff_get_color_opt(opt, DIFF_METAINFO);
	const char *c_reset = diff_get_color_opt(opt, DIFF_RESET);
906 907 908 909 910 911 912 913 914
	const char *abb;
	int added = 0;
	int deleted = 0;
	int i;

	if (rev->loginfo && !rev->no_commit_id)
		show_log(rev);

	dump_quoted_path(dense ? "diff --cc " : "diff --combined ",
915 916
			 "", elem->path, line_prefix, c_meta, c_reset);
	printf("%s%sindex ", line_prefix, c_meta);
917
	for (i = 0; i < num_parent; i++) {
918
		abb = find_unique_abbrev(&elem->parent[i].oid,
919 920 921
					 abbrev);
		printf("%s%s", i ? "," : "", abb);
	}
922
	abb = find_unique_abbrev(&elem->oid, abbrev);
923 924 925 926 927 928 929 930 931 932 933 934
	printf("..%s%s\n", abb, c_reset);

	if (mode_differs) {
		deleted = !elem->mode;

		/* We say it was added if nobody had it */
		added = !deleted;
		for (i = 0; added && i < num_parent; i++)
			if (elem->parent[i].status !=
			    DIFF_STATUS_ADDED)
				added = 0;
		if (added)
935 936
			printf("%s%snew file mode %06o",
			       line_prefix, c_meta, elem->mode);
937 938
		else {
			if (deleted)
939 940
				printf("%s%sdeleted file ",
				       line_prefix, c_meta);
941 942 943 944 945 946 947 948 949 950 951
			printf("mode ");
			for (i = 0; i < num_parent; i++) {
				printf("%s%06o", i ? "," : "",
				       elem->parent[i].mode);
			}
			if (elem->mode)
				printf("..%06o", elem->mode);
		}
		printf("%s\n", c_reset);
	}

952 953 954
	if (!show_file_header)
		return;

955 956
	if (added)
		dump_quoted_path("--- ", "", "/dev/null",
957
				 line_prefix, c_meta, c_reset);
958 959
	else
		dump_quoted_path("--- ", a_prefix, elem->path,
960
				 line_prefix, c_meta, c_reset);
961 962
	if (deleted)
		dump_quoted_path("+++ ", "", "/dev/null",
963
				 line_prefix, c_meta, c_reset);
964 965
	else
		dump_quoted_path("+++ ", b_prefix, elem->path,
966
				 line_prefix, c_meta, c_reset);
967 968
}

969
static void show_patch_diff(struct combine_diff_path *elem, int num_parent,
970 971
			    int dense, int working_tree_file,
			    struct rev_info *rev)
972
{
973
	struct diff_options *opt = &rev->diffopt;
974
	unsigned long result_size, cnt, lno;
975
	int result_deleted = 0;
976
	char *result, *cp;
977
	struct sline *sline; /* survived lines */
978
	int mode_differs = 0;
979
	int i, show_hunks;
980
	mmfile_t result_file;
981
	struct userdiff_driver *userdiff;
982
	struct userdiff_driver *textconv = NULL;
983
	int is_binary;
984
	const char *line_prefix = diff_line_prefix(opt);
985

986
	context = opt->context;
987 988 989
	userdiff = userdiff_find_by_path(elem->path);
	if (!userdiff)
		userdiff = userdiff_find_by_name("default");
990
	if (opt->flags.allow_textconv)
991
		textconv = userdiff_get_textconv(userdiff);
992

993
	/* Read the result of merge first */
994
	if (!working_tree_file)
995
		result = grab_blob(&elem->oid, elem->mode, &result_size,
996
				   textconv, elem->path);
997
	else {
998
		/* Used by diff-tree to read from the working tree */
999
		struct stat st;
1000 1001 1002 1003 1004 1005
		int fd = -1;

		if (lstat(elem->path, &st) < 0)
			goto deleted_file;

		if (S_ISLNK(st.st_mode)) {
1006 1007 1008
			struct strbuf buf = STRBUF_INIT;

			if (strbuf_readlink(&buf, elem->path, st.st_size) < 0) {
1009
				error_errno("readlink(%s)", elem->path);
1010 1011
				return;
			}
1012 1013
			result_size = buf.len;
			result = strbuf_detach(&buf, NULL);
1014
			elem->mode = canon_mode(st.st_mode);
1015
		} else if (S_ISDIR(st.st_mode)) {
1016
			struct object_id oid;
1017
			if (resolve_gitlink_ref(elem->path, "HEAD", &oid) < 0)
1018
				result = grab_blob(&elem->oid, elem->mode,
1019
						   &result_size, NULL, NULL);
1020
			else
1021
				result = grab_blob(&oid, elem->mode,
1022 1023 1024
						   &result_size, NULL, NULL);
		} else if (textconv) {
			struct diff_filespec *df = alloc_filespec(elem->path);
1025
			fill_filespec(df, &null_oid, 0, st.st_mode);
1026 1027
			result_size = fill_textconv(textconv, df, &result);
			free_filespec(df);
1028
		} else if (0 <= (fd = open(elem->path, O_RDONLY))) {
1029
			size_t len = xsize_t(st.st_size);
1030
			ssize_t done;
1031
			int is_file, i;
1032

Junio C Hamano's avatar
Junio C Hamano committed
1033
			elem->mode = canon_mode(st.st_mode);
1034 1035 1036 1037 1038 1039 1040 1041 1042
			/* if symlinks don't work, assume symlink if all parents
			 * are symlinks
			 */
			is_file = has_symlinks;
			for (i = 0; !is_file && i < num_parent; i++)
				is_file = !S_ISLNK(elem->parent[i].mode);
			if (!is_file)
				elem->mode = canon_mode(S_IFLNK);

1043
			result_size = len;
1044
			result = xmallocz(len);
1045 1046 1047

			done = read_in_full(fd, result, len);
			if (done < 0)
1048
				die_errno("read error '%s'", elem->path);
1049 1050 1051
			else if (done < len)
				die("early EOF '%s'", elem->path);

1052 1053
			/* If not a fake symlink, apply filters, e.g. autocrlf */
			if (is_file) {
1054
				struct strbuf buf = STRBUF_INIT;
1055

1056
				if (convert_to_git(&the_index, elem->path, result, len, &buf, global_conv_flags_eol)) {
1057 1058 1059 1060 1061
					free(result);
					result = strbuf_detach(&buf, &len);
					result_size = len;
				}
			}
1062 1063
		}
		else {
1064
		deleted_file:
1065
			result_deleted = 1;
1066
			result_size = 0;
1067
			elem->mode = 0;
1068
			result = xcalloc(1, 1);
1069
		}
1070

1071 1072 1073
		if (0 <= fd)
			close(fd);
	}
1074

1075 1076 1077 1078 1079 1080 1081
	for (i = 0; i < num_parent; i++) {
		if (elem->parent[i].mode != elem->mode) {
			mode_differs = 1;
			break;
		}
	}

1082 1083 1084
	if (textconv)
		is_binary = 0;
	else if (userdiff->binary != -1)
1085 1086 1087 1088 1089 1090
		is_binary = userdiff->binary;
	else {
		is_binary = buffer_is_binary(result, result_size);
		for (i = 0; !is_binary && i < num_parent; i++) {
			char *buf;
			unsigned long size;
1091
			buf = grab_blob(&elem->parent[i].oid,
1092
					elem->parent[i].mode,
1093
					&size, NULL, NULL);
1094 1095 1096 1097 1098 1099 1100
			if (buffer_is_binary(buf, size))
				is_binary = 1;
			free(buf);
		}
	}
	if (is_binary) {
		show_combined_header(elem, num_parent, dense, rev,
1101
				     line_prefix, mode_differs, 0);