fetch: Speed up fetch of large numbers of refs
[gitweb.git] / builtin-grep.c
index eda2c884870abde5f9d943259916de45d9425a6c..761799d7d0afd62ecba99d703a8595664faa4723 100644 (file)
@@ -10,6 +10,8 @@
 #include "tag.h"
 #include "tree-walk.h"
 #include "builtin.h"
+#include "parse-options.h"
+#include "userdiff.h"
 #include "grep.h"
 #include "quote.h"
 
 #endif
 #endif
 
-static int builtin_grep;
+static char const * const grep_usage[] = {
+       "git grep [options] [-e] <pattern> [<rev>...] [[--] path...]",
+       NULL
+};
 
 static int grep_config(const char *var, const char *value, void *cb)
 {
        struct grep_opt *opt = cb;
 
+       switch (userdiff_config(var, value)) {
+       case 0: break;
+       case -1: return -1;
+       default: return 0;
+       }
+
        if (!strcmp(var, "color.grep")) {
                opt->color = git_config_colorbool(var, value, -1);
                return 0;
@@ -42,26 +53,58 @@ static int grep_config(const char *var, const char *value, void *cb)
        return git_color_default_config(var, value, cb);
 }
 
+/*
+ * Return non-zero if max_depth is negative or path has no more then max_depth
+ * slashes.
+ */
+static int accept_subdir(const char *path, int max_depth)
+{
+       if (max_depth < 0)
+               return 1;
+
+       while ((path = strchr(path, '/')) != NULL) {
+               max_depth--;
+               if (max_depth < 0)
+                       return 0;
+               path++;
+       }
+       return 1;
+}
+
+/*
+ * Return non-zero if name is a subdirectory of match and is not too deep.
+ */
+static int is_subdir(const char *name, int namelen,
+               const char *match, int matchlen, int max_depth)
+{
+       if (matchlen > namelen || strncmp(name, match, matchlen))
+               return 0;
+
+       if (name[matchlen] == '\0') /* exact match */
+               return 1;
+
+       if (!matchlen || match[matchlen-1] == '/' || name[matchlen] == '/')
+               return accept_subdir(name + matchlen + 1, max_depth);
+
+       return 0;
+}
+
 /*
  * git grep pathspecs are somewhat different from diff-tree pathspecs;
  * pathname wildcards are allowed.
  */
-static int pathspec_matches(const char **paths, const char *name)
+static int pathspec_matches(const char **paths, const char *name, int max_depth)
 {
        int namelen, i;
        if (!paths || !*paths)
-               return 1;
+               return accept_subdir(name, max_depth);
        namelen = strlen(name);
        for (i = 0; paths[i]; i++) {
                const char *match = paths[i];
                int matchlen = strlen(match);
                const char *cp, *meta;
 
-               if (!matchlen ||
-                   ((matchlen <= namelen) &&
-                    !strncmp(name, match, matchlen) &&
-                    (match[matchlen-1] == '/' ||
-                     name[matchlen] == '\0' || name[matchlen] == '/')))
+               if (is_subdir(name, namelen, match, matchlen, max_depth))
                        return 1;
                if (!fnmatch(match, name, 0))
                        return 1;
@@ -264,6 +307,17 @@ static int flush_grep(struct grep_opt *opt,
                argc -= 2;
        }
 
+       if (opt->pre_context || opt->post_context) {
+               /*
+                * grep handles hunk marks between files, but we need to
+                * do that ourselves between multiple calls.
+                */
+               if (opt->show_hunk_mark)
+                       write_or_die(1, "--\n", 3);
+               else
+                       opt->show_hunk_mark = 1;
+       }
+
        status = exec_grep(argc, argv);
 
        if (kept_0) {
@@ -389,7 +443,7 @@ static int external_grep(struct grep_opt *opt, const char **paths, int cached)
                int kept;
                if (!S_ISREG(ce->ce_mode))
                        continue;
-               if (!pathspec_matches(paths, ce->name))
+               if (!pathspec_matches(paths, ce->name, opt->max_depth))
                        continue;
                name = ce->name;
                if (name[0] == '-') {
@@ -422,7 +476,8 @@ static int external_grep(struct grep_opt *opt, const char **paths, int cached)
 }
 #endif
 
-static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
+static int grep_cache(struct grep_opt *opt, const char **paths, int cached,
+                     int external_grep_allowed)
 {
        int hit = 0;
        int nr;
@@ -434,7 +489,7 @@ static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
         * we grep through the checked-out files. It tends to
         * be a lot more optimized
         */
-       if (!cached && !builtin_grep) {
+       if (!cached && external_grep_allowed) {
                hit = external_grep(opt, paths, cached);
                if (hit >= 0)
                        return hit;
@@ -446,7 +501,7 @@ static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
                struct cache_entry *ce = active_cache[nr];
                if (!S_ISREG(ce->ce_mode))
                        continue;
-               if (!pathspec_matches(paths, ce->name))
+               if (!pathspec_matches(paths, ce->name, opt->max_depth))
                        continue;
                /*
                 * If CE_VALID is on, we assume worktree file and its cache entry
@@ -506,7 +561,7 @@ static int grep_tree(struct grep_opt *opt, const char **paths,
                        strbuf_addch(&pathbuf, '/');
 
                down = pathbuf.buf + tn_len;
-               if (!pathspec_matches(paths, down))
+               if (!pathspec_matches(paths, down, opt->max_depth))
                        ;
                else if (S_ISREG(entry.mode))
                        hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
@@ -551,25 +606,187 @@ static int grep_object(struct grep_opt *opt, const char **paths,
        die("unable to grep from object of type %s", typename(obj->type));
 }
 
-static const char builtin_grep_usage[] =
-"git grep <option>* [-e] <pattern> <rev>* [[--] <path>...]";
+static int context_callback(const struct option *opt, const char *arg,
+                           int unset)
+{
+       struct grep_opt *grep_opt = opt->value;
+       int value;
+       const char *endp;
 
-static const char emsg_invalid_context_len[] =
-"%s: invalid context length argument";
-static const char emsg_missing_context_len[] =
-"missing context length argument";
-static const char emsg_missing_argument[] =
-"option requires an argument -%s";
+       if (unset) {
+               grep_opt->pre_context = grep_opt->post_context = 0;
+               return 0;
+       }
+       value = strtol(arg, (char **)&endp, 10);
+       if (*endp) {
+               return error("switch `%c' expects a numerical value",
+                            opt->short_name);
+       }
+       grep_opt->pre_context = grep_opt->post_context = value;
+       return 0;
+}
+
+static int file_callback(const struct option *opt, const char *arg, int unset)
+{
+       struct grep_opt *grep_opt = opt->value;
+       FILE *patterns;
+       int lno = 0;
+       struct strbuf sb;
+
+       patterns = fopen(arg, "r");
+       if (!patterns)
+               die_errno("cannot open '%s'", arg);
+       while (strbuf_getline(&sb, patterns, '\n') == 0) {
+               /* ignore empty line like grep does */
+               if (sb.len == 0)
+                       continue;
+               append_grep_pattern(grep_opt, strbuf_detach(&sb, NULL), arg,
+                                   ++lno, GREP_PATTERN);
+       }
+       fclose(patterns);
+       strbuf_release(&sb);
+       return 0;
+}
+
+static int not_callback(const struct option *opt, const char *arg, int unset)
+{
+       struct grep_opt *grep_opt = opt->value;
+       append_grep_pattern(grep_opt, "--not", "command line", 0, GREP_NOT);
+       return 0;
+}
+
+static int and_callback(const struct option *opt, const char *arg, int unset)
+{
+       struct grep_opt *grep_opt = opt->value;
+       append_grep_pattern(grep_opt, "--and", "command line", 0, GREP_AND);
+       return 0;
+}
+
+static int open_callback(const struct option *opt, const char *arg, int unset)
+{
+       struct grep_opt *grep_opt = opt->value;
+       append_grep_pattern(grep_opt, "(", "command line", 0, GREP_OPEN_PAREN);
+       return 0;
+}
+
+static int close_callback(const struct option *opt, const char *arg, int unset)
+{
+       struct grep_opt *grep_opt = opt->value;
+       append_grep_pattern(grep_opt, ")", "command line", 0, GREP_CLOSE_PAREN);
+       return 0;
+}
+
+static int pattern_callback(const struct option *opt, const char *arg,
+                           int unset)
+{
+       struct grep_opt *grep_opt = opt->value;
+       append_grep_pattern(grep_opt, arg, "-e option", 0, GREP_PATTERN);
+       return 0;
+}
+
+static int help_callback(const struct option *opt, const char *arg, int unset)
+{
+       return -1;
+}
 
 int cmd_grep(int argc, const char **argv, const char *prefix)
 {
        int hit = 0;
        int cached = 0;
+       int external_grep_allowed = 1;
        int seen_dashdash = 0;
        struct grep_opt opt;
        struct object_array list = { 0, 0, NULL };
        const char **paths = NULL;
        int i;
+       int dummy;
+       struct option options[] = {
+               OPT_BOOLEAN(0, "cached", &cached,
+                       "search in index instead of in the work tree"),
+               OPT_GROUP(""),
+               OPT_BOOLEAN('v', "invert-match", &opt.invert,
+                       "show non-matching lines"),
+               OPT_BIT('i', "ignore-case", &opt.regflags,
+                       "case insensitive matching", REG_ICASE),
+               OPT_BOOLEAN('w', "word-regexp", &opt.word_regexp,
+                       "match patterns only at word boundaries"),
+               OPT_SET_INT('a', "text", &opt.binary,
+                       "process binary files as text", GREP_BINARY_TEXT),
+               OPT_SET_INT('I', NULL, &opt.binary,
+                       "don't match patterns in binary files",
+                       GREP_BINARY_NOMATCH),
+               { OPTION_INTEGER, 0, "max-depth", &opt.max_depth, "depth",
+                       "descend at most <depth> levels", PARSE_OPT_NONEG,
+                       NULL, 1 },
+               OPT_GROUP(""),
+               OPT_BIT('E', "extended-regexp", &opt.regflags,
+                       "use extended POSIX regular expressions", REG_EXTENDED),
+               OPT_NEGBIT('G', "basic-regexp", &opt.regflags,
+                       "use basic POSIX regular expressions (default)",
+                       REG_EXTENDED),
+               OPT_BOOLEAN('F', "fixed-strings", &opt.fixed,
+                       "interpret patterns as fixed strings"),
+               OPT_GROUP(""),
+               OPT_BOOLEAN('n', NULL, &opt.linenum, "show line numbers"),
+               OPT_NEGBIT('h', NULL, &opt.pathname, "don't show filenames", 1),
+               OPT_BIT('H', NULL, &opt.pathname, "show filenames", 1),
+               OPT_NEGBIT(0, "full-name", &opt.relative,
+                       "show filenames relative to top directory", 1),
+               OPT_BOOLEAN('l', "files-with-matches", &opt.name_only,
+                       "show only filenames instead of matching lines"),
+               OPT_BOOLEAN(0, "name-only", &opt.name_only,
+                       "synonym for --files-with-matches"),
+               OPT_BOOLEAN('L', "files-without-match",
+                       &opt.unmatch_name_only,
+                       "show only the names of files without match"),
+               OPT_BOOLEAN('z', "null", &opt.null_following_name,
+                       "print NUL after filenames"),
+               OPT_BOOLEAN('c', "count", &opt.count,
+                       "show the number of matches instead of matching lines"),
+               OPT_SET_INT(0, "color", &opt.color, "highlight matches", 1),
+               OPT_GROUP(""),
+               OPT_CALLBACK('C', NULL, &opt, "n",
+                       "show <n> context lines before and after matches",
+                       context_callback),
+               OPT_INTEGER('B', NULL, &opt.pre_context,
+                       "show <n> context lines before matches"),
+               OPT_INTEGER('A', NULL, &opt.post_context,
+                       "show <n> context lines after matches"),
+               OPT_NUMBER_CALLBACK(&opt, "shortcut for -C NUM",
+                       context_callback),
+               OPT_BOOLEAN('p', "show-function", &opt.funcname,
+                       "show a line with the function name before matches"),
+               OPT_GROUP(""),
+               OPT_CALLBACK('f', NULL, &opt, "file",
+                       "read patterns from file", file_callback),
+               { OPTION_CALLBACK, 'e', NULL, &opt, "pattern",
+                       "match <pattern>", PARSE_OPT_NONEG, pattern_callback },
+               { OPTION_CALLBACK, 0, "and", &opt, NULL,
+                 "combine patterns specified with -e",
+                 PARSE_OPT_NOARG | PARSE_OPT_NONEG, and_callback },
+               OPT_BOOLEAN(0, "or", &dummy, ""),
+               { OPTION_CALLBACK, 0, "not", &opt, NULL, "",
+                 PARSE_OPT_NOARG | PARSE_OPT_NONEG, not_callback },
+               { OPTION_CALLBACK, '(', NULL, &opt, NULL, "",
+                 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
+                 open_callback },
+               { OPTION_CALLBACK, ')', NULL, &opt, NULL, "",
+                 PARSE_OPT_NOARG | PARSE_OPT_NONEG | PARSE_OPT_NODASH,
+                 close_callback },
+               OPT_BOOLEAN(0, "all-match", &opt.all_match,
+                       "show only matches from files that match all patterns"),
+               OPT_GROUP(""),
+#if NO_EXTERNAL_GREP
+               OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed,
+                       "allow calling of grep(1) (ignored by this build)"),
+#else
+               OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed,
+                       "allow calling of grep(1) (default)"),
+#endif
+               { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
+                 PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
+               OPT_END()
+       };
 
        memset(&opt, 0, sizeof(opt));
        opt.prefix = prefix;
@@ -578,6 +795,7 @@ int cmd_grep(int argc, const char **argv, const char *prefix)
        opt.pathname = 1;
        opt.pattern_tail = &opt.pattern_list;
        opt.regflags = REG_NEWLINE;
+       opt.max_depth = -1;
 
        strcpy(opt.color_match, GIT_COLOR_RED GIT_COLOR_BOLD);
        opt.color = -1;
@@ -595,227 +813,21 @@ int cmd_grep(int argc, const char **argv, const char *prefix)
         * unrecognized non option is the beginning of the refs list
         * that continues up to the -- (if exists), and then paths.
         */
-
-       while (1 < argc) {
-               const char *arg = argv[1];
-               argc--; argv++;
-               if (!strcmp("--cached", arg)) {
-                       cached = 1;
-                       continue;
-               }
-               if (!strcmp("--no-ext-grep", arg)) {
-                       builtin_grep = 1;
-                       continue;
-               }
-               if (!strcmp("-a", arg) ||
-                   !strcmp("--text", arg)) {
-                       opt.binary = GREP_BINARY_TEXT;
-                       continue;
-               }
-               if (!strcmp("-i", arg) ||
-                   !strcmp("--ignore-case", arg)) {
-                       opt.regflags |= REG_ICASE;
-                       continue;
-               }
-               if (!strcmp("-I", arg)) {
-                       opt.binary = GREP_BINARY_NOMATCH;
-                       continue;
-               }
-               if (!strcmp("-v", arg) ||
-                   !strcmp("--invert-match", arg)) {
-                       opt.invert = 1;
-                       continue;
-               }
-               if (!strcmp("-E", arg) ||
-                   !strcmp("--extended-regexp", arg)) {
-                       opt.regflags |= REG_EXTENDED;
-                       continue;
-               }
-               if (!strcmp("-F", arg) ||
-                   !strcmp("--fixed-strings", arg)) {
-                       opt.fixed = 1;
-                       continue;
-               }
-               if (!strcmp("-G", arg) ||
-                   !strcmp("--basic-regexp", arg)) {
-                       opt.regflags &= ~REG_EXTENDED;
-                       continue;
-               }
-               if (!strcmp("-n", arg)) {
-                       opt.linenum = 1;
-                       continue;
-               }
-               if (!strcmp("-h", arg)) {
-                       opt.pathname = 0;
-                       continue;
-               }
-               if (!strcmp("-H", arg)) {
-                       opt.pathname = 1;
-                       continue;
-               }
-               if (!strcmp("-l", arg) ||
-                   !strcmp("--name-only", arg) ||
-                   !strcmp("--files-with-matches", arg)) {
-                       opt.name_only = 1;
-                       continue;
-               }
-               if (!strcmp("-L", arg) ||
-                   !strcmp("--files-without-match", arg)) {
-                       opt.unmatch_name_only = 1;
-                       continue;
-               }
-               if (!strcmp("-z", arg) ||
-                   !strcmp("--null", arg)) {
-                       opt.null_following_name = 1;
-                       continue;
-               }
-               if (!strcmp("-c", arg) ||
-                   !strcmp("--count", arg)) {
-                       opt.count = 1;
-                       continue;
-               }
-               if (!strcmp("-w", arg) ||
-                   !strcmp("--word-regexp", arg)) {
-                       opt.word_regexp = 1;
-                       continue;
-               }
-               if (!prefixcmp(arg, "-A") ||
-                   !prefixcmp(arg, "-B") ||
-                   !prefixcmp(arg, "-C") ||
-                   (arg[0] == '-' && '1' <= arg[1] && arg[1] <= '9')) {
-                       unsigned num;
-                       const char *scan;
-                       switch (arg[1]) {
-                       case 'A': case 'B': case 'C':
-                               if (!arg[2]) {
-                                       if (argc <= 1)
-                                               die(emsg_missing_context_len);
-                                       scan = *++argv;
-                                       argc--;
-                               }
-                               else
-                                       scan = arg + 2;
-                               break;
-                       default:
-                               scan = arg + 1;
-                               break;
-                       }
-                       if (strtoul_ui(scan, 10, &num))
-                               die(emsg_invalid_context_len, scan);
-                       switch (arg[1]) {
-                       case 'A':
-                               opt.post_context = num;
-                               break;
-                       default:
-                       case 'C':
-                               opt.post_context = num;
-                       case 'B':
-                               opt.pre_context = num;
-                               break;
-                       }
-                       continue;
-               }
-               if (!strcmp("-f", arg)) {
-                       FILE *patterns;
-                       int lno = 0;
-                       char buf[1024];
-                       if (argc <= 1)
-                               die(emsg_missing_argument, arg);
-                       patterns = fopen(argv[1], "r");
-                       if (!patterns)
-                               die("'%s': %s", argv[1], strerror(errno));
-                       while (fgets(buf, sizeof(buf), patterns)) {
-                               int len = strlen(buf);
-                               if (len && buf[len-1] == '\n')
-                                       buf[len-1] = 0;
-                               /* ignore empty line like grep does */
-                               if (!buf[0])
-                                       continue;
-                               append_grep_pattern(&opt, xstrdup(buf),
-                                                   argv[1], ++lno,
-                                                   GREP_PATTERN);
-                       }
-                       fclose(patterns);
-                       argv++;
-                       argc--;
-                       continue;
-               }
-               if (!strcmp("--not", arg)) {
-                       append_grep_pattern(&opt, arg, "command line", 0,
-                                           GREP_NOT);
-                       continue;
-               }
-               if (!strcmp("--and", arg)) {
-                       append_grep_pattern(&opt, arg, "command line", 0,
-                                           GREP_AND);
-                       continue;
-               }
-               if (!strcmp("--or", arg))
-                       continue; /* no-op */
-               if (!strcmp("(", arg)) {
-                       append_grep_pattern(&opt, arg, "command line", 0,
-                                           GREP_OPEN_PAREN);
-                       continue;
-               }
-               if (!strcmp(")", arg)) {
-                       append_grep_pattern(&opt, arg, "command line", 0,
-                                           GREP_CLOSE_PAREN);
-                       continue;
-               }
-               if (!strcmp("--all-match", arg)) {
-                       opt.all_match = 1;
-                       continue;
-               }
-               if (!strcmp("-e", arg)) {
-                       if (1 < argc) {
-                               append_grep_pattern(&opt, argv[1],
-                                                   "-e option", 0,
-                                                   GREP_PATTERN);
-                               argv++;
-                               argc--;
-                               continue;
-                       }
-                       die(emsg_missing_argument, arg);
-               }
-               if (!strcmp("--full-name", arg)) {
-                       opt.relative = 0;
-                       continue;
-               }
-               if (!strcmp("--color", arg)) {
-                       opt.color = 1;
-                       continue;
-               }
-               if (!strcmp("--no-color", arg)) {
-                       opt.color = 0;
-                       continue;
-               }
-               if (!strcmp("--", arg)) {
-                       /* later processing wants to have this at argv[1] */
-                       argv--;
-                       argc++;
-                       break;
-               }
-               if (*arg == '-')
-                       usage(builtin_grep_usage);
-
-               /* First unrecognized non-option token */
-               if (!opt.pattern_list) {
-                       append_grep_pattern(&opt, arg, "command line", 0,
-                                           GREP_PATTERN);
-                       break;
-               }
-               else {
-                       /* We are looking at the first path or rev;
-                        * it is found at argv[1] after leaving the
-                        * loop.
-                        */
-                       argc++; argv--;
-                       break;
-               }
+       argc = parse_options(argc, argv, prefix, options, grep_usage,
+                            PARSE_OPT_KEEP_DASHDASH |
+                            PARSE_OPT_STOP_AT_NON_OPTION |
+                            PARSE_OPT_NO_INTERNAL_HELP);
+
+       /* First unrecognized non-option token */
+       if (argc > 0 && !opt.pattern_list) {
+               append_grep_pattern(&opt, argv[0], "command line", 0,
+                                   GREP_PATTERN);
+               argv++;
+               argc--;
        }
 
-       if (opt.color && !opt.color_external)
-               builtin_grep = 1;
+       if ((opt.color && !opt.color_external) || opt.funcname)
+               external_grep_allowed = 0;
        if (!opt.pattern_list)
                die("no pattern given.");
        if ((opt.regflags != REG_NEWLINE) && opt.fixed)
@@ -823,7 +835,7 @@ int cmd_grep(int argc, const char **argv, const char *prefix)
        compile_grep_patterns(&opt);
 
        /* Check revs and then paths */
-       for (i = 1; i < argc; i++) {
+       for (i = 0; i < argc; i++) {
                const char *arg = argv[i];
                unsigned char sha1[20];
                /* Is it a rev? */
@@ -859,7 +871,7 @@ int cmd_grep(int argc, const char **argv, const char *prefix)
        if (!list.nr) {
                if (!cached)
                        setup_work_tree();
-               return !grep_cache(&opt, paths, cached);
+               return !grep_cache(&opt, paths, cached, external_grep_allowed);
        }
 
        if (cached)