Merge branch 'ew/instaweb'
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include <sys/types.h>
5 #include <sys/wait.h>
6 #include <signal.h>
7 #include "cache.h"
8 #include "quote.h"
9 #include "diff.h"
10 #include "diffcore.h"
11 #include "delta.h"
12 #include "xdiff-interface.h"
13
14 static int use_size_cache;
15
16 static int diff_rename_limit_default = -1;
17 static int diff_use_color_default = 0;
18
19 enum color_diff {
20         DIFF_RESET = 0,
21         DIFF_PLAIN = 1,
22         DIFF_METAINFO = 2,
23         DIFF_FRAGINFO = 3,
24         DIFF_FILE_OLD = 4,
25         DIFF_FILE_NEW = 5,
26 };
27
28 #define COLOR_NORMAL  ""
29 #define COLOR_BOLD    "\033[1m"
30 #define COLOR_DIM     "\033[2m"
31 #define COLOR_UL      "\033[4m"
32 #define COLOR_BLINK   "\033[5m"
33 #define COLOR_REVERSE "\033[7m"
34 #define COLOR_RESET   "\033[m"
35
36 #define COLOR_BLACK   "\033[30m"
37 #define COLOR_RED     "\033[31m"
38 #define COLOR_GREEN   "\033[32m"
39 #define COLOR_YELLOW  "\033[33m"
40 #define COLOR_BLUE    "\033[34m"
41 #define COLOR_MAGENTA "\033[35m"
42 #define COLOR_CYAN    "\033[36m"
43 #define COLOR_WHITE   "\033[37m"
44
45 static const char *diff_colors[] = {
46         [DIFF_RESET]    = COLOR_RESET,
47         [DIFF_PLAIN]    = COLOR_NORMAL,
48         [DIFF_METAINFO] = COLOR_BOLD,
49         [DIFF_FRAGINFO] = COLOR_CYAN,
50         [DIFF_FILE_OLD] = COLOR_RED,
51         [DIFF_FILE_NEW] = COLOR_GREEN,
52 };
53
54 static int parse_diff_color_slot(const char *var, int ofs)
55 {
56         if (!strcasecmp(var+ofs, "plain"))
57                 return DIFF_PLAIN;
58         if (!strcasecmp(var+ofs, "meta"))
59                 return DIFF_METAINFO;
60         if (!strcasecmp(var+ofs, "frag"))
61                 return DIFF_FRAGINFO;
62         if (!strcasecmp(var+ofs, "old"))
63                 return DIFF_FILE_OLD;
64         if (!strcasecmp(var+ofs, "new"))
65                 return DIFF_FILE_NEW;
66         die("bad config variable '%s'", var);
67 }
68
69 static const char *parse_diff_color_value(const char *value, const char *var)
70 {
71         if (!strcasecmp(value, "normal"))
72                 return COLOR_NORMAL;
73         if (!strcasecmp(value, "bold"))
74                 return COLOR_BOLD;
75         if (!strcasecmp(value, "dim"))
76                 return COLOR_DIM;
77         if (!strcasecmp(value, "ul"))
78                 return COLOR_UL;
79         if (!strcasecmp(value, "blink"))
80                 return COLOR_BLINK;
81         if (!strcasecmp(value, "reverse"))
82                 return COLOR_REVERSE;
83         if (!strcasecmp(value, "reset"))
84                 return COLOR_RESET;
85         if (!strcasecmp(value, "black"))
86                 return COLOR_BLACK;
87         if (!strcasecmp(value, "red"))
88                 return COLOR_RED;
89         if (!strcasecmp(value, "green"))
90                 return COLOR_GREEN;
91         if (!strcasecmp(value, "yellow"))
92                 return COLOR_YELLOW;
93         if (!strcasecmp(value, "blue"))
94                 return COLOR_BLUE;
95         if (!strcasecmp(value, "magenta"))
96                 return COLOR_MAGENTA;
97         if (!strcasecmp(value, "cyan"))
98                 return COLOR_CYAN;
99         if (!strcasecmp(value, "white"))
100                 return COLOR_WHITE;
101         die("bad config value '%s' for variable '%s'", value, var);
102 }
103
104 int git_diff_config(const char *var, const char *value)
105 {
106         if (!strcmp(var, "diff.renamelimit")) {
107                 diff_rename_limit_default = git_config_int(var, value);
108                 return 0;
109         }
110         if (!strcmp(var, "diff.color")) {
111                 if (!value)
112                         diff_use_color_default = 1; /* bool */
113                 else if (!strcasecmp(value, "auto"))
114                         diff_use_color_default = isatty(1);
115                 else if (!strcasecmp(value, "never"))
116                         diff_use_color_default = 0;
117                 else if (!strcasecmp(value, "always"))
118                         diff_use_color_default = 1;
119                 else
120                         diff_use_color_default = git_config_bool(var, value);
121                 return 0;
122         }
123         if (!strncmp(var, "diff.color.", 11)) {
124                 int slot = parse_diff_color_slot(var, 11);
125                 diff_colors[slot] = parse_diff_color_value(value, var);
126                 return 0;
127         }
128         return git_default_config(var, value);
129 }
130
131 static char *quote_one(const char *str)
132 {
133         int needlen;
134         char *xp;
135
136         if (!str)
137                 return NULL;
138         needlen = quote_c_style(str, NULL, NULL, 0);
139         if (!needlen)
140                 return strdup(str);
141         xp = xmalloc(needlen + 1);
142         quote_c_style(str, xp, NULL, 0);
143         return xp;
144 }
145
146 static char *quote_two(const char *one, const char *two)
147 {
148         int need_one = quote_c_style(one, NULL, NULL, 1);
149         int need_two = quote_c_style(two, NULL, NULL, 1);
150         char *xp;
151
152         if (need_one + need_two) {
153                 if (!need_one) need_one = strlen(one);
154                 if (!need_two) need_one = strlen(two);
155
156                 xp = xmalloc(need_one + need_two + 3);
157                 xp[0] = '"';
158                 quote_c_style(one, xp + 1, NULL, 1);
159                 quote_c_style(two, xp + need_one + 1, NULL, 1);
160                 strcpy(xp + need_one + need_two + 1, "\"");
161                 return xp;
162         }
163         need_one = strlen(one);
164         need_two = strlen(two);
165         xp = xmalloc(need_one + need_two + 1);
166         strcpy(xp, one);
167         strcpy(xp + need_one, two);
168         return xp;
169 }
170
171 static const char *external_diff(void)
172 {
173         static const char *external_diff_cmd = NULL;
174         static int done_preparing = 0;
175
176         if (done_preparing)
177                 return external_diff_cmd;
178         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
179         done_preparing = 1;
180         return external_diff_cmd;
181 }
182
183 #define TEMPFILE_PATH_LEN               50
184
185 static struct diff_tempfile {
186         const char *name; /* filename external diff should read from */
187         char hex[41];
188         char mode[10];
189         char tmp_path[TEMPFILE_PATH_LEN];
190 } diff_temp[2];
191
192 static int count_lines(const char *data, int size)
193 {
194         int count, ch, completely_empty = 1, nl_just_seen = 0;
195         count = 0;
196         while (0 < size--) {
197                 ch = *data++;
198                 if (ch == '\n') {
199                         count++;
200                         nl_just_seen = 1;
201                         completely_empty = 0;
202                 }
203                 else {
204                         nl_just_seen = 0;
205                         completely_empty = 0;
206                 }
207         }
208         if (completely_empty)
209                 return 0;
210         if (!nl_just_seen)
211                 count++; /* no trailing newline */
212         return count;
213 }
214
215 static void print_line_count(int count)
216 {
217         switch (count) {
218         case 0:
219                 printf("0,0");
220                 break;
221         case 1:
222                 printf("1");
223                 break;
224         default:
225                 printf("1,%d", count);
226                 break;
227         }
228 }
229
230 static void copy_file(int prefix, const char *data, int size)
231 {
232         int ch, nl_just_seen = 1;
233         while (0 < size--) {
234                 ch = *data++;
235                 if (nl_just_seen)
236                         putchar(prefix);
237                 putchar(ch);
238                 if (ch == '\n')
239                         nl_just_seen = 1;
240                 else
241                         nl_just_seen = 0;
242         }
243         if (!nl_just_seen)
244                 printf("\n\\ No newline at end of file\n");
245 }
246
247 static void emit_rewrite_diff(const char *name_a,
248                               const char *name_b,
249                               struct diff_filespec *one,
250                               struct diff_filespec *two)
251 {
252         int lc_a, lc_b;
253         diff_populate_filespec(one, 0);
254         diff_populate_filespec(two, 0);
255         lc_a = count_lines(one->data, one->size);
256         lc_b = count_lines(two->data, two->size);
257         printf("--- %s\n+++ %s\n@@ -", name_a, name_b);
258         print_line_count(lc_a);
259         printf(" +");
260         print_line_count(lc_b);
261         printf(" @@\n");
262         if (lc_a)
263                 copy_file('-', one->data, one->size);
264         if (lc_b)
265                 copy_file('+', two->data, two->size);
266 }
267
268 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
269 {
270         if (!DIFF_FILE_VALID(one)) {
271                 mf->ptr = (char *)""; /* does not matter */
272                 mf->size = 0;
273                 return 0;
274         }
275         else if (diff_populate_filespec(one, 0))
276                 return -1;
277         mf->ptr = one->data;
278         mf->size = one->size;
279         return 0;
280 }
281
282 struct emit_callback {
283         struct xdiff_emit_state xm;
284         int nparents, color_diff;
285         const char **label_path;
286 };
287
288 static inline const char *get_color(int diff_use_color, enum color_diff ix)
289 {
290         if (diff_use_color)
291                 return diff_colors[ix];
292         return "";
293 }
294
295 static void fn_out_consume(void *priv, char *line, unsigned long len)
296 {
297         int i;
298         struct emit_callback *ecbdata = priv;
299         const char *set = get_color(ecbdata->color_diff, DIFF_METAINFO);
300         const char *reset = get_color(ecbdata->color_diff, DIFF_RESET);
301
302         if (ecbdata->label_path[0]) {
303                 printf("%s--- %s%s\n", set, ecbdata->label_path[0], reset);
304                 printf("%s+++ %s%s\n", set, ecbdata->label_path[1], reset);
305                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
306         }
307
308         /* This is not really necessary for now because
309          * this codepath only deals with two-way diffs.
310          */
311         for (i = 0; i < len && line[i] == '@'; i++)
312                 ;
313         if (2 <= i && i < len && line[i] == ' ') {
314                 ecbdata->nparents = i - 1;
315                 set = get_color(ecbdata->color_diff, DIFF_FRAGINFO);
316         }
317         else if (len < ecbdata->nparents)
318                 set = reset;
319         else {
320                 int nparents = ecbdata->nparents;
321                 int color = DIFF_PLAIN;
322                 for (i = 0; i < nparents && len; i++) {
323                         if (line[i] == '-')
324                                 color = DIFF_FILE_OLD;
325                         else if (line[i] == '+')
326                                 color = DIFF_FILE_NEW;
327                 }
328                 set = get_color(ecbdata->color_diff, color);
329         }
330         if (len > 0 && line[len-1] == '\n')
331                 len--;
332         printf("%s%.*s%s\n", set, (int) len, line, reset);
333 }
334
335 static char *pprint_rename(const char *a, const char *b)
336 {
337         const char *old = a;
338         const char *new = b;
339         char *name = NULL;
340         int pfx_length, sfx_length;
341         int len_a = strlen(a);
342         int len_b = strlen(b);
343
344         /* Find common prefix */
345         pfx_length = 0;
346         while (*old && *new && *old == *new) {
347                 if (*old == '/')
348                         pfx_length = old - a + 1;
349                 old++;
350                 new++;
351         }
352
353         /* Find common suffix */
354         old = a + len_a;
355         new = b + len_b;
356         sfx_length = 0;
357         while (a <= old && b <= new && *old == *new) {
358                 if (*old == '/')
359                         sfx_length = len_a - (old - a);
360                 old--;
361                 new--;
362         }
363
364         /*
365          * pfx{mid-a => mid-b}sfx
366          * {pfx-a => pfx-b}sfx
367          * pfx{sfx-a => sfx-b}
368          * name-a => name-b
369          */
370         if (pfx_length + sfx_length) {
371                 int a_midlen = len_a - pfx_length - sfx_length;
372                 int b_midlen = len_b - pfx_length - sfx_length;
373                 if (a_midlen < 0) a_midlen = 0;
374                 if (b_midlen < 0) b_midlen = 0;
375
376                 name = xmalloc(pfx_length + a_midlen + b_midlen + sfx_length + 7);
377                 sprintf(name, "%.*s{%.*s => %.*s}%s",
378                         pfx_length, a,
379                         a_midlen, a + pfx_length,
380                         b_midlen, b + pfx_length,
381                         a + len_a - sfx_length);
382         }
383         else {
384                 name = xmalloc(len_a + len_b + 5);
385                 sprintf(name, "%s => %s", a, b);
386         }
387         return name;
388 }
389
390 struct diffstat_t {
391         struct xdiff_emit_state xm;
392
393         int nr;
394         int alloc;
395         struct diffstat_file {
396                 char *name;
397                 unsigned is_unmerged:1;
398                 unsigned is_binary:1;
399                 unsigned is_renamed:1;
400                 unsigned int added, deleted;
401         } **files;
402 };
403
404 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
405                                           const char *name_a,
406                                           const char *name_b)
407 {
408         struct diffstat_file *x;
409         x = xcalloc(sizeof (*x), 1);
410         if (diffstat->nr == diffstat->alloc) {
411                 diffstat->alloc = alloc_nr(diffstat->alloc);
412                 diffstat->files = xrealloc(diffstat->files,
413                                 diffstat->alloc * sizeof(x));
414         }
415         diffstat->files[diffstat->nr++] = x;
416         if (name_b) {
417                 x->name = pprint_rename(name_a, name_b);
418                 x->is_renamed = 1;
419         }
420         else
421                 x->name = strdup(name_a);
422         return x;
423 }
424
425 static void diffstat_consume(void *priv, char *line, unsigned long len)
426 {
427         struct diffstat_t *diffstat = priv;
428         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
429
430         if (line[0] == '+')
431                 x->added++;
432         else if (line[0] == '-')
433                 x->deleted++;
434 }
435
436 static const char pluses[] = "++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++";
437 static const char minuses[]= "----------------------------------------------------------------------";
438 const char mime_boundary_leader[] = "------------";
439
440 static void show_stats(struct diffstat_t* data)
441 {
442         int i, len, add, del, total, adds = 0, dels = 0;
443         int max, max_change = 0, max_len = 0;
444         int total_files = data->nr;
445
446         if (data->nr == 0)
447                 return;
448
449         for (i = 0; i < data->nr; i++) {
450                 struct diffstat_file *file = data->files[i];
451
452                 len = strlen(file->name);
453                 if (max_len < len)
454                         max_len = len;
455
456                 if (file->is_binary || file->is_unmerged)
457                         continue;
458                 if (max_change < file->added + file->deleted)
459                         max_change = file->added + file->deleted;
460         }
461
462         for (i = 0; i < data->nr; i++) {
463                 const char *prefix = "";
464                 char *name = data->files[i]->name;
465                 int added = data->files[i]->added;
466                 int deleted = data->files[i]->deleted;
467
468                 if (0 < (len = quote_c_style(name, NULL, NULL, 0))) {
469                         char *qname = xmalloc(len + 1);
470                         quote_c_style(name, qname, NULL, 0);
471                         free(name);
472                         data->files[i]->name = name = qname;
473                 }
474
475                 /*
476                  * "scale" the filename
477                  */
478                 len = strlen(name);
479                 max = max_len;
480                 if (max > 50)
481                         max = 50;
482                 if (len > max) {
483                         char *slash;
484                         prefix = "...";
485                         max -= 3;
486                         name += len - max;
487                         slash = strchr(name, '/');
488                         if (slash)
489                                 name = slash;
490                 }
491                 len = max;
492
493                 /*
494                  * scale the add/delete
495                  */
496                 max = max_change;
497                 if (max + len > 70)
498                         max = 70 - len;
499
500                 if (data->files[i]->is_binary) {
501                         printf(" %s%-*s |  Bin\n", prefix, len, name);
502                         goto free_diffstat_file;
503                 }
504                 else if (data->files[i]->is_unmerged) {
505                         printf(" %s%-*s |  Unmerged\n", prefix, len, name);
506                         goto free_diffstat_file;
507                 }
508                 else if (!data->files[i]->is_renamed &&
509                          (added + deleted == 0)) {
510                         total_files--;
511                         goto free_diffstat_file;
512                 }
513
514                 add = added;
515                 del = deleted;
516                 total = add + del;
517                 adds += add;
518                 dels += del;
519
520                 if (max_change > 0) {
521                         total = (total * max + max_change / 2) / max_change;
522                         add = (add * max + max_change / 2) / max_change;
523                         del = total - add;
524                 }
525                 printf(" %s%-*s |%5d %.*s%.*s\n", prefix,
526                                 len, name, added + deleted,
527                                 add, pluses, del, minuses);
528         free_diffstat_file:
529                 free(data->files[i]->name);
530                 free(data->files[i]);
531         }
532         free(data->files);
533         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
534                         total_files, adds, dels);
535 }
536
537 struct checkdiff_t {
538         struct xdiff_emit_state xm;
539         const char *filename;
540         int lineno;
541 };
542
543 static void checkdiff_consume(void *priv, char *line, unsigned long len)
544 {
545         struct checkdiff_t *data = priv;
546
547         if (line[0] == '+') {
548                 int i, spaces = 0;
549
550                 data->lineno++;
551
552                 /* check space before tab */
553                 for (i = 1; i < len && (line[i] == ' ' || line[i] == '\t'); i++)
554                         if (line[i] == ' ')
555                                 spaces++;
556                 if (line[i - 1] == '\t' && spaces)
557                         printf("%s:%d: space before tab:%.*s\n",
558                                 data->filename, data->lineno, (int)len, line);
559
560                 /* check white space at line end */
561                 if (line[len - 1] == '\n')
562                         len--;
563                 if (isspace(line[len - 1]))
564                         printf("%s:%d: white space at end: %.*s\n",
565                                 data->filename, data->lineno, (int)len, line);
566         } else if (line[0] == ' ')
567                 data->lineno++;
568         else if (line[0] == '@') {
569                 char *plus = strchr(line, '+');
570                 if (plus)
571                         data->lineno = strtol(plus, NULL, 10);
572                 else
573                         die("invalid diff");
574         }
575 }
576
577 static unsigned char *deflate_it(char *data,
578                                  unsigned long size,
579                                  unsigned long *result_size)
580 {
581         int bound;
582         unsigned char *deflated;
583         z_stream stream;
584
585         memset(&stream, 0, sizeof(stream));
586         deflateInit(&stream, zlib_compression_level);
587         bound = deflateBound(&stream, size);
588         deflated = xmalloc(bound);
589         stream.next_out = deflated;
590         stream.avail_out = bound;
591
592         stream.next_in = (unsigned char *)data;
593         stream.avail_in = size;
594         while (deflate(&stream, Z_FINISH) == Z_OK)
595                 ; /* nothing */
596         deflateEnd(&stream);
597         *result_size = stream.total_out;
598         return deflated;
599 }
600
601 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
602 {
603         void *cp;
604         void *delta;
605         void *deflated;
606         void *data;
607         unsigned long orig_size;
608         unsigned long delta_size;
609         unsigned long deflate_size;
610         unsigned long data_size;
611
612         printf("GIT binary patch\n");
613         /* We could do deflated delta, or we could do just deflated two,
614          * whichever is smaller.
615          */
616         delta = NULL;
617         deflated = deflate_it(two->ptr, two->size, &deflate_size);
618         if (one->size && two->size) {
619                 delta = diff_delta(one->ptr, one->size,
620                                    two->ptr, two->size,
621                                    &delta_size, deflate_size);
622                 if (delta) {
623                         void *to_free = delta;
624                         orig_size = delta_size;
625                         delta = deflate_it(delta, delta_size, &delta_size);
626                         free(to_free);
627                 }
628         }
629
630         if (delta && delta_size < deflate_size) {
631                 printf("delta %lu\n", orig_size);
632                 free(deflated);
633                 data = delta;
634                 data_size = delta_size;
635         }
636         else {
637                 printf("literal %lu\n", two->size);
638                 free(delta);
639                 data = deflated;
640                 data_size = deflate_size;
641         }
642
643         /* emit data encoded in base85 */
644         cp = data;
645         while (data_size) {
646                 int bytes = (52 < data_size) ? 52 : data_size;
647                 char line[70];
648                 data_size -= bytes;
649                 if (bytes <= 26)
650                         line[0] = bytes + 'A' - 1;
651                 else
652                         line[0] = bytes - 26 + 'a' - 1;
653                 encode_85(line + 1, cp, bytes);
654                 cp = (char *) cp + bytes;
655                 puts(line);
656         }
657         printf("\n");
658         free(data);
659 }
660
661 #define FIRST_FEW_BYTES 8000
662 static int mmfile_is_binary(mmfile_t *mf)
663 {
664         long sz = mf->size;
665         if (FIRST_FEW_BYTES < sz)
666                 sz = FIRST_FEW_BYTES;
667         if (memchr(mf->ptr, 0, sz))
668                 return 1;
669         return 0;
670 }
671
672 static void builtin_diff(const char *name_a,
673                          const char *name_b,
674                          struct diff_filespec *one,
675                          struct diff_filespec *two,
676                          const char *xfrm_msg,
677                          struct diff_options *o,
678                          int complete_rewrite)
679 {
680         mmfile_t mf1, mf2;
681         const char *lbl[2];
682         char *a_one, *b_two;
683         const char *set = get_color(o->color_diff, DIFF_METAINFO);
684         const char *reset = get_color(o->color_diff, DIFF_RESET);
685
686         a_one = quote_two("a/", name_a);
687         b_two = quote_two("b/", name_b);
688         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
689         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
690         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
691         if (lbl[0][0] == '/') {
692                 /* /dev/null */
693                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
694                 if (xfrm_msg && xfrm_msg[0])
695                         printf("%s%s%s\n", set, xfrm_msg, reset);
696         }
697         else if (lbl[1][0] == '/') {
698                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
699                 if (xfrm_msg && xfrm_msg[0])
700                         printf("%s%s%s\n", set, xfrm_msg, reset);
701         }
702         else {
703                 if (one->mode != two->mode) {
704                         printf("%sold mode %06o%s\n", set, one->mode, reset);
705                         printf("%snew mode %06o%s\n", set, two->mode, reset);
706                 }
707                 if (xfrm_msg && xfrm_msg[0])
708                         printf("%s%s%s\n", set, xfrm_msg, reset);
709                 /*
710                  * we do not run diff between different kind
711                  * of objects.
712                  */
713                 if ((one->mode ^ two->mode) & S_IFMT)
714                         goto free_ab_and_return;
715                 if (complete_rewrite) {
716                         emit_rewrite_diff(name_a, name_b, one, two);
717                         goto free_ab_and_return;
718                 }
719         }
720
721         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
722                 die("unable to read files to diff");
723
724         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2)) {
725                 /* Quite common confusing case */
726                 if (mf1.size == mf2.size &&
727                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
728                         goto free_ab_and_return;
729                 if (o->binary)
730                         emit_binary_diff(&mf1, &mf2);
731                 else
732                         printf("Binary files %s and %s differ\n",
733                                lbl[0], lbl[1]);
734         }
735         else {
736                 /* Crazy xdl interfaces.. */
737                 const char *diffopts = getenv("GIT_DIFF_OPTS");
738                 xpparam_t xpp;
739                 xdemitconf_t xecfg;
740                 xdemitcb_t ecb;
741                 struct emit_callback ecbdata;
742
743                 memset(&ecbdata, 0, sizeof(ecbdata));
744                 ecbdata.label_path = lbl;
745                 ecbdata.color_diff = o->color_diff;
746                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
747                 xecfg.ctxlen = o->context;
748                 xecfg.flags = XDL_EMIT_FUNCNAMES;
749                 if (!diffopts)
750                         ;
751                 else if (!strncmp(diffopts, "--unified=", 10))
752                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
753                 else if (!strncmp(diffopts, "-u", 2))
754                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
755                 ecb.outf = xdiff_outf;
756                 ecb.priv = &ecbdata;
757                 ecbdata.xm.consume = fn_out_consume;
758                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
759         }
760
761  free_ab_and_return:
762         free(a_one);
763         free(b_two);
764         return;
765 }
766
767 static void builtin_diffstat(const char *name_a, const char *name_b,
768                              struct diff_filespec *one,
769                              struct diff_filespec *two,
770                              struct diffstat_t *diffstat,
771                              struct diff_options *o,
772                              int complete_rewrite)
773 {
774         mmfile_t mf1, mf2;
775         struct diffstat_file *data;
776
777         data = diffstat_add(diffstat, name_a, name_b);
778
779         if (!one || !two) {
780                 data->is_unmerged = 1;
781                 return;
782         }
783         if (complete_rewrite) {
784                 diff_populate_filespec(one, 0);
785                 diff_populate_filespec(two, 0);
786                 data->deleted = count_lines(one->data, one->size);
787                 data->added = count_lines(two->data, two->size);
788                 return;
789         }
790         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
791                 die("unable to read files to diff");
792
793         if (mmfile_is_binary(&mf1) || mmfile_is_binary(&mf2))
794                 data->is_binary = 1;
795         else {
796                 /* Crazy xdl interfaces.. */
797                 xpparam_t xpp;
798                 xdemitconf_t xecfg;
799                 xdemitcb_t ecb;
800
801                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
802                 xecfg.ctxlen = 0;
803                 xecfg.flags = 0;
804                 ecb.outf = xdiff_outf;
805                 ecb.priv = diffstat;
806                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
807         }
808 }
809
810 static void builtin_checkdiff(const char *name_a, const char *name_b,
811                              struct diff_filespec *one,
812                              struct diff_filespec *two)
813 {
814         mmfile_t mf1, mf2;
815         struct checkdiff_t data;
816
817         if (!two)
818                 return;
819
820         memset(&data, 0, sizeof(data));
821         data.xm.consume = checkdiff_consume;
822         data.filename = name_b ? name_b : name_a;
823         data.lineno = 0;
824
825         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
826                 die("unable to read files to diff");
827
828         if (mmfile_is_binary(&mf2))
829                 return;
830         else {
831                 /* Crazy xdl interfaces.. */
832                 xpparam_t xpp;
833                 xdemitconf_t xecfg;
834                 xdemitcb_t ecb;
835
836                 xpp.flags = XDF_NEED_MINIMAL;
837                 xecfg.ctxlen = 0;
838                 xecfg.flags = 0;
839                 ecb.outf = xdiff_outf;
840                 ecb.priv = &data;
841                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
842         }
843 }
844
845 struct diff_filespec *alloc_filespec(const char *path)
846 {
847         int namelen = strlen(path);
848         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
849
850         memset(spec, 0, sizeof(*spec));
851         spec->path = (char *)(spec + 1);
852         memcpy(spec->path, path, namelen+1);
853         return spec;
854 }
855
856 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
857                    unsigned short mode)
858 {
859         if (mode) {
860                 spec->mode = canon_mode(mode);
861                 memcpy(spec->sha1, sha1, 20);
862                 spec->sha1_valid = !!memcmp(sha1, null_sha1, 20);
863         }
864 }
865
866 /*
867  * Given a name and sha1 pair, if the dircache tells us the file in
868  * the work tree has that object contents, return true, so that
869  * prepare_temp_file() does not have to inflate and extract.
870  */
871 static int work_tree_matches(const char *name, const unsigned char *sha1)
872 {
873         struct cache_entry *ce;
874         struct stat st;
875         int pos, len;
876
877         /* We do not read the cache ourselves here, because the
878          * benchmark with my previous version that always reads cache
879          * shows that it makes things worse for diff-tree comparing
880          * two linux-2.6 kernel trees in an already checked out work
881          * tree.  This is because most diff-tree comparisons deal with
882          * only a small number of files, while reading the cache is
883          * expensive for a large project, and its cost outweighs the
884          * savings we get by not inflating the object to a temporary
885          * file.  Practically, this code only helps when we are used
886          * by diff-cache --cached, which does read the cache before
887          * calling us.
888          */
889         if (!active_cache)
890                 return 0;
891
892         len = strlen(name);
893         pos = cache_name_pos(name, len);
894         if (pos < 0)
895                 return 0;
896         ce = active_cache[pos];
897         if ((lstat(name, &st) < 0) ||
898             !S_ISREG(st.st_mode) || /* careful! */
899             ce_match_stat(ce, &st, 0) ||
900             memcmp(sha1, ce->sha1, 20))
901                 return 0;
902         /* we return 1 only when we can stat, it is a regular file,
903          * stat information matches, and sha1 recorded in the cache
904          * matches.  I.e. we know the file in the work tree really is
905          * the same as the <name, sha1> pair.
906          */
907         return 1;
908 }
909
910 static struct sha1_size_cache {
911         unsigned char sha1[20];
912         unsigned long size;
913 } **sha1_size_cache;
914 static int sha1_size_cache_nr, sha1_size_cache_alloc;
915
916 static struct sha1_size_cache *locate_size_cache(unsigned char *sha1,
917                                                  int find_only,
918                                                  unsigned long size)
919 {
920         int first, last;
921         struct sha1_size_cache *e;
922
923         first = 0;
924         last = sha1_size_cache_nr;
925         while (last > first) {
926                 int cmp, next = (last + first) >> 1;
927                 e = sha1_size_cache[next];
928                 cmp = memcmp(e->sha1, sha1, 20);
929                 if (!cmp)
930                         return e;
931                 if (cmp < 0) {
932                         last = next;
933                         continue;
934                 }
935                 first = next+1;
936         }
937         /* not found */
938         if (find_only)
939                 return NULL;
940         /* insert to make it at "first" */
941         if (sha1_size_cache_alloc <= sha1_size_cache_nr) {
942                 sha1_size_cache_alloc = alloc_nr(sha1_size_cache_alloc);
943                 sha1_size_cache = xrealloc(sha1_size_cache,
944                                            sha1_size_cache_alloc *
945                                            sizeof(*sha1_size_cache));
946         }
947         sha1_size_cache_nr++;
948         if (first < sha1_size_cache_nr)
949                 memmove(sha1_size_cache + first + 1, sha1_size_cache + first,
950                         (sha1_size_cache_nr - first - 1) *
951                         sizeof(*sha1_size_cache));
952         e = xmalloc(sizeof(struct sha1_size_cache));
953         sha1_size_cache[first] = e;
954         memcpy(e->sha1, sha1, 20);
955         e->size = size;
956         return e;
957 }
958
959 /*
960  * While doing rename detection and pickaxe operation, we may need to
961  * grab the data for the blob (or file) for our own in-core comparison.
962  * diff_filespec has data and size fields for this purpose.
963  */
964 int diff_populate_filespec(struct diff_filespec *s, int size_only)
965 {
966         int err = 0;
967         if (!DIFF_FILE_VALID(s))
968                 die("internal error: asking to populate invalid file.");
969         if (S_ISDIR(s->mode))
970                 return -1;
971
972         if (!use_size_cache)
973                 size_only = 0;
974
975         if (s->data)
976                 return err;
977         if (!s->sha1_valid ||
978             work_tree_matches(s->path, s->sha1)) {
979                 struct stat st;
980                 int fd;
981                 if (lstat(s->path, &st) < 0) {
982                         if (errno == ENOENT) {
983                         err_empty:
984                                 err = -1;
985                         empty:
986                                 s->data = (char *)"";
987                                 s->size = 0;
988                                 return err;
989                         }
990                 }
991                 s->size = st.st_size;
992                 if (!s->size)
993                         goto empty;
994                 if (size_only)
995                         return 0;
996                 if (S_ISLNK(st.st_mode)) {
997                         int ret;
998                         s->data = xmalloc(s->size);
999                         s->should_free = 1;
1000                         ret = readlink(s->path, s->data, s->size);
1001                         if (ret < 0) {
1002                                 free(s->data);
1003                                 goto err_empty;
1004                         }
1005                         return 0;
1006                 }
1007                 fd = open(s->path, O_RDONLY);
1008                 if (fd < 0)
1009                         goto err_empty;
1010                 s->data = mmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1011                 close(fd);
1012                 if (s->data == MAP_FAILED)
1013                         goto err_empty;
1014                 s->should_munmap = 1;
1015         }
1016         else {
1017                 char type[20];
1018                 struct sha1_size_cache *e;
1019
1020                 if (size_only) {
1021                         e = locate_size_cache(s->sha1, 1, 0);
1022                         if (e) {
1023                                 s->size = e->size;
1024                                 return 0;
1025                         }
1026                         if (!sha1_object_info(s->sha1, type, &s->size))
1027                                 locate_size_cache(s->sha1, 0, s->size);
1028                 }
1029                 else {
1030                         s->data = read_sha1_file(s->sha1, type, &s->size);
1031                         s->should_free = 1;
1032                 }
1033         }
1034         return 0;
1035 }
1036
1037 void diff_free_filespec_data(struct diff_filespec *s)
1038 {
1039         if (s->should_free)
1040                 free(s->data);
1041         else if (s->should_munmap)
1042                 munmap(s->data, s->size);
1043         s->should_free = s->should_munmap = 0;
1044         s->data = NULL;
1045         free(s->cnt_data);
1046         s->cnt_data = NULL;
1047 }
1048
1049 static void prep_temp_blob(struct diff_tempfile *temp,
1050                            void *blob,
1051                            unsigned long size,
1052                            const unsigned char *sha1,
1053                            int mode)
1054 {
1055         int fd;
1056
1057         fd = git_mkstemp(temp->tmp_path, TEMPFILE_PATH_LEN, ".diff_XXXXXX");
1058         if (fd < 0)
1059                 die("unable to create temp-file");
1060         if (write(fd, blob, size) != size)
1061                 die("unable to write temp-file");
1062         close(fd);
1063         temp->name = temp->tmp_path;
1064         strcpy(temp->hex, sha1_to_hex(sha1));
1065         temp->hex[40] = 0;
1066         sprintf(temp->mode, "%06o", mode);
1067 }
1068
1069 static void prepare_temp_file(const char *name,
1070                               struct diff_tempfile *temp,
1071                               struct diff_filespec *one)
1072 {
1073         if (!DIFF_FILE_VALID(one)) {
1074         not_a_valid_file:
1075                 /* A '-' entry produces this for file-2, and
1076                  * a '+' entry produces this for file-1.
1077                  */
1078                 temp->name = "/dev/null";
1079                 strcpy(temp->hex, ".");
1080                 strcpy(temp->mode, ".");
1081                 return;
1082         }
1083
1084         if (!one->sha1_valid ||
1085             work_tree_matches(name, one->sha1)) {
1086                 struct stat st;
1087                 if (lstat(name, &st) < 0) {
1088                         if (errno == ENOENT)
1089                                 goto not_a_valid_file;
1090                         die("stat(%s): %s", name, strerror(errno));
1091                 }
1092                 if (S_ISLNK(st.st_mode)) {
1093                         int ret;
1094                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1095                         if (sizeof(buf) <= st.st_size)
1096                                 die("symlink too long: %s", name);
1097                         ret = readlink(name, buf, st.st_size);
1098                         if (ret < 0)
1099                                 die("readlink(%s)", name);
1100                         prep_temp_blob(temp, buf, st.st_size,
1101                                        (one->sha1_valid ?
1102                                         one->sha1 : null_sha1),
1103                                        (one->sha1_valid ?
1104                                         one->mode : S_IFLNK));
1105                 }
1106                 else {
1107                         /* we can borrow from the file in the work tree */
1108                         temp->name = name;
1109                         if (!one->sha1_valid)
1110                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1111                         else
1112                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1113                         /* Even though we may sometimes borrow the
1114                          * contents from the work tree, we always want
1115                          * one->mode.  mode is trustworthy even when
1116                          * !(one->sha1_valid), as long as
1117                          * DIFF_FILE_VALID(one).
1118                          */
1119                         sprintf(temp->mode, "%06o", one->mode);
1120                 }
1121                 return;
1122         }
1123         else {
1124                 if (diff_populate_filespec(one, 0))
1125                         die("cannot read data blob for %s", one->path);
1126                 prep_temp_blob(temp, one->data, one->size,
1127                                one->sha1, one->mode);
1128         }
1129 }
1130
1131 static void remove_tempfile(void)
1132 {
1133         int i;
1134
1135         for (i = 0; i < 2; i++)
1136                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1137                         unlink(diff_temp[i].name);
1138                         diff_temp[i].name = NULL;
1139                 }
1140 }
1141
1142 static void remove_tempfile_on_signal(int signo)
1143 {
1144         remove_tempfile();
1145         signal(SIGINT, SIG_DFL);
1146         raise(signo);
1147 }
1148
1149 static int spawn_prog(const char *pgm, const char **arg)
1150 {
1151         pid_t pid;
1152         int status;
1153
1154         fflush(NULL);
1155         pid = fork();
1156         if (pid < 0)
1157                 die("unable to fork");
1158         if (!pid) {
1159                 execvp(pgm, (char *const*) arg);
1160                 exit(255);
1161         }
1162
1163         while (waitpid(pid, &status, 0) < 0) {
1164                 if (errno == EINTR)
1165                         continue;
1166                 return -1;
1167         }
1168
1169         /* Earlier we did not check the exit status because
1170          * diff exits non-zero if files are different, and
1171          * we are not interested in knowing that.  It was a
1172          * mistake which made it harder to quit a diff-*
1173          * session that uses the git-apply-patch-script as
1174          * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
1175          * should also exit non-zero only when it wants to
1176          * abort the entire diff-* session.
1177          */
1178         if (WIFEXITED(status) && !WEXITSTATUS(status))
1179                 return 0;
1180         return -1;
1181 }
1182
1183 /* An external diff command takes:
1184  *
1185  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1186  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1187  *
1188  */
1189 static void run_external_diff(const char *pgm,
1190                               const char *name,
1191                               const char *other,
1192                               struct diff_filespec *one,
1193                               struct diff_filespec *two,
1194                               const char *xfrm_msg,
1195                               int complete_rewrite)
1196 {
1197         const char *spawn_arg[10];
1198         struct diff_tempfile *temp = diff_temp;
1199         int retval;
1200         static int atexit_asked = 0;
1201         const char *othername;
1202         const char **arg = &spawn_arg[0];
1203
1204         othername = (other? other : name);
1205         if (one && two) {
1206                 prepare_temp_file(name, &temp[0], one);
1207                 prepare_temp_file(othername, &temp[1], two);
1208                 if (! atexit_asked &&
1209                     (temp[0].name == temp[0].tmp_path ||
1210                      temp[1].name == temp[1].tmp_path)) {
1211                         atexit_asked = 1;
1212                         atexit(remove_tempfile);
1213                 }
1214                 signal(SIGINT, remove_tempfile_on_signal);
1215         }
1216
1217         if (one && two) {
1218                 *arg++ = pgm;
1219                 *arg++ = name;
1220                 *arg++ = temp[0].name;
1221                 *arg++ = temp[0].hex;
1222                 *arg++ = temp[0].mode;
1223                 *arg++ = temp[1].name;
1224                 *arg++ = temp[1].hex;
1225                 *arg++ = temp[1].mode;
1226                 if (other) {
1227                         *arg++ = other;
1228                         *arg++ = xfrm_msg;
1229                 }
1230         } else {
1231                 *arg++ = pgm;
1232                 *arg++ = name;
1233         }
1234         *arg = NULL;
1235         retval = spawn_prog(pgm, spawn_arg);
1236         remove_tempfile();
1237         if (retval) {
1238                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1239                 exit(1);
1240         }
1241 }
1242
1243 static void run_diff_cmd(const char *pgm,
1244                          const char *name,
1245                          const char *other,
1246                          struct diff_filespec *one,
1247                          struct diff_filespec *two,
1248                          const char *xfrm_msg,
1249                          struct diff_options *o,
1250                          int complete_rewrite)
1251 {
1252         if (pgm) {
1253                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1254                                   complete_rewrite);
1255                 return;
1256         }
1257         if (one && two)
1258                 builtin_diff(name, other ? other : name,
1259                              one, two, xfrm_msg, o, complete_rewrite);
1260         else
1261                 printf("* Unmerged path %s\n", name);
1262 }
1263
1264 static void diff_fill_sha1_info(struct diff_filespec *one)
1265 {
1266         if (DIFF_FILE_VALID(one)) {
1267                 if (!one->sha1_valid) {
1268                         struct stat st;
1269                         if (lstat(one->path, &st) < 0)
1270                                 die("stat %s", one->path);
1271                         if (index_path(one->sha1, one->path, &st, 0))
1272                                 die("cannot hash %s\n", one->path);
1273                 }
1274         }
1275         else
1276                 memset(one->sha1, 0, 20);
1277 }
1278
1279 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1280 {
1281         const char *pgm = external_diff();
1282         char msg[PATH_MAX*2+300], *xfrm_msg;
1283         struct diff_filespec *one;
1284         struct diff_filespec *two;
1285         const char *name;
1286         const char *other;
1287         char *name_munged, *other_munged;
1288         int complete_rewrite = 0;
1289         int len;
1290
1291         if (DIFF_PAIR_UNMERGED(p)) {
1292                 /* unmerged */
1293                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1294                 return;
1295         }
1296
1297         name = p->one->path;
1298         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1299         name_munged = quote_one(name);
1300         other_munged = quote_one(other);
1301         one = p->one; two = p->two;
1302
1303         diff_fill_sha1_info(one);
1304         diff_fill_sha1_info(two);
1305
1306         len = 0;
1307         switch (p->status) {
1308         case DIFF_STATUS_COPIED:
1309                 len += snprintf(msg + len, sizeof(msg) - len,
1310                                 "similarity index %d%%\n"
1311                                 "copy from %s\n"
1312                                 "copy to %s\n",
1313                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1314                                 name_munged, other_munged);
1315                 break;
1316         case DIFF_STATUS_RENAMED:
1317                 len += snprintf(msg + len, sizeof(msg) - len,
1318                                 "similarity index %d%%\n"
1319                                 "rename from %s\n"
1320                                 "rename to %s\n",
1321                                 (int)(0.5 + p->score * 100.0/MAX_SCORE),
1322                                 name_munged, other_munged);
1323                 break;
1324         case DIFF_STATUS_MODIFIED:
1325                 if (p->score) {
1326                         len += snprintf(msg + len, sizeof(msg) - len,
1327                                         "dissimilarity index %d%%\n",
1328                                         (int)(0.5 + p->score *
1329                                               100.0/MAX_SCORE));
1330                         complete_rewrite = 1;
1331                         break;
1332                 }
1333                 /* fallthru */
1334         default:
1335                 /* nothing */
1336                 ;
1337         }
1338
1339         if (memcmp(one->sha1, two->sha1, 20)) {
1340                 int abbrev = o->full_index ? 40 : DEFAULT_ABBREV;
1341
1342                 len += snprintf(msg + len, sizeof(msg) - len,
1343                                 "index %.*s..%.*s",
1344                                 abbrev, sha1_to_hex(one->sha1),
1345                                 abbrev, sha1_to_hex(two->sha1));
1346                 if (one->mode == two->mode)
1347                         len += snprintf(msg + len, sizeof(msg) - len,
1348                                         " %06o", one->mode);
1349                 len += snprintf(msg + len, sizeof(msg) - len, "\n");
1350         }
1351
1352         if (len)
1353                 msg[--len] = 0;
1354         xfrm_msg = len ? msg : NULL;
1355
1356         if (!pgm &&
1357             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
1358             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
1359                 /* a filepair that changes between file and symlink
1360                  * needs to be split into deletion and creation.
1361                  */
1362                 struct diff_filespec *null = alloc_filespec(two->path);
1363                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
1364                 free(null);
1365                 null = alloc_filespec(one->path);
1366                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
1367                 free(null);
1368         }
1369         else
1370                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
1371                              complete_rewrite);
1372
1373         free(name_munged);
1374         free(other_munged);
1375 }
1376
1377 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
1378                          struct diffstat_t *diffstat)
1379 {
1380         const char *name;
1381         const char *other;
1382         int complete_rewrite = 0;
1383
1384         if (DIFF_PAIR_UNMERGED(p)) {
1385                 /* unmerged */
1386                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
1387                 return;
1388         }
1389
1390         name = p->one->path;
1391         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1392
1393         diff_fill_sha1_info(p->one);
1394         diff_fill_sha1_info(p->two);
1395
1396         if (p->status == DIFF_STATUS_MODIFIED && p->score)
1397                 complete_rewrite = 1;
1398         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
1399 }
1400
1401 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
1402 {
1403         const char *name;
1404         const char *other;
1405
1406         if (DIFF_PAIR_UNMERGED(p)) {
1407                 /* unmerged */
1408                 return;
1409         }
1410
1411         name = p->one->path;
1412         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1413
1414         diff_fill_sha1_info(p->one);
1415         diff_fill_sha1_info(p->two);
1416
1417         builtin_checkdiff(name, other, p->one, p->two);
1418 }
1419
1420 void diff_setup(struct diff_options *options)
1421 {
1422         memset(options, 0, sizeof(*options));
1423         options->line_termination = '\n';
1424         options->break_opt = -1;
1425         options->rename_limit = -1;
1426         options->context = 3;
1427         options->msg_sep = "";
1428
1429         options->change = diff_change;
1430         options->add_remove = diff_addremove;
1431         options->color_diff = diff_use_color_default;
1432 }
1433
1434 int diff_setup_done(struct diff_options *options)
1435 {
1436         if ((options->find_copies_harder &&
1437              options->detect_rename != DIFF_DETECT_COPY) ||
1438             (0 <= options->rename_limit && !options->detect_rename))
1439                 return -1;
1440
1441         if (options->output_format & (DIFF_FORMAT_NAME |
1442                                       DIFF_FORMAT_NAME_STATUS |
1443                                       DIFF_FORMAT_CHECKDIFF |
1444                                       DIFF_FORMAT_NO_OUTPUT))
1445                 options->output_format &= ~(DIFF_FORMAT_RAW |
1446                                             DIFF_FORMAT_DIFFSTAT |
1447                                             DIFF_FORMAT_SUMMARY |
1448                                             DIFF_FORMAT_PATCH);
1449
1450         /*
1451          * These cases always need recursive; we do not drop caller-supplied
1452          * recursive bits for other formats here.
1453          */
1454         if (options->output_format & (DIFF_FORMAT_PATCH |
1455                                       DIFF_FORMAT_DIFFSTAT |
1456                                       DIFF_FORMAT_CHECKDIFF))
1457                 options->recursive = 1;
1458         /*
1459          * Also pickaxe would not work very well if you do not say recursive
1460          */
1461         if (options->pickaxe)
1462                 options->recursive = 1;
1463
1464         if (options->detect_rename && options->rename_limit < 0)
1465                 options->rename_limit = diff_rename_limit_default;
1466         if (options->setup & DIFF_SETUP_USE_CACHE) {
1467                 if (!active_cache)
1468                         /* read-cache does not die even when it fails
1469                          * so it is safe for us to do this here.  Also
1470                          * it does not smudge active_cache or active_nr
1471                          * when it fails, so we do not have to worry about
1472                          * cleaning it up ourselves either.
1473                          */
1474                         read_cache();
1475         }
1476         if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
1477                 use_size_cache = 1;
1478         if (options->abbrev <= 0 || 40 < options->abbrev)
1479                 options->abbrev = 40; /* full */
1480
1481         return 0;
1482 }
1483
1484 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
1485 {
1486         char c, *eq;
1487         int len;
1488
1489         if (*arg != '-')
1490                 return 0;
1491         c = *++arg;
1492         if (!c)
1493                 return 0;
1494         if (c == arg_short) {
1495                 c = *++arg;
1496                 if (!c)
1497                         return 1;
1498                 if (val && isdigit(c)) {
1499                         char *end;
1500                         int n = strtoul(arg, &end, 10);
1501                         if (*end)
1502                                 return 0;
1503                         *val = n;
1504                         return 1;
1505                 }
1506                 return 0;
1507         }
1508         if (c != '-')
1509                 return 0;
1510         arg++;
1511         eq = strchr(arg, '=');
1512         if (eq)
1513                 len = eq - arg;
1514         else
1515                 len = strlen(arg);
1516         if (!len || strncmp(arg, arg_long, len))
1517                 return 0;
1518         if (eq) {
1519                 int n;
1520                 char *end;
1521                 if (!isdigit(*++eq))
1522                         return 0;
1523                 n = strtoul(eq, &end, 10);
1524                 if (*end)
1525                         return 0;
1526                 *val = n;
1527         }
1528         return 1;
1529 }
1530
1531 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
1532 {
1533         const char *arg = av[0];
1534         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
1535                 options->output_format |= DIFF_FORMAT_PATCH;
1536         else if (opt_arg(arg, 'U', "unified", &options->context))
1537                 options->output_format |= DIFF_FORMAT_PATCH;
1538         else if (!strcmp(arg, "--raw"))
1539                 options->output_format |= DIFF_FORMAT_RAW;
1540         else if (!strcmp(arg, "--patch-with-raw")) {
1541                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
1542         }
1543         else if (!strcmp(arg, "--stat"))
1544                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
1545         else if (!strcmp(arg, "--check"))
1546                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
1547         else if (!strcmp(arg, "--summary"))
1548                 options->output_format |= DIFF_FORMAT_SUMMARY;
1549         else if (!strcmp(arg, "--patch-with-stat")) {
1550                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
1551         }
1552         else if (!strcmp(arg, "-z"))
1553                 options->line_termination = 0;
1554         else if (!strncmp(arg, "-l", 2))
1555                 options->rename_limit = strtoul(arg+2, NULL, 10);
1556         else if (!strcmp(arg, "--full-index"))
1557                 options->full_index = 1;
1558         else if (!strcmp(arg, "--binary")) {
1559                 options->output_format |= DIFF_FORMAT_PATCH;
1560                 options->full_index = options->binary = 1;
1561         }
1562         else if (!strcmp(arg, "--name-only"))
1563                 options->output_format |= DIFF_FORMAT_NAME;
1564         else if (!strcmp(arg, "--name-status"))
1565                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
1566         else if (!strcmp(arg, "-R"))
1567                 options->reverse_diff = 1;
1568         else if (!strncmp(arg, "-S", 2))
1569                 options->pickaxe = arg + 2;
1570         else if (!strcmp(arg, "-s")) {
1571                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
1572         }
1573         else if (!strncmp(arg, "-O", 2))
1574                 options->orderfile = arg + 2;
1575         else if (!strncmp(arg, "--diff-filter=", 14))
1576                 options->filter = arg + 14;
1577         else if (!strcmp(arg, "--pickaxe-all"))
1578                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
1579         else if (!strcmp(arg, "--pickaxe-regex"))
1580                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
1581         else if (!strncmp(arg, "-B", 2)) {
1582                 if ((options->break_opt =
1583                      diff_scoreopt_parse(arg)) == -1)
1584                         return -1;
1585         }
1586         else if (!strncmp(arg, "-M", 2)) {
1587                 if ((options->rename_score =
1588                      diff_scoreopt_parse(arg)) == -1)
1589                         return -1;
1590                 options->detect_rename = DIFF_DETECT_RENAME;
1591         }
1592         else if (!strncmp(arg, "-C", 2)) {
1593                 if ((options->rename_score =
1594                      diff_scoreopt_parse(arg)) == -1)
1595                         return -1;
1596                 options->detect_rename = DIFF_DETECT_COPY;
1597         }
1598         else if (!strcmp(arg, "--find-copies-harder"))
1599                 options->find_copies_harder = 1;
1600         else if (!strcmp(arg, "--abbrev"))
1601                 options->abbrev = DEFAULT_ABBREV;
1602         else if (!strncmp(arg, "--abbrev=", 9)) {
1603                 options->abbrev = strtoul(arg + 9, NULL, 10);
1604                 if (options->abbrev < MINIMUM_ABBREV)
1605                         options->abbrev = MINIMUM_ABBREV;
1606                 else if (40 < options->abbrev)
1607                         options->abbrev = 40;
1608         }
1609         else if (!strcmp(arg, "--color"))
1610                 options->color_diff = 1;
1611         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
1612                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
1613         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
1614                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
1615         else
1616                 return 0;
1617         return 1;
1618 }
1619
1620 static int parse_num(const char **cp_p)
1621 {
1622         unsigned long num, scale;
1623         int ch, dot;
1624         const char *cp = *cp_p;
1625
1626         num = 0;
1627         scale = 1;
1628         dot = 0;
1629         for(;;) {
1630                 ch = *cp;
1631                 if ( !dot && ch == '.' ) {
1632                         scale = 1;
1633                         dot = 1;
1634                 } else if ( ch == '%' ) {
1635                         scale = dot ? scale*100 : 100;
1636                         cp++;   /* % is always at the end */
1637                         break;
1638                 } else if ( ch >= '0' && ch <= '9' ) {
1639                         if ( scale < 100000 ) {
1640                                 scale *= 10;
1641                                 num = (num*10) + (ch-'0');
1642                         }
1643                 } else {
1644                         break;
1645                 }
1646                 cp++;
1647         }
1648         *cp_p = cp;
1649
1650         /* user says num divided by scale and we say internally that
1651          * is MAX_SCORE * num / scale.
1652          */
1653         return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
1654 }
1655
1656 int diff_scoreopt_parse(const char *opt)
1657 {
1658         int opt1, opt2, cmd;
1659
1660         if (*opt++ != '-')
1661                 return -1;
1662         cmd = *opt++;
1663         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
1664                 return -1; /* that is not a -M, -C nor -B option */
1665
1666         opt1 = parse_num(&opt);
1667         if (cmd != 'B')
1668                 opt2 = 0;
1669         else {
1670                 if (*opt == 0)
1671                         opt2 = 0;
1672                 else if (*opt != '/')
1673                         return -1; /* we expect -B80/99 or -B80 */
1674                 else {
1675                         opt++;
1676                         opt2 = parse_num(&opt);
1677                 }
1678         }
1679         if (*opt != 0)
1680                 return -1;
1681         return opt1 | (opt2 << 16);
1682 }
1683
1684 struct diff_queue_struct diff_queued_diff;
1685
1686 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
1687 {
1688         if (queue->alloc <= queue->nr) {
1689                 queue->alloc = alloc_nr(queue->alloc);
1690                 queue->queue = xrealloc(queue->queue,
1691                                         sizeof(dp) * queue->alloc);
1692         }
1693         queue->queue[queue->nr++] = dp;
1694 }
1695
1696 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
1697                                  struct diff_filespec *one,
1698                                  struct diff_filespec *two)
1699 {
1700         struct diff_filepair *dp = xmalloc(sizeof(*dp));
1701         dp->one = one;
1702         dp->two = two;
1703         dp->score = 0;
1704         dp->status = 0;
1705         dp->source_stays = 0;
1706         dp->broken_pair = 0;
1707         if (queue)
1708                 diff_q(queue, dp);
1709         return dp;
1710 }
1711
1712 void diff_free_filepair(struct diff_filepair *p)
1713 {
1714         diff_free_filespec_data(p->one);
1715         diff_free_filespec_data(p->two);
1716         free(p->one);
1717         free(p->two);
1718         free(p);
1719 }
1720
1721 /* This is different from find_unique_abbrev() in that
1722  * it stuffs the result with dots for alignment.
1723  */
1724 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
1725 {
1726         int abblen;
1727         const char *abbrev;
1728         if (len == 40)
1729                 return sha1_to_hex(sha1);
1730
1731         abbrev = find_unique_abbrev(sha1, len);
1732         if (!abbrev)
1733                 return sha1_to_hex(sha1);
1734         abblen = strlen(abbrev);
1735         if (abblen < 37) {
1736                 static char hex[41];
1737                 if (len < abblen && abblen <= len + 2)
1738                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
1739                 else
1740                         sprintf(hex, "%s...", abbrev);
1741                 return hex;
1742         }
1743         return sha1_to_hex(sha1);
1744 }
1745
1746 static void diff_flush_raw(struct diff_filepair *p,
1747                            struct diff_options *options)
1748 {
1749         int two_paths;
1750         char status[10];
1751         int abbrev = options->abbrev;
1752         const char *path_one, *path_two;
1753         int inter_name_termination = '\t';
1754         int line_termination = options->line_termination;
1755
1756         if (!line_termination)
1757                 inter_name_termination = 0;
1758
1759         path_one = p->one->path;
1760         path_two = p->two->path;
1761         if (line_termination) {
1762                 path_one = quote_one(path_one);
1763                 path_two = quote_one(path_two);
1764         }
1765
1766         if (p->score)
1767                 sprintf(status, "%c%03d", p->status,
1768                         (int)(0.5 + p->score * 100.0/MAX_SCORE));
1769         else {
1770                 status[0] = p->status;
1771                 status[1] = 0;
1772         }
1773         switch (p->status) {
1774         case DIFF_STATUS_COPIED:
1775         case DIFF_STATUS_RENAMED:
1776                 two_paths = 1;
1777                 break;
1778         case DIFF_STATUS_ADDED:
1779         case DIFF_STATUS_DELETED:
1780                 two_paths = 0;
1781                 break;
1782         default:
1783                 two_paths = 0;
1784                 break;
1785         }
1786         if (!(options->output_format & DIFF_FORMAT_NAME_STATUS)) {
1787                 printf(":%06o %06o %s ",
1788                        p->one->mode, p->two->mode,
1789                        diff_unique_abbrev(p->one->sha1, abbrev));
1790                 printf("%s ",
1791                        diff_unique_abbrev(p->two->sha1, abbrev));
1792         }
1793         printf("%s%c%s", status, inter_name_termination, path_one);
1794         if (two_paths)
1795                 printf("%c%s", inter_name_termination, path_two);
1796         putchar(line_termination);
1797         if (path_one != p->one->path)
1798                 free((void*)path_one);
1799         if (path_two != p->two->path)
1800                 free((void*)path_two);
1801 }
1802
1803 static void diff_flush_name(struct diff_filepair *p, int line_termination)
1804 {
1805         char *path = p->two->path;
1806
1807         if (line_termination)
1808                 path = quote_one(p->two->path);
1809         printf("%s%c", path, line_termination);
1810         if (p->two->path != path)
1811                 free(path);
1812 }
1813
1814 int diff_unmodified_pair(struct diff_filepair *p)
1815 {
1816         /* This function is written stricter than necessary to support
1817          * the currently implemented transformers, but the idea is to
1818          * let transformers to produce diff_filepairs any way they want,
1819          * and filter and clean them up here before producing the output.
1820          */
1821         struct diff_filespec *one, *two;
1822
1823         if (DIFF_PAIR_UNMERGED(p))
1824                 return 0; /* unmerged is interesting */
1825
1826         one = p->one;
1827         two = p->two;
1828
1829         /* deletion, addition, mode or type change
1830          * and rename are all interesting.
1831          */
1832         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
1833             DIFF_PAIR_MODE_CHANGED(p) ||
1834             strcmp(one->path, two->path))
1835                 return 0;
1836
1837         /* both are valid and point at the same path.  that is, we are
1838          * dealing with a change.
1839          */
1840         if (one->sha1_valid && two->sha1_valid &&
1841             !memcmp(one->sha1, two->sha1, sizeof(one->sha1)))
1842                 return 1; /* no change */
1843         if (!one->sha1_valid && !two->sha1_valid)
1844                 return 1; /* both look at the same file on the filesystem. */
1845         return 0;
1846 }
1847
1848 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
1849 {
1850         if (diff_unmodified_pair(p))
1851                 return;
1852
1853         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1854             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1855                 return; /* no tree diffs in patch format */
1856
1857         run_diff(p, o);
1858 }
1859
1860 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
1861                             struct diffstat_t *diffstat)
1862 {
1863         if (diff_unmodified_pair(p))
1864                 return;
1865
1866         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1867             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1868                 return; /* no tree diffs in patch format */
1869
1870         run_diffstat(p, o, diffstat);
1871 }
1872
1873 static void diff_flush_checkdiff(struct diff_filepair *p,
1874                 struct diff_options *o)
1875 {
1876         if (diff_unmodified_pair(p))
1877                 return;
1878
1879         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
1880             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
1881                 return; /* no tree diffs in patch format */
1882
1883         run_checkdiff(p, o);
1884 }
1885
1886 int diff_queue_is_empty(void)
1887 {
1888         struct diff_queue_struct *q = &diff_queued_diff;
1889         int i;
1890         for (i = 0; i < q->nr; i++)
1891                 if (!diff_unmodified_pair(q->queue[i]))
1892                         return 0;
1893         return 1;
1894 }
1895
1896 #if DIFF_DEBUG
1897 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
1898 {
1899         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
1900                 x, one ? one : "",
1901                 s->path,
1902                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
1903                 s->mode,
1904                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
1905         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
1906                 x, one ? one : "",
1907                 s->size, s->xfrm_flags);
1908 }
1909
1910 void diff_debug_filepair(const struct diff_filepair *p, int i)
1911 {
1912         diff_debug_filespec(p->one, i, "one");
1913         diff_debug_filespec(p->two, i, "two");
1914         fprintf(stderr, "score %d, status %c stays %d broken %d\n",
1915                 p->score, p->status ? p->status : '?',
1916                 p->source_stays, p->broken_pair);
1917 }
1918
1919 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
1920 {
1921         int i;
1922         if (msg)
1923                 fprintf(stderr, "%s\n", msg);
1924         fprintf(stderr, "q->nr = %d\n", q->nr);
1925         for (i = 0; i < q->nr; i++) {
1926                 struct diff_filepair *p = q->queue[i];
1927                 diff_debug_filepair(p, i);
1928         }
1929 }
1930 #endif
1931
1932 static void diff_resolve_rename_copy(void)
1933 {
1934         int i, j;
1935         struct diff_filepair *p, *pp;
1936         struct diff_queue_struct *q = &diff_queued_diff;
1937
1938         diff_debug_queue("resolve-rename-copy", q);
1939
1940         for (i = 0; i < q->nr; i++) {
1941                 p = q->queue[i];
1942                 p->status = 0; /* undecided */
1943                 if (DIFF_PAIR_UNMERGED(p))
1944                         p->status = DIFF_STATUS_UNMERGED;
1945                 else if (!DIFF_FILE_VALID(p->one))
1946                         p->status = DIFF_STATUS_ADDED;
1947                 else if (!DIFF_FILE_VALID(p->two))
1948                         p->status = DIFF_STATUS_DELETED;
1949                 else if (DIFF_PAIR_TYPE_CHANGED(p))
1950                         p->status = DIFF_STATUS_TYPE_CHANGED;
1951
1952                 /* from this point on, we are dealing with a pair
1953                  * whose both sides are valid and of the same type, i.e.
1954                  * either in-place edit or rename/copy edit.
1955                  */
1956                 else if (DIFF_PAIR_RENAME(p)) {
1957                         if (p->source_stays) {
1958                                 p->status = DIFF_STATUS_COPIED;
1959                                 continue;
1960                         }
1961                         /* See if there is some other filepair that
1962                          * copies from the same source as us.  If so
1963                          * we are a copy.  Otherwise we are either a
1964                          * copy if the path stays, or a rename if it
1965                          * does not, but we already handled "stays" case.
1966                          */
1967                         for (j = i + 1; j < q->nr; j++) {
1968                                 pp = q->queue[j];
1969                                 if (strcmp(pp->one->path, p->one->path))
1970                                         continue; /* not us */
1971                                 if (!DIFF_PAIR_RENAME(pp))
1972                                         continue; /* not a rename/copy */
1973                                 /* pp is a rename/copy from the same source */
1974                                 p->status = DIFF_STATUS_COPIED;
1975                                 break;
1976                         }
1977                         if (!p->status)
1978                                 p->status = DIFF_STATUS_RENAMED;
1979                 }
1980                 else if (memcmp(p->one->sha1, p->two->sha1, 20) ||
1981                          p->one->mode != p->two->mode)
1982                         p->status = DIFF_STATUS_MODIFIED;
1983                 else {
1984                         /* This is a "no-change" entry and should not
1985                          * happen anymore, but prepare for broken callers.
1986                          */
1987                         error("feeding unmodified %s to diffcore",
1988                               p->one->path);
1989                         p->status = DIFF_STATUS_UNKNOWN;
1990                 }
1991         }
1992         diff_debug_queue("resolve-rename-copy done", q);
1993 }
1994
1995 static int check_pair_status(struct diff_filepair *p)
1996 {
1997         switch (p->status) {
1998         case DIFF_STATUS_UNKNOWN:
1999                 return 0;
2000         case 0:
2001                 die("internal error in diff-resolve-rename-copy");
2002         default:
2003                 return 1;
2004         }
2005 }
2006
2007 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2008 {
2009         int fmt = opt->output_format;
2010
2011         if (fmt & DIFF_FORMAT_CHECKDIFF)
2012                 diff_flush_checkdiff(p, opt);
2013         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2014                 diff_flush_raw(p, opt);
2015         else if (fmt & DIFF_FORMAT_NAME)
2016                 diff_flush_name(p, opt->line_termination);
2017 }
2018
2019 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2020 {
2021         if (fs->mode)
2022                 printf(" %s mode %06o %s\n", newdelete, fs->mode, fs->path);
2023         else
2024                 printf(" %s %s\n", newdelete, fs->path);
2025 }
2026
2027
2028 static void show_mode_change(struct diff_filepair *p, int show_name)
2029 {
2030         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2031                 if (show_name)
2032                         printf(" mode change %06o => %06o %s\n",
2033                                p->one->mode, p->two->mode, p->two->path);
2034                 else
2035                         printf(" mode change %06o => %06o\n",
2036                                p->one->mode, p->two->mode);
2037         }
2038 }
2039
2040 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2041 {
2042         const char *old, *new;
2043
2044         /* Find common prefix */
2045         old = p->one->path;
2046         new = p->two->path;
2047         while (1) {
2048                 const char *slash_old, *slash_new;
2049                 slash_old = strchr(old, '/');
2050                 slash_new = strchr(new, '/');
2051                 if (!slash_old ||
2052                     !slash_new ||
2053                     slash_old - old != slash_new - new ||
2054                     memcmp(old, new, slash_new - new))
2055                         break;
2056                 old = slash_old + 1;
2057                 new = slash_new + 1;
2058         }
2059         /* p->one->path thru old is the common prefix, and old and new
2060          * through the end of names are renames
2061          */
2062         if (old != p->one->path)
2063                 printf(" %s %.*s{%s => %s} (%d%%)\n", renamecopy,
2064                        (int)(old - p->one->path), p->one->path,
2065                        old, new, (int)(0.5 + p->score * 100.0/MAX_SCORE));
2066         else
2067                 printf(" %s %s => %s (%d%%)\n", renamecopy,
2068                        p->one->path, p->two->path,
2069                        (int)(0.5 + p->score * 100.0/MAX_SCORE));
2070         show_mode_change(p, 0);
2071 }
2072
2073 static void diff_summary(struct diff_filepair *p)
2074 {
2075         switch(p->status) {
2076         case DIFF_STATUS_DELETED:
2077                 show_file_mode_name("delete", p->one);
2078                 break;
2079         case DIFF_STATUS_ADDED:
2080                 show_file_mode_name("create", p->two);
2081                 break;
2082         case DIFF_STATUS_COPIED:
2083                 show_rename_copy("copy", p);
2084                 break;
2085         case DIFF_STATUS_RENAMED:
2086                 show_rename_copy("rename", p);
2087                 break;
2088         default:
2089                 if (p->score) {
2090                         printf(" rewrite %s (%d%%)\n", p->two->path,
2091                                 (int)(0.5 + p->score * 100.0/MAX_SCORE));
2092                         show_mode_change(p, 0);
2093                 } else  show_mode_change(p, 1);
2094                 break;
2095         }
2096 }
2097
2098 struct patch_id_t {
2099         struct xdiff_emit_state xm;
2100         SHA_CTX *ctx;
2101         int patchlen;
2102 };
2103
2104 static int remove_space(char *line, int len)
2105 {
2106         int i;
2107         char *dst = line;
2108         unsigned char c;
2109
2110         for (i = 0; i < len; i++)
2111                 if (!isspace((c = line[i])))
2112                         *dst++ = c;
2113
2114         return dst - line;
2115 }
2116
2117 static void patch_id_consume(void *priv, char *line, unsigned long len)
2118 {
2119         struct patch_id_t *data = priv;
2120         int new_len;
2121
2122         /* Ignore line numbers when computing the SHA1 of the patch */
2123         if (!strncmp(line, "@@ -", 4))
2124                 return;
2125
2126         new_len = remove_space(line, len);
2127
2128         SHA1_Update(data->ctx, line, new_len);
2129         data->patchlen += new_len;
2130 }
2131
2132 /* returns 0 upon success, and writes result into sha1 */
2133 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2134 {
2135         struct diff_queue_struct *q = &diff_queued_diff;
2136         int i;
2137         SHA_CTX ctx;
2138         struct patch_id_t data;
2139         char buffer[PATH_MAX * 4 + 20];
2140
2141         SHA1_Init(&ctx);
2142         memset(&data, 0, sizeof(struct patch_id_t));
2143         data.ctx = &ctx;
2144         data.xm.consume = patch_id_consume;
2145
2146         for (i = 0; i < q->nr; i++) {
2147                 xpparam_t xpp;
2148                 xdemitconf_t xecfg;
2149                 xdemitcb_t ecb;
2150                 mmfile_t mf1, mf2;
2151                 struct diff_filepair *p = q->queue[i];
2152                 int len1, len2;
2153
2154                 if (p->status == 0)
2155                         return error("internal diff status error");
2156                 if (p->status == DIFF_STATUS_UNKNOWN)
2157                         continue;
2158                 if (diff_unmodified_pair(p))
2159                         continue;
2160                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2161                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2162                         continue;
2163                 if (DIFF_PAIR_UNMERGED(p))
2164                         continue;
2165
2166                 diff_fill_sha1_info(p->one);
2167                 diff_fill_sha1_info(p->two);
2168                 if (fill_mmfile(&mf1, p->one) < 0 ||
2169                                 fill_mmfile(&mf2, p->two) < 0)
2170                         return error("unable to read files to diff");
2171
2172                 /* Maybe hash p->two? into the patch id? */
2173                 if (mmfile_is_binary(&mf2))
2174                         continue;
2175
2176                 len1 = remove_space(p->one->path, strlen(p->one->path));
2177                 len2 = remove_space(p->two->path, strlen(p->two->path));
2178                 if (p->one->mode == 0)
2179                         len1 = snprintf(buffer, sizeof(buffer),
2180                                         "diff--gita/%.*sb/%.*s"
2181                                         "newfilemode%06o"
2182                                         "---/dev/null"
2183                                         "+++b/%.*s",
2184                                         len1, p->one->path,
2185                                         len2, p->two->path,
2186                                         p->two->mode,
2187                                         len2, p->two->path);
2188                 else if (p->two->mode == 0)
2189                         len1 = snprintf(buffer, sizeof(buffer),
2190                                         "diff--gita/%.*sb/%.*s"
2191                                         "deletedfilemode%06o"
2192                                         "---a/%.*s"
2193                                         "+++/dev/null",
2194                                         len1, p->one->path,
2195                                         len2, p->two->path,
2196                                         p->one->mode,
2197                                         len1, p->one->path);
2198                 else
2199                         len1 = snprintf(buffer, sizeof(buffer),
2200                                         "diff--gita/%.*sb/%.*s"
2201                                         "---a/%.*s"
2202                                         "+++b/%.*s",
2203                                         len1, p->one->path,
2204                                         len2, p->two->path,
2205                                         len1, p->one->path,
2206                                         len2, p->two->path);
2207                 SHA1_Update(&ctx, buffer, len1);
2208
2209                 xpp.flags = XDF_NEED_MINIMAL;
2210                 xecfg.ctxlen = 3;
2211                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2212                 ecb.outf = xdiff_outf;
2213                 ecb.priv = &data;
2214                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2215         }
2216
2217         SHA1_Final(sha1, &ctx);
2218         return 0;
2219 }
2220
2221 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2222 {
2223         struct diff_queue_struct *q = &diff_queued_diff;
2224         int i;
2225         int result = diff_get_patch_id(options, sha1);
2226
2227         for (i = 0; i < q->nr; i++)
2228                 diff_free_filepair(q->queue[i]);
2229
2230         free(q->queue);
2231         q->queue = NULL;
2232         q->nr = q->alloc = 0;
2233
2234         return result;
2235 }
2236
2237 static int is_summary_empty(const struct diff_queue_struct *q)
2238 {
2239         int i;
2240
2241         for (i = 0; i < q->nr; i++) {
2242                 const struct diff_filepair *p = q->queue[i];
2243
2244                 switch (p->status) {
2245                 case DIFF_STATUS_DELETED:
2246                 case DIFF_STATUS_ADDED:
2247                 case DIFF_STATUS_COPIED:
2248                 case DIFF_STATUS_RENAMED:
2249                         return 0;
2250                 default:
2251                         if (p->score)
2252                                 return 0;
2253                         if (p->one->mode && p->two->mode &&
2254                             p->one->mode != p->two->mode)
2255                                 return 0;
2256                         break;
2257                 }
2258         }
2259         return 1;
2260 }
2261
2262 void diff_flush(struct diff_options *options)
2263 {
2264         struct diff_queue_struct *q = &diff_queued_diff;
2265         int i, output_format = options->output_format;
2266         int separator = 0;
2267
2268         /*
2269          * Order: raw, stat, summary, patch
2270          * or:    name/name-status/checkdiff (other bits clear)
2271          */
2272         if (!q->nr)
2273                 goto free_queue;
2274
2275         if (output_format & (DIFF_FORMAT_RAW |
2276                              DIFF_FORMAT_NAME |
2277                              DIFF_FORMAT_NAME_STATUS |
2278                              DIFF_FORMAT_CHECKDIFF)) {
2279                 for (i = 0; i < q->nr; i++) {
2280                         struct diff_filepair *p = q->queue[i];
2281                         if (check_pair_status(p))
2282                                 flush_one_pair(p, options);
2283                 }
2284                 separator++;
2285         }
2286
2287         if (output_format & DIFF_FORMAT_DIFFSTAT) {
2288                 struct diffstat_t diffstat;
2289
2290                 memset(&diffstat, 0, sizeof(struct diffstat_t));
2291                 diffstat.xm.consume = diffstat_consume;
2292                 for (i = 0; i < q->nr; i++) {
2293                         struct diff_filepair *p = q->queue[i];
2294                         if (check_pair_status(p))
2295                                 diff_flush_stat(p, options, &diffstat);
2296                 }
2297                 show_stats(&diffstat);
2298                 separator++;
2299         }
2300
2301         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2302                 for (i = 0; i < q->nr; i++)
2303                         diff_summary(q->queue[i]);
2304                 separator++;
2305         }
2306
2307         if (output_format & DIFF_FORMAT_PATCH) {
2308                 if (separator) {
2309                         if (options->stat_sep) {
2310                                 /* attach patch instead of inline */
2311                                 fputs(options->stat_sep, stdout);
2312                         } else {
2313                                 putchar(options->line_termination);
2314                         }
2315                 }
2316
2317                 for (i = 0; i < q->nr; i++) {
2318                         struct diff_filepair *p = q->queue[i];
2319                         if (check_pair_status(p))
2320                                 diff_flush_patch(p, options);
2321                 }
2322         }
2323
2324         for (i = 0; i < q->nr; i++)
2325                 diff_free_filepair(q->queue[i]);
2326 free_queue:
2327         free(q->queue);
2328         q->queue = NULL;
2329         q->nr = q->alloc = 0;
2330 }
2331
2332 static void diffcore_apply_filter(const char *filter)
2333 {
2334         int i;
2335         struct diff_queue_struct *q = &diff_queued_diff;
2336         struct diff_queue_struct outq;
2337         outq.queue = NULL;
2338         outq.nr = outq.alloc = 0;
2339
2340         if (!filter)
2341                 return;
2342
2343         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
2344                 int found;
2345                 for (i = found = 0; !found && i < q->nr; i++) {
2346                         struct diff_filepair *p = q->queue[i];
2347                         if (((p->status == DIFF_STATUS_MODIFIED) &&
2348                              ((p->score &&
2349                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2350                               (!p->score &&
2351                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2352                             ((p->status != DIFF_STATUS_MODIFIED) &&
2353                              strchr(filter, p->status)))
2354                                 found++;
2355                 }
2356                 if (found)
2357                         return;
2358
2359                 /* otherwise we will clear the whole queue
2360                  * by copying the empty outq at the end of this
2361                  * function, but first clear the current entries
2362                  * in the queue.
2363                  */
2364                 for (i = 0; i < q->nr; i++)
2365                         diff_free_filepair(q->queue[i]);
2366         }
2367         else {
2368                 /* Only the matching ones */
2369                 for (i = 0; i < q->nr; i++) {
2370                         struct diff_filepair *p = q->queue[i];
2371
2372                         if (((p->status == DIFF_STATUS_MODIFIED) &&
2373                              ((p->score &&
2374                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
2375                               (!p->score &&
2376                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
2377                             ((p->status != DIFF_STATUS_MODIFIED) &&
2378                              strchr(filter, p->status)))
2379                                 diff_q(&outq, p);
2380                         else
2381                                 diff_free_filepair(p);
2382                 }
2383         }
2384         free(q->queue);
2385         *q = outq;
2386 }
2387
2388 void diffcore_std(struct diff_options *options)
2389 {
2390         if (options->break_opt != -1)
2391                 diffcore_break(options->break_opt);
2392         if (options->detect_rename)
2393                 diffcore_rename(options);
2394         if (options->break_opt != -1)
2395                 diffcore_merge_broken();
2396         if (options->pickaxe)
2397                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2398         if (options->orderfile)
2399                 diffcore_order(options->orderfile);
2400         diff_resolve_rename_copy();
2401         diffcore_apply_filter(options->filter);
2402 }
2403
2404
2405 void diffcore_std_no_resolve(struct diff_options *options)
2406 {
2407         if (options->pickaxe)
2408                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
2409         if (options->orderfile)
2410                 diffcore_order(options->orderfile);
2411         diffcore_apply_filter(options->filter);
2412 }
2413
2414 void diff_addremove(struct diff_options *options,
2415                     int addremove, unsigned mode,
2416                     const unsigned char *sha1,
2417                     const char *base, const char *path)
2418 {
2419         char concatpath[PATH_MAX];
2420         struct diff_filespec *one, *two;
2421
2422         /* This may look odd, but it is a preparation for
2423          * feeding "there are unchanged files which should
2424          * not produce diffs, but when you are doing copy
2425          * detection you would need them, so here they are"
2426          * entries to the diff-core.  They will be prefixed
2427          * with something like '=' or '*' (I haven't decided
2428          * which but should not make any difference).
2429          * Feeding the same new and old to diff_change() 
2430          * also has the same effect.
2431          * Before the final output happens, they are pruned after
2432          * merged into rename/copy pairs as appropriate.
2433          */
2434         if (options->reverse_diff)
2435                 addremove = (addremove == '+' ? '-' :
2436                              addremove == '-' ? '+' : addremove);
2437
2438         if (!path) path = "";
2439         sprintf(concatpath, "%s%s", base, path);
2440         one = alloc_filespec(concatpath);
2441         two = alloc_filespec(concatpath);
2442
2443         if (addremove != '+')
2444                 fill_filespec(one, sha1, mode);
2445         if (addremove != '-')
2446                 fill_filespec(two, sha1, mode);
2447
2448         diff_queue(&diff_queued_diff, one, two);
2449 }
2450
2451 void diff_change(struct diff_options *options,
2452                  unsigned old_mode, unsigned new_mode,
2453                  const unsigned char *old_sha1,
2454                  const unsigned char *new_sha1,
2455                  const char *base, const char *path) 
2456 {
2457         char concatpath[PATH_MAX];
2458         struct diff_filespec *one, *two;
2459
2460         if (options->reverse_diff) {
2461                 unsigned tmp;
2462                 const unsigned char *tmp_c;
2463                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
2464                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
2465         }
2466         if (!path) path = "";
2467         sprintf(concatpath, "%s%s", base, path);
2468         one = alloc_filespec(concatpath);
2469         two = alloc_filespec(concatpath);
2470         fill_filespec(one, old_sha1, old_mode);
2471         fill_filespec(two, new_sha1, new_mode);
2472
2473         diff_queue(&diff_queued_diff, one, two);
2474 }
2475
2476 void diff_unmerge(struct diff_options *options,
2477                   const char *path)
2478 {
2479         struct diff_filespec *one, *two;
2480         one = alloc_filespec(path);
2481         two = alloc_filespec(path);
2482         diff_queue(&diff_queued_diff, one, two);
2483 }