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 <limits.h>
   8 #include "cache.h"
   9 #include "diff.h"
  10
  11 static const char *diff_opts = "-pu";
  12
  13 static const char *external_diff(void)
  14 {
  15         static const char *external_diff_cmd = NULL;
  16         static int done_preparing = 0;
  17
  18         if (done_preparing)
  19                 return external_diff_cmd;
  20
  21         /*
  22          * Default values above are meant to match the
  23          * Linux kernel development style.  Examples of
  24          * alternative styles you can specify via environment
  25          * variables are:
  26          *
  27          * GIT_DIFF_OPTS="-c";
  28          */
  29         if (gitenv("GIT_EXTERNAL_DIFF"))
  30                 external_diff_cmd = gitenv("GIT_EXTERNAL_DIFF");
  31
  32         /* In case external diff fails... */
  33         diff_opts = gitenv("GIT_DIFF_OPTS") ? : diff_opts;
  34
  35         done_preparing = 1;
  36         return external_diff_cmd;
  37 }
  38
  39 /* Help to copy the thing properly quoted for the shell safety.
  40  * any single quote is replaced with '\'', and the caller is
  41  * expected to enclose the result within a single quote pair.
  42  *
  43  * E.g.
  44  *  original     sq_expand     result
  45  *  name     ==> name      ==> 'name'
  46  *  a b      ==> a b       ==> 'a b'
  47  *  a'b      ==> a'\''b    ==> 'a'\''b'
  48  */
  49 static char *sq_expand(const char *src)
  50 {
  51         static char *buf = NULL;
  52         int cnt, c;
  53         const char *cp;
  54         char *bp;
  55
  56         /* count bytes needed to store the quoted string. */
  57         for (cnt = 1, cp = src; *cp; cnt++, cp++)
  58                 if (*cp == '\'')
  59                         cnt += 3;
  60
  61         buf = xmalloc(cnt);
  62         bp = buf;
  63         while ((c = *src++)) {
  64                 if (c != '\'')
  65                         *bp++ = c;
  66                 else {
  67                         bp = strcpy(bp, "'\\''");
  68                         bp += 4;
  69                 }
  70         }
  71         *bp = 0;
  72         return buf;
  73 }
  74
  75 static struct diff_tempfile {
  76         const char *name;
  77         char hex[41];
  78         char mode[10];
  79         char tmp_path[50];
  80 } diff_temp[2];
  81
  82 static void builtin_diff(const char *name_a,
  83                          const char *name_b,
  84                          struct diff_tempfile *temp)
  85 {
  86         int i, next_at;
  87         const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
  88         const char *diff_arg  = "'%s' '%s'||:"; /* "||:" is to return 0 */
  89         const char *input_name_sq[2];
  90         const char *path0[2];
  91         const char *path1[2];
  92         const char *name_sq[2];
  93         char *cmd;
  94
  95         name_sq[0] = sq_expand(name_a);
  96         name_sq[1] = sq_expand(name_b);
  97
  98         /* diff_cmd and diff_arg have 6 %s in total which makes
  99          * the sum of these strings 12 bytes larger than required.
 100          * we use 2 spaces around diff-opts, and we need to count
 101          * terminating NUL, so we subtract 9 here.
 102          */
 103         int cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
 104                         strlen(diff_arg) - 9);
 105         for (i = 0; i < 2; i++) {
 106                 input_name_sq[i] = sq_expand(temp[i].name);
 107                 if (!strcmp(temp[i].name, "/dev/null")) {
 108                         path0[i] = "/dev/null";
 109                         path1[i] = "";
 110                 } else {
 111                         path0[i] = i ? "b/" : "a/";
 112                         path1[i] = name_sq[i];
 113                 }
 114                 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
 115                              strlen(input_name_sq[i]));
 116         }
 117
 118         cmd = xmalloc(cmd_size);
 119
 120         next_at = 0;
 121         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 122                             diff_cmd,
 123                             path0[0], path1[0], path0[1], path1[1]);
 124         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 125                             " %s ", diff_opts);
 126         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 127                             diff_arg, input_name_sq[0], input_name_sq[1]);
 128
 129         printf("diff --git a/%s b/%s\n", name_a, name_b);
 130         if (!path1[0][0])
 131                 printf("new file mode %s\n", temp[1].mode);
 132         else if (!path1[1][0])
 133                 printf("deleted file mode %s\n", temp[0].mode);
 134         else {
 135                 if (strcmp(temp[0].mode, temp[1].mode)) {
 136                         printf("old mode %s\n", temp[0].mode);
 137                         printf("new mode %s\n", temp[1].mode);
 138                 }
 139                 if (strcmp(name_a, name_b)) {
 140                         printf("rename old %s\n", name_a);
 141                         printf("rename new %s\n", name_b);
 142                 }
 143                 if (strncmp(temp[0].mode, temp[1].mode, 3))
 144                         /* we do not run diff between different kind
 145                          * of objects.
 146                          */
 147                         exit(0);
 148         }
 149         fflush(NULL);
 150         execlp("/bin/sh","sh", "-c", cmd, NULL);
 151 }
 152
 153 /*
 154  * Given a name and sha1 pair, if the dircache tells us the file in
 155  * the work tree has that object contents, return true, so that
 156  * prepare_temp_file() does not have to inflate and extract.
 157  */
 158 static int work_tree_matches(const char *name, const unsigned char *sha1)
 159 {
 160         struct cache_entry *ce;
 161         struct stat st;
 162         int pos, len;
 163
 164         /* We do not read the cache ourselves here, because the
 165          * benchmark with my previous version that always reads cache
 166          * shows that it makes things worse for diff-tree comparing
 167          * two linux-2.6 kernel trees in an already checked out work
 168          * tree.  This is because most diff-tree comparisons deal with
 169          * only a small number of files, while reading the cache is
 170          * expensive for a large project, and its cost outweighs the
 171          * savings we get by not inflating the object to a temporary
 172          * file.  Practically, this code only helps when we are used
 173          * by diff-cache --cached, which does read the cache before
 174          * calling us.
 175          */
 176         if (!active_cache)
 177                 return 0;
 178
 179         len = strlen(name);
 180         pos = cache_name_pos(name, len);
 181         if (pos < 0)
 182                 return 0;
 183         ce = active_cache[pos];
 184         if ((lstat(name, &st) < 0) ||
 185             !S_ISREG(st.st_mode) ||
 186             ce_match_stat(ce, &st) ||
 187             memcmp(sha1, ce->sha1, 20))
 188                 return 0;
 189         return 1;
 190 }
 191
 192 static void prep_temp_blob(struct diff_tempfile *temp,
 193                            void *blob,
 194                            unsigned long size,
 195                            unsigned char *sha1,
 196                            int mode)
 197 {
 198         int fd;
 199
 200         strcpy(temp->tmp_path, ".diff_XXXXXX");
 201         fd = mkstemp(temp->tmp_path);
 202         if (fd < 0)
 203                 die("unable to create temp-file");
 204         if (write(fd, blob, size) != size)
 205                 die("unable to write temp-file");
 206         close(fd);
 207         temp->name = temp->tmp_path;
 208         strcpy(temp->hex, sha1_to_hex(sha1));
 209         temp->hex[40] = 0;
 210         sprintf(temp->mode, "%06o", mode);
 211 }
 212
 213 static void prepare_temp_file(const char *name,
 214                               struct diff_tempfile *temp,
 215                               struct diff_spec *one)
 216 {
 217         static unsigned char null_sha1[20] = { 0, };
 218         int use_work_tree = 0;
 219
 220         if (!one->file_valid) {
 221         not_a_valid_file:
 222                 /* A '-' entry produces this for file-2, and
 223                  * a '+' entry produces this for file-1.
 224                  */
 225                 temp->name = "/dev/null";
 226                 strcpy(temp->hex, ".");
 227                 strcpy(temp->mode, ".");
 228                 return;
 229         }
 230
 231         if (one->sha1_valid &&
 232             (!memcmp(one->blob_sha1, null_sha1, sizeof(null_sha1)) ||
 233              work_tree_matches(name, one->blob_sha1)))
 234                 use_work_tree = 1;
 235
 236         if (!one->sha1_valid || use_work_tree) {
 237                 struct stat st;
 238                 temp->name = name;
 239                 if (lstat(temp->name, &st) < 0) {
 240                         if (errno == ENOENT)
 241                                 goto not_a_valid_file;
 242                         die("stat(%s): %s", temp->name, strerror(errno));
 243                 }
 244                 if (S_ISLNK(st.st_mode)) {
 245                         int ret;
 246                         char *buf, buf_[1024];
 247                         buf = ((sizeof(buf_) < st.st_size) ?
 248                                xmalloc(st.st_size) : buf_);
 249                         ret = readlink(name, buf, st.st_size);
 250                         if (ret < 0)
 251                                 die("readlink(%s)", name);
 252                         prep_temp_blob(temp, buf, st.st_size,
 253                                        (one->sha1_valid ?
 254                                         one->blob_sha1 : null_sha1),
 255                                        (one->sha1_valid ?
 256                                         one->mode : S_IFLNK));
 257                 }
 258                 else {
 259                         if (!one->sha1_valid)
 260                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 261                         else
 262                                 strcpy(temp->hex, sha1_to_hex(one->blob_sha1));
 263                         sprintf(temp->mode, "%06o",
 264                                 S_IFREG |ce_permissions(st.st_mode));
 265                 }
 266                 return;
 267         }
 268         else {
 269                 void *blob;
 270                 char type[20];
 271                 unsigned long size;
 272
 273                 blob = read_sha1_file(one->blob_sha1, type, &size);
 274                 if (!blob || strcmp(type, "blob"))
 275                         die("unable to read blob object for %s (%s)",
 276                             name, sha1_to_hex(one->blob_sha1));
 277                 prep_temp_blob(temp, blob, size, one->blob_sha1, one->mode);
 278                 free(blob);
 279         }
 280 }
 281
 282 static void remove_tempfile(void)
 283 {
 284         int i;
 285
 286         for (i = 0; i < 2; i++)
 287                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 288                         unlink(diff_temp[i].name);
 289                         diff_temp[i].name = NULL;
 290                 }
 291 }
 292
 293 static void remove_tempfile_on_signal(int signo)
 294 {
 295         remove_tempfile();
 296 }
 297
 298 /* An external diff command takes:
 299  *
 300  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 301  *               infile2 infile2-sha1 infile2-mode.
 302  *
 303  */
 304 void run_external_diff(const char *name,
 305                        const char *other,
 306                        struct diff_spec *one,
 307                        struct diff_spec *two)
 308 {
 309         struct diff_tempfile *temp = diff_temp;
 310         pid_t pid;
 311         int status;
 312         static int atexit_asked = 0;
 313
 314         if (one && two) {
 315                 prepare_temp_file(name, &temp[0], one);
 316                 prepare_temp_file(other ? : name, &temp[1], two);
 317                 if (! atexit_asked &&
 318                     (temp[0].name == temp[0].tmp_path ||
 319                      temp[1].name == temp[1].tmp_path)) {
 320                         atexit_asked = 1;
 321                         atexit(remove_tempfile);
 322                 }
 323                 signal(SIGINT, remove_tempfile_on_signal);
 324         }
 325
 326         fflush(NULL);
 327         pid = fork();
 328         if (pid < 0)
 329                 die("unable to fork");
 330         if (!pid) {
 331                 const char *pgm = external_diff();
 332                 /* not passing rename patch to external ones */
 333                 if (!other && pgm) {
 334                         if (one && two)
 335                                 execlp(pgm, pgm,
 336                                        name,
 337                                        temp[0].name, temp[0].hex, temp[0].mode,
 338                                        temp[1].name, temp[1].hex, temp[1].mode,
 339                                        NULL);
 340                         else
 341                                 execlp(pgm, pgm, name, NULL);
 342                 }
 343                 /*
 344                  * otherwise we use the built-in one.
 345                  */
 346                 if (one && two)
 347                         builtin_diff(name, other ? : name, temp);
 348                 else
 349                         printf("* Unmerged path %s\n", name);
 350                 exit(0);
 351         }
 352         if (waitpid(pid, &status, 0) < 0 ||
 353             !WIFEXITED(status) || WEXITSTATUS(status)) {
 354                 /* Earlier we did not check the exit status because
 355                  * diff exits non-zero if files are different, and
 356                  * we are not interested in knowing that.  It was a
 357                  * mistake which made it harder to quit a diff-*
 358                  * session that uses the git-apply-patch-script as
 359                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 360                  * should also exit non-zero only when it wants to
 361                  * abort the entire diff-* session.
 362                  */
 363                 remove_tempfile();
 364                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 365                 exit(1);
 366         }
 367         remove_tempfile();
 368 }
 369
 370 void diff_addremove(int addremove, unsigned mode,
 371                     const unsigned char *sha1,
 372                     const char *base, const char *path)
 373 {
 374         char concatpath[PATH_MAX];
 375         struct diff_spec spec[2], *one, *two;
 376
 377         memcpy(spec[0].blob_sha1, sha1, 20);
 378         spec[0].mode = mode;
 379         spec[0].sha1_valid = spec[0].file_valid = 1;
 380         spec[1].file_valid = 0;
 381
 382         if (addremove == '+') {
 383                 one = spec + 1; two = spec;
 384         } else {
 385                 one = spec; two = one + 1;
 386         }
 387
 388         if (path) {
 389                 strcpy(concatpath, base);
 390                 strcat(concatpath, path);
 391         }
 392         run_external_diff(path ? concatpath : base, NULL, one, two);
 393 }
 394
 395 void diff_change(unsigned old_mode, unsigned new_mode,
 396                  const unsigned char *old_sha1,
 397                  const unsigned char *new_sha1,
 398                  const char *base, const char *path) {
 399         char concatpath[PATH_MAX];
 400         struct diff_spec spec[2];
 401
 402         memcpy(spec[0].blob_sha1, old_sha1, 20);
 403         spec[0].mode = old_mode;
 404         memcpy(spec[1].blob_sha1, new_sha1, 20);
 405         spec[1].mode = new_mode;
 406         spec[0].sha1_valid = spec[0].file_valid = 1;
 407         spec[1].sha1_valid = spec[1].file_valid = 1;
 408
 409         if (path) {
 410                 strcpy(concatpath, base);
 411                 strcat(concatpath, path);
 412         }
 413         run_external_diff(path ? concatpath : base, NULL, &spec[0], &spec[1]);
 414 }
 415
 416 void diff_unmerge(const char *path)
 417 {
 418         run_external_diff(path, NULL, NULL, NULL);
 419 }