git-diff-cache: add "-m" flag to match all non-checked-out files with the index.
[git/git.git] / diff.c
1 /*
2 * Copyright (C) 2005 Junio C Hamano
3 */
4 #include <sys/types.h>
5 #include <sys/wait.h>
6 #include <signal.h>
7 #include "cache.h"
8 #include "diff.h"
9
10 static char *diff_opts = "-pu";
11
12 static const char *external_diff(void)
13 {
14 static char *external_diff_cmd = NULL;
15 static int done_preparing = 0;
16
17 if (done_preparing)
18 return external_diff_cmd;
19
20 /*
21 * Default values above are meant to match the
22 * Linux kernel development style. Examples of
23 * alternative styles you can specify via environment
24 * variables are:
25 *
26 * GIT_DIFF_OPTS="-c";
27 */
28 if (getenv("GIT_EXTERNAL_DIFF"))
29 external_diff_cmd = getenv("GIT_EXTERNAL_DIFF");
30
31 /* In case external diff fails... */
32 diff_opts = getenv("GIT_DIFF_OPTS") ? : diff_opts;
33
34 done_preparing = 1;
35 return external_diff_cmd;
36 }
37
38 /* Help to copy the thing properly quoted for the shell safety.
39 * any single quote is replaced with '\'', and the caller is
40 * expected to enclose the result within a single quote pair.
41 *
42 * E.g.
43 * original sq_expand result
44 * name ==> name ==> 'name'
45 * a b ==> a b ==> 'a b'
46 * a'b ==> a'\''b ==> 'a'\''b'
47 */
48 static char *sq_expand(const char *src)
49 {
50 static char *buf = NULL;
51 int cnt, c;
52 const char *cp;
53 char *bp;
54
55 /* count bytes needed to store the quoted string. */
56 for (cnt = 1, cp = src; *cp; cnt++, cp++)
57 if (*cp == '\'')
58 cnt += 3;
59
60 buf = xmalloc(cnt);
61 bp = buf;
62 while ((c = *src++)) {
63 if (c != '\'')
64 *bp++ = c;
65 else {
66 bp = strcpy(bp, "'\\''");
67 bp += 4;
68 }
69 }
70 *bp = 0;
71 return buf;
72 }
73
74 static struct diff_tempfile {
75 const char *name;
76 char hex[41];
77 char mode[10];
78 char tmp_path[50];
79 } diff_temp[2];
80
81 static void builtin_diff(const char *name,
82 struct diff_tempfile *temp)
83 {
84 int i, next_at;
85 const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
86 const char *diff_arg = "'%s' '%s'||:"; /* "||:" is to return 0 */
87 const char *input_name_sq[2];
88 const char *path0[2];
89 const char *path1[2];
90 const char *name_sq = sq_expand(name);
91 char *cmd;
92
93 /* diff_cmd and diff_arg have 6 %s in total which makes
94 * the sum of these strings 12 bytes larger than required.
95 * we use 2 spaces around diff-opts, and we need to count
96 * terminating NUL, so we subtract 9 here.
97 */
98 int cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
99 strlen(diff_arg) - 9);
100 for (i = 0; i < 2; i++) {
101 input_name_sq[i] = sq_expand(temp[i].name);
102 if (!strcmp(temp[i].name, "/dev/null")) {
103 path0[i] = "/dev/null";
104 path1[i] = "";
105 } else {
106 path0[i] = i ? "b/" : "a/";
107 path1[i] = name_sq;
108 }
109 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
110 strlen(input_name_sq[i]));
111 }
112
113 cmd = xmalloc(cmd_size);
114
115 next_at = 0;
116 next_at += snprintf(cmd+next_at, cmd_size-next_at,
117 diff_cmd,
118 path0[0], path1[0], path0[1], path1[1]);
119 next_at += snprintf(cmd+next_at, cmd_size-next_at,
120 " %s ", diff_opts);
121 next_at += snprintf(cmd+next_at, cmd_size-next_at,
122 diff_arg, input_name_sq[0], input_name_sq[1]);
123
124 if (!path1[0][0])
125 printf("Created: %s (mode:%s)\n", name, temp[1].mode);
126 else if (!path1[1][0])
127 printf("Deleted: %s\n", name);
128 else if (strcmp(temp[0].mode, temp[1].mode))
129 printf("Mode changed: %s (%s->%s)\n", name,
130 temp[0].mode, temp[1].mode);
131 fflush(NULL);
132 execlp("/bin/sh","sh", "-c", cmd, NULL);
133 }
134
135 /*
136 * Given a name and sha1 pair, if the dircache tells us the file in
137 * the work tree has that object contents, return true, so that
138 * prepare_temp_file() does not have to inflate and extract.
139 */
140 static int work_tree_matches(const char *name, const unsigned char *sha1)
141 {
142 struct cache_entry *ce;
143 struct stat st;
144 int pos, len;
145
146 /* We do not read the cache ourselves here, because the
147 * benchmark with my previous version that always reads cache
148 * shows that it makes things worse for diff-tree comparing
149 * two linux-2.6 kernel trees in an already checked out work
150 * tree. This is because most diff-tree comparison deals with
151 * only a small number of files, while reading the cache is
152 * expensive for a large project, and its cost outweighs the
153 * savings we get by not inflating the object to a temporary
154 * file. Practically, this code only helps when we are used
155 * by diff-cache --cached, which does read the cache before
156 * calling us.
157 */
158 if (!active_cache)
159 return 0;
160
161 len = strlen(name);
162 pos = cache_name_pos(name, len);
163 if (pos < 0)
164 return 0;
165 ce = active_cache[pos];
166 if ((stat(name, &st) < 0) ||
167 cache_match_stat(ce, &st) ||
168 memcmp(sha1, ce->sha1, 20))
169 return 0;
170 return 1;
171 }
172
173 static void prepare_temp_file(const char *name,
174 struct diff_tempfile *temp,
175 struct diff_spec *one)
176 {
177 static unsigned char null_sha1[20] = { 0, };
178 int use_work_tree = 0;
179
180 if (!one->file_valid) {
181 not_a_valid_file:
182 /* A '-' entry produces this for file-2, and
183 * a '+' entry produces this for file-1.
184 */
185 temp->name = "/dev/null";
186 strcpy(temp->hex, ".");
187 strcpy(temp->mode, ".");
188 return;
189 }
190
191 if (one->sha1_valid &&
192 (!memcmp(one->blob_sha1, null_sha1, sizeof(null_sha1)) ||
193 work_tree_matches(name, one->blob_sha1)))
194 use_work_tree = 1;
195
196 if (!one->sha1_valid || use_work_tree) {
197 struct stat st;
198 temp->name = name;
199 if (stat(temp->name, &st) < 0) {
200 if (errno == ENOENT)
201 goto not_a_valid_file;
202 die("stat(%s): %s", temp->name, strerror(errno));
203 }
204 if (!one->sha1_valid)
205 strcpy(temp->hex, sha1_to_hex(null_sha1));
206 else
207 strcpy(temp->hex, sha1_to_hex(one->blob_sha1));
208 sprintf(temp->mode, "%06o",
209 S_IFREG |ce_permissions(st.st_mode));
210 }
211 else {
212 int fd;
213 void *blob;
214 char type[20];
215 unsigned long size;
216
217 blob = read_sha1_file(one->blob_sha1, type, &size);
218 if (!blob || strcmp(type, "blob"))
219 die("unable to read blob object for %s (%s)",
220 name, sha1_to_hex(one->blob_sha1));
221
222 strcpy(temp->tmp_path, ".diff_XXXXXX");
223 fd = mkstemp(temp->tmp_path);
224 if (fd < 0)
225 die("unable to create temp-file");
226 if (write(fd, blob, size) != size)
227 die("unable to write temp-file");
228 close(fd);
229 free(blob);
230 temp->name = temp->tmp_path;
231 strcpy(temp->hex, sha1_to_hex(one->blob_sha1));
232 temp->hex[40] = 0;
233 sprintf(temp->mode, "%06o", one->mode);
234 }
235 }
236
237 static void remove_tempfile(void)
238 {
239 int i;
240
241 for (i = 0; i < 2; i++)
242 if (diff_temp[i].name == diff_temp[i].tmp_path) {
243 unlink(diff_temp[i].name);
244 diff_temp[i].name = NULL;
245 }
246 }
247
248 static void remove_tempfile_on_signal(int signo)
249 {
250 remove_tempfile();
251 }
252
253 /* An external diff command takes:
254 *
255 * diff-cmd name infile1 infile1-sha1 infile1-mode \
256 * infile2 infile2-sha1 infile2-mode.
257 *
258 */
259 void run_external_diff(const char *name,
260 struct diff_spec *one,
261 struct diff_spec *two)
262 {
263 struct diff_tempfile *temp = diff_temp;
264 pid_t pid;
265 int status;
266 static int atexit_asked = 0;
267
268 if (one && two) {
269 prepare_temp_file(name, &temp[0], one);
270 prepare_temp_file(name, &temp[1], two);
271 if (! atexit_asked &&
272 (temp[0].name == temp[0].tmp_path ||
273 temp[1].name == temp[1].tmp_path)) {
274 atexit_asked = 1;
275 atexit(remove_tempfile);
276 }
277 signal(SIGINT, remove_tempfile_on_signal);
278 }
279
280 fflush(NULL);
281 pid = fork();
282 if (pid < 0)
283 die("unable to fork");
284 if (!pid) {
285 const char *pgm = external_diff();
286 if (pgm) {
287 if (one && two)
288 execlp(pgm, pgm,
289 name,
290 temp[0].name, temp[0].hex, temp[0].mode,
291 temp[1].name, temp[1].hex, temp[1].mode,
292 NULL);
293 else
294 execlp(pgm, pgm, name, NULL);
295 }
296 /*
297 * otherwise we use the built-in one.
298 */
299 if (one && two)
300 builtin_diff(name, temp);
301 else
302 printf("* Unmerged path %s\n", name);
303 exit(0);
304 }
305 if (waitpid(pid, &status, 0) < 0 ||
306 !WIFEXITED(status) || WEXITSTATUS(status)) {
307 /* Earlier we did not check the exit status because
308 * diff exits non-zero if files are different, and
309 * we are not interested in knowing that. It was a
310 * mistake which made it harder to quit a diff-*
311 * session that uses the git-apply-patch-script as
312 * the GIT_EXTERNAL_DIFF. A custom GIT_EXTERNAL_DIFF
313 * should also exit non-zero only when it wants to
314 * abort the entire diff-* session.
315 */
316 remove_tempfile();
317 fprintf(stderr, "external diff died, stopping at %s.\n", name);
318 exit(1);
319 }
320 remove_tempfile();
321 }
322
323 void diff_addremove(int addremove, unsigned mode,
324 const unsigned char *sha1,
325 const char *base, const char *path)
326 {
327 char concatpath[PATH_MAX];
328 struct diff_spec spec[2], *one, *two;
329
330 memcpy(spec[0].blob_sha1, sha1, 20);
331 spec[0].mode = mode;
332 spec[0].sha1_valid = spec[0].file_valid = 1;
333 spec[1].file_valid = 0;
334
335 if (addremove == '+') {
336 one = spec + 1; two = spec;
337 } else {
338 one = spec; two = one + 1;
339 }
340
341 if (path) {
342 strcpy(concatpath, base);
343 strcat(concatpath, path);
344 }
345 run_external_diff(path ? concatpath : base, one, two);
346 }
347
348 void diff_change(unsigned old_mode, unsigned new_mode,
349 const unsigned char *old_sha1,
350 const unsigned char *new_sha1,
351 const char *base, const char *path) {
352 char concatpath[PATH_MAX];
353 struct diff_spec spec[2];
354
355 memcpy(spec[0].blob_sha1, old_sha1, 20);
356 spec[0].mode = old_mode;
357 memcpy(spec[1].blob_sha1, new_sha1, 20);
358 spec[1].mode = new_mode;
359 spec[0].sha1_valid = spec[0].file_valid = 1;
360 spec[1].sha1_valid = spec[1].file_valid = 1;
361
362 if (path) {
363 strcpy(concatpath, base);
364 strcat(concatpath, path);
365 }
366 run_external_diff(path ? concatpath : base, &spec[0], &spec[1]);
367 }
368
369 void diff_unmerge(const char *path)
370 {
371 run_external_diff(path, NULL, NULL);
372 }