combine-diff: better hunk splitting.
[gitweb.git] / diff.c
diff --git a/diff.c b/diff.c
index cbb86320a6b98ea57712a78c558c7dc84613087b..17d68fa699f7c88aef8e6c1a0812575bfaec70e1 100644 (file)
--- a/diff.c
+++ b/diff.c
@@ -13,6 +13,58 @@ static const char *diff_opts = "-pu";
 
 static int use_size_cache;
 
+int diff_rename_limit_default = -1;
+
+int git_diff_config(const char *var, const char *value)
+{
+       if (!strcmp(var, "diff.renamelimit")) {
+               diff_rename_limit_default = git_config_int(var, value);
+               return 0;
+       }
+
+       return git_default_config(var, value);
+}
+
+static char *quote_one(const char *str)
+{
+       int needlen;
+       char *xp;
+
+       if (!str)
+               return NULL;
+       needlen = quote_c_style(str, NULL, NULL, 0);
+       if (!needlen)
+               return strdup(str);
+       xp = xmalloc(needlen + 1);
+       quote_c_style(str, xp, NULL, 0);
+       return xp;
+}
+
+static char *quote_two(const char *one, const char *two)
+{
+       int need_one = quote_c_style(one, NULL, NULL, 1);
+       int need_two = quote_c_style(two, NULL, NULL, 1);
+       char *xp;
+
+       if (need_one + need_two) {
+               if (!need_one) need_one = strlen(one);
+               if (!need_two) need_one = strlen(two);
+
+               xp = xmalloc(need_one + need_two + 3);
+               xp[0] = '"';
+               quote_c_style(one, xp + 1, NULL, 1);
+               quote_c_style(two, xp + need_one + 1, NULL, 1);
+               strcpy(xp + need_one + need_two + 1, "\"");
+               return xp;
+       }
+       need_one = strlen(one);
+       need_two = strlen(two);
+       xp = xmalloc(need_one + need_two + 1);
+       strcpy(xp, one);
+       strcpy(xp + need_one, two);
+       return xp;
+}
+
 static const char *external_diff(void)
 {
        static const char *external_diff_cmd = NULL;
@@ -133,55 +185,52 @@ static void builtin_diff(const char *name_a,
                         int complete_rewrite)
 {
        int i, next_at, cmd_size;
-       const char *const diff_cmd = "diff -L%s%s -L%s%s";
-       const char *const diff_arg  = "%s %s||:"; /* "||:" is to return 0 */
+       const char *const diff_cmd = "diff -L%s -L%s";
+       const char *const diff_arg  = "-- %s %s||:"; /* "||:" is to return 0 */
        const char *input_name_sq[2];
-       const char *path0[2];
-       const char *path1[2];
-       const char *name_sq[2];
+       const char *label_path[2];
        char *cmd;
 
-       name_sq[0] = sq_quote(name_a);
-       name_sq[1] = sq_quote(name_b);
-
-       /* diff_cmd and diff_arg have 6 %s in total which makes
-        * the sum of these strings 12 bytes larger than required.
+       /* diff_cmd and diff_arg have 4 %s in total which makes
+        * the sum of these strings 8 bytes larger than required.
         * we use 2 spaces around diff-opts, and we need to count
-        * terminating NUL, so we subtract 9 here.
+        * terminating NUL; we used to subtract 5 here, but we do not
+        * care about small leaks in this subprocess that is about
+        * to exec "diff" anymore.
         */
-       cmd_size = (strlen(diff_cmd) + strlen(diff_opts) +
-                       strlen(diff_arg) - 9);
+       cmd_size = (strlen(diff_cmd) + strlen(diff_opts) + strlen(diff_arg)
+                   + 128);
+
        for (i = 0; i < 2; i++) {
                input_name_sq[i] = sq_quote(temp[i].name);
-               if (!strcmp(temp[i].name, "/dev/null")) {
-                       path0[i] = "/dev/null";
-                       path1[i] = "";
-               } else {
-                       path0[i] = i ? "b/" : "a/";
-                       path1[i] = name_sq[i];
-               }
-               cmd_size += (strlen(path0[i]) + strlen(path1[i]) +
-                            strlen(input_name_sq[i]));
+               if (!strcmp(temp[i].name, "/dev/null"))
+                       label_path[i] = "/dev/null";
+               else if (!i)
+                       label_path[i] = sq_quote(quote_two("a/", name_a));
+               else
+                       label_path[i] = sq_quote(quote_two("b/", name_b));
+               cmd_size += (strlen(label_path[i]) + strlen(input_name_sq[i]));
        }
 
        cmd = xmalloc(cmd_size);
 
        next_at = 0;
        next_at += snprintf(cmd+next_at, cmd_size-next_at,
-                           diff_cmd,
-                           path0[0], path1[0], path0[1], path1[1]);
+                           diff_cmd, label_path[0], label_path[1]);
        next_at += snprintf(cmd+next_at, cmd_size-next_at,
                            " %s ", diff_opts);
        next_at += snprintf(cmd+next_at, cmd_size-next_at,
                            diff_arg, input_name_sq[0], input_name_sq[1]);
 
-       printf("diff --git a/%s b/%s\n", name_a, name_b);
-       if (!path1[0][0]) {
+       printf("diff --git %s %s\n",
+              quote_two("a/", name_a), quote_two("b/", name_b));
+       if (label_path[0][0] == '/') {
+               /* dev/null */
                printf("new file mode %s\n", temp[1].mode);
                if (xfrm_msg && xfrm_msg[0])
                        puts(xfrm_msg);
        }
-       else if (!path1[1][0]) {
+       else if (label_path[1][0] == '/') {
                printf("deleted file mode %s\n", temp[0].mode);
                if (xfrm_msg && xfrm_msg[0])
                        puts(xfrm_msg);
@@ -455,9 +504,9 @@ static void prepare_temp_file(const char *name,
                }
                if (S_ISLNK(st.st_mode)) {
                        int ret;
-                       char *buf, buf_[1024];
-                       buf = ((sizeof(buf_) < st.st_size) ?
-                              xmalloc(st.st_size) : buf_);
+                       char buf[PATH_MAX + 1]; /* ought to be SYMLINK_MAX */
+                       if (sizeof(buf) <= st.st_size)
+                               die("symlink too long: %s", name);
                        ret = readlink(name, buf, st.st_size);
                        if (ret < 0)
                                die("readlink(%s)", name);
@@ -601,7 +650,7 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
        if (DIFF_FILE_VALID(one)) {
                if (!one->sha1_valid) {
                        struct stat st;
-                       if (stat(one->path, &st) < 0)
+                       if (lstat(one->path, &st) < 0)
                                die("stat %s", one->path);
                        if (index_path(one->sha1, one->path, &st, 0))
                                die("cannot hash %s\n", one->path);
@@ -611,7 +660,7 @@ static void diff_fill_sha1_info(struct diff_filespec *one)
                memset(one->sha1, 0, 20);
 }
 
-static void run_diff(struct diff_filepair *p)
+static void run_diff(struct diff_filepair *p, struct diff_options *o)
 {
        const char *pgm = external_diff();
        char msg[PATH_MAX*2+300], *xfrm_msg;
@@ -619,6 +668,7 @@ static void run_diff(struct diff_filepair *p)
        struct diff_filespec *two;
        const char *name;
        const char *other;
+       char *name_munged, *other_munged;
        int complete_rewrite = 0;
        int len;
 
@@ -631,6 +681,8 @@ static void run_diff(struct diff_filepair *p)
 
        name = p->one->path;
        other = (strcmp(name, p->two->path) ? p->two->path : NULL);
+       name_munged = quote_one(name);
+       other_munged = quote_one(other);
        one = p->one; two = p->two;
 
        diff_fill_sha1_info(one);
@@ -644,7 +696,7 @@ static void run_diff(struct diff_filepair *p)
                                "copy from %s\n"
                                "copy to %s\n",
                                (int)(0.5 + p->score * 100.0/MAX_SCORE),
-                               name, other);
+                               name_munged, other_munged);
                break;
        case DIFF_STATUS_RENAMED:
                len += snprintf(msg + len, sizeof(msg) - len,
@@ -652,7 +704,7 @@ static void run_diff(struct diff_filepair *p)
                                "rename from %s\n"
                                "rename to %s\n",
                                (int)(0.5 + p->score * 100.0/MAX_SCORE),
-                               name, other);
+                               name_munged, other_munged);
                break;
        case DIFF_STATUS_MODIFIED:
                if (p->score) {
@@ -671,10 +723,12 @@ static void run_diff(struct diff_filepair *p)
 
        if (memcmp(one->sha1, two->sha1, 20)) {
                char one_sha1[41];
+               int abbrev = o->full_index ? 40 : DIFF_DEFAULT_INDEX_ABBREV;
                memcpy(one_sha1, sha1_to_hex(one->sha1), 41);
 
                len += snprintf(msg + len, sizeof(msg) - len,
-                               "index %.7s..%.7s", one_sha1,
+                               "index %.*s..%.*s",
+                               abbrev, one_sha1, abbrev,
                                sha1_to_hex(two->sha1));
                if (one->mode == two->mode)
                        len += snprintf(msg + len, sizeof(msg) - len,
@@ -702,6 +756,9 @@ static void run_diff(struct diff_filepair *p)
        else
                run_external_diff(pgm, name, other, one, two, xfrm_msg,
                                  complete_rewrite);
+
+       free(name_munged);
+       free(other_munged);
 }
 
 void diff_setup(struct diff_options *options)
@@ -711,25 +768,33 @@ void diff_setup(struct diff_options *options)
        options->line_termination = '\n';
        options->break_opt = -1;
        options->rename_limit = -1;
+
+       options->change = diff_change;
+       options->add_remove = diff_addremove;
 }
 
 int diff_setup_done(struct diff_options *options)
 {
-       if ((options->find_copies_harder || 0 <= options->rename_limit) &&
-           options->detect_rename != DIFF_DETECT_COPY)
+       if ((options->find_copies_harder &&
+            options->detect_rename != DIFF_DETECT_COPY) ||
+           (0 <= options->rename_limit && !options->detect_rename))
                return -1;
+       if (options->detect_rename && options->rename_limit < 0)
+               options->rename_limit = diff_rename_limit_default;
        if (options->setup & DIFF_SETUP_USE_CACHE) {
                if (!active_cache)
                        /* read-cache does not die even when it fails
                         * so it is safe for us to do this here.  Also
                         * it does not smudge active_cache or active_nr
                         * when it fails, so we do not have to worry about
-                        * cleaning it up oufselves either.
+                        * cleaning it up ourselves either.
                         */
                        read_cache();
        }
        if (options->setup & DIFF_SETUP_USE_SIZE_CACHE)
                use_size_cache = 1;
+       if (options->abbrev <= 0 || 40 < options->abbrev)
+               options->abbrev = 40; /* full */
 
        return 0;
 }
@@ -743,6 +808,8 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
                options->line_termination = 0;
        else if (!strncmp(arg, "-l", 2))
                options->rename_limit = strtoul(arg+2, NULL, 10);
+       else if (!strcmp(arg, "--full-index"))
+               options->full_index = 1;
        else if (!strcmp(arg, "--name-only"))
                options->output_format = DIFF_FORMAT_NAME;
        else if (!strcmp(arg, "--name-status"))
@@ -778,6 +845,10 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
        }
        else if (!strcmp(arg, "--find-copies-harder"))
                options->find_copies_harder = 1;
+       else if (!strcmp(arg, "--abbrev"))
+               options->abbrev = DIFF_DEFAULT_ABBREV;
+       else if (!strncmp(arg, "--abbrev=", 9))
+               options->abbrev = strtoul(arg + 9, NULL, 10);
        else
                return 0;
        return 1;
@@ -785,16 +856,29 @@ int diff_opt_parse(struct diff_options *options, const char **av, int ac)
 
 static int parse_num(const char **cp_p)
 {
-       int num, scale, ch, cnt;
+       unsigned long num, scale;
+       int ch, dot;
        const char *cp = *cp_p;
 
-       cnt = num = 0;
+       num = 0;
        scale = 1;
-       while ('0' <= (ch = *cp) && ch <= '9') {
-               if (cnt++ < 5) {
-                       /* We simply ignore more than 5 digits precision. */
-                       scale *= 10;
-                       num = num * 10 + ch - '0';
+       dot = 0;
+       for(;;) {
+               ch = *cp;
+               if ( !dot && ch == '.' ) {
+                       scale = 1;
+                       dot = 1;
+               } else if ( ch == '%' ) {
+                       scale = dot ? scale*100 : 100;
+                       cp++;   /* % is always at the end */
+                       break;
+               } else if ( ch >= '0' && ch <= '9' ) {
+                       if ( scale < 100000 ) {
+                               scale *= 10;
+                               num = (num*10) + (ch-'0');
+                       }
+               } else {
+                       break;
                }
                cp++;
        }
@@ -803,7 +887,7 @@ static int parse_num(const char **cp_p)
        /* user says num divided by scale and we say internally that
         * is MAX_SCORE * num / scale.
         */
-       return (MAX_SCORE * num / scale);
+       return (num >= scale) ? MAX_SCORE : (MAX_SCORE * num / scale);
 }
 
 int diff_scoreopt_parse(const char *opt)
@@ -871,23 +955,55 @@ void diff_free_filepair(struct diff_filepair *p)
        free(p);
 }
 
+/* This is different from find_unique_abbrev() in that
+ * it needs to deal with 0{40} SHA1.
+ */
+const char *diff_unique_abbrev(const unsigned char *sha1, int len)
+{
+       int abblen;
+       const char *abbrev;
+       if (len == 40)
+               return sha1_to_hex(sha1);
+
+       abbrev = find_unique_abbrev(sha1, len);
+       if (!abbrev) {
+               if (!memcmp(sha1, null_sha1, 20)) {
+                       char *buf = sha1_to_hex(null_sha1);
+                       if (len < 37)
+                               strcpy(buf + len, "...");
+                       return buf;
+               }
+               else 
+                       return sha1_to_hex(sha1);
+       }
+       abblen = strlen(abbrev);
+       if (abblen < 37) {
+               static char hex[41];
+               if (len < abblen && abblen <= len + 2)
+                       sprintf(hex, "%s%.*s", abbrev, len+3-abblen, "..");
+               else
+                       sprintf(hex, "%s...", abbrev);
+               return hex;
+       }
+       return sha1_to_hex(sha1);
+}
+
 static void diff_flush_raw(struct diff_filepair *p,
                           int line_termination,
                           int inter_name_termination,
-                          int output_format)
+                          struct diff_options *options)
 {
        int two_paths;
        char status[10];
+       int abbrev = options->abbrev;
+       const char *path_one, *path_two;
+       int output_format = options->output_format;
 
+       path_one = p->one->path;
+       path_two = p->two->path;
        if (line_termination) {
-               const char *const err =
-                       "path %s cannot be expressed without -z";
-               if (strchr(p->one->path, line_termination) ||
-                   strchr(p->one->path, inter_name_termination))
-                       die(err, p->one->path);
-               if (strchr(p->two->path, line_termination) ||
-                   strchr(p->two->path, inter_name_termination))
-                       die(err, p->two->path);
+               path_one = quote_one(path_one);
+               path_two = quote_one(path_two);
        }
 
        if (p->score)
@@ -912,19 +1028,34 @@ static void diff_flush_raw(struct diff_filepair *p,
        }
        if (output_format != DIFF_FORMAT_NAME_STATUS) {
                printf(":%06o %06o %s ",
-                      p->one->mode, p->two->mode, sha1_to_hex(p->one->sha1));
-               printf("%s ", sha1_to_hex(p->two->sha1));
+                      p->one->mode, p->two->mode,
+                      diff_unique_abbrev(p->one->sha1, abbrev));
+               printf("%s ",
+                      diff_unique_abbrev(p->two->sha1, abbrev));
        }
-       printf("%s%c%s",status, inter_name_termination, p->one->path);
+       printf("%s%c%s", status, inter_name_termination, path_one);
        if (two_paths)
-               printf("%c%s", inter_name_termination, p->two->path);
+               printf("%c%s", inter_name_termination, path_two);
        putchar(line_termination);
+       if (path_one != p->one->path)
+               free((void*)path_one);
+       if (path_two != p->two->path)
+               free((void*)path_two);
 }
 
 static void diff_flush_name(struct diff_filepair *p,
+                           int inter_name_termination,
                            int line_termination)
 {
-       printf("%s%c", p->two->path, line_termination);
+       char *path = p->two->path;
+
+       if (line_termination)
+               path = quote_one(p->two->path);
+       else
+               path = p->two->path;
+       printf("%s%c", path, line_termination);
+       if (p->two->path != path)
+               free(path);
 }
 
 int diff_unmodified_pair(struct diff_filepair *p)
@@ -961,7 +1092,7 @@ int diff_unmodified_pair(struct diff_filepair *p)
        return 0;
 }
 
-static void diff_flush_patch(struct diff_filepair *p)
+static void diff_flush_patch(struct diff_filepair *p, struct diff_options *o)
 {
        if (diff_unmodified_pair(p))
                return;
@@ -970,7 +1101,7 @@ static void diff_flush_patch(struct diff_filepair *p)
            (DIFF_FILE_VALID(p->two) && S_ISDIR(p->two->mode)))
                return; /* no tree diffs in patch format */ 
 
-       run_diff(p);
+       run_diff(p, o);
 }
 
 int diff_queue_is_empty(void)
@@ -1102,16 +1233,18 @@ void diff_flush(struct diff_options *options)
                        die("internal error in diff-resolve-rename-copy");
                switch (diff_output_format) {
                case DIFF_FORMAT_PATCH:
-                       diff_flush_patch(p);
+                       diff_flush_patch(p, options);
                        break;
                case DIFF_FORMAT_RAW:
                case DIFF_FORMAT_NAME_STATUS:
                        diff_flush_raw(p, line_termination,
                                       inter_name_termination,
-                                      diff_output_format);
+                                      options);
                        break;
                case DIFF_FORMAT_NAME:
-                       diff_flush_name(p, line_termination);
+                       diff_flush_name(p,
+                                       inter_name_termination,
+                                       line_termination);
                        break;
                }
                diff_free_filepair(q->queue[i]);