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,
  83                          struct diff_tempfile *temp)
  84 {
  85         int i, next_at;
  86         const char *diff_cmd = "diff -L'%s%s' -L'%s%s'";
  87         const char *diff_arg  = "'%s' '%s'||:"; /* "||:" is to return 0 */
  88         const char *input_name_sq[2];
  89         const char *path0[2];
  90         const char *path1[2];
  91         const char *name_sq = sq_expand(name);
  92         char *cmd;
  93
  94         /* diff_cmd and diff_arg have 6 %s in total which makes
  95          * the sum of these strings 12 bytes larger than required.
  96          * we use 2 spaces around diff-opts, and we need to count
  97          * terminating NUL, so we subtract 9 here.
  98          */
  99         int cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
 100                         strlen(diff_arg) - 9);
 101         for (i = 0; i < 2; i++) {
 102                 input_name_sq[i] = sq_expand(temp[i].name);
 103                 if (!strcmp(temp[i].name, "/dev/null")) {
 104                         path0[i] = "/dev/null";
 105                         path1[i] = "";
 106                 } else {
 107                         path0[i] = i ? "b/" : "a/";
 108                         path1[i] = name_sq;
 109                 }
 110                 cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
 111                              strlen(input_name_sq[i]));
 112         }
 113
 114         cmd = xmalloc(cmd_size);
 115
 116         next_at = 0;
 117         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 118                             diff_cmd,
 119                             path0[0], path1[0], path0[1], path1[1]);
 120         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 121                             " %s ", diff_opts);
 122         next_at += snprintf(cmd+next_at, cmd_size-next_at,
 123                             diff_arg, input_name_sq[0], input_name_sq[1]);
 124
 125         if (!path1[0][0])
 126                 printf("Created: %s (mode:%s)\n", name, temp[1].mode);
 127         else if (!path1[1][0])
 128                 printf("Deleted: %s\n", name);
 129         else if (strcmp(temp[0].mode, temp[1].mode)) {
 130                 printf("Mode changed: %s (%s->%s)\n", name,
 131                        temp[0].mode, temp[1].mode);
 132                 /* Be careful.  We do not want to diff between
 133                  * symlink and a file.
 134                  */
 135                 if (strncmp(temp[0].mode, "120", 3) !=
 136                     strncmp(temp[1].mode, "120", 3))
 137                         exit(0);
 138         }
 139         fflush(NULL);
 140         execlp("/bin/sh","sh", "-c", cmd, NULL);
 141 }
 142
 143 /*
 144  * Given a name and sha1 pair, if the dircache tells us the file in
 145  * the work tree has that object contents, return true, so that
 146  * prepare_temp_file() does not have to inflate and extract.
 147  */
 148 static int work_tree_matches(const char *name, const unsigned char *sha1)
 149 {
 150         struct cache_entry *ce;
 151         struct stat st;
 152         int pos, len;
 153
 154         /* We do not read the cache ourselves here, because the
 155          * benchmark with my previous version that always reads cache
 156          * shows that it makes things worse for diff-tree comparing
 157          * two linux-2.6 kernel trees in an already checked out work
 158          * tree.  This is because most diff-tree comparison deals with
 159          * only a small number of files, while reading the cache is
 160          * expensive for a large project, and its cost outweighs the
 161          * savings we get by not inflating the object to a temporary
 162          * file.  Practically, this code only helps when we are used
 163          * by diff-cache --cached, which does read the cache before
 164          * calling us.
 165          */
 166         if (!active_cache)
 167                 return 0;
 168
 169         len = strlen(name);
 170         pos = cache_name_pos(name, len);
 171         if (pos < 0)
 172                 return 0;
 173         ce = active_cache[pos];
 174         if ((lstat(name, &st) < 0) ||
 175             !S_ISREG(st.st_mode) ||
 176             cache_match_stat(ce, &st) ||
 177             memcmp(sha1, ce->sha1, 20))
 178                 return 0;
 179         return 1;
 180 }
 181
 182 static void prep_temp_blob(struct diff_tempfile *temp,
 183                            void *blob,
 184                            unsigned long size,
 185                            unsigned char *sha1,
 186                            int mode)
 187 {
 188         int fd;
 189
 190         strcpy(temp->tmp_path, ".diff_XXXXXX");
 191         fd = mkstemp(temp->tmp_path);
 192         if (fd < 0)
 193                 die("unable to create temp-file");
 194         if (write(fd, blob, size) != size)
 195                 die("unable to write temp-file");
 196         close(fd);
 197         temp->name = temp->tmp_path;
 198         strcpy(temp->hex, sha1_to_hex(sha1));
 199         temp->hex[40] = 0;
 200         sprintf(temp->mode, "%06o", mode);
 201 }
 202
 203 static void prepare_temp_file(const char *name,
 204                               struct diff_tempfile *temp,
 205                               struct diff_spec *one)
 206 {
 207         static unsigned char null_sha1[20] = { 0, };
 208         int use_work_tree = 0;
 209
 210         if (!one->file_valid) {
 211         not_a_valid_file:
 212                 /* A '-' entry produces this for file-2, and
 213                  * a '+' entry produces this for file-1.
 214                  */
 215                 temp->name = "/dev/null";
 216                 strcpy(temp->hex, ".");
 217                 strcpy(temp->mode, ".");
 218                 return;
 219         }
 220
 221         if (one->sha1_valid &&
 222             (!memcmp(one->blob_sha1, null_sha1, sizeof(null_sha1)) ||
 223              work_tree_matches(name, one->blob_sha1)))
 224                 use_work_tree = 1;
 225
 226         if (!one->sha1_valid || use_work_tree) {
 227                 struct stat st;
 228                 temp->name = name;
 229                 if (lstat(temp->name, &st) < 0) {
 230                         if (errno == ENOENT)
 231                                 goto not_a_valid_file;
 232                         die("stat(%s): %s", temp->name, strerror(errno));
 233                 }
 234                 if (S_ISLNK(st.st_mode)) {
 235                         int ret;
 236                         char *buf, buf_[1024];
 237                         buf = ((sizeof(buf_) < st.st_size) ?
 238                                xmalloc(st.st_size) : buf_);
 239                         ret = readlink(name, buf, st.st_size);
 240                         if (ret < 0)
 241                                 die("readlink(%s)", name);
 242                         prep_temp_blob(temp, buf, st.st_size,
 243                                        (one->sha1_valid ?
 244                                         one->blob_sha1 : null_sha1),
 245                                        (one->sha1_valid ?
 246                                         one->mode : S_IFLNK));
 247                 }
 248                 else {
 249                         if (!one->sha1_valid)
 250                                 strcpy(temp->hex, sha1_to_hex(null_sha1));
 251                         else
 252                                 strcpy(temp->hex, sha1_to_hex(one->blob_sha1));
 253                         sprintf(temp->mode, "%06o",
 254                                 S_IFREG |ce_permissions(st.st_mode));
 255                 }
 256                 return;
 257         }
 258         else {
 259                 void *blob;
 260                 char type[20];
 261                 unsigned long size;
 262
 263                 blob = read_sha1_file(one->blob_sha1, type, &size);
 264                 if (!blob || strcmp(type, "blob"))
 265                         die("unable to read blob object for %s (%s)",
 266                             name, sha1_to_hex(one->blob_sha1));
 267                 prep_temp_blob(temp, blob, size, one->blob_sha1, one->mode);
 268                 free(blob);
 269         }
 270 }
 271
 272 static void remove_tempfile(void)
 273 {
 274         int i;
 275
 276         for (i = 0; i < 2; i++)
 277                 if (diff_temp[i].name == diff_temp[i].tmp_path) {
 278                         unlink(diff_temp[i].name);
 279                         diff_temp[i].name = NULL;
 280                 }
 281 }
 282
 283 static void remove_tempfile_on_signal(int signo)
 284 {
 285         remove_tempfile();
 286 }
 287
 288 /* An external diff command takes:
 289  *
 290  * diff-cmd name infile1 infile1-sha1 infile1-mode \
 291  *               infile2 infile2-sha1 infile2-mode.
 292  *
 293  */
 294 void run_external_diff(const char *name,
 295                        struct diff_spec *one,
 296                        struct diff_spec *two)
 297 {
 298         struct diff_tempfile *temp = diff_temp;
 299         pid_t pid;
 300         int status;
 301         static int atexit_asked = 0;
 302
 303         if (one && two) {
 304                 prepare_temp_file(name, &temp[0], one);
 305                 prepare_temp_file(name, &temp[1], two);
 306                 if (! atexit_asked &&
 307                     (temp[0].name == temp[0].tmp_path ||
 308                      temp[1].name == temp[1].tmp_path)) {
 309                         atexit_asked = 1;
 310                         atexit(remove_tempfile);
 311                 }
 312                 signal(SIGINT, remove_tempfile_on_signal);
 313         }
 314
 315         fflush(NULL);
 316         pid = fork();
 317         if (pid < 0)
 318                 die("unable to fork");
 319         if (!pid) {
 320                 const char *pgm = external_diff();
 321                 if (pgm) {
 322                         if (one && two)
 323                                 execlp(pgm, pgm,
 324                                        name,
 325                                        temp[0].name, temp[0].hex, temp[0].mode,
 326                                        temp[1].name, temp[1].hex, temp[1].mode,
 327                                        NULL);
 328                         else
 329                                 execlp(pgm, pgm, name, NULL);
 330                 }
 331                 /*
 332                  * otherwise we use the built-in one.
 333                  */
 334                 if (one && two)
 335                         builtin_diff(name, temp);
 336                 else
 337                         printf("* Unmerged path %s\n", name);
 338                 exit(0);
 339         }
 340         if (waitpid(pid, &status, 0) < 0 ||
 341             !WIFEXITED(status) || WEXITSTATUS(status)) {
 342                 /* Earlier we did not check the exit status because
 343                  * diff exits non-zero if files are different, and
 344                  * we are not interested in knowing that.  It was a
 345                  * mistake which made it harder to quit a diff-*
 346                  * session that uses the git-apply-patch-script as
 347                  * the GIT_EXTERNAL_DIFF.  A custom GIT_EXTERNAL_DIFF
 348                  * should also exit non-zero only when it wants to
 349                  * abort the entire diff-* session.
 350                  */
 351                 remove_tempfile();
 352                 fprintf(stderr, "external diff died, stopping at %s.\n", name);
 353                 exit(1);
 354         }
 355         remove_tempfile();
 356 }
 357
 358 void diff_addremove(int addremove, unsigned mode,
 359                     const unsigned char *sha1,
 360                     const char *base, const char *path)
 361 {
 362         char concatpath[PATH_MAX];
 363         struct diff_spec spec[2], *one, *two;
 364
 365         memcpy(spec[0].blob_sha1, sha1, 20);
 366         spec[0].mode = mode;
 367         spec[0].sha1_valid = spec[0].file_valid = 1;
 368         spec[1].file_valid = 0;
 369
 370         if (addremove == '+') {
 371                 one = spec + 1; two = spec;
 372         } else {
 373                 one = spec; two = one + 1;
 374         }
 375
 376         if (path) {
 377                 strcpy(concatpath, base);
 378                 strcat(concatpath, path);
 379         }
 380         run_external_diff(path ? concatpath : base, one, two);
 381 }
 382
 383 void diff_change(unsigned old_mode, unsigned new_mode,
 384                  const unsigned char *old_sha1,
 385                  const unsigned char *new_sha1,
 386                  const char *base, const char *path) {
 387         char concatpath[PATH_MAX];
 388         struct diff_spec spec[2];
 389
 390         memcpy(spec[0].blob_sha1, old_sha1, 20);
 391         spec[0].mode = old_mode;
 392         memcpy(spec[1].blob_sha1, new_sha1, 20);
 393         spec[1].mode = new_mode;
 394         spec[0].sha1_valid = spec[0].file_valid = 1;
 395         spec[1].sha1_valid = spec[1].file_valid = 1;
 396
 397         if (path) {
 398                 strcpy(concatpath, base);
 399                 strcat(concatpath, path);
 400         }
 401         run_external_diff(path ? concatpath : base, &spec[0], &spec[1]);
 402 }
 403
 404 void diff_unmerge(const char *path)
 405 {
 406         run_external_diff(path, NULL, NULL);
 407 }