grep: rip out pessimization to use fixmatch()
[gitweb.git] / builtin-grep.c
index a551d3402171878ea0b82263529c015caa850380..3d6ebb586dd0ff4db513dcc77f9d37c75b79e19b 100644 (file)
 #include "tag.h"
 #include "tree-walk.h"
 #include "builtin.h"
-#include <regex.h>
-#include <fnmatch.h>
+#include "parse-options.h"
+#include "userdiff.h"
+#include "grep.h"
+#include "quote.h"
+
+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;
+       }
+       if (!strcmp(var, "color.grep.match")) {
+               if (!value)
+                       return config_error_nonbool(var);
+               color_parse(value, var, opt->color_match);
+               return 0;
+       }
+       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 *slash, *cp;
+               const char *cp, *meta;
 
-               if ((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;
                if (name[namelen-1] != '/')
                        continue;
 
-               /* We are being asked if the name directory is worth
+               /* We are being asked if the directory ("name") is worth
                 * descending into.
                 *
                 * Find the longest leading directory name that does
                 * not have metacharacter in the pathspec; the name
                 * we are looking at must overlap with that directory.
                 */
-               for (cp = match, slash = NULL; cp - match < matchlen; cp++) {
+               for (cp = match, meta = NULL; cp - match < matchlen; cp++) {
                        char ch = *cp;
-                       if (ch == '/')
-                               slash = cp;
-                       if (ch == '*' || ch == '[')
+                       if (ch == '*' || ch == '[' || ch == '?') {
+                               meta = cp;
                                break;
+                       }
                }
-               if (!slash)
-                       slash = match; /* toplevel */
-               else
-                       slash++;
-               if (namelen <= slash - match) {
+               if (!meta)
+                       meta = cp; /* fully literal */
+
+               if (namelen <= meta - match) {
                        /* Looking at "Documentation/" and
                         * the pattern says "Documentation/howto/", or
-                        * "Documentation/diff*.txt".
+                        * "Documentation/diff*.txt".  The name we
+                        * have should match prefix.
                         */
                        if (!memcmp(match, name, namelen))
                                return 1;
+                       continue;
                }
-               else {
+
+               if (meta - match < namelen) {
                        /* Looking at "Documentation/howto/" and
-                        * the pattern says "Documentation/h*".
+                        * the pattern says "Documentation/h*";
+                        * match up to "Do.../h"; this avoids descending
+                        * into "Documentation/technical/".
                         */
-                       if (!memcmp(match, name, slash - match))
+                       if (!memcmp(match, name, meta - match))
                                return 1;
+                       continue;
                }
        }
        return 0;
 }
 
-struct grep_opt {
-       const char *pattern;
-       regex_t regexp;
-       unsigned linenum:1;
-       unsigned invert:1;
-       unsigned name_only:1;
-       int regflags;
-       unsigned pre_context;
-       unsigned post_context;
-};
-
-static char *end_of_line(char *cp, unsigned long *left)
-{
-       unsigned long l = *left;
-       while (l && *cp != '\n') {
-               l--;
-               cp++;
-       }
-       *left = l;
-       return cp;
-}
-
-static void show_line(struct grep_opt *opt, const char *bol, const char *eol,
-                     const char *name, unsigned lno, char sign)
-{
-       printf("%s%c", name, sign);
-       if (opt->linenum)
-               printf("%d%c", lno, sign);
-       printf("%.*s\n", (int)(eol-bol), bol);
-}
-
-static int grep_buffer(struct grep_opt *opt, const char *name,
-                      char *buf, unsigned long size)
-{
-       char *bol = buf;
-       unsigned long left = size;
-       unsigned lno = 1;
-       struct pre_context_line {
-               char *bol;
-               char *eol;
-       } *prev = NULL, *pcl;
-       unsigned last_hit = 0;
-       unsigned last_shown = 0;
-       const char *hunk_mark = "";
-
-       if (opt->pre_context)
-               prev = xcalloc(opt->pre_context, sizeof(*prev));
-       if (opt->pre_context || opt->post_context)
-               hunk_mark = "--\n";
-
-       while (left) {
-               regmatch_t pmatch[10];
-               char *eol, ch;
-               int hit;
-
-               eol = end_of_line(bol, &left);
-               ch = *eol;
-               *eol = 0;
-
-               hit = !regexec(&opt->regexp, bol, ARRAY_SIZE(pmatch),
-                              pmatch, 0);
-               if (opt->invert)
-                       hit = !hit;
-               if (hit) {
-                       if (opt->name_only) {
-                               printf("%s\n", name);
-                               return 1;
-                       }
-                       /* Hit at this line.  If we haven't shown the
-                        * pre-context lines, we would need to show them.
-                        */
-                       if (opt->pre_context) {
-                               unsigned from;
-                               if (opt->pre_context < lno)
-                                       from = lno - opt->pre_context;
-                               else
-                                       from = 1;
-                               if (from <= last_shown)
-                                       from = last_shown + 1;
-                               if (last_shown && from != last_shown + 1)
-                                       printf(hunk_mark);
-                               while (from < lno) {
-                                       pcl = &prev[lno-from-1];
-                                       show_line(opt, pcl->bol, pcl->eol,
-                                                 name, from, '-');
-                                       from++;
-                               }
-                               last_shown = lno-1;
-                       }
-                       if (last_shown && lno != last_shown + 1)
-                               printf(hunk_mark);
-                       show_line(opt, bol, eol, name, lno, ':');
-                       last_shown = last_hit = lno;
-               }
-               else if (last_hit &&
-                        lno <= last_hit + opt->post_context) {
-                       /* If the last hit is within the post context,
-                        * we need to show this line.
-                        */
-                       if (last_shown && lno != last_shown + 1)
-                               printf(hunk_mark);
-                       show_line(opt, bol, eol, name, lno, '-');
-                       last_shown = lno;
-               }
-               if (opt->pre_context) {
-                       memmove(prev+1, prev,
-                               (opt->pre_context-1) * sizeof(*prev));
-                       prev->bol = bol;
-                       prev->eol = eol;
-               }
-               *eol = ch;
-               bol = eol + 1;
-               left--;
-               lno++;
-       }
-       return !!last_hit;
-}
-
-static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name)
+static int grep_sha1(struct grep_opt *opt, const unsigned char *sha1, const char *name, int tree_name_len)
 {
        unsigned long size;
        char *data;
-       char type[20];
+       enum object_type type;
        int hit;
-       data = read_sha1_file(sha1, type, &size);
+       struct strbuf pathbuf = STRBUF_INIT;
+
+       data = read_sha1_file(sha1, &type, &size);
        if (!data) {
                error("'%s': unable to read %s", name, sha1_to_hex(sha1));
                return 0;
        }
+       if (opt->relative && opt->prefix_length) {
+               quote_path_relative(name + tree_name_len, -1, &pathbuf, opt->prefix);
+               strbuf_insert(&pathbuf, 0, name, tree_name_len);
+               name = pathbuf.buf;
+       }
        hit = grep_buffer(opt, name, data, size);
+       strbuf_release(&pathbuf);
        free(data);
        return hit;
 }
@@ -214,6 +172,9 @@ static int grep_file(struct grep_opt *opt, const char *filename)
        struct stat st;
        int i;
        char *data;
+       size_t sz;
+       struct strbuf buf = STRBUF_INIT;
+
        if (lstat(filename, &st) < 0) {
        err_ret:
                if (errno != ENOENT)
@@ -224,18 +185,22 @@ static int grep_file(struct grep_opt *opt, const char *filename)
                return 0; /* empty file -- no grep hit */
        if (!S_ISREG(st.st_mode))
                return 0;
+       sz = xsize_t(st.st_size);
        i = open(filename, O_RDONLY);
        if (i < 0)
                goto err_ret;
-       data = xmalloc(st.st_size + 1);
-       if (st.st_size != xread(i, data, st.st_size)) {
+       data = xmalloc(sz + 1);
+       if (st.st_size != read_in_full(i, data, sz)) {
                error("'%s': short read %s", filename, strerror(errno));
                close(i);
                free(data);
                return 0;
        }
        close(i);
-       i = grep_buffer(opt, filename, data, st.st_size);
+       if (opt->relative && opt->prefix_length)
+               filename = quote_path_relative(filename, -1, &buf, opt->prefix);
+       i = grep_buffer(opt, filename, data, sz);
+       strbuf_release(&buf);
        free(data);
        return i;
 }
@@ -248,15 +213,31 @@ static int grep_cache(struct grep_opt *opt, const char **paths, int cached)
 
        for (nr = 0; nr < active_nr; nr++) {
                struct cache_entry *ce = active_cache[nr];
-               if (ce_stage(ce) || !S_ISREG(ntohl(ce->ce_mode)))
+               if (!S_ISREG(ce->ce_mode))
                        continue;
-               if (!pathspec_matches(paths, ce->name))
+               if (!pathspec_matches(paths, ce->name, opt->max_depth))
                        continue;
-               if (cached)
-                       hit |= grep_sha1(opt, ce->sha1, ce->name);
+               /*
+                * If CE_VALID is on, we assume worktree file and its cache entry
+                * are identical, even if worktree file has been modified, so use
+                * cache version instead
+                */
+               if (cached || (ce->ce_flags & CE_VALID)) {
+                       if (ce_stage(ce))
+                               continue;
+                       hit |= grep_sha1(opt, ce->sha1, ce->name, 0);
+               }
                else
                        hit |= grep_file(opt, ce->name);
+               if (ce_stage(ce)) {
+                       do {
+                               nr++;
+                       } while (nr < active_nr &&
+                                !strcmp(ce->name, active_cache[nr]->name));
+                       nr--; /* compensate for loop control */
+               }
        }
+       free_grep_patterns(opt);
        return hit;
 }
 
@@ -264,269 +245,360 @@ static int grep_tree(struct grep_opt *opt, const char **paths,
                     struct tree_desc *tree,
                     const char *tree_name, const char *base)
 {
-       unsigned mode;
        int len;
        int hit = 0;
-       const char *path;
-       const unsigned char *sha1;
+       struct name_entry entry;
        char *down;
-       char *path_buf = xmalloc(PATH_MAX + strlen(tree_name) + 100);
+       int tn_len = strlen(tree_name);
+       struct strbuf pathbuf;
 
-       if (tree_name[0]) {
-               int offset = sprintf(path_buf, "%s:", tree_name);
-               down = path_buf + offset;
-               strcat(down, base);
-       }
-       else {
-               down = path_buf;
-               strcpy(down, base);
+       strbuf_init(&pathbuf, PATH_MAX + tn_len);
+
+       if (tn_len) {
+               strbuf_add(&pathbuf, tree_name, tn_len);
+               strbuf_addch(&pathbuf, ':');
+               tn_len = pathbuf.len;
        }
-       len = strlen(path_buf);
+       strbuf_addstr(&pathbuf, base);
+       len = pathbuf.len;
 
-       while (tree->size) {
-               int pathlen;
-               sha1 = tree_entry_extract(tree, &path, &mode);
-               pathlen = strlen(path);
-               strcpy(path_buf + len, path);
+       while (tree_entry(tree, &entry)) {
+               int te_len = tree_entry_len(entry.path, entry.sha1);
+               pathbuf.len = len;
+               strbuf_add(&pathbuf, entry.path, te_len);
 
-               if (S_ISDIR(mode))
+               if (S_ISDIR(entry.mode))
                        /* Match "abc/" against pathspec to
                         * decide if we want to descend into "abc"
                         * directory.
                         */
-                       strcpy(path_buf + len + pathlen, "/");
+                       strbuf_addch(&pathbuf, '/');
 
-               if (!pathspec_matches(paths, down))
+               down = pathbuf.buf + tn_len;
+               if (!pathspec_matches(paths, down, opt->max_depth))
                        ;
-               else if (S_ISREG(mode))
-                       hit |= grep_sha1(opt, sha1, path_buf);
-               else if (S_ISDIR(mode)) {
-                       char type[20];
+               else if (S_ISREG(entry.mode))
+                       hit |= grep_sha1(opt, entry.sha1, pathbuf.buf, tn_len);
+               else if (S_ISDIR(entry.mode)) {
+                       enum object_type type;
                        struct tree_desc sub;
                        void *data;
-                       data = read_sha1_file(sha1, type, &sub.size);
+                       unsigned long size;
+
+                       data = read_sha1_file(entry.sha1, &type, &size);
                        if (!data)
                                die("unable to read tree (%s)",
-                                   sha1_to_hex(sha1));
-                       sub.buf = data;
+                                   sha1_to_hex(entry.sha1));
+                       init_tree_desc(&sub, data, size);
                        hit |= grep_tree(opt, paths, &sub, tree_name, down);
                        free(data);
                }
-               update_tree_entry(tree);
        }
+       strbuf_release(&pathbuf);
        return hit;
 }
 
 static int grep_object(struct grep_opt *opt, const char **paths,
                       struct object *obj, const char *name)
 {
-       if (!strcmp(obj->type, blob_type))
-               return grep_sha1(opt, obj->sha1, name);
-       if (!strcmp(obj->type, commit_type) ||
-           !strcmp(obj->type, tree_type)) {
+       if (obj->type == OBJ_BLOB)
+               return grep_sha1(opt, obj->sha1, name, 0);
+       if (obj->type == OBJ_COMMIT || obj->type == OBJ_TREE) {
                struct tree_desc tree;
                void *data;
+               unsigned long size;
                int hit;
                data = read_object_with_reference(obj->sha1, tree_type,
-                                                 &tree.size, NULL);
+                                                 &size, NULL);
                if (!data)
                        die("unable to read tree (%s)", sha1_to_hex(obj->sha1));
-               tree.buf = data;
+               init_tree_desc(&tree, data, size);
                hit = grep_tree(opt, paths, &tree, name, "");
                free(data);
                return hit;
        }
-       die("unable to grep from object of type %s", obj->type);
+       die("unable to grep from object of type %s", typename(obj->type));
+}
+
+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;
+
+       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 = STRBUF_INIT;
+
+       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 const char builtin_grep_usage[] =
-"git-grep <option>* <rev>* [-e] <pattern> [<path>...]";
+static int help_callback(const struct option *opt, const char *arg, int unset)
+{
+       return -1;
+}
 
-int cmd_grep(int argc, const char **argv, char **envp)
+int cmd_grep(int argc, const char **argv, const char *prefix)
 {
-       int err;
        int hit = 0;
-       int no_more_flags = 0;
-       int seen_noncommit = 0;
        int cached = 0;
+       int seen_dashdash = 0;
+       int external_grep_allowed__ignored;
        struct grep_opt opt;
-       struct object_list *list, **tail, *object_list = NULL;
-       const char *prefix = setup_git_directory();
+       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_BOOLEAN('i', "ignore-case", &opt.ignore_case,
+                       "case insensitive matching"),
+               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(""),
+               OPT_BOOLEAN(0, "ext-grep", &external_grep_allowed__ignored,
+                           "allow calling of grep(1) (ignored by this build)"),
+               { OPTION_CALLBACK, 0, "help-all", &options, NULL, "show usage",
+                 PARSE_OPT_HIDDEN | PARSE_OPT_NOARG, help_callback },
+               OPT_END()
+       };
+
+       /*
+        * 'git grep -h', unlike 'git grep -h <pattern>', is a request
+        * to show usage information and exit.
+        */
+       if (argc == 2 && !strcmp(argv[1], "-h"))
+               usage_with_options(grep_usage, options);
 
        memset(&opt, 0, sizeof(opt));
+       opt.prefix = prefix;
+       opt.prefix_length = (prefix && *prefix) ? strlen(prefix) : 0;
+       opt.relative = 1;
+       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;
+       git_config(grep_config, &opt);
+       if (opt.color == -1)
+               opt.color = git_use_color_default;
 
        /*
-        * No point using rev_info, really.
+        * If there is no -- then the paths must exist in the working
+        * tree.  If there is no explicit pattern specified with -e or
+        * -f, we take the first unrecognized non option to be the
+        * pattern, but then what follows it must be zero or more
+        * valid refs up to the -- (if exists), and then existing
+        * paths.  If there is an explicit pattern, then the first
+        * 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("-i", arg) ||
-                   !strcmp("--ignore-case", arg)) {
-                       opt.regflags |= REG_ICASE;
-                       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("-G", arg) ||
-                   !strcmp("--basic-regexp", arg)) {
-                       opt.regflags &= ~REG_EXTENDED;
-                       continue;
-               }
-               if (!strcmp("-n", arg)) {
-                       opt.linenum = 1;
-                       continue;
-               }
-               if (!strcmp("-H", arg)) {
-                       /* We always show the pathname, so this
-                        * is a noop.
-                        */
-                       continue;
-               }
-               if (!strcmp("-l", arg) ||
-                   !strcmp("--files-with-matches", arg)) {
-                       opt.name_only = 1;
-                       continue;
-               }
-               if (!strncmp("-A", arg, 2) ||
-                   !strncmp("-B", arg, 2) ||
-                   !strncmp("-C", arg, 2) ||
-                   (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)
-                                               usage(builtin_grep_usage);
-                                       scan = *++argv;
-                                       argc--;
-                               }
-                               else
-                                       scan = arg + 2;
-                               break;
-                       default:
-                               scan = arg + 1;
-                               break;
-                       }
-                       if (sscanf(scan, "%u", &num) != 1)
-                               usage(builtin_grep_usage);
-                       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("-e", arg)) {
-                       if (1 < argc) {
-                               /* We probably would want to do
-                                * -e pat1 -e pat2 as well later...
-                                */
-                               if (opt.pattern)
-                                       die("more than one pattern?");
-                               opt.pattern = *++argv;
-                               argc--;
-                               continue;
-                       }
-                       usage(builtin_grep_usage);
-               }
-               if (!strcmp("--", arg)) {
-                       no_more_flags = 1;
+       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.pattern_list)
+               die("no pattern given.");
+       if (!opt.fixed && opt.ignore_case)
+               opt.regflags |= REG_ICASE;
+       if ((opt.regflags != REG_NEWLINE) && opt.fixed)
+               die("cannot mix --fixed-strings and regexp");
+       compile_grep_patterns(&opt);
+
+       /* Check revs and then paths */
+       for (i = 0; i < argc; i++) {
+               const char *arg = argv[i];
+               unsigned char sha1[20];
+               /* Is it a rev? */
+               if (!get_sha1(arg, sha1)) {
+                       struct object *object = parse_object(sha1);
+                       if (!object)
+                               die("bad object %s", arg);
+                       add_object_array(object, arg, &list);
                        continue;
                }
-               /* Either unrecognized option or a single pattern */
-               if (!no_more_flags && *arg == '-')
-                       usage(builtin_grep_usage);
-               if (!opt.pattern) {
-                       opt.pattern = arg;
-                       break;
-               }
-               else {
-                       /* We are looking at the first path or rev;
-                        * it is found at argv[0] after leaving the
-                        * loop.
-                        */
-                       argc++; argv--;
-                       break;
+               if (!strcmp(arg, "--")) {
+                       i++;
+                       seen_dashdash = 1;
                }
+               break;
        }
-       if (!opt.pattern)
-               die("no pattern given.");
-       err = regcomp(&opt.regexp, opt.pattern, opt.regflags);
-       if (err) {
-               char errbuf[1024];
-               regerror(err, &opt.regexp, errbuf, 1024);
-               regfree(&opt.regexp);
-               die("'%s': %s", opt.pattern, errbuf);
-       }
-       tail = &object_list;
-       while (1 < argc) {
-               struct object *object;
-               struct object_list *elem;
-               const char *arg = argv[1];
-               unsigned char sha1[20];
-               if (get_sha1(arg, sha1) < 0)
-                       break;
-               object = parse_object(sha1);
-               if (!object)
-                       die("bad object %s", arg);
-               elem = object_list_insert(object, tail);
-               elem->name = arg;
-               tail = &elem->next;
-               argc--; argv++;
+
+       /* The rest are paths */
+       if (!seen_dashdash) {
+               int j;
+               for (j = i; j < argc; j++)
+                       verify_filename(prefix, argv[j]);
        }
-       if (1 < argc)
-               paths = get_pathspec(prefix, argv + 1);
+
+       if (i < argc)
+               paths = get_pathspec(prefix, argv + i);
        else if (prefix) {
                paths = xcalloc(2, sizeof(const char *));
                paths[0] = prefix;
                paths[1] = NULL;
        }
 
-       if (!object_list)
+       if (!list.nr) {
+               if (!cached)
+                       setup_work_tree();
                return !grep_cache(&opt, paths, cached);
-       /*
-        * Do not walk "grep -e foo master next pu -- Documentation/"
-        * but do walk "grep -e foo master..next -- Documentation/".
-        * Ranged request mixed with a blob or tree object, like
-        * "grep -e foo v1.0.0:Documentation/ master..next"
-        * so detect that and complain.
-        */
-       for (list = object_list; list; list = list->next) {
-               struct object *real_obj;
-               real_obj = deref_tag(list->item, NULL, 0);
-               if (strcmp(real_obj->type, commit_type))
-                       seen_noncommit = 1;
        }
+
        if (cached)
-               die("both --cached and revisions given.");
+               die("both --cached and trees are given.");
 
-       for (list = object_list; list; list = list->next) {
+       for (i = 0; i < list.nr; i++) {
                struct object *real_obj;
-               real_obj = deref_tag(list->item, NULL, 0);
-               if (grep_object(&opt, paths, real_obj, list->name))
+               real_obj = deref_tag(list.objects[i].item, NULL, 0);
+               if (grep_object(&opt, paths, real_obj, list.objects[i].name))
                        hit = 1;
        }
+       free_grep_patterns(&opt);
        return !hit;
 }