Merge branch 'cc/help'
[git] / diff.c
1 /*
2  * Copyright (C) 2005 Junio C Hamano
3  */
4 #include "cache.h"
5 #include "quote.h"
6 #include "diff.h"
7 #include "diffcore.h"
8 #include "delta.h"
9 #include "xdiff-interface.h"
10 #include "color.h"
11 #include "attr.h"
12 #include "run-command.h"
13
14 #ifdef NO_FAST_WORKING_DIRECTORY
15 #define FAST_WORKING_DIRECTORY 0
16 #else
17 #define FAST_WORKING_DIRECTORY 1
18 #endif
19
20 static int diff_detect_rename_default;
21 static int diff_rename_limit_default = 100;
22 static int diff_use_color_default;
23 int diff_auto_refresh_index = 1;
24
25 static char diff_colors[][COLOR_MAXLEN] = {
26         "\033[m",       /* reset */
27         "",             /* PLAIN (normal) */
28         "\033[1m",      /* METAINFO (bold) */
29         "\033[36m",     /* FRAGINFO (cyan) */
30         "\033[31m",     /* OLD (red) */
31         "\033[32m",     /* NEW (green) */
32         "\033[33m",     /* COMMIT (yellow) */
33         "\033[41m",     /* WHITESPACE (red background) */
34 };
35
36 static int parse_diff_color_slot(const char *var, int ofs)
37 {
38         if (!strcasecmp(var+ofs, "plain"))
39                 return DIFF_PLAIN;
40         if (!strcasecmp(var+ofs, "meta"))
41                 return DIFF_METAINFO;
42         if (!strcasecmp(var+ofs, "frag"))
43                 return DIFF_FRAGINFO;
44         if (!strcasecmp(var+ofs, "old"))
45                 return DIFF_FILE_OLD;
46         if (!strcasecmp(var+ofs, "new"))
47                 return DIFF_FILE_NEW;
48         if (!strcasecmp(var+ofs, "commit"))
49                 return DIFF_COMMIT;
50         if (!strcasecmp(var+ofs, "whitespace"))
51                 return DIFF_WHITESPACE;
52         die("bad config variable '%s'", var);
53 }
54
55 static struct ll_diff_driver {
56         const char *name;
57         struct ll_diff_driver *next;
58         char *cmd;
59 } *user_diff, **user_diff_tail;
60
61 static void read_config_if_needed(void)
62 {
63         if (!user_diff_tail) {
64                 user_diff_tail = &user_diff;
65                 git_config(git_diff_ui_config);
66         }
67 }
68
69 /*
70  * Currently there is only "diff.<drivername>.command" variable;
71  * because there are "diff.color.<slot>" variables, we are parsing
72  * this in a bit convoluted way to allow low level diff driver
73  * called "color".
74  */
75 static int parse_lldiff_command(const char *var, const char *ep, const char *value)
76 {
77         const char *name;
78         int namelen;
79         struct ll_diff_driver *drv;
80
81         name = var + 5;
82         namelen = ep - name;
83         for (drv = user_diff; drv; drv = drv->next)
84                 if (!strncmp(drv->name, name, namelen) && !drv->name[namelen])
85                         break;
86         if (!drv) {
87                 drv = xcalloc(1, sizeof(struct ll_diff_driver));
88                 drv->name = xmemdupz(name, namelen);
89                 if (!user_diff_tail)
90                         user_diff_tail = &user_diff;
91                 *user_diff_tail = drv;
92                 user_diff_tail = &(drv->next);
93         }
94
95         if (!value)
96                 return error("%s: lacks value", var);
97         drv->cmd = strdup(value);
98         return 0;
99 }
100
101 /*
102  * 'diff.<what>.funcname' attribute can be specified in the configuration
103  * to define a customized regexp to find the beginning of a function to
104  * be used for hunk header lines of "diff -p" style output.
105  */
106 static struct funcname_pattern {
107         char *name;
108         char *pattern;
109         struct funcname_pattern *next;
110 } *funcname_pattern_list;
111
112 static int parse_funcname_pattern(const char *var, const char *ep, const char *value)
113 {
114         const char *name;
115         int namelen;
116         struct funcname_pattern *pp;
117
118         name = var + 5; /* "diff." */
119         namelen = ep - name;
120
121         for (pp = funcname_pattern_list; pp; pp = pp->next)
122                 if (!strncmp(pp->name, name, namelen) && !pp->name[namelen])
123                         break;
124         if (!pp) {
125                 pp = xcalloc(1, sizeof(*pp));
126                 pp->name = xmemdupz(name, namelen);
127                 pp->next = funcname_pattern_list;
128                 funcname_pattern_list = pp;
129         }
130         if (pp->pattern)
131                 free(pp->pattern);
132         pp->pattern = xstrdup(value);
133         return 0;
134 }
135
136 /*
137  * These are to give UI layer defaults.
138  * The core-level commands such as git-diff-files should
139  * never be affected by the setting of diff.renames
140  * the user happens to have in the configuration file.
141  */
142 int git_diff_ui_config(const char *var, const char *value)
143 {
144         if (!strcmp(var, "diff.renamelimit")) {
145                 diff_rename_limit_default = git_config_int(var, value);
146                 return 0;
147         }
148         if (!strcmp(var, "diff.color") || !strcmp(var, "color.diff")) {
149                 diff_use_color_default = git_config_colorbool(var, value, -1);
150                 return 0;
151         }
152         if (!strcmp(var, "diff.renames")) {
153                 if (!value)
154                         diff_detect_rename_default = DIFF_DETECT_RENAME;
155                 else if (!strcasecmp(value, "copies") ||
156                          !strcasecmp(value, "copy"))
157                         diff_detect_rename_default = DIFF_DETECT_COPY;
158                 else if (git_config_bool(var,value))
159                         diff_detect_rename_default = DIFF_DETECT_RENAME;
160                 return 0;
161         }
162         if (!strcmp(var, "diff.autorefreshindex")) {
163                 diff_auto_refresh_index = git_config_bool(var, value);
164                 return 0;
165         }
166         if (!prefixcmp(var, "diff.")) {
167                 const char *ep = strrchr(var, '.');
168
169                 if (ep != var + 4) {
170                         if (!strcmp(ep, ".command"))
171                                 return parse_lldiff_command(var, ep, value);
172                         if (!strcmp(ep, ".funcname"))
173                                 return parse_funcname_pattern(var, ep, value);
174                 }
175         }
176         if (!prefixcmp(var, "diff.color.") || !prefixcmp(var, "color.diff.")) {
177                 int slot = parse_diff_color_slot(var, 11);
178                 color_parse(value, var, diff_colors[slot]);
179                 return 0;
180         }
181
182         return git_default_config(var, value);
183 }
184
185 static char *quote_two(const char *one, const char *two)
186 {
187         int need_one = quote_c_style(one, NULL, NULL, 1);
188         int need_two = quote_c_style(two, NULL, NULL, 1);
189         struct strbuf res;
190
191         strbuf_init(&res, 0);
192         if (need_one + need_two) {
193                 strbuf_addch(&res, '"');
194                 quote_c_style(one, &res, NULL, 1);
195                 quote_c_style(two, &res, NULL, 1);
196                 strbuf_addch(&res, '"');
197         } else {
198                 strbuf_addstr(&res, one);
199                 strbuf_addstr(&res, two);
200         }
201         return strbuf_detach(&res, NULL);
202 }
203
204 static const char *external_diff(void)
205 {
206         static const char *external_diff_cmd = NULL;
207         static int done_preparing = 0;
208
209         if (done_preparing)
210                 return external_diff_cmd;
211         external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
212         done_preparing = 1;
213         return external_diff_cmd;
214 }
215
216 static struct diff_tempfile {
217         const char *name; /* filename external diff should read from */
218         char hex[41];
219         char mode[10];
220         char tmp_path[PATH_MAX];
221 } diff_temp[2];
222
223 static int count_lines(const char *data, int size)
224 {
225         int count, ch, completely_empty = 1, nl_just_seen = 0;
226         count = 0;
227         while (0 < size--) {
228                 ch = *data++;
229                 if (ch == '\n') {
230                         count++;
231                         nl_just_seen = 1;
232                         completely_empty = 0;
233                 }
234                 else {
235                         nl_just_seen = 0;
236                         completely_empty = 0;
237                 }
238         }
239         if (completely_empty)
240                 return 0;
241         if (!nl_just_seen)
242                 count++; /* no trailing newline */
243         return count;
244 }
245
246 static void print_line_count(int count)
247 {
248         switch (count) {
249         case 0:
250                 printf("0,0");
251                 break;
252         case 1:
253                 printf("1");
254                 break;
255         default:
256                 printf("1,%d", count);
257                 break;
258         }
259 }
260
261 static void copy_file(int prefix, const char *data, int size,
262                 const char *set, const char *reset)
263 {
264         int ch, nl_just_seen = 1;
265         while (0 < size--) {
266                 ch = *data++;
267                 if (nl_just_seen) {
268                         fputs(set, stdout);
269                         putchar(prefix);
270                 }
271                 if (ch == '\n') {
272                         nl_just_seen = 1;
273                         fputs(reset, stdout);
274                 } else
275                         nl_just_seen = 0;
276                 putchar(ch);
277         }
278         if (!nl_just_seen)
279                 printf("%s\n\\ No newline at end of file\n", reset);
280 }
281
282 static void emit_rewrite_diff(const char *name_a,
283                               const char *name_b,
284                               struct diff_filespec *one,
285                               struct diff_filespec *two,
286                               int color_diff)
287 {
288         int lc_a, lc_b;
289         const char *name_a_tab, *name_b_tab;
290         const char *metainfo = diff_get_color(color_diff, DIFF_METAINFO);
291         const char *fraginfo = diff_get_color(color_diff, DIFF_FRAGINFO);
292         const char *old = diff_get_color(color_diff, DIFF_FILE_OLD);
293         const char *new = diff_get_color(color_diff, DIFF_FILE_NEW);
294         const char *reset = diff_get_color(color_diff, DIFF_RESET);
295
296         name_a += (*name_a == '/');
297         name_b += (*name_b == '/');
298         name_a_tab = strchr(name_a, ' ') ? "\t" : "";
299         name_b_tab = strchr(name_b, ' ') ? "\t" : "";
300
301         diff_populate_filespec(one, 0);
302         diff_populate_filespec(two, 0);
303         lc_a = count_lines(one->data, one->size);
304         lc_b = count_lines(two->data, two->size);
305         printf("%s--- a/%s%s%s\n%s+++ b/%s%s%s\n%s@@ -",
306                metainfo, name_a, name_a_tab, reset,
307                metainfo, name_b, name_b_tab, reset, fraginfo);
308         print_line_count(lc_a);
309         printf(" +");
310         print_line_count(lc_b);
311         printf(" @@%s\n", reset);
312         if (lc_a)
313                 copy_file('-', one->data, one->size, old, reset);
314         if (lc_b)
315                 copy_file('+', two->data, two->size, new, reset);
316 }
317
318 static int fill_mmfile(mmfile_t *mf, struct diff_filespec *one)
319 {
320         if (!DIFF_FILE_VALID(one)) {
321                 mf->ptr = (char *)""; /* does not matter */
322                 mf->size = 0;
323                 return 0;
324         }
325         else if (diff_populate_filespec(one, 0))
326                 return -1;
327         mf->ptr = one->data;
328         mf->size = one->size;
329         return 0;
330 }
331
332 struct diff_words_buffer {
333         mmfile_t text;
334         long alloc;
335         long current; /* output pointer */
336         int suppressed_newline;
337 };
338
339 static void diff_words_append(char *line, unsigned long len,
340                 struct diff_words_buffer *buffer)
341 {
342         if (buffer->text.size + len > buffer->alloc) {
343                 buffer->alloc = (buffer->text.size + len) * 3 / 2;
344                 buffer->text.ptr = xrealloc(buffer->text.ptr, buffer->alloc);
345         }
346         line++;
347         len--;
348         memcpy(buffer->text.ptr + buffer->text.size, line, len);
349         buffer->text.size += len;
350 }
351
352 struct diff_words_data {
353         struct xdiff_emit_state xm;
354         struct diff_words_buffer minus, plus;
355 };
356
357 static void print_word(struct diff_words_buffer *buffer, int len, int color,
358                 int suppress_newline)
359 {
360         const char *ptr;
361         int eol = 0;
362
363         if (len == 0)
364                 return;
365
366         ptr  = buffer->text.ptr + buffer->current;
367         buffer->current += len;
368
369         if (ptr[len - 1] == '\n') {
370                 eol = 1;
371                 len--;
372         }
373
374         fputs(diff_get_color(1, color), stdout);
375         fwrite(ptr, len, 1, stdout);
376         fputs(diff_get_color(1, DIFF_RESET), stdout);
377
378         if (eol) {
379                 if (suppress_newline)
380                         buffer->suppressed_newline = 1;
381                 else
382                         putchar('\n');
383         }
384 }
385
386 static void fn_out_diff_words_aux(void *priv, char *line, unsigned long len)
387 {
388         struct diff_words_data *diff_words = priv;
389
390         if (diff_words->minus.suppressed_newline) {
391                 if (line[0] != '+')
392                         putchar('\n');
393                 diff_words->minus.suppressed_newline = 0;
394         }
395
396         len--;
397         switch (line[0]) {
398                 case '-':
399                         print_word(&diff_words->minus, len, DIFF_FILE_OLD, 1);
400                         break;
401                 case '+':
402                         print_word(&diff_words->plus, len, DIFF_FILE_NEW, 0);
403                         break;
404                 case ' ':
405                         print_word(&diff_words->plus, len, DIFF_PLAIN, 0);
406                         diff_words->minus.current += len;
407                         break;
408         }
409 }
410
411 /* this executes the word diff on the accumulated buffers */
412 static void diff_words_show(struct diff_words_data *diff_words)
413 {
414         xpparam_t xpp;
415         xdemitconf_t xecfg;
416         xdemitcb_t ecb;
417         mmfile_t minus, plus;
418         int i;
419
420         memset(&xecfg, 0, sizeof(xecfg));
421         minus.size = diff_words->minus.text.size;
422         minus.ptr = xmalloc(minus.size);
423         memcpy(minus.ptr, diff_words->minus.text.ptr, minus.size);
424         for (i = 0; i < minus.size; i++)
425                 if (isspace(minus.ptr[i]))
426                         minus.ptr[i] = '\n';
427         diff_words->minus.current = 0;
428
429         plus.size = diff_words->plus.text.size;
430         plus.ptr = xmalloc(plus.size);
431         memcpy(plus.ptr, diff_words->plus.text.ptr, plus.size);
432         for (i = 0; i < plus.size; i++)
433                 if (isspace(plus.ptr[i]))
434                         plus.ptr[i] = '\n';
435         diff_words->plus.current = 0;
436
437         xpp.flags = XDF_NEED_MINIMAL;
438         xecfg.ctxlen = diff_words->minus.alloc + diff_words->plus.alloc;
439         ecb.outf = xdiff_outf;
440         ecb.priv = diff_words;
441         diff_words->xm.consume = fn_out_diff_words_aux;
442         xdl_diff(&minus, &plus, &xpp, &xecfg, &ecb);
443
444         free(minus.ptr);
445         free(plus.ptr);
446         diff_words->minus.text.size = diff_words->plus.text.size = 0;
447
448         if (diff_words->minus.suppressed_newline) {
449                 putchar('\n');
450                 diff_words->minus.suppressed_newline = 0;
451         }
452 }
453
454 struct emit_callback {
455         struct xdiff_emit_state xm;
456         int nparents, color_diff;
457         unsigned ws_rule;
458         const char **label_path;
459         struct diff_words_data *diff_words;
460         int *found_changesp;
461 };
462
463 static void free_diff_words_data(struct emit_callback *ecbdata)
464 {
465         if (ecbdata->diff_words) {
466                 /* flush buffers */
467                 if (ecbdata->diff_words->minus.text.size ||
468                                 ecbdata->diff_words->plus.text.size)
469                         diff_words_show(ecbdata->diff_words);
470
471                 if (ecbdata->diff_words->minus.text.ptr)
472                         free (ecbdata->diff_words->minus.text.ptr);
473                 if (ecbdata->diff_words->plus.text.ptr)
474                         free (ecbdata->diff_words->plus.text.ptr);
475                 free(ecbdata->diff_words);
476                 ecbdata->diff_words = NULL;
477         }
478 }
479
480 const char *diff_get_color(int diff_use_color, enum color_diff ix)
481 {
482         if (diff_use_color)
483                 return diff_colors[ix];
484         return "";
485 }
486
487 static void emit_line(const char *set, const char *reset, const char *line, int len)
488 {
489         if (len > 0 && line[len-1] == '\n')
490                 len--;
491         fputs(set, stdout);
492         fwrite(line, len, 1, stdout);
493         puts(reset);
494 }
495
496 static void emit_line_with_ws(int nparents,
497                               const char *set, const char *reset, const char *ws,
498                               const char *line, int len, unsigned ws_rule)
499 {
500         int col0 = nparents;
501         int last_tab_in_indent = -1;
502         int last_space_in_indent = -1;
503         int i;
504         int tail = len;
505         int need_highlight_leading_space = 0;
506         /*
507          * The line is a newly added line.  Does it have funny leading
508          * whitespaces?  In indent, SP should never precede a TAB.  In
509          * addition, under "indent with non tab" rule, there should not
510          * be more than 8 consecutive spaces.
511          */
512         for (i = col0; i < len; i++) {
513                 if (line[i] == '\t') {
514                         last_tab_in_indent = i;
515                         if ((ws_rule & WS_SPACE_BEFORE_TAB) &&
516                             0 <= last_space_in_indent)
517                                 need_highlight_leading_space = 1;
518                 }
519                 else if (line[i] == ' ')
520                         last_space_in_indent = i;
521                 else
522                         break;
523         }
524         if ((ws_rule & WS_INDENT_WITH_NON_TAB) &&
525             0 <= last_space_in_indent &&
526             last_tab_in_indent < 0 &&
527             8 <= (i - col0)) {
528                 last_tab_in_indent = i;
529                 need_highlight_leading_space = 1;
530         }
531         fputs(set, stdout);
532         fwrite(line, col0, 1, stdout);
533         fputs(reset, stdout);
534         if (((i == len) || line[i] == '\n') && i != col0) {
535                 /* The whole line was indent */
536                 emit_line(ws, reset, line + col0, len - col0);
537                 return;
538         }
539         i = col0;
540         if (need_highlight_leading_space) {
541                 while (i < last_tab_in_indent) {
542                         if (line[i] == ' ') {
543                                 fputs(ws, stdout);
544                                 putchar(' ');
545                                 fputs(reset, stdout);
546                         }
547                         else
548                                 putchar(line[i]);
549                         i++;
550                 }
551         }
552         tail = len - 1;
553         if (line[tail] == '\n' && i < tail)
554                 tail--;
555         if (ws_rule & WS_TRAILING_SPACE) {
556                 while (i < tail) {
557                         if (!isspace(line[tail]))
558                                 break;
559                         tail--;
560                 }
561         }
562         if ((i < tail && line[tail + 1] != '\n')) {
563                 /* This has whitespace between tail+1..len */
564                 fputs(set, stdout);
565                 fwrite(line + i, tail - i + 1, 1, stdout);
566                 fputs(reset, stdout);
567                 emit_line(ws, reset, line + tail + 1, len - tail - 1);
568         }
569         else
570                 emit_line(set, reset, line + i, len - i);
571 }
572
573 static void emit_add_line(const char *reset, struct emit_callback *ecbdata, const char *line, int len)
574 {
575         const char *ws = diff_get_color(ecbdata->color_diff, DIFF_WHITESPACE);
576         const char *set = diff_get_color(ecbdata->color_diff, DIFF_FILE_NEW);
577
578         if (!*ws)
579                 emit_line(set, reset, line, len);
580         else
581                 emit_line_with_ws(ecbdata->nparents, set, reset, ws,
582                                   line, len, ecbdata->ws_rule);
583 }
584
585 static void fn_out_consume(void *priv, char *line, unsigned long len)
586 {
587         int i;
588         int color;
589         struct emit_callback *ecbdata = priv;
590         const char *set = diff_get_color(ecbdata->color_diff, DIFF_METAINFO);
591         const char *reset = diff_get_color(ecbdata->color_diff, DIFF_RESET);
592
593         *(ecbdata->found_changesp) = 1;
594
595         if (ecbdata->label_path[0]) {
596                 const char *name_a_tab, *name_b_tab;
597
598                 name_a_tab = strchr(ecbdata->label_path[0], ' ') ? "\t" : "";
599                 name_b_tab = strchr(ecbdata->label_path[1], ' ') ? "\t" : "";
600
601                 printf("%s--- %s%s%s\n",
602                        set, ecbdata->label_path[0], reset, name_a_tab);
603                 printf("%s+++ %s%s%s\n",
604                        set, ecbdata->label_path[1], reset, name_b_tab);
605                 ecbdata->label_path[0] = ecbdata->label_path[1] = NULL;
606         }
607
608         /* This is not really necessary for now because
609          * this codepath only deals with two-way diffs.
610          */
611         for (i = 0; i < len && line[i] == '@'; i++)
612                 ;
613         if (2 <= i && i < len && line[i] == ' ') {
614                 ecbdata->nparents = i - 1;
615                 emit_line(diff_get_color(ecbdata->color_diff, DIFF_FRAGINFO),
616                           reset, line, len);
617                 return;
618         }
619
620         if (len < ecbdata->nparents) {
621                 set = reset;
622                 emit_line(reset, reset, line, len);
623                 return;
624         }
625
626         color = DIFF_PLAIN;
627         if (ecbdata->diff_words && ecbdata->nparents != 1)
628                 /* fall back to normal diff */
629                 free_diff_words_data(ecbdata);
630         if (ecbdata->diff_words) {
631                 if (line[0] == '-') {
632                         diff_words_append(line, len,
633                                           &ecbdata->diff_words->minus);
634                         return;
635                 } else if (line[0] == '+') {
636                         diff_words_append(line, len,
637                                           &ecbdata->diff_words->plus);
638                         return;
639                 }
640                 if (ecbdata->diff_words->minus.text.size ||
641                     ecbdata->diff_words->plus.text.size)
642                         diff_words_show(ecbdata->diff_words);
643                 line++;
644                 len--;
645                 emit_line(set, reset, line, len);
646                 return;
647         }
648         for (i = 0; i < ecbdata->nparents && len; i++) {
649                 if (line[i] == '-')
650                         color = DIFF_FILE_OLD;
651                 else if (line[i] == '+')
652                         color = DIFF_FILE_NEW;
653         }
654
655         if (color != DIFF_FILE_NEW) {
656                 emit_line(diff_get_color(ecbdata->color_diff, color),
657                           reset, line, len);
658                 return;
659         }
660         emit_add_line(reset, ecbdata, line, len);
661 }
662
663 static char *pprint_rename(const char *a, const char *b)
664 {
665         const char *old = a;
666         const char *new = b;
667         struct strbuf name;
668         int pfx_length, sfx_length;
669         int len_a = strlen(a);
670         int len_b = strlen(b);
671         int a_midlen, b_midlen;
672         int qlen_a = quote_c_style(a, NULL, NULL, 0);
673         int qlen_b = quote_c_style(b, NULL, NULL, 0);
674
675         strbuf_init(&name, 0);
676         if (qlen_a || qlen_b) {
677                 quote_c_style(a, &name, NULL, 0);
678                 strbuf_addstr(&name, " => ");
679                 quote_c_style(b, &name, NULL, 0);
680                 return strbuf_detach(&name, NULL);
681         }
682
683         /* Find common prefix */
684         pfx_length = 0;
685         while (*old && *new && *old == *new) {
686                 if (*old == '/')
687                         pfx_length = old - a + 1;
688                 old++;
689                 new++;
690         }
691
692         /* Find common suffix */
693         old = a + len_a;
694         new = b + len_b;
695         sfx_length = 0;
696         while (a <= old && b <= new && *old == *new) {
697                 if (*old == '/')
698                         sfx_length = len_a - (old - a);
699                 old--;
700                 new--;
701         }
702
703         /*
704          * pfx{mid-a => mid-b}sfx
705          * {pfx-a => pfx-b}sfx
706          * pfx{sfx-a => sfx-b}
707          * name-a => name-b
708          */
709         a_midlen = len_a - pfx_length - sfx_length;
710         b_midlen = len_b - pfx_length - sfx_length;
711         if (a_midlen < 0)
712                 a_midlen = 0;
713         if (b_midlen < 0)
714                 b_midlen = 0;
715
716         strbuf_grow(&name, pfx_length + a_midlen + b_midlen + sfx_length + 7);
717         if (pfx_length + sfx_length) {
718                 strbuf_add(&name, a, pfx_length);
719                 strbuf_addch(&name, '{');
720         }
721         strbuf_add(&name, a + pfx_length, a_midlen);
722         strbuf_addstr(&name, " => ");
723         strbuf_add(&name, b + pfx_length, b_midlen);
724         if (pfx_length + sfx_length) {
725                 strbuf_addch(&name, '}');
726                 strbuf_add(&name, a + len_a - sfx_length, sfx_length);
727         }
728         return strbuf_detach(&name, NULL);
729 }
730
731 struct diffstat_t {
732         struct xdiff_emit_state xm;
733
734         int nr;
735         int alloc;
736         struct diffstat_file {
737                 char *from_name;
738                 char *name;
739                 char *print_name;
740                 unsigned is_unmerged:1;
741                 unsigned is_binary:1;
742                 unsigned is_renamed:1;
743                 unsigned int added, deleted;
744         } **files;
745 };
746
747 static struct diffstat_file *diffstat_add(struct diffstat_t *diffstat,
748                                           const char *name_a,
749                                           const char *name_b)
750 {
751         struct diffstat_file *x;
752         x = xcalloc(sizeof (*x), 1);
753         if (diffstat->nr == diffstat->alloc) {
754                 diffstat->alloc = alloc_nr(diffstat->alloc);
755                 diffstat->files = xrealloc(diffstat->files,
756                                 diffstat->alloc * sizeof(x));
757         }
758         diffstat->files[diffstat->nr++] = x;
759         if (name_b) {
760                 x->from_name = xstrdup(name_a);
761                 x->name = xstrdup(name_b);
762                 x->is_renamed = 1;
763         }
764         else {
765                 x->from_name = NULL;
766                 x->name = xstrdup(name_a);
767         }
768         return x;
769 }
770
771 static void diffstat_consume(void *priv, char *line, unsigned long len)
772 {
773         struct diffstat_t *diffstat = priv;
774         struct diffstat_file *x = diffstat->files[diffstat->nr - 1];
775
776         if (line[0] == '+')
777                 x->added++;
778         else if (line[0] == '-')
779                 x->deleted++;
780 }
781
782 const char mime_boundary_leader[] = "------------";
783
784 static int scale_linear(int it, int width, int max_change)
785 {
786         /*
787          * make sure that at least one '-' is printed if there were deletions,
788          * and likewise for '+'.
789          */
790         if (max_change < 2)
791                 return it;
792         return ((it - 1) * (width - 1) + max_change - 1) / (max_change - 1);
793 }
794
795 static void show_name(const char *prefix, const char *name, int len,
796                       const char *reset, const char *set)
797 {
798         printf(" %s%s%-*s%s |", set, prefix, len, name, reset);
799 }
800
801 static void show_graph(char ch, int cnt, const char *set, const char *reset)
802 {
803         if (cnt <= 0)
804                 return;
805         printf("%s", set);
806         while (cnt--)
807                 putchar(ch);
808         printf("%s", reset);
809 }
810
811 static void fill_print_name(struct diffstat_file *file)
812 {
813         char *pname;
814
815         if (file->print_name)
816                 return;
817
818         if (!file->is_renamed) {
819                 struct strbuf buf;
820                 strbuf_init(&buf, 0);
821                 if (quote_c_style(file->name, &buf, NULL, 0)) {
822                         pname = strbuf_detach(&buf, NULL);
823                 } else {
824                         pname = file->name;
825                         strbuf_release(&buf);
826                 }
827         } else {
828                 pname = pprint_rename(file->from_name, file->name);
829         }
830         file->print_name = pname;
831 }
832
833 static void show_stats(struct diffstat_t* data, struct diff_options *options)
834 {
835         int i, len, add, del, total, adds = 0, dels = 0;
836         int max_change = 0, max_len = 0;
837         int total_files = data->nr;
838         int width, name_width;
839         const char *reset, *set, *add_c, *del_c;
840
841         if (data->nr == 0)
842                 return;
843
844         width = options->stat_width ? options->stat_width : 80;
845         name_width = options->stat_name_width ? options->stat_name_width : 50;
846
847         /* Sanity: give at least 5 columns to the graph,
848          * but leave at least 10 columns for the name.
849          */
850         if (width < name_width + 15) {
851                 if (name_width <= 25)
852                         width = name_width + 15;
853                 else
854                         name_width = width - 15;
855         }
856
857         /* Find the longest filename and max number of changes */
858         reset = diff_get_color_opt(options, DIFF_RESET);
859         set   = diff_get_color_opt(options, DIFF_PLAIN);
860         add_c = diff_get_color_opt(options, DIFF_FILE_NEW);
861         del_c = diff_get_color_opt(options, DIFF_FILE_OLD);
862
863         for (i = 0; i < data->nr; i++) {
864                 struct diffstat_file *file = data->files[i];
865                 int change = file->added + file->deleted;
866                 fill_print_name(file);
867                 len = strlen(file->print_name);
868                 if (max_len < len)
869                         max_len = len;
870
871                 if (file->is_binary || file->is_unmerged)
872                         continue;
873                 if (max_change < change)
874                         max_change = change;
875         }
876
877         /* Compute the width of the graph part;
878          * 10 is for one blank at the beginning of the line plus
879          * " | count " between the name and the graph.
880          *
881          * From here on, name_width is the width of the name area,
882          * and width is the width of the graph area.
883          */
884         name_width = (name_width < max_len) ? name_width : max_len;
885         if (width < (name_width + 10) + max_change)
886                 width = width - (name_width + 10);
887         else
888                 width = max_change;
889
890         for (i = 0; i < data->nr; i++) {
891                 const char *prefix = "";
892                 char *name = data->files[i]->print_name;
893                 int added = data->files[i]->added;
894                 int deleted = data->files[i]->deleted;
895                 int name_len;
896
897                 /*
898                  * "scale" the filename
899                  */
900                 len = name_width;
901                 name_len = strlen(name);
902                 if (name_width < name_len) {
903                         char *slash;
904                         prefix = "...";
905                         len -= 3;
906                         name += name_len - len;
907                         slash = strchr(name, '/');
908                         if (slash)
909                                 name = slash;
910                 }
911
912                 if (data->files[i]->is_binary) {
913                         show_name(prefix, name, len, reset, set);
914                         printf("  Bin ");
915                         printf("%s%d%s", del_c, deleted, reset);
916                         printf(" -> ");
917                         printf("%s%d%s", add_c, added, reset);
918                         printf(" bytes");
919                         printf("\n");
920                         continue;
921                 }
922                 else if (data->files[i]->is_unmerged) {
923                         show_name(prefix, name, len, reset, set);
924                         printf("  Unmerged\n");
925                         continue;
926                 }
927                 else if (!data->files[i]->is_renamed &&
928                          (added + deleted == 0)) {
929                         total_files--;
930                         continue;
931                 }
932
933                 /*
934                  * scale the add/delete
935                  */
936                 add = added;
937                 del = deleted;
938                 total = add + del;
939                 adds += add;
940                 dels += del;
941
942                 if (width <= max_change) {
943                         add = scale_linear(add, width, max_change);
944                         del = scale_linear(del, width, max_change);
945                         total = add + del;
946                 }
947                 show_name(prefix, name, len, reset, set);
948                 printf("%5d ", added + deleted);
949                 show_graph('+', add, add_c, reset);
950                 show_graph('-', del, del_c, reset);
951                 putchar('\n');
952         }
953         printf("%s %d files changed, %d insertions(+), %d deletions(-)%s\n",
954                set, total_files, adds, dels, reset);
955 }
956
957 static void show_shortstats(struct diffstat_t* data)
958 {
959         int i, adds = 0, dels = 0, total_files = data->nr;
960
961         if (data->nr == 0)
962                 return;
963
964         for (i = 0; i < data->nr; i++) {
965                 if (!data->files[i]->is_binary &&
966                     !data->files[i]->is_unmerged) {
967                         int added = data->files[i]->added;
968                         int deleted= data->files[i]->deleted;
969                         if (!data->files[i]->is_renamed &&
970                             (added + deleted == 0)) {
971                                 total_files--;
972                         } else {
973                                 adds += added;
974                                 dels += deleted;
975                         }
976                 }
977         }
978         printf(" %d files changed, %d insertions(+), %d deletions(-)\n",
979                total_files, adds, dels);
980 }
981
982 static void show_numstat(struct diffstat_t* data, struct diff_options *options)
983 {
984         int i;
985
986         if (data->nr == 0)
987                 return;
988
989         for (i = 0; i < data->nr; i++) {
990                 struct diffstat_file *file = data->files[i];
991
992                 if (file->is_binary)
993                         printf("-\t-\t");
994                 else
995                         printf("%d\t%d\t", file->added, file->deleted);
996                 if (options->line_termination) {
997                         fill_print_name(file);
998                         if (!file->is_renamed)
999                                 write_name_quoted(file->name, stdout,
1000                                                   options->line_termination);
1001                         else {
1002                                 fputs(file->print_name, stdout);
1003                                 putchar(options->line_termination);
1004                         }
1005                 } else {
1006                         if (file->is_renamed) {
1007                                 putchar('\0');
1008                                 write_name_quoted(file->from_name, stdout, '\0');
1009                         }
1010                         write_name_quoted(file->name, stdout, '\0');
1011                 }
1012         }
1013 }
1014
1015 static void free_diffstat_info(struct diffstat_t *diffstat)
1016 {
1017         int i;
1018         for (i = 0; i < diffstat->nr; i++) {
1019                 struct diffstat_file *f = diffstat->files[i];
1020                 if (f->name != f->print_name)
1021                         free(f->print_name);
1022                 free(f->name);
1023                 free(f->from_name);
1024                 free(f);
1025         }
1026         free(diffstat->files);
1027 }
1028
1029 struct checkdiff_t {
1030         struct xdiff_emit_state xm;
1031         const char *filename;
1032         int lineno, color_diff;
1033         unsigned ws_rule;
1034 };
1035
1036 static void checkdiff_consume(void *priv, char *line, unsigned long len)
1037 {
1038         struct checkdiff_t *data = priv;
1039         const char *ws = diff_get_color(data->color_diff, DIFF_WHITESPACE);
1040         const char *reset = diff_get_color(data->color_diff, DIFF_RESET);
1041         const char *set = diff_get_color(data->color_diff, DIFF_FILE_NEW);
1042
1043         if (line[0] == '+') {
1044                 int i, spaces = 0, space_before_tab = 0, white_space_at_end = 0;
1045
1046                 /* check space before tab */
1047                 for (i = 1; i < len; i++) {
1048                         if (line[i] == ' ')
1049                                 spaces++;
1050                         else if (line[i] == '\t') {
1051                                 if (spaces) {
1052                                         space_before_tab = 1;
1053                                         break;
1054                                 }
1055                         }
1056                         else
1057                                 break;
1058                 }
1059
1060                 /* check whitespace at line end */
1061                 if (line[len - 1] == '\n')
1062                         len--;
1063                 if (isspace(line[len - 1]))
1064                         white_space_at_end = 1;
1065
1066                 if (space_before_tab || white_space_at_end) {
1067                         printf("%s:%d: %s", data->filename, data->lineno, ws);
1068                         if (space_before_tab) {
1069                                 printf("space before tab");
1070                                 if (white_space_at_end)
1071                                         putchar(',');
1072                         }
1073                         if (white_space_at_end)
1074                                 printf("whitespace at end");
1075                         printf(":%s ", reset);
1076                         emit_line_with_ws(1, set, reset, ws, line, len,
1077                                           data->ws_rule);
1078                 }
1079
1080                 data->lineno++;
1081         } else if (line[0] == ' ')
1082                 data->lineno++;
1083         else if (line[0] == '@') {
1084                 char *plus = strchr(line, '+');
1085                 if (plus)
1086                         data->lineno = strtol(plus, NULL, 10);
1087                 else
1088                         die("invalid diff");
1089         }
1090 }
1091
1092 static unsigned char *deflate_it(char *data,
1093                                  unsigned long size,
1094                                  unsigned long *result_size)
1095 {
1096         int bound;
1097         unsigned char *deflated;
1098         z_stream stream;
1099
1100         memset(&stream, 0, sizeof(stream));
1101         deflateInit(&stream, zlib_compression_level);
1102         bound = deflateBound(&stream, size);
1103         deflated = xmalloc(bound);
1104         stream.next_out = deflated;
1105         stream.avail_out = bound;
1106
1107         stream.next_in = (unsigned char *)data;
1108         stream.avail_in = size;
1109         while (deflate(&stream, Z_FINISH) == Z_OK)
1110                 ; /* nothing */
1111         deflateEnd(&stream);
1112         *result_size = stream.total_out;
1113         return deflated;
1114 }
1115
1116 static void emit_binary_diff_body(mmfile_t *one, mmfile_t *two)
1117 {
1118         void *cp;
1119         void *delta;
1120         void *deflated;
1121         void *data;
1122         unsigned long orig_size;
1123         unsigned long delta_size;
1124         unsigned long deflate_size;
1125         unsigned long data_size;
1126
1127         /* We could do deflated delta, or we could do just deflated two,
1128          * whichever is smaller.
1129          */
1130         delta = NULL;
1131         deflated = deflate_it(two->ptr, two->size, &deflate_size);
1132         if (one->size && two->size) {
1133                 delta = diff_delta(one->ptr, one->size,
1134                                    two->ptr, two->size,
1135                                    &delta_size, deflate_size);
1136                 if (delta) {
1137                         void *to_free = delta;
1138                         orig_size = delta_size;
1139                         delta = deflate_it(delta, delta_size, &delta_size);
1140                         free(to_free);
1141                 }
1142         }
1143
1144         if (delta && delta_size < deflate_size) {
1145                 printf("delta %lu\n", orig_size);
1146                 free(deflated);
1147                 data = delta;
1148                 data_size = delta_size;
1149         }
1150         else {
1151                 printf("literal %lu\n", two->size);
1152                 free(delta);
1153                 data = deflated;
1154                 data_size = deflate_size;
1155         }
1156
1157         /* emit data encoded in base85 */
1158         cp = data;
1159         while (data_size) {
1160                 int bytes = (52 < data_size) ? 52 : data_size;
1161                 char line[70];
1162                 data_size -= bytes;
1163                 if (bytes <= 26)
1164                         line[0] = bytes + 'A' - 1;
1165                 else
1166                         line[0] = bytes - 26 + 'a' - 1;
1167                 encode_85(line + 1, cp, bytes);
1168                 cp = (char *) cp + bytes;
1169                 puts(line);
1170         }
1171         printf("\n");
1172         free(data);
1173 }
1174
1175 static void emit_binary_diff(mmfile_t *one, mmfile_t *two)
1176 {
1177         printf("GIT binary patch\n");
1178         emit_binary_diff_body(one, two);
1179         emit_binary_diff_body(two, one);
1180 }
1181
1182 static void setup_diff_attr_check(struct git_attr_check *check)
1183 {
1184         static struct git_attr *attr_diff;
1185
1186         if (!attr_diff) {
1187                 attr_diff = git_attr("diff", 4);
1188         }
1189         check[0].attr = attr_diff;
1190 }
1191
1192 static void diff_filespec_check_attr(struct diff_filespec *one)
1193 {
1194         struct git_attr_check attr_diff_check;
1195         int check_from_data = 0;
1196
1197         if (one->checked_attr)
1198                 return;
1199
1200         setup_diff_attr_check(&attr_diff_check);
1201         one->is_binary = 0;
1202         one->funcname_pattern_ident = NULL;
1203
1204         if (!git_checkattr(one->path, 1, &attr_diff_check)) {
1205                 const char *value;
1206
1207                 /* binaryness */
1208                 value = attr_diff_check.value;
1209                 if (ATTR_TRUE(value))
1210                         ;
1211                 else if (ATTR_FALSE(value))
1212                         one->is_binary = 1;
1213                 else
1214                         check_from_data = 1;
1215
1216                 /* funcname pattern ident */
1217                 if (ATTR_TRUE(value) || ATTR_FALSE(value) || ATTR_UNSET(value))
1218                         ;
1219                 else
1220                         one->funcname_pattern_ident = value;
1221         }
1222
1223         if (check_from_data) {
1224                 if (!one->data && DIFF_FILE_VALID(one))
1225                         diff_populate_filespec(one, 0);
1226
1227                 if (one->data)
1228                         one->is_binary = buffer_is_binary(one->data, one->size);
1229         }
1230 }
1231
1232 int diff_filespec_is_binary(struct diff_filespec *one)
1233 {
1234         diff_filespec_check_attr(one);
1235         return one->is_binary;
1236 }
1237
1238 static const char *funcname_pattern(const char *ident)
1239 {
1240         struct funcname_pattern *pp;
1241
1242         read_config_if_needed();
1243         for (pp = funcname_pattern_list; pp; pp = pp->next)
1244                 if (!strcmp(ident, pp->name))
1245                         return pp->pattern;
1246         return NULL;
1247 }
1248
1249 static struct builtin_funcname_pattern {
1250         const char *name;
1251         const char *pattern;
1252 } builtin_funcname_pattern[] = {
1253         { "java", "!^[  ]*\\(catch\\|do\\|for\\|if\\|instanceof\\|"
1254                         "new\\|return\\|switch\\|throw\\|while\\)\n"
1255                         "^[     ]*\\(\\([       ]*"
1256                         "[A-Za-z_][A-Za-z_0-9]*\\)\\{2,\\}"
1257                         "[      ]*([^;]*$\\)" },
1258         { "tex", "^\\(\\\\\\(sub\\)*section{.*\\)$" },
1259 };
1260
1261 static const char *diff_funcname_pattern(struct diff_filespec *one)
1262 {
1263         const char *ident, *pattern;
1264         int i;
1265
1266         diff_filespec_check_attr(one);
1267         ident = one->funcname_pattern_ident;
1268
1269         if (!ident)
1270                 /*
1271                  * If the config file has "funcname.default" defined, that
1272                  * regexp is used; otherwise NULL is returned and xemit uses
1273                  * the built-in default.
1274                  */
1275                 return funcname_pattern("default");
1276
1277         /* Look up custom "funcname.$ident" regexp from config. */
1278         pattern = funcname_pattern(ident);
1279         if (pattern)
1280                 return pattern;
1281
1282         /*
1283          * And define built-in fallback patterns here.  Note that
1284          * these can be overriden by the user's config settings.
1285          */
1286         for (i = 0; i < ARRAY_SIZE(builtin_funcname_pattern); i++)
1287                 if (!strcmp(ident, builtin_funcname_pattern[i].name))
1288                         return builtin_funcname_pattern[i].pattern;
1289
1290         return NULL;
1291 }
1292
1293 static void builtin_diff(const char *name_a,
1294                          const char *name_b,
1295                          struct diff_filespec *one,
1296                          struct diff_filespec *two,
1297                          const char *xfrm_msg,
1298                          struct diff_options *o,
1299                          int complete_rewrite)
1300 {
1301         mmfile_t mf1, mf2;
1302         const char *lbl[2];
1303         char *a_one, *b_two;
1304         const char *set = diff_get_color_opt(o, DIFF_METAINFO);
1305         const char *reset = diff_get_color_opt(o, DIFF_RESET);
1306
1307         a_one = quote_two("a/", name_a + (*name_a == '/'));
1308         b_two = quote_two("b/", name_b + (*name_b == '/'));
1309         lbl[0] = DIFF_FILE_VALID(one) ? a_one : "/dev/null";
1310         lbl[1] = DIFF_FILE_VALID(two) ? b_two : "/dev/null";
1311         printf("%sdiff --git %s %s%s\n", set, a_one, b_two, reset);
1312         if (lbl[0][0] == '/') {
1313                 /* /dev/null */
1314                 printf("%snew file mode %06o%s\n", set, two->mode, reset);
1315                 if (xfrm_msg && xfrm_msg[0])
1316                         printf("%s%s%s\n", set, xfrm_msg, reset);
1317         }
1318         else if (lbl[1][0] == '/') {
1319                 printf("%sdeleted file mode %06o%s\n", set, one->mode, reset);
1320                 if (xfrm_msg && xfrm_msg[0])
1321                         printf("%s%s%s\n", set, xfrm_msg, reset);
1322         }
1323         else {
1324                 if (one->mode != two->mode) {
1325                         printf("%sold mode %06o%s\n", set, one->mode, reset);
1326                         printf("%snew mode %06o%s\n", set, two->mode, reset);
1327                 }
1328                 if (xfrm_msg && xfrm_msg[0])
1329                         printf("%s%s%s\n", set, xfrm_msg, reset);
1330                 /*
1331                  * we do not run diff between different kind
1332                  * of objects.
1333                  */
1334                 if ((one->mode ^ two->mode) & S_IFMT)
1335                         goto free_ab_and_return;
1336                 if (complete_rewrite) {
1337                         emit_rewrite_diff(name_a, name_b, one, two,
1338                                         DIFF_OPT_TST(o, COLOR_DIFF));
1339                         o->found_changes = 1;
1340                         goto free_ab_and_return;
1341                 }
1342         }
1343
1344         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1345                 die("unable to read files to diff");
1346
1347         if (!DIFF_OPT_TST(o, TEXT) &&
1348             (diff_filespec_is_binary(one) || diff_filespec_is_binary(two))) {
1349                 /* Quite common confusing case */
1350                 if (mf1.size == mf2.size &&
1351                     !memcmp(mf1.ptr, mf2.ptr, mf1.size))
1352                         goto free_ab_and_return;
1353                 if (DIFF_OPT_TST(o, BINARY))
1354                         emit_binary_diff(&mf1, &mf2);
1355                 else
1356                         printf("Binary files %s and %s differ\n",
1357                                lbl[0], lbl[1]);
1358                 o->found_changes = 1;
1359         }
1360         else {
1361                 /* Crazy xdl interfaces.. */
1362                 const char *diffopts = getenv("GIT_DIFF_OPTS");
1363                 xpparam_t xpp;
1364                 xdemitconf_t xecfg;
1365                 xdemitcb_t ecb;
1366                 struct emit_callback ecbdata;
1367                 const char *funcname_pattern;
1368
1369                 funcname_pattern = diff_funcname_pattern(one);
1370                 if (!funcname_pattern)
1371                         funcname_pattern = diff_funcname_pattern(two);
1372
1373                 memset(&xecfg, 0, sizeof(xecfg));
1374                 memset(&ecbdata, 0, sizeof(ecbdata));
1375                 ecbdata.label_path = lbl;
1376                 ecbdata.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1377                 ecbdata.found_changesp = &o->found_changes;
1378                 ecbdata.ws_rule = whitespace_rule(name_b ? name_b : name_a);
1379                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1380                 xecfg.ctxlen = o->context;
1381                 xecfg.flags = XDL_EMIT_FUNCNAMES;
1382                 if (funcname_pattern)
1383                         xdiff_set_find_func(&xecfg, funcname_pattern);
1384                 if (!diffopts)
1385                         ;
1386                 else if (!prefixcmp(diffopts, "--unified="))
1387                         xecfg.ctxlen = strtoul(diffopts + 10, NULL, 10);
1388                 else if (!prefixcmp(diffopts, "-u"))
1389                         xecfg.ctxlen = strtoul(diffopts + 2, NULL, 10);
1390                 ecb.outf = xdiff_outf;
1391                 ecb.priv = &ecbdata;
1392                 ecbdata.xm.consume = fn_out_consume;
1393                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1394                         ecbdata.diff_words =
1395                                 xcalloc(1, sizeof(struct diff_words_data));
1396                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1397                 if (DIFF_OPT_TST(o, COLOR_DIFF_WORDS))
1398                         free_diff_words_data(&ecbdata);
1399         }
1400
1401  free_ab_and_return:
1402         diff_free_filespec_data(one);
1403         diff_free_filespec_data(two);
1404         free(a_one);
1405         free(b_two);
1406         return;
1407 }
1408
1409 static void builtin_diffstat(const char *name_a, const char *name_b,
1410                              struct diff_filespec *one,
1411                              struct diff_filespec *two,
1412                              struct diffstat_t *diffstat,
1413                              struct diff_options *o,
1414                              int complete_rewrite)
1415 {
1416         mmfile_t mf1, mf2;
1417         struct diffstat_file *data;
1418
1419         data = diffstat_add(diffstat, name_a, name_b);
1420
1421         if (!one || !two) {
1422                 data->is_unmerged = 1;
1423                 return;
1424         }
1425         if (complete_rewrite) {
1426                 diff_populate_filespec(one, 0);
1427                 diff_populate_filespec(two, 0);
1428                 data->deleted = count_lines(one->data, one->size);
1429                 data->added = count_lines(two->data, two->size);
1430                 goto free_and_return;
1431         }
1432         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1433                 die("unable to read files to diff");
1434
1435         if (diff_filespec_is_binary(one) || diff_filespec_is_binary(two)) {
1436                 data->is_binary = 1;
1437                 data->added = mf2.size;
1438                 data->deleted = mf1.size;
1439         } else {
1440                 /* Crazy xdl interfaces.. */
1441                 xpparam_t xpp;
1442                 xdemitconf_t xecfg;
1443                 xdemitcb_t ecb;
1444
1445                 memset(&xecfg, 0, sizeof(xecfg));
1446                 xpp.flags = XDF_NEED_MINIMAL | o->xdl_opts;
1447                 ecb.outf = xdiff_outf;
1448                 ecb.priv = diffstat;
1449                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1450         }
1451
1452  free_and_return:
1453         diff_free_filespec_data(one);
1454         diff_free_filespec_data(two);
1455 }
1456
1457 static void builtin_checkdiff(const char *name_a, const char *name_b,
1458                              struct diff_filespec *one,
1459                              struct diff_filespec *two, struct diff_options *o)
1460 {
1461         mmfile_t mf1, mf2;
1462         struct checkdiff_t data;
1463
1464         if (!two)
1465                 return;
1466
1467         memset(&data, 0, sizeof(data));
1468         data.xm.consume = checkdiff_consume;
1469         data.filename = name_b ? name_b : name_a;
1470         data.lineno = 0;
1471         data.color_diff = DIFF_OPT_TST(o, COLOR_DIFF);
1472         data.ws_rule = whitespace_rule(data.filename);
1473
1474         if (fill_mmfile(&mf1, one) < 0 || fill_mmfile(&mf2, two) < 0)
1475                 die("unable to read files to diff");
1476
1477         if (diff_filespec_is_binary(two))
1478                 goto free_and_return;
1479         else {
1480                 /* Crazy xdl interfaces.. */
1481                 xpparam_t xpp;
1482                 xdemitconf_t xecfg;
1483                 xdemitcb_t ecb;
1484
1485                 memset(&xecfg, 0, sizeof(xecfg));
1486                 xpp.flags = XDF_NEED_MINIMAL;
1487                 ecb.outf = xdiff_outf;
1488                 ecb.priv = &data;
1489                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
1490         }
1491  free_and_return:
1492         diff_free_filespec_data(one);
1493         diff_free_filespec_data(two);
1494 }
1495
1496 struct diff_filespec *alloc_filespec(const char *path)
1497 {
1498         int namelen = strlen(path);
1499         struct diff_filespec *spec = xmalloc(sizeof(*spec) + namelen + 1);
1500
1501         memset(spec, 0, sizeof(*spec));
1502         spec->path = (char *)(spec + 1);
1503         memcpy(spec->path, path, namelen+1);
1504         spec->count = 1;
1505         return spec;
1506 }
1507
1508 void free_filespec(struct diff_filespec *spec)
1509 {
1510         if (!--spec->count) {
1511                 diff_free_filespec_data(spec);
1512                 free(spec);
1513         }
1514 }
1515
1516 void fill_filespec(struct diff_filespec *spec, const unsigned char *sha1,
1517                    unsigned short mode)
1518 {
1519         if (mode) {
1520                 spec->mode = canon_mode(mode);
1521                 hashcpy(spec->sha1, sha1);
1522                 spec->sha1_valid = !is_null_sha1(sha1);
1523         }
1524 }
1525
1526 /*
1527  * Given a name and sha1 pair, if the index tells us the file in
1528  * the work tree has that object contents, return true, so that
1529  * prepare_temp_file() does not have to inflate and extract.
1530  */
1531 static int reuse_worktree_file(const char *name, const unsigned char *sha1, int want_file)
1532 {
1533         struct cache_entry *ce;
1534         struct stat st;
1535         int pos, len;
1536
1537         /* We do not read the cache ourselves here, because the
1538          * benchmark with my previous version that always reads cache
1539          * shows that it makes things worse for diff-tree comparing
1540          * two linux-2.6 kernel trees in an already checked out work
1541          * tree.  This is because most diff-tree comparisons deal with
1542          * only a small number of files, while reading the cache is
1543          * expensive for a large project, and its cost outweighs the
1544          * savings we get by not inflating the object to a temporary
1545          * file.  Practically, this code only helps when we are used
1546          * by diff-cache --cached, which does read the cache before
1547          * calling us.
1548          */
1549         if (!active_cache)
1550                 return 0;
1551
1552         /* We want to avoid the working directory if our caller
1553          * doesn't need the data in a normal file, this system
1554          * is rather slow with its stat/open/mmap/close syscalls,
1555          * and the object is contained in a pack file.  The pack
1556          * is probably already open and will be faster to obtain
1557          * the data through than the working directory.  Loose
1558          * objects however would tend to be slower as they need
1559          * to be individually opened and inflated.
1560          */
1561         if (!FAST_WORKING_DIRECTORY && !want_file && has_sha1_pack(sha1, NULL))
1562                 return 0;
1563
1564         len = strlen(name);
1565         pos = cache_name_pos(name, len);
1566         if (pos < 0)
1567                 return 0;
1568         ce = active_cache[pos];
1569         if ((lstat(name, &st) < 0) ||
1570             !S_ISREG(st.st_mode) || /* careful! */
1571             ce_match_stat(ce, &st, 0) ||
1572             hashcmp(sha1, ce->sha1))
1573                 return 0;
1574         /* we return 1 only when we can stat, it is a regular file,
1575          * stat information matches, and sha1 recorded in the cache
1576          * matches.  I.e. we know the file in the work tree really is
1577          * the same as the <name, sha1> pair.
1578          */
1579         return 1;
1580 }
1581
1582 static int populate_from_stdin(struct diff_filespec *s)
1583 {
1584         struct strbuf buf;
1585         size_t size = 0;
1586
1587         strbuf_init(&buf, 0);
1588         if (strbuf_read(&buf, 0, 0) < 0)
1589                 return error("error while reading from stdin %s",
1590                                      strerror(errno));
1591
1592         s->should_munmap = 0;
1593         s->data = strbuf_detach(&buf, &size);
1594         s->size = size;
1595         s->should_free = 1;
1596         return 0;
1597 }
1598
1599 static int diff_populate_gitlink(struct diff_filespec *s, int size_only)
1600 {
1601         int len;
1602         char *data = xmalloc(100);
1603         len = snprintf(data, 100,
1604                 "Subproject commit %s\n", sha1_to_hex(s->sha1));
1605         s->data = data;
1606         s->size = len;
1607         s->should_free = 1;
1608         if (size_only) {
1609                 s->data = NULL;
1610                 free(data);
1611         }
1612         return 0;
1613 }
1614
1615 /*
1616  * While doing rename detection and pickaxe operation, we may need to
1617  * grab the data for the blob (or file) for our own in-core comparison.
1618  * diff_filespec has data and size fields for this purpose.
1619  */
1620 int diff_populate_filespec(struct diff_filespec *s, int size_only)
1621 {
1622         int err = 0;
1623         if (!DIFF_FILE_VALID(s))
1624                 die("internal error: asking to populate invalid file.");
1625         if (S_ISDIR(s->mode))
1626                 return -1;
1627
1628         if (s->data)
1629                 return 0;
1630
1631         if (size_only && 0 < s->size)
1632                 return 0;
1633
1634         if (S_ISGITLINK(s->mode))
1635                 return diff_populate_gitlink(s, size_only);
1636
1637         if (!s->sha1_valid ||
1638             reuse_worktree_file(s->path, s->sha1, 0)) {
1639                 struct strbuf buf;
1640                 struct stat st;
1641                 int fd;
1642
1643                 if (!strcmp(s->path, "-"))
1644                         return populate_from_stdin(s);
1645
1646                 if (lstat(s->path, &st) < 0) {
1647                         if (errno == ENOENT) {
1648                         err_empty:
1649                                 err = -1;
1650                         empty:
1651                                 s->data = (char *)"";
1652                                 s->size = 0;
1653                                 return err;
1654                         }
1655                 }
1656                 s->size = xsize_t(st.st_size);
1657                 if (!s->size)
1658                         goto empty;
1659                 if (size_only)
1660                         return 0;
1661                 if (S_ISLNK(st.st_mode)) {
1662                         int ret;
1663                         s->data = xmalloc(s->size);
1664                         s->should_free = 1;
1665                         ret = readlink(s->path, s->data, s->size);
1666                         if (ret < 0) {
1667                                 free(s->data);
1668                                 goto err_empty;
1669                         }
1670                         return 0;
1671                 }
1672                 fd = open(s->path, O_RDONLY);
1673                 if (fd < 0)
1674                         goto err_empty;
1675                 s->data = xmmap(NULL, s->size, PROT_READ, MAP_PRIVATE, fd, 0);
1676                 close(fd);
1677                 s->should_munmap = 1;
1678
1679                 /*
1680                  * Convert from working tree format to canonical git format
1681                  */
1682                 strbuf_init(&buf, 0);
1683                 if (convert_to_git(s->path, s->data, s->size, &buf)) {
1684                         size_t size = 0;
1685                         munmap(s->data, s->size);
1686                         s->should_munmap = 0;
1687                         s->data = strbuf_detach(&buf, &size);
1688                         s->size = size;
1689                         s->should_free = 1;
1690                 }
1691         }
1692         else {
1693                 enum object_type type;
1694                 if (size_only)
1695                         type = sha1_object_info(s->sha1, &s->size);
1696                 else {
1697                         s->data = read_sha1_file(s->sha1, &type, &s->size);
1698                         s->should_free = 1;
1699                 }
1700         }
1701         return 0;
1702 }
1703
1704 void diff_free_filespec_blob(struct diff_filespec *s)
1705 {
1706         if (s->should_free)
1707                 free(s->data);
1708         else if (s->should_munmap)
1709                 munmap(s->data, s->size);
1710
1711         if (s->should_free || s->should_munmap) {
1712                 s->should_free = s->should_munmap = 0;
1713                 s->data = NULL;
1714         }
1715 }
1716
1717 void diff_free_filespec_data(struct diff_filespec *s)
1718 {
1719         diff_free_filespec_blob(s);
1720         free(s->cnt_data);
1721         s->cnt_data = NULL;
1722 }
1723
1724 static void prep_temp_blob(struct diff_tempfile *temp,
1725                            void *blob,
1726                            unsigned long size,
1727                            const unsigned char *sha1,
1728                            int mode)
1729 {
1730         int fd;
1731
1732         fd = git_mkstemp(temp->tmp_path, PATH_MAX, ".diff_XXXXXX");
1733         if (fd < 0)
1734                 die("unable to create temp-file: %s", strerror(errno));
1735         if (write_in_full(fd, blob, size) != size)
1736                 die("unable to write temp-file");
1737         close(fd);
1738         temp->name = temp->tmp_path;
1739         strcpy(temp->hex, sha1_to_hex(sha1));
1740         temp->hex[40] = 0;
1741         sprintf(temp->mode, "%06o", mode);
1742 }
1743
1744 static void prepare_temp_file(const char *name,
1745                               struct diff_tempfile *temp,
1746                               struct diff_filespec *one)
1747 {
1748         if (!DIFF_FILE_VALID(one)) {
1749         not_a_valid_file:
1750                 /* A '-' entry produces this for file-2, and
1751                  * a '+' entry produces this for file-1.
1752                  */
1753                 temp->name = "/dev/null";
1754                 strcpy(temp->hex, ".");
1755                 strcpy(temp->mode, ".");
1756                 return;
1757         }
1758
1759         if (!one->sha1_valid ||
1760             reuse_worktree_file(name, one->sha1, 1)) {
1761                 struct stat st;
1762                 if (lstat(name, &st) < 0) {
1763                         if (errno == ENOENT)
1764                                 goto not_a_valid_file;
1765                         die("stat(%s): %s", name, strerror(errno));
1766                 }
1767                 if (S_ISLNK(st.st_mode)) {
1768                         int ret;
1769                         char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
1770                         size_t sz = xsize_t(st.st_size);
1771                         if (sizeof(buf) <= st.st_size)
1772                                 die("symlink too long: %s", name);
1773                         ret = readlink(name, buf, sz);
1774                         if (ret < 0)
1775                                 die("readlink(%s)", name);
1776                         prep_temp_blob(temp, buf, sz,
1777                                        (one->sha1_valid ?
1778                                         one->sha1 : null_sha1),
1779                                        (one->sha1_valid ?
1780                                         one->mode : S_IFLNK));
1781                 }
1782                 else {
1783                         /* we can borrow from the file in the work tree */
1784                         temp->name = name;
1785                         if (!one->sha1_valid)
1786                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
1787                         else
1788                                 strcpy(temp->hex, sha1_to_hex(one->sha1));
1789                         /* Even though we may sometimes borrow the
1790                          * contents from the work tree, we always want
1791                          * one->mode.  mode is trustworthy even when
1792                          * !(one->sha1_valid), as long as
1793                          * DIFF_FILE_VALID(one).
1794                          */
1795                         sprintf(temp->mode, "%06o", one->mode);
1796                 }
1797                 return;
1798         }
1799         else {
1800                 if (diff_populate_filespec(one, 0))
1801                         die("cannot read data blob for %s", one->path);
1802                 prep_temp_blob(temp, one->data, one->size,
1803                                one->sha1, one->mode);
1804         }
1805 }
1806
1807 static void remove_tempfile(void)
1808 {
1809         int i;
1810
1811         for (i = 0; i < 2; i++)
1812                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
1813                         unlink(diff_temp[i].name);
1814                         diff_temp[i].name = NULL;
1815                 }
1816 }
1817
1818 static void remove_tempfile_on_signal(int signo)
1819 {
1820         remove_tempfile();
1821         signal(SIGINT, SIG_DFL);
1822         raise(signo);
1823 }
1824
1825 /* An external diff command takes:
1826  *
1827  * diff-cmd name infile1 infile1-sha1 infile1-mode \
1828  *               infile2 infile2-sha1 infile2-mode [ rename-to ]
1829  *
1830  */
1831 static void run_external_diff(const char *pgm,
1832                               const char *name,
1833                               const char *other,
1834                               struct diff_filespec *one,
1835                               struct diff_filespec *two,
1836                               const char *xfrm_msg,
1837                               int complete_rewrite)
1838 {
1839         const char *spawn_arg[10];
1840         struct diff_tempfile *temp = diff_temp;
1841         int retval;
1842         static int atexit_asked = 0;
1843         const char *othername;
1844         const char **arg = &spawn_arg[0];
1845
1846         othername = (other? other : name);
1847         if (one && two) {
1848                 prepare_temp_file(name, &temp[0], one);
1849                 prepare_temp_file(othername, &temp[1], two);
1850                 if (! atexit_asked &&
1851                     (temp[0].name == temp[0].tmp_path ||
1852                      temp[1].name == temp[1].tmp_path)) {
1853                         atexit_asked = 1;
1854                         atexit(remove_tempfile);
1855                 }
1856                 signal(SIGINT, remove_tempfile_on_signal);
1857         }
1858
1859         if (one && two) {
1860                 *arg++ = pgm;
1861                 *arg++ = name;
1862                 *arg++ = temp[0].name;
1863                 *arg++ = temp[0].hex;
1864                 *arg++ = temp[0].mode;
1865                 *arg++ = temp[1].name;
1866                 *arg++ = temp[1].hex;
1867                 *arg++ = temp[1].mode;
1868                 if (other) {
1869                         *arg++ = other;
1870                         *arg++ = xfrm_msg;
1871                 }
1872         } else {
1873                 *arg++ = pgm;
1874                 *arg++ = name;
1875         }
1876         *arg = NULL;
1877         fflush(NULL);
1878         retval = run_command_v_opt(spawn_arg, 0);
1879         remove_tempfile();
1880         if (retval) {
1881                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
1882                 exit(1);
1883         }
1884 }
1885
1886 static const char *external_diff_attr(const char *name)
1887 {
1888         struct git_attr_check attr_diff_check;
1889
1890         setup_diff_attr_check(&attr_diff_check);
1891         if (!git_checkattr(name, 1, &attr_diff_check)) {
1892                 const char *value = attr_diff_check.value;
1893                 if (!ATTR_TRUE(value) &&
1894                     !ATTR_FALSE(value) &&
1895                     !ATTR_UNSET(value)) {
1896                         struct ll_diff_driver *drv;
1897
1898                         read_config_if_needed();
1899                         for (drv = user_diff; drv; drv = drv->next)
1900                                 if (!strcmp(drv->name, value))
1901                                         return drv->cmd;
1902                 }
1903         }
1904         return NULL;
1905 }
1906
1907 static void run_diff_cmd(const char *pgm,
1908                          const char *name,
1909                          const char *other,
1910                          struct diff_filespec *one,
1911                          struct diff_filespec *two,
1912                          const char *xfrm_msg,
1913                          struct diff_options *o,
1914                          int complete_rewrite)
1915 {
1916         if (!DIFF_OPT_TST(o, ALLOW_EXTERNAL))
1917                 pgm = NULL;
1918         else {
1919                 const char *cmd = external_diff_attr(name);
1920                 if (cmd)
1921                         pgm = cmd;
1922         }
1923
1924         if (pgm) {
1925                 run_external_diff(pgm, name, other, one, two, xfrm_msg,
1926                                   complete_rewrite);
1927                 return;
1928         }
1929         if (one && two)
1930                 builtin_diff(name, other ? other : name,
1931                              one, two, xfrm_msg, o, complete_rewrite);
1932         else
1933                 printf("* Unmerged path %s\n", name);
1934 }
1935
1936 static void diff_fill_sha1_info(struct diff_filespec *one)
1937 {
1938         if (DIFF_FILE_VALID(one)) {
1939                 if (!one->sha1_valid) {
1940                         struct stat st;
1941                         if (!strcmp(one->path, "-")) {
1942                                 hashcpy(one->sha1, null_sha1);
1943                                 return;
1944                         }
1945                         if (lstat(one->path, &st) < 0)
1946                                 die("stat %s", one->path);
1947                         if (index_path(one->sha1, one->path, &st, 0))
1948                                 die("cannot hash %s\n", one->path);
1949                 }
1950         }
1951         else
1952                 hashclr(one->sha1);
1953 }
1954
1955 static int similarity_index(struct diff_filepair *p)
1956 {
1957         return p->score * 100 / MAX_SCORE;
1958 }
1959
1960 static void run_diff(struct diff_filepair *p, struct diff_options *o)
1961 {
1962         const char *pgm = external_diff();
1963         struct strbuf msg;
1964         char *xfrm_msg;
1965         struct diff_filespec *one = p->one;
1966         struct diff_filespec *two = p->two;
1967         const char *name;
1968         const char *other;
1969         int complete_rewrite = 0;
1970
1971
1972         if (DIFF_PAIR_UNMERGED(p)) {
1973                 run_diff_cmd(pgm, p->one->path, NULL, NULL, NULL, NULL, o, 0);
1974                 return;
1975         }
1976
1977         name  = p->one->path;
1978         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
1979         diff_fill_sha1_info(one);
1980         diff_fill_sha1_info(two);
1981
1982         strbuf_init(&msg, PATH_MAX * 2 + 300);
1983         switch (p->status) {
1984         case DIFF_STATUS_COPIED:
1985                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1986                 strbuf_addstr(&msg, "\ncopy from ");
1987                 quote_c_style(name, &msg, NULL, 0);
1988                 strbuf_addstr(&msg, "\ncopy to ");
1989                 quote_c_style(other, &msg, NULL, 0);
1990                 strbuf_addch(&msg, '\n');
1991                 break;
1992         case DIFF_STATUS_RENAMED:
1993                 strbuf_addf(&msg, "similarity index %d%%", similarity_index(p));
1994                 strbuf_addstr(&msg, "\nrename from ");
1995                 quote_c_style(name, &msg, NULL, 0);
1996                 strbuf_addstr(&msg, "\nrename to ");
1997                 quote_c_style(other, &msg, NULL, 0);
1998                 strbuf_addch(&msg, '\n');
1999                 break;
2000         case DIFF_STATUS_MODIFIED:
2001                 if (p->score) {
2002                         strbuf_addf(&msg, "dissimilarity index %d%%\n",
2003                                         similarity_index(p));
2004                         complete_rewrite = 1;
2005                         break;
2006                 }
2007                 /* fallthru */
2008         default:
2009                 /* nothing */
2010                 ;
2011         }
2012
2013         if (hashcmp(one->sha1, two->sha1)) {
2014                 int abbrev = DIFF_OPT_TST(o, FULL_INDEX) ? 40 : DEFAULT_ABBREV;
2015
2016                 if (DIFF_OPT_TST(o, BINARY)) {
2017                         mmfile_t mf;
2018                         if ((!fill_mmfile(&mf, one) && diff_filespec_is_binary(one)) ||
2019                             (!fill_mmfile(&mf, two) && diff_filespec_is_binary(two)))
2020                                 abbrev = 40;
2021                 }
2022                 strbuf_addf(&msg, "index %.*s..%.*s",
2023                                 abbrev, sha1_to_hex(one->sha1),
2024                                 abbrev, sha1_to_hex(two->sha1));
2025                 if (one->mode == two->mode)
2026                         strbuf_addf(&msg, " %06o", one->mode);
2027                 strbuf_addch(&msg, '\n');
2028         }
2029
2030         if (msg.len)
2031                 strbuf_setlen(&msg, msg.len - 1);
2032         xfrm_msg = msg.len ? msg.buf : NULL;
2033
2034         if (!pgm &&
2035             DIFF_FILE_VALID(one) && DIFF_FILE_VALID(two) &&
2036             (S_IFMT & one->mode) != (S_IFMT & two->mode)) {
2037                 /* a filepair that changes between file and symlink
2038                  * needs to be split into deletion and creation.
2039                  */
2040                 struct diff_filespec *null = alloc_filespec(two->path);
2041                 run_diff_cmd(NULL, name, other, one, null, xfrm_msg, o, 0);
2042                 free(null);
2043                 null = alloc_filespec(one->path);
2044                 run_diff_cmd(NULL, name, other, null, two, xfrm_msg, o, 0);
2045                 free(null);
2046         }
2047         else
2048                 run_diff_cmd(pgm, name, other, one, two, xfrm_msg, o,
2049                              complete_rewrite);
2050
2051         strbuf_release(&msg);
2052 }
2053
2054 static void run_diffstat(struct diff_filepair *p, struct diff_options *o,
2055                          struct diffstat_t *diffstat)
2056 {
2057         const char *name;
2058         const char *other;
2059         int complete_rewrite = 0;
2060
2061         if (DIFF_PAIR_UNMERGED(p)) {
2062                 /* unmerged */
2063                 builtin_diffstat(p->one->path, NULL, NULL, NULL, diffstat, o, 0);
2064                 return;
2065         }
2066
2067         name = p->one->path;
2068         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2069
2070         diff_fill_sha1_info(p->one);
2071         diff_fill_sha1_info(p->two);
2072
2073         if (p->status == DIFF_STATUS_MODIFIED && p->score)
2074                 complete_rewrite = 1;
2075         builtin_diffstat(name, other, p->one, p->two, diffstat, o, complete_rewrite);
2076 }
2077
2078 static void run_checkdiff(struct diff_filepair *p, struct diff_options *o)
2079 {
2080         const char *name;
2081         const char *other;
2082
2083         if (DIFF_PAIR_UNMERGED(p)) {
2084                 /* unmerged */
2085                 return;
2086         }
2087
2088         name = p->one->path;
2089         other = (strcmp(name, p->two->path) ? p->two->path : NULL);
2090
2091         diff_fill_sha1_info(p->one);
2092         diff_fill_sha1_info(p->two);
2093
2094         builtin_checkdiff(name, other, p->one, p->two, o);
2095 }
2096
2097 void diff_setup(struct diff_options *options)
2098 {
2099         memset(options, 0, sizeof(*options));
2100         options->line_termination = '\n';
2101         options->break_opt = -1;
2102         options->rename_limit = -1;
2103         options->context = 3;
2104         options->msg_sep = "";
2105
2106         options->change = diff_change;
2107         options->add_remove = diff_addremove;
2108         if (diff_use_color_default)
2109                 DIFF_OPT_SET(options, COLOR_DIFF);
2110         else
2111                 DIFF_OPT_CLR(options, COLOR_DIFF);
2112         options->detect_rename = diff_detect_rename_default;
2113 }
2114
2115 int diff_setup_done(struct diff_options *options)
2116 {
2117         int count = 0;
2118
2119         if (options->output_format & DIFF_FORMAT_NAME)
2120                 count++;
2121         if (options->output_format & DIFF_FORMAT_NAME_STATUS)
2122                 count++;
2123         if (options->output_format & DIFF_FORMAT_CHECKDIFF)
2124                 count++;
2125         if (options->output_format & DIFF_FORMAT_NO_OUTPUT)
2126                 count++;
2127         if (count > 1)
2128                 die("--name-only, --name-status, --check and -s are mutually exclusive");
2129
2130         if (DIFF_OPT_TST(options, FIND_COPIES_HARDER))
2131                 options->detect_rename = DIFF_DETECT_COPY;
2132
2133         if (options->output_format & (DIFF_FORMAT_NAME |
2134                                       DIFF_FORMAT_NAME_STATUS |
2135                                       DIFF_FORMAT_CHECKDIFF |
2136                                       DIFF_FORMAT_NO_OUTPUT))
2137                 options->output_format &= ~(DIFF_FORMAT_RAW |
2138                                             DIFF_FORMAT_NUMSTAT |
2139                                             DIFF_FORMAT_DIFFSTAT |
2140                                             DIFF_FORMAT_SHORTSTAT |
2141                                             DIFF_FORMAT_SUMMARY |
2142                                             DIFF_FORMAT_PATCH);
2143
2144         /*
2145          * These cases always need recursive; we do not drop caller-supplied
2146          * recursive bits for other formats here.
2147          */
2148         if (options->output_format & (DIFF_FORMAT_PATCH |
2149                                       DIFF_FORMAT_NUMSTAT |
2150                                       DIFF_FORMAT_DIFFSTAT |
2151                                       DIFF_FORMAT_SHORTSTAT |
2152                                       DIFF_FORMAT_SUMMARY |
2153                                       DIFF_FORMAT_CHECKDIFF))
2154                 DIFF_OPT_SET(options, RECURSIVE);
2155         /*
2156          * Also pickaxe would not work very well if you do not say recursive
2157          */
2158         if (options->pickaxe)
2159                 DIFF_OPT_SET(options, RECURSIVE);
2160
2161         if (options->detect_rename && options->rename_limit < 0)
2162                 options->rename_limit = diff_rename_limit_default;
2163         if (options->setup & DIFF_SETUP_USE_CACHE) {
2164                 if (!active_cache)
2165                         /* read-cache does not die even when it fails
2166                          * so it is safe for us to do this here.  Also
2167                          * it does not smudge active_cache or active_nr
2168                          * when it fails, so we do not have to worry about
2169                          * cleaning it up ourselves either.
2170                          */
2171                         read_cache();
2172         }
2173         if (options->abbrev <= 0 || 40 < options->abbrev)
2174                 options->abbrev = 40; /* full */
2175
2176         /*
2177          * It does not make sense to show the first hit we happened
2178          * to have found.  It does not make sense not to return with
2179          * exit code in such a case either.
2180          */
2181         if (DIFF_OPT_TST(options, QUIET)) {
2182                 options->output_format = DIFF_FORMAT_NO_OUTPUT;
2183                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2184         }
2185
2186         /*
2187          * If we postprocess in diffcore, we cannot simply return
2188          * upon the first hit.  We need to run diff as usual.
2189          */
2190         if (options->pickaxe || options->filter)
2191                 DIFF_OPT_CLR(options, QUIET);
2192
2193         return 0;
2194 }
2195
2196 static int opt_arg(const char *arg, int arg_short, const char *arg_long, int *val)
2197 {
2198         char c, *eq;
2199         int len;
2200
2201         if (*arg != '-')
2202                 return 0;
2203         c = *++arg;
2204         if (!c)
2205                 return 0;
2206         if (c == arg_short) {
2207                 c = *++arg;
2208                 if (!c)
2209                         return 1;
2210                 if (val && isdigit(c)) {
2211                         char *end;
2212                         int n = strtoul(arg, &end, 10);
2213                         if (*end)
2214                                 return 0;
2215                         *val = n;
2216                         return 1;
2217                 }
2218                 return 0;
2219         }
2220         if (c != '-')
2221                 return 0;
2222         arg++;
2223         eq = strchr(arg, '=');
2224         if (eq)
2225                 len = eq - arg;
2226         else
2227                 len = strlen(arg);
2228         if (!len || strncmp(arg, arg_long, len))
2229                 return 0;
2230         if (eq) {
2231                 int n;
2232                 char *end;
2233                 if (!isdigit(*++eq))
2234                         return 0;
2235                 n = strtoul(eq, &end, 10);
2236                 if (*end)
2237                         return 0;
2238                 *val = n;
2239         }
2240         return 1;
2241 }
2242
2243 static int diff_scoreopt_parse(const char *opt);
2244
2245 int diff_opt_parse(struct diff_options *options, const char **av, int ac)
2246 {
2247         const char *arg = av[0];
2248
2249         /* Output format options */
2250         if (!strcmp(arg, "-p") || !strcmp(arg, "-u"))
2251                 options->output_format |= DIFF_FORMAT_PATCH;
2252         else if (opt_arg(arg, 'U', "unified", &options->context))
2253                 options->output_format |= DIFF_FORMAT_PATCH;
2254         else if (!strcmp(arg, "--raw"))
2255                 options->output_format |= DIFF_FORMAT_RAW;
2256         else if (!strcmp(arg, "--patch-with-raw"))
2257                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_RAW;
2258         else if (!strcmp(arg, "--numstat"))
2259                 options->output_format |= DIFF_FORMAT_NUMSTAT;
2260         else if (!strcmp(arg, "--shortstat"))
2261                 options->output_format |= DIFF_FORMAT_SHORTSTAT;
2262         else if (!strcmp(arg, "--check"))
2263                 options->output_format |= DIFF_FORMAT_CHECKDIFF;
2264         else if (!strcmp(arg, "--summary"))
2265                 options->output_format |= DIFF_FORMAT_SUMMARY;
2266         else if (!strcmp(arg, "--patch-with-stat"))
2267                 options->output_format |= DIFF_FORMAT_PATCH | DIFF_FORMAT_DIFFSTAT;
2268         else if (!strcmp(arg, "--name-only"))
2269                 options->output_format |= DIFF_FORMAT_NAME;
2270         else if (!strcmp(arg, "--name-status"))
2271                 options->output_format |= DIFF_FORMAT_NAME_STATUS;
2272         else if (!strcmp(arg, "-s"))
2273                 options->output_format |= DIFF_FORMAT_NO_OUTPUT;
2274         else if (!prefixcmp(arg, "--stat")) {
2275                 char *end;
2276                 int width = options->stat_width;
2277                 int name_width = options->stat_name_width;
2278                 arg += 6;
2279                 end = (char *)arg;
2280
2281                 switch (*arg) {
2282                 case '-':
2283                         if (!prefixcmp(arg, "-width="))
2284                                 width = strtoul(arg + 7, &end, 10);
2285                         else if (!prefixcmp(arg, "-name-width="))
2286                                 name_width = strtoul(arg + 12, &end, 10);
2287                         break;
2288                 case '=':
2289                         width = strtoul(arg+1, &end, 10);
2290                         if (*end == ',')
2291                                 name_width = strtoul(end+1, &end, 10);
2292                 }
2293
2294                 /* Important! This checks all the error cases! */
2295                 if (*end)
2296                         return 0;
2297                 options->output_format |= DIFF_FORMAT_DIFFSTAT;
2298                 options->stat_name_width = name_width;
2299                 options->stat_width = width;
2300         }
2301
2302         /* renames options */
2303         else if (!prefixcmp(arg, "-B")) {
2304                 if ((options->break_opt = diff_scoreopt_parse(arg)) == -1)
2305                         return -1;
2306         }
2307         else if (!prefixcmp(arg, "-M")) {
2308                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2309                         return -1;
2310                 options->detect_rename = DIFF_DETECT_RENAME;
2311         }
2312         else if (!prefixcmp(arg, "-C")) {
2313                 if (options->detect_rename == DIFF_DETECT_COPY)
2314                         DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2315                 if ((options->rename_score = diff_scoreopt_parse(arg)) == -1)
2316                         return -1;
2317                 options->detect_rename = DIFF_DETECT_COPY;
2318         }
2319         else if (!strcmp(arg, "--no-renames"))
2320                 options->detect_rename = 0;
2321
2322         /* xdiff options */
2323         else if (!strcmp(arg, "-w") || !strcmp(arg, "--ignore-all-space"))
2324                 options->xdl_opts |= XDF_IGNORE_WHITESPACE;
2325         else if (!strcmp(arg, "-b") || !strcmp(arg, "--ignore-space-change"))
2326                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_CHANGE;
2327         else if (!strcmp(arg, "--ignore-space-at-eol"))
2328                 options->xdl_opts |= XDF_IGNORE_WHITESPACE_AT_EOL;
2329
2330         /* flags options */
2331         else if (!strcmp(arg, "--binary")) {
2332                 options->output_format |= DIFF_FORMAT_PATCH;
2333                 DIFF_OPT_SET(options, BINARY);
2334         }
2335         else if (!strcmp(arg, "--full-index"))
2336                 DIFF_OPT_SET(options, FULL_INDEX);
2337         else if (!strcmp(arg, "-a") || !strcmp(arg, "--text"))
2338                 DIFF_OPT_SET(options, TEXT);
2339         else if (!strcmp(arg, "-R"))
2340                 DIFF_OPT_SET(options, REVERSE_DIFF);
2341         else if (!strcmp(arg, "--find-copies-harder"))
2342                 DIFF_OPT_SET(options, FIND_COPIES_HARDER);
2343         else if (!strcmp(arg, "--follow"))
2344                 DIFF_OPT_SET(options, FOLLOW_RENAMES);
2345         else if (!strcmp(arg, "--color"))
2346                 DIFF_OPT_SET(options, COLOR_DIFF);
2347         else if (!strcmp(arg, "--no-color"))
2348                 DIFF_OPT_CLR(options, COLOR_DIFF);
2349         else if (!strcmp(arg, "--color-words"))
2350                 options->flags |= DIFF_OPT_COLOR_DIFF | DIFF_OPT_COLOR_DIFF_WORDS;
2351         else if (!strcmp(arg, "--exit-code"))
2352                 DIFF_OPT_SET(options, EXIT_WITH_STATUS);
2353         else if (!strcmp(arg, "--quiet"))
2354                 DIFF_OPT_SET(options, QUIET);
2355         else if (!strcmp(arg, "--ext-diff"))
2356                 DIFF_OPT_SET(options, ALLOW_EXTERNAL);
2357         else if (!strcmp(arg, "--no-ext-diff"))
2358                 DIFF_OPT_CLR(options, ALLOW_EXTERNAL);
2359
2360         /* misc options */
2361         else if (!strcmp(arg, "-z"))
2362                 options->line_termination = 0;
2363         else if (!prefixcmp(arg, "-l"))
2364                 options->rename_limit = strtoul(arg+2, NULL, 10);
2365         else if (!prefixcmp(arg, "-S"))
2366                 options->pickaxe = arg + 2;
2367         else if (!strcmp(arg, "--pickaxe-all"))
2368                 options->pickaxe_opts = DIFF_PICKAXE_ALL;
2369         else if (!strcmp(arg, "--pickaxe-regex"))
2370                 options->pickaxe_opts = DIFF_PICKAXE_REGEX;
2371         else if (!prefixcmp(arg, "-O"))
2372                 options->orderfile = arg + 2;
2373         else if (!prefixcmp(arg, "--diff-filter="))
2374                 options->filter = arg + 14;
2375         else if (!strcmp(arg, "--abbrev"))
2376                 options->abbrev = DEFAULT_ABBREV;
2377         else if (!prefixcmp(arg, "--abbrev=")) {
2378                 options->abbrev = strtoul(arg + 9, NULL, 10);
2379                 if (options->abbrev < MINIMUM_ABBREV)
2380                         options->abbrev = MINIMUM_ABBREV;
2381                 else if (40 < options->abbrev)
2382                         options->abbrev = 40;
2383         }
2384         else
2385                 return 0;
2386         return 1;
2387 }
2388
2389 static int parse_num(const char **cp_p)
2390 {
2391         unsigned long num, scale;
2392         int ch, dot;
2393         const char *cp = *cp_p;
2394
2395         num = 0;
2396         scale = 1;
2397         dot = 0;
2398         for(;;) {
2399                 ch = *cp;
2400                 if ( !dot && ch == '.' ) {
2401                         scale = 1;
2402                         dot = 1;
2403                 } else if ( ch == '%' ) {
2404                         scale = dot ? scale*100 : 100;
2405                         cp++;   /* % is always at the end */
2406                         break;
2407                 } else if ( ch >= '0' && ch <= '9' ) {
2408                         if ( scale < 100000 ) {
2409                                 scale *= 10;
2410                                 num = (num*10) + (ch-'0');
2411                         }
2412                 } else {
2413                         break;
2414                 }
2415                 cp++;
2416         }
2417         *cp_p = cp;
2418
2419         /* user says num divided by scale and we say internally that
2420          * is MAX_SCORE * num / scale.
2421          */
2422         return (int)((num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale));
2423 }
2424
2425 static int diff_scoreopt_parse(const char *opt)
2426 {
2427         int opt1, opt2, cmd;
2428
2429         if (*opt++ != '-')
2430                 return -1;
2431         cmd = *opt++;
2432         if (cmd != 'M' && cmd != 'C' && cmd != 'B')
2433                 return -1; /* that is not a -M, -C nor -B option */
2434
2435         opt1 = parse_num(&opt);
2436         if (cmd != 'B')
2437                 opt2 = 0;
2438         else {
2439                 if (*opt == 0)
2440                         opt2 = 0;
2441                 else if (*opt != '/')
2442                         return -1; /* we expect -B80/99 or -B80 */
2443                 else {
2444                         opt++;
2445                         opt2 = parse_num(&opt);
2446                 }
2447         }
2448         if (*opt != 0)
2449                 return -1;
2450         return opt1 | (opt2 << 16);
2451 }
2452
2453 struct diff_queue_struct diff_queued_diff;
2454
2455 void diff_q(struct diff_queue_struct *queue, struct diff_filepair *dp)
2456 {
2457         if (queue->alloc <= queue->nr) {
2458                 queue->alloc = alloc_nr(queue->alloc);
2459                 queue->queue = xrealloc(queue->queue,
2460                                         sizeof(dp) * queue->alloc);
2461         }
2462         queue->queue[queue->nr++] = dp;
2463 }
2464
2465 struct diff_filepair *diff_queue(struct diff_queue_struct *queue,
2466                                  struct diff_filespec *one,
2467                                  struct diff_filespec *two)
2468 {
2469         struct diff_filepair *dp = xcalloc(1, sizeof(*dp));
2470         dp->one = one;
2471         dp->two = two;
2472         if (queue)
2473                 diff_q(queue, dp);
2474         return dp;
2475 }
2476
2477 void diff_free_filepair(struct diff_filepair *p)
2478 {
2479         free_filespec(p->one);
2480         free_filespec(p->two);
2481         free(p);
2482 }
2483
2484 /* This is different from find_unique_abbrev() in that
2485  * it stuffs the result with dots for alignment.
2486  */
2487 const char *diff_unique_abbrev(const unsigned char *sha1, int len)
2488 {
2489         int abblen;
2490         const char *abbrev;
2491         if (len == 40)
2492                 return sha1_to_hex(sha1);
2493
2494         abbrev = find_unique_abbrev(sha1, len);
2495         if (!abbrev)
2496                 return sha1_to_hex(sha1);
2497         abblen = strlen(abbrev);
2498         if (abblen < 37) {
2499                 static char hex[41];
2500                 if (len < abblen && abblen <= len + 2)
2501                         sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
2502                 else
2503                         sprintf(hex, "%s...", abbrev);
2504                 return hex;
2505         }
2506         return sha1_to_hex(sha1);
2507 }
2508
2509 static void diff_flush_raw(struct diff_filepair *p, struct diff_options *opt)
2510 {
2511         int line_termination = opt->line_termination;
2512         int inter_name_termination = line_termination ? '\t' : '\0';
2513
2514         if (!(opt->output_format & DIFF_FORMAT_NAME_STATUS)) {
2515                 printf(":%06o %06o %s ", p->one->mode, p->two->mode,
2516                        diff_unique_abbrev(p->one->sha1, opt->abbrev));
2517                 printf("%s ", diff_unique_abbrev(p->two->sha1, opt->abbrev));
2518         }
2519         if (p->score) {
2520                 printf("%c%03d%c", p->status, similarity_index(p),
2521                            inter_name_termination);
2522         } else {
2523                 printf("%c%c", p->status, inter_name_termination);
2524         }
2525
2526         if (p->status == DIFF_STATUS_COPIED || p->status == DIFF_STATUS_RENAMED) {
2527                 write_name_quoted(p->one->path, stdout, inter_name_termination);
2528                 write_name_quoted(p->two->path, stdout, line_termination);
2529         } else {
2530                 const char *path = p->one->mode ? p->one->path : p->two->path;
2531                 write_name_quoted(path, stdout, line_termination);
2532         }
2533 }
2534
2535 int diff_unmodified_pair(struct diff_filepair *p)
2536 {
2537         /* This function is written stricter than necessary to support
2538          * the currently implemented transformers, but the idea is to
2539          * let transformers to produce diff_filepairs any way they want,
2540          * and filter and clean them up here before producing the output.
2541          */
2542         struct diff_filespec *one = p->one, *two = p->two;
2543
2544         if (DIFF_PAIR_UNMERGED(p))
2545                 return 0; /* unmerged is interesting */
2546
2547         /* deletion, addition, mode or type change
2548          * and rename are all interesting.
2549          */
2550         if (DIFF_FILE_VALID(one) != DIFF_FILE_VALID(two) ||
2551             DIFF_PAIR_MODE_CHANGED(p) ||
2552             strcmp(one->path, two->path))
2553                 return 0;
2554
2555         /* both are valid and point at the same path.  that is, we are
2556          * dealing with a change.
2557          */
2558         if (one->sha1_valid && two->sha1_valid &&
2559             !hashcmp(one->sha1, two->sha1))
2560                 return 1; /* no change */
2561         if (!one->sha1_valid && !two->sha1_valid)
2562                 return 1; /* both look at the same file on the filesystem. */
2563         return 0;
2564 }
2565
2566 static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
2567 {
2568         if (diff_unmodified_pair(p))
2569                 return;
2570
2571         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2572             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2573                 return; /* no tree diffs in patch format */
2574
2575         run_diff(p, o);
2576 }
2577
2578 static void diff_flush_stat(struct diff_filepair *p, struct diff_options *o,
2579                             struct diffstat_t *diffstat)
2580 {
2581         if (diff_unmodified_pair(p))
2582                 return;
2583
2584         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2585             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2586                 return; /* no tree diffs in patch format */
2587
2588         run_diffstat(p, o, diffstat);
2589 }
2590
2591 static void diff_flush_checkdiff(struct diff_filepair *p,
2592                 struct diff_options *o)
2593 {
2594         if (diff_unmodified_pair(p))
2595                 return;
2596
2597         if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2598             (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2599                 return; /* no tree diffs in patch format */
2600
2601         run_checkdiff(p, o);
2602 }
2603
2604 int diff_queue_is_empty(void)
2605 {
2606         struct diff_queue_struct *q = &diff_queued_diff;
2607         int i;
2608         for (i = 0; i < q->nr; i++)
2609                 if (!diff_unmodified_pair(q->queue[i]))
2610                         return 0;
2611         return 1;
2612 }
2613
2614 #if DIFF_DEBUG
2615 void diff_debug_filespec(struct diff_filespec *s, int x, const char *one)
2616 {
2617         fprintf(stderr, "queue[%d] %s (%s) %s %06o %s\n",
2618                 x, one ? one : "",
2619                 s->path,
2620                 DIFF_FILE_VALID(s) ? "valid" : "invalid",
2621                 s->mode,
2622                 s->sha1_valid ? sha1_to_hex(s->sha1) : "");
2623         fprintf(stderr, "queue[%d] %s size %lu flags %d\n",
2624                 x, one ? one : "",
2625                 s->size, s->xfrm_flags);
2626 }
2627
2628 void diff_debug_filepair(const struct diff_filepair *p, int i)
2629 {
2630         diff_debug_filespec(p->one, i, "one");
2631         diff_debug_filespec(p->two, i, "two");
2632         fprintf(stderr, "score %d, status %c rename_used %d broken %d\n",
2633                 p->score, p->status ? p->status : '?',
2634                 p->one->rename_used, p->broken_pair);
2635 }
2636
2637 void diff_debug_queue(const char *msg, struct diff_queue_struct *q)
2638 {
2639         int i;
2640         if (msg)
2641                 fprintf(stderr, "%s\n", msg);
2642         fprintf(stderr, "q->nr = %d\n", q->nr);
2643         for (i = 0; i < q->nr; i++) {
2644                 struct diff_filepair *p = q->queue[i];
2645                 diff_debug_filepair(p, i);
2646         }
2647 }
2648 #endif
2649
2650 static void diff_resolve_rename_copy(void)
2651 {
2652         int i;
2653         struct diff_filepair *p;
2654         struct diff_queue_struct *q = &diff_queued_diff;
2655
2656         diff_debug_queue("resolve-rename-copy", q);
2657
2658         for (i = 0; i < q->nr; i++) {
2659                 p = q->queue[i];
2660                 p->status = 0; /* undecided */
2661                 if (DIFF_PAIR_UNMERGED(p))
2662                         p->status = DIFF_STATUS_UNMERGED;
2663                 else if (!DIFF_FILE_VALID(p->one))
2664                         p->status = DIFF_STATUS_ADDED;
2665                 else if (!DIFF_FILE_VALID(p->two))
2666                         p->status = DIFF_STATUS_DELETED;
2667                 else if (DIFF_PAIR_TYPE_CHANGED(p))
2668                         p->status = DIFF_STATUS_TYPE_CHANGED;
2669
2670                 /* from this point on, we are dealing with a pair
2671                  * whose both sides are valid and of the same type, i.e.
2672                  * either in-place edit or rename/copy edit.
2673                  */
2674                 else if (DIFF_PAIR_RENAME(p)) {
2675                         /*
2676                          * A rename might have re-connected a broken
2677                          * pair up, causing the pathnames to be the
2678                          * same again. If so, that's not a rename at
2679                          * all, just a modification..
2680                          *
2681                          * Otherwise, see if this source was used for
2682                          * multiple renames, in which case we decrement
2683                          * the count, and call it a copy.
2684                          */
2685                         if (!strcmp(p->one->path, p->two->path))
2686                                 p->status = DIFF_STATUS_MODIFIED;
2687                         else if (--p->one->rename_used > 0)
2688                                 p->status = DIFF_STATUS_COPIED;
2689                         else
2690                                 p->status = DIFF_STATUS_RENAMED;
2691                 }
2692                 else if (hashcmp(p->one->sha1, p->two->sha1) ||
2693                          p->one->mode != p->two->mode ||
2694                          is_null_sha1(p->one->sha1))
2695                         p->status = DIFF_STATUS_MODIFIED;
2696                 else {
2697                         /* This is a "no-change" entry and should not
2698                          * happen anymore, but prepare for broken callers.
2699                          */
2700                         error("feeding unmodified %s to diffcore",
2701                               p->one->path);
2702                         p->status = DIFF_STATUS_UNKNOWN;
2703                 }
2704         }
2705         diff_debug_queue("resolve-rename-copy done", q);
2706 }
2707
2708 static int check_pair_status(struct diff_filepair *p)
2709 {
2710         switch (p->status) {
2711         case DIFF_STATUS_UNKNOWN:
2712                 return 0;
2713         case 0:
2714                 die("internal error in diff-resolve-rename-copy");
2715         default:
2716                 return 1;
2717         }
2718 }
2719
2720 static void flush_one_pair(struct diff_filepair *p, struct diff_options *opt)
2721 {
2722         int fmt = opt->output_format;
2723
2724         if (fmt & DIFF_FORMAT_CHECKDIFF)
2725                 diff_flush_checkdiff(p, opt);
2726         else if (fmt & (DIFF_FORMAT_RAW | DIFF_FORMAT_NAME_STATUS))
2727                 diff_flush_raw(p, opt);
2728         else if (fmt & DIFF_FORMAT_NAME)
2729                 write_name_quoted(p->two->path, stdout, opt->line_termination);
2730 }
2731
2732 static void show_file_mode_name(const char *newdelete, struct diff_filespec *fs)
2733 {
2734         if (fs->mode)
2735                 printf(" %s mode %06o ", newdelete, fs->mode);
2736         else
2737                 printf(" %s ", newdelete);
2738         write_name_quoted(fs->path, stdout, '\n');
2739 }
2740
2741
2742 static void show_mode_change(struct diff_filepair *p, int show_name)
2743 {
2744         if (p->one->mode && p->two->mode && p->one->mode != p->two->mode) {
2745                 printf(" mode change %06o => %06o%c", p->one->mode, p->two->mode,
2746                         show_name ? ' ' : '\n');
2747                 if (show_name) {
2748                         write_name_quoted(p->two->path, stdout, '\n');
2749                 }
2750         }
2751 }
2752
2753 static void show_rename_copy(const char *renamecopy, struct diff_filepair *p)
2754 {
2755         char *names = pprint_rename(p->one->path, p->two->path);
2756
2757         printf(" %s %s (%d%%)\n", renamecopy, names, similarity_index(p));
2758         free(names);
2759         show_mode_change(p, 0);
2760 }
2761
2762 static void diff_summary(struct diff_filepair *p)
2763 {
2764         switch(p->status) {
2765         case DIFF_STATUS_DELETED:
2766                 show_file_mode_name("delete", p->one);
2767                 break;
2768         case DIFF_STATUS_ADDED:
2769                 show_file_mode_name("create", p->two);
2770                 break;
2771         case DIFF_STATUS_COPIED:
2772                 show_rename_copy("copy", p);
2773                 break;
2774         case DIFF_STATUS_RENAMED:
2775                 show_rename_copy("rename", p);
2776                 break;
2777         default:
2778                 if (p->score) {
2779                         fputs(" rewrite ", stdout);
2780                         write_name_quoted(p->two->path, stdout, ' ');
2781                         printf("(%d%%)\n", similarity_index(p));
2782                 }
2783                 show_mode_change(p, !p->score);
2784                 break;
2785         }
2786 }
2787
2788 struct patch_id_t {
2789         struct xdiff_emit_state xm;
2790         SHA_CTX *ctx;
2791         int patchlen;
2792 };
2793
2794 static int remove_space(char *line, int len)
2795 {
2796         int i;
2797         char *dst = line;
2798         unsigned char c;
2799
2800         for (i = 0; i < len; i++)
2801                 if (!isspace((c = line[i])))
2802                         *dst++ = c;
2803
2804         return dst - line;
2805 }
2806
2807 static void patch_id_consume(void *priv, char *line, unsigned long len)
2808 {
2809         struct patch_id_t *data = priv;
2810         int new_len;
2811
2812         /* Ignore line numbers when computing the SHA1 of the patch */
2813         if (!prefixcmp(line, "@@ -"))
2814                 return;
2815
2816         new_len = remove_space(line, len);
2817
2818         SHA1_Update(data->ctx, line, new_len);
2819         data->patchlen += new_len;
2820 }
2821
2822 /* returns 0 upon success, and writes result into sha1 */
2823 static int diff_get_patch_id(struct diff_options *options, unsigned char *sha1)
2824 {
2825         struct diff_queue_struct *q = &diff_queued_diff;
2826         int i;
2827         SHA_CTX ctx;
2828         struct patch_id_t data;
2829         char buffer[PATH_MAX * 4 + 20];
2830
2831         SHA1_Init(&ctx);
2832         memset(&data, 0, sizeof(struct patch_id_t));
2833         data.ctx = &ctx;
2834         data.xm.consume = patch_id_consume;
2835
2836         for (i = 0; i < q->nr; i++) {
2837                 xpparam_t xpp;
2838                 xdemitconf_t xecfg;
2839                 xdemitcb_t ecb;
2840                 mmfile_t mf1, mf2;
2841                 struct diff_filepair *p = q->queue[i];
2842                 int len1, len2;
2843
2844                 memset(&xecfg, 0, sizeof(xecfg));
2845                 if (p->status == 0)
2846                         return error("internal diff status error");
2847                 if (p->status == DIFF_STATUS_UNKNOWN)
2848                         continue;
2849                 if (diff_unmodified_pair(p))
2850                         continue;
2851                 if ((DIFF_FILE_VALID(p->one) && S_ISDIR(p->one->mode)) ||
2852                     (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
2853                         continue;
2854                 if (DIFF_PAIR_UNMERGED(p))
2855                         continue;
2856
2857                 diff_fill_sha1_info(p->one);
2858                 diff_fill_sha1_info(p->two);
2859                 if (fill_mmfile(&mf1, p->one) < 0 ||
2860                                 fill_mmfile(&mf2, p->two) < 0)
2861                         return error("unable to read files to diff");
2862
2863                 len1 = remove_space(p->one->path, strlen(p->one->path));
2864                 len2 = remove_space(p->two->path, strlen(p->two->path));
2865                 if (p->one->mode == 0)
2866                         len1 = snprintf(buffer, sizeof(buffer),
2867                                         "diff--gita/%.*sb/%.*s"
2868                                         "newfilemode%06o"
2869                                         "---/dev/null"
2870                                         "+++b/%.*s",
2871                                         len1, p->one->path,
2872                                         len2, p->two->path,
2873                                         p->two->mode,
2874                                         len2, p->two->path);
2875                 else if (p->two->mode == 0)
2876                         len1 = snprintf(buffer, sizeof(buffer),
2877                                         "diff--gita/%.*sb/%.*s"
2878                                         "deletedfilemode%06o"
2879                                         "---a/%.*s"
2880                                         "+++/dev/null",
2881                                         len1, p->one->path,
2882                                         len2, p->two->path,
2883                                         p->one->mode,
2884                                         len1, p->one->path);
2885                 else
2886                         len1 = snprintf(buffer, sizeof(buffer),
2887                                         "diff--gita/%.*sb/%.*s"
2888                                         "---a/%.*s"
2889                                         "+++b/%.*s",
2890                                         len1, p->one->path,
2891                                         len2, p->two->path,
2892                                         len1, p->one->path,
2893                                         len2, p->two->path);
2894                 SHA1_Update(&ctx, buffer, len1);
2895
2896                 xpp.flags = XDF_NEED_MINIMAL;
2897                 xecfg.ctxlen = 3;
2898                 xecfg.flags = XDL_EMIT_FUNCNAMES;
2899                 ecb.outf = xdiff_outf;
2900                 ecb.priv = &data;
2901                 xdl_diff(&mf1, &mf2, &xpp, &xecfg, &ecb);
2902         }
2903
2904         SHA1_Final(sha1, &ctx);
2905         return 0;
2906 }
2907
2908 int diff_flush_patch_id(struct diff_options *options, unsigned char *sha1)
2909 {
2910         struct diff_queue_struct *q = &diff_queued_diff;
2911         int i;
2912         int result = diff_get_patch_id(options, sha1);
2913
2914         for (i = 0; i < q->nr; i++)
2915                 diff_free_filepair(q->queue[i]);
2916
2917         free(q->queue);
2918         q->queue = NULL;
2919         q->nr = q->alloc = 0;
2920
2921         return result;
2922 }
2923
2924 static int is_summary_empty(const struct diff_queue_struct *q)
2925 {
2926         int i;
2927
2928         for (i = 0; i < q->nr; i++) {
2929                 const struct diff_filepair *p = q->queue[i];
2930
2931                 switch (p->status) {
2932                 case DIFF_STATUS_DELETED:
2933                 case DIFF_STATUS_ADDED:
2934                 case DIFF_STATUS_COPIED:
2935                 case DIFF_STATUS_RENAMED:
2936                         return 0;
2937                 default:
2938                         if (p->score)
2939                                 return 0;
2940                         if (p->one->mode && p->two->mode &&
2941                             p->one->mode != p->two->mode)
2942                                 return 0;
2943                         break;
2944                 }
2945         }
2946         return 1;
2947 }
2948
2949 void diff_flush(struct diff_options *options)
2950 {
2951         struct diff_queue_struct *q = &diff_queued_diff;
2952         int i, output_format = options->output_format;
2953         int separator = 0;
2954
2955         /*
2956          * Order: raw, stat, summary, patch
2957          * or:    name/name-status/checkdiff (other bits clear)
2958          */
2959         if (!q->nr)
2960                 goto free_queue;
2961
2962         if (output_format & (DIFF_FORMAT_RAW |
2963                              DIFF_FORMAT_NAME |
2964                              DIFF_FORMAT_NAME_STATUS |
2965                              DIFF_FORMAT_CHECKDIFF)) {
2966                 for (i = 0; i < q->nr; i++) {
2967                         struct diff_filepair *p = q->queue[i];
2968                         if (check_pair_status(p))
2969                                 flush_one_pair(p, options);
2970                 }
2971                 separator++;
2972         }
2973
2974         if (output_format & (DIFF_FORMAT_DIFFSTAT|DIFF_FORMAT_SHORTSTAT|DIFF_FORMAT_NUMSTAT)) {
2975                 struct diffstat_t diffstat;
2976
2977                 memset(&diffstat, 0, sizeof(struct diffstat_t));
2978                 diffstat.xm.consume = diffstat_consume;
2979                 for (i = 0; i < q->nr; i++) {
2980                         struct diff_filepair *p = q->queue[i];
2981                         if (check_pair_status(p))
2982                                 diff_flush_stat(p, options, &diffstat);
2983                 }
2984                 if (output_format & DIFF_FORMAT_NUMSTAT)
2985                         show_numstat(&diffstat, options);
2986                 if (output_format & DIFF_FORMAT_DIFFSTAT)
2987                         show_stats(&diffstat, options);
2988                 if (output_format & DIFF_FORMAT_SHORTSTAT)
2989                         show_shortstats(&diffstat);
2990                 free_diffstat_info(&diffstat);
2991                 separator++;
2992         }
2993
2994         if (output_format & DIFF_FORMAT_SUMMARY && !is_summary_empty(q)) {
2995                 for (i = 0; i < q->nr; i++)
2996                         diff_summary(q->queue[i]);
2997                 separator++;
2998         }
2999
3000         if (output_format & DIFF_FORMAT_PATCH) {
3001                 if (separator) {
3002                         if (options->stat_sep) {
3003                                 /* attach patch instead of inline */
3004                                 fputs(options->stat_sep, stdout);
3005                         } else {
3006                                 putchar(options->line_termination);
3007                         }
3008                 }
3009
3010                 for (i = 0; i < q->nr; i++) {
3011                         struct diff_filepair *p = q->queue[i];
3012                         if (check_pair_status(p))
3013                                 diff_flush_patch(p, options);
3014                 }
3015         }
3016
3017         if (output_format & DIFF_FORMAT_CALLBACK)
3018                 options->format_callback(q, options, options->format_callback_data);
3019
3020         for (i = 0; i < q->nr; i++)
3021                 diff_free_filepair(q->queue[i]);
3022 free_queue:
3023         free(q->queue);
3024         q->queue = NULL;
3025         q->nr = q->alloc = 0;
3026 }
3027
3028 static void diffcore_apply_filter(const char *filter)
3029 {
3030         int i;
3031         struct diff_queue_struct *q = &diff_queued_diff;
3032         struct diff_queue_struct outq;
3033         outq.queue = NULL;
3034         outq.nr = outq.alloc = 0;
3035
3036         if (!filter)
3037                 return;
3038
3039         if (strchr(filter, DIFF_STATUS_FILTER_AON)) {
3040                 int found;
3041                 for (i = found = 0; !found && i < q->nr; i++) {
3042                         struct diff_filepair *p = q->queue[i];
3043                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3044                              ((p->score &&
3045                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3046                               (!p->score &&
3047                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3048                             ((p->status != DIFF_STATUS_MODIFIED) &&
3049                              strchr(filter, p->status)))
3050                                 found++;
3051                 }
3052                 if (found)
3053                         return;
3054
3055                 /* otherwise we will clear the whole queue
3056                  * by copying the empty outq at the end of this
3057                  * function, but first clear the current entries
3058                  * in the queue.
3059                  */
3060                 for (i = 0; i < q->nr; i++)
3061                         diff_free_filepair(q->queue[i]);
3062         }
3063         else {
3064                 /* Only the matching ones */
3065                 for (i = 0; i < q->nr; i++) {
3066                         struct diff_filepair *p = q->queue[i];
3067
3068                         if (((p->status == DIFF_STATUS_MODIFIED) &&
3069                              ((p->score &&
3070                                strchr(filter, DIFF_STATUS_FILTER_BROKEN)) ||
3071                               (!p->score &&
3072                                strchr(filter, DIFF_STATUS_MODIFIED)))) ||
3073                             ((p->status != DIFF_STATUS_MODIFIED) &&
3074                              strchr(filter, p->status)))
3075                                 diff_q(&outq, p);
3076                         else
3077                                 diff_free_filepair(p);
3078                 }
3079         }
3080         free(q->queue);
3081         *q = outq;
3082 }
3083
3084 /* Check whether two filespecs with the same mode and size are identical */
3085 static int diff_filespec_is_identical(struct diff_filespec *one,
3086                                       struct diff_filespec *two)
3087 {
3088         if (S_ISGITLINK(one->mode)) {
3089                 diff_fill_sha1_info(one);
3090                 diff_fill_sha1_info(two);
3091                 return !hashcmp(one->sha1, two->sha1);
3092         }
3093         if (diff_populate_filespec(one, 0))
3094                 return 0;
3095         if (diff_populate_filespec(two, 0))
3096                 return 0;
3097         return !memcmp(one->data, two->data, one->size);
3098 }
3099
3100 static void diffcore_skip_stat_unmatch(struct diff_options *diffopt)
3101 {
3102         int i;
3103         struct diff_queue_struct *q = &diff_queued_diff;
3104         struct diff_queue_struct outq;
3105         outq.queue = NULL;
3106         outq.nr = outq.alloc = 0;
3107
3108         for (i = 0; i < q->nr; i++) {
3109                 struct diff_filepair *p = q->queue[i];
3110
3111                 /*
3112                  * 1. Entries that come from stat info dirtyness
3113                  *    always have both sides (iow, not create/delete),
3114                  *    one side of the object name is unknown, with
3115                  *    the same mode and size.  Keep the ones that
3116                  *    do not match these criteria.  They have real
3117                  *    differences.
3118                  *
3119                  * 2. At this point, the file is known to be modified,
3120                  *    with the same mode and size, and the object
3121                  *    name of one side is unknown.  Need to inspect
3122                  *    the identical contents.
3123                  */
3124                 if (!DIFF_FILE_VALID(p->one) || /* (1) */
3125                     !DIFF_FILE_VALID(p->two) ||
3126                     (p->one->sha1_valid && p->two->sha1_valid) ||
3127                     (p->one->mode != p->two->mode) ||
3128                     diff_populate_filespec(p->one, 1) ||
3129                     diff_populate_filespec(p->two, 1) ||
3130                     (p->one->size != p->two->size) ||
3131                     !diff_filespec_is_identical(p->one, p->two)) /* (2) */
3132                         diff_q(&outq, p);
3133                 else {
3134                         /*
3135                          * The caller can subtract 1 from skip_stat_unmatch
3136                          * to determine how many paths were dirty only
3137                          * due to stat info mismatch.
3138                          */
3139                         if (!DIFF_OPT_TST(diffopt, NO_INDEX))
3140                                 diffopt->skip_stat_unmatch++;
3141                         diff_free_filepair(p);
3142                 }
3143         }
3144         free(q->queue);
3145         *q = outq;
3146 }
3147
3148 void diffcore_std(struct diff_options *options)
3149 {
3150         if (DIFF_OPT_TST(options, QUIET))
3151                 return;
3152
3153         if (options->skip_stat_unmatch && !DIFF_OPT_TST(options, FIND_COPIES_HARDER))
3154                 diffcore_skip_stat_unmatch(options);
3155         if (options->break_opt != -1)
3156                 diffcore_break(options->break_opt);
3157         if (options->detect_rename)
3158                 diffcore_rename(options);
3159         if (options->break_opt != -1)
3160                 diffcore_merge_broken();
3161         if (options->pickaxe)
3162                 diffcore_pickaxe(options->pickaxe, options->pickaxe_opts);
3163         if (options->orderfile)
3164                 diffcore_order(options->orderfile);
3165         diff_resolve_rename_copy();
3166         diffcore_apply_filter(options->filter);
3167
3168         if (diff_queued_diff.nr)
3169                 DIFF_OPT_SET(options, HAS_CHANGES);
3170         else
3171                 DIFF_OPT_CLR(options, HAS_CHANGES);
3172 }
3173
3174
3175 void diff_addremove(struct diff_options *options,
3176                     int addremove, unsigned mode,
3177                     const unsigned char *sha1,
3178                     const char *base, const char *path)
3179 {
3180         char concatpath[PATH_MAX];
3181         struct diff_filespec *one, *two;
3182
3183         /* This may look odd, but it is a preparation for
3184          * feeding "there are unchanged files which should
3185          * not produce diffs, but when you are doing copy
3186          * detection you would need them, so here they are"
3187          * entries to the diff-core.  They will be prefixed
3188          * with something like '=' or '*' (I haven't decided
3189          * which but should not make any difference).
3190          * Feeding the same new and old to diff_change()
3191          * also has the same effect.
3192          * Before the final output happens, they are pruned after
3193          * merged into rename/copy pairs as appropriate.
3194          */
3195         if (DIFF_OPT_TST(options, REVERSE_DIFF))
3196                 addremove = (addremove == '+' ? '-' :
3197                              addremove == '-' ? '+' : addremove);
3198
3199         if (!path) path = "";
3200         sprintf(concatpath, "%s%s", base, path);
3201         one = alloc_filespec(concatpath);
3202         two = alloc_filespec(concatpath);
3203
3204         if (addremove != '+')
3205                 fill_filespec(one, sha1, mode);
3206         if (addremove != '-')
3207                 fill_filespec(two, sha1, mode);
3208
3209         diff_queue(&diff_queued_diff, one, two);
3210         DIFF_OPT_SET(options, HAS_CHANGES);
3211 }
3212
3213 void diff_change(struct diff_options *options,
3214                  unsigned old_mode, unsigned new_mode,
3215                  const unsigned char *old_sha1,
3216                  const unsigned char *new_sha1,
3217                  const char *base, const char *path)
3218 {
3219         char concatpath[PATH_MAX];
3220         struct diff_filespec *one, *two;
3221
3222         if (DIFF_OPT_TST(options, REVERSE_DIFF)) {
3223                 unsigned tmp;
3224                 const unsigned char *tmp_c;
3225                 tmp = old_mode; old_mode = new_mode; new_mode = tmp;
3226                 tmp_c = old_sha1; old_sha1 = new_sha1; new_sha1 = tmp_c;
3227         }
3228         if (!path) path = "";
3229         sprintf(concatpath, "%s%s", base, path);
3230         one = alloc_filespec(concatpath);
3231         two = alloc_filespec(concatpath);
3232         fill_filespec(one, old_sha1, old_mode);
3233         fill_filespec(two, new_sha1, new_mode);
3234
3235         diff_queue(&diff_queued_diff, one, two);
3236         DIFF_OPT_SET(options, HAS_CHANGES);
3237 }
3238
3239 void diff_unmerge(struct diff_options *options,
3240                   const char *path,
3241                   unsigned mode, const unsigned char *sha1)
3242 {
3243         struct diff_filespec *one, *two;
3244         one = alloc_filespec(path);
3245         two = alloc_filespec(path);
3246         fill_filespec(one, sha1, mode);
3247         diff_queue(&diff_queued_diff, one, two)->is_unmerged = 1;
3248 }