git-ls-tree: add "-t" option to always show the tree entries
[gitweb.git] / ls-files.c
index c6c32d94b0755550a97544708ce71885c02a4b68..f3f1a6a6633881bf43887731fdcbe08af9ad2f4a 100644 (file)
@@ -9,6 +9,7 @@
 #include <fnmatch.h>
 
 #include "cache.h"
+#include "quote.h"
 
 static int show_deleted = 0;
 static int show_cached = 0;
@@ -16,29 +17,61 @@ static int show_others = 0;
 static int show_ignored = 0;
 static int show_stage = 0;
 static int show_unmerged = 0;
+static int show_modified = 0;
 static int show_killed = 0;
 static int line_terminator = '\n';
 
+static int prefix_len = 0, prefix_offset = 0;
+static const char *prefix = NULL;
+static const char **pathspec = NULL;
+
 static const char *tag_cached = "";
 static const char *tag_unmerged = "";
 static const char *tag_removed = "";
 static const char *tag_other = "";
 static const char *tag_killed = "";
+static const char *tag_modified = "";
 
-static int nr_excludes;
-static const char **excludes;
-static int excludes_alloc;
+static const char *exclude_per_dir = NULL;
 
-static void add_exclude(const char *string)
+/* We maintain three exclude pattern lists:
+ * EXC_CMDL lists patterns explicitly given on the command line.
+ * EXC_DIRS lists patterns obtained from per-directory ignore files.
+ * EXC_FILE lists patterns from fallback ignore files.
+ */
+#define EXC_CMDL 0
+#define EXC_DIRS 1
+#define EXC_FILE 2
+static struct exclude_list {
+       int nr;
+       int alloc;
+       struct exclude {
+               const char *pattern;
+               const char *base;
+               int baselen;
+       } **excludes;
+} exclude_list[3];
+
+static void add_exclude(const char *string, const char *base,
+                       int baselen, struct exclude_list *which)
 {
-       if (nr_excludes == excludes_alloc) {
-               excludes_alloc = alloc_nr(excludes_alloc);
-               excludes = realloc(excludes, excludes_alloc*sizeof(char *));
+       struct exclude *x = xmalloc(sizeof (*x));
+
+       x->pattern = string;
+       x->base = base;
+       x->baselen = baselen;
+       if (which->nr == which->alloc) {
+               which->alloc = alloc_nr(which->alloc);
+               which->excludes = realloc(which->excludes,
+                                         which->alloc * sizeof(x));
        }
-       excludes[nr_excludes++] = string;
+       which->excludes[which->nr++] = x;
 }
 
-static void add_excludes_from_file(const char *fname)
+static int add_excludes_from_file_1(const char *fname,
+                                   const char *base,
+                                   int baselen,
+                                   struct exclude_list *which)
 {
        int fd, i;
        long size;
@@ -53,7 +86,7 @@ static void add_excludes_from_file(const char *fname)
        lseek(fd, 0, SEEK_SET);
        if (size == 0) {
                close(fd);
-               return;
+               return 0;
        }
        buf = xmalloc(size);
        if (read(fd, buf, size) != size)
@@ -63,28 +96,112 @@ static void add_excludes_from_file(const char *fname)
        entry = buf;
        for (i = 0; i < size; i++) {
                if (buf[i] == '\n') {
-                       if (entry != buf + i) {
-                               buf[i] = 0;
-                               add_exclude(entry);
+                       if (entry != buf + i && entry[0] != '#') {
+                               buf[i - (i && buf[i-1] == '\r')] = 0;
+                               add_exclude(entry, base, baselen, which);
                        }
                        entry = buf + i + 1;
                }
        }
-       return;
+       return 0;
 
-err:   perror(fname);
-       exit(1);
+ err:
+       if (0 <= fd)
+               close(fd);
+       return -1;
 }
 
-static int excluded(const char *pathname)
+static void add_excludes_from_file(const char *fname)
+{
+       if (add_excludes_from_file_1(fname, "", 0,
+                                    &exclude_list[EXC_FILE]) < 0)
+               die("cannot use %s as an exclude file", fname);
+}
+
+static int push_exclude_per_directory(const char *base, int baselen)
+{
+       char exclude_file[PATH_MAX];
+       struct exclude_list *el = &exclude_list[EXC_DIRS];
+       int current_nr = el->nr;
+
+       if (exclude_per_dir) {
+               memcpy(exclude_file, base, baselen);
+               strcpy(exclude_file + baselen, exclude_per_dir);
+               add_excludes_from_file_1(exclude_file, base, baselen, el);
+       }
+       return current_nr;
+}
+
+static void pop_exclude_per_directory(int stk)
+{
+       struct exclude_list *el = &exclude_list[EXC_DIRS];
+
+       while (stk < el->nr)
+               free(el->excludes[--el->nr]);
+}
+
+/* Scan the list and let the last match determines the fate.
+ * Return 1 for exclude, 0 for include and -1 for undecided.
+ */
+static int excluded_1(const char *pathname,
+                     int pathlen,
+                     struct exclude_list *el)
 {
        int i;
-       if (nr_excludes) {
-               const char *basename = strrchr(pathname, '/');
-               basename = (basename) ? basename+1 : pathname;
-               for (i = 0; i < nr_excludes; i++)
-                       if (fnmatch(excludes[i], basename, 0) == 0)
-                               return 1;
+
+       if (el->nr) {
+               for (i = el->nr - 1; 0 <= i; i--) {
+                       struct exclude *x = el->excludes[i];
+                       const char *exclude = x->pattern;
+                       int to_exclude = 1;
+
+                       if (*exclude == '!') {
+                               to_exclude = 0;
+                               exclude++;
+                       }
+
+                       if (!strchr(exclude, '/')) {
+                               /* match basename */
+                               const char *basename = strrchr(pathname, '/');
+                               basename = (basename) ? basename+1 : pathname;
+                               if (fnmatch(exclude, basename, 0) == 0)
+                                       return to_exclude;
+                       }
+                       else {
+                               /* match with FNM_PATHNAME:
+                                * exclude has base (baselen long) inplicitly
+                                * in front of it.
+                                */
+                               int baselen = x->baselen;
+                               if (*exclude == '/')
+                                       exclude++;
+
+                               if (pathlen < baselen ||
+                                   (baselen && pathname[baselen-1] != '/') ||
+                                   strncmp(pathname, x->base, baselen))
+                                   continue;
+
+                               if (fnmatch(exclude, pathname+baselen,
+                                           FNM_PATHNAME) == 0)
+                                       return to_exclude;
+                       }
+               }
+       }
+       return -1; /* undecided */
+}
+
+static int excluded(const char *pathname)
+{
+       int pathlen = strlen(pathname);
+       int st;
+
+       for (st = EXC_CMDL; st <= EXC_FILE; st++) {
+               switch (excluded_1(pathname, pathlen, &exclude_list[st])) {
+               case 0:
+                       return 0;
+               case 1:
+                       return 1;
+               }
        }
        return 0;
 }
@@ -112,6 +229,7 @@ static void add_name(const char *pathname, int len)
        ent = xmalloc(sizeof(*ent) + len + 1);
        ent->len = len;
        memcpy(ent->name, pathname, len);
+       ent->name[len] = 0;
        dir[nr_dir++] = ent;
 }
 
@@ -121,7 +239,7 @@ static void add_name(const char *pathname, int len)
  * doesn't handle them at all yet. Maybe that will change some
  * day.
  *
- * Also, we currently ignore all names starting with a dot.
+ * Also, we ignore the name ".git" (even if it is not a directory).
  * That likely will not change.
  */
 static void read_directory(const char *path, const char *base, int baselen)
@@ -129,10 +247,13 @@ static void read_directory(const char *path, const char *base, int baselen)
        DIR *dir = opendir(path);
 
        if (dir) {
+               int exclude_stk;
                struct dirent *de;
                char fullname[MAXPATHLEN + 1];
                memcpy(fullname, base, baselen);
 
+               exclude_stk = push_exclude_per_directory(base, baselen);
+
                while ((de = readdir(dir)) != NULL) {
                        int len;
 
@@ -141,10 +262,10 @@ static void read_directory(const char *path, const char *base, int baselen)
                             !strcmp(de->d_name + 1, ".") ||
                             !strcmp(de->d_name + 1, "git")))
                                continue;
-                       if (excluded(de->d_name) != show_ignored)
-                               continue;
                        len = strlen(de->d_name);
                        memcpy(fullname + baselen, de->d_name, len+1);
+                       if (excluded(fullname) != show_ignored)
+                               continue;
 
                        switch (DTYPE(de)) {
                        struct stat st;
@@ -170,6 +291,8 @@ static void read_directory(const char *path, const char *base, int baselen)
                        add_name(fullname, baselen + len);
                }
                closedir(dir);
+
+               pop_exclude_per_directory(exclude_stk);
        }
 }
 
@@ -182,6 +305,72 @@ static int cmp_name(const void *p1, const void *p2)
                                  e2->name, e2->len);
 }
 
+/*
+ * Match a pathspec against a filename. The first "len" characters
+ * are the common prefix
+ */
+static int match(const char **spec, const char *filename, int len)
+{
+       const char *m;
+
+       while ((m = *spec++) != NULL) {
+               int matchlen = strlen(m + len);
+
+               if (!matchlen)
+                       return 1;
+               if (!strncmp(m + len, filename + len, matchlen)) {
+                       if (m[len + matchlen - 1] == '/')
+                               return 1;
+                       switch (filename[len + matchlen]) {
+                       case '/': case '\0':
+                               return 1;
+                       }
+               }
+               if (!fnmatch(m + len, filename + len, 0))
+                       return 1;
+       }
+       return 0;
+}
+
+static void show_dir_entry(const char *tag, struct nond_on_fs *ent)
+{
+       int len = prefix_len;
+       int offset = prefix_offset;
+
+       if (len >= ent->len)
+               die("git-ls-files: internal error - directory entry not superset of prefix");
+
+       if (pathspec && !match(pathspec, ent->name, len))
+               return;
+
+       fputs(tag, stdout);
+       write_name_quoted("", 0, ent->name + offset, line_terminator, stdout);
+       putchar(line_terminator);
+}
+
+static void show_other_files(void)
+{
+       int i;
+       for (i = 0; i < nr_dir; i++) {
+               /* We should not have a matching entry, but we
+                * may have an unmerged entry for this path.
+                */
+               struct nond_on_fs *ent = dir[i];
+               int pos = cache_name_pos(ent->name, ent->len);
+               struct cache_entry *ce;
+               if (0 <= pos)
+                       die("bug in show-other-files");
+               pos = -pos - 1;
+               if (pos < active_nr) { 
+                       ce = active_cache[pos];
+                       if (ce_namelen(ce) == ent->len &&
+                           !memcmp(ce->name, ent->name, ent->len))
+                               continue; /* Yup, this one exists unmerged */
+               }
+               show_dir_entry(tag_other, ent);
+       }
+}
+
 static void show_killed_files(void)
 {
        int i;
@@ -227,9 +416,36 @@ static void show_killed_files(void)
                        }
                }
                if (killed)
-                       printf("%s%.*s%c", tag_killed,
-                              dir[i]->len, dir[i]->name,
-                              line_terminator);
+                       show_dir_entry(tag_killed, dir[i]);
+       }
+}
+
+static void show_ce_entry(const char *tag, struct cache_entry *ce)
+{
+       int len = prefix_len;
+       int offset = prefix_offset;
+
+       if (len >= ce_namelen(ce))
+               die("git-ls-files: internal error - cache entry not superset of prefix");
+
+       if (pathspec && !match(pathspec, ce->name, len))
+               return;
+
+       if (!show_stage) {
+               fputs(tag, stdout);
+               write_name_quoted("", 0, ce->name + offset,
+                                 line_terminator, stdout);
+               putchar(line_terminator);
+       }
+       else {
+               printf("%s%06o %s %d\t",
+                      tag,
+                      ntohl(ce->ce_mode),
+                      sha1_to_hex(ce->sha1),
+                      ce_stage(ce));
+               write_name_quoted("", 0, ce->name + offset,
+                                 line_terminator, stdout);
+               putchar(line_terminator);
        }
 }
 
@@ -239,13 +455,15 @@ static void show_files(void)
 
        /* For cached/deleted files we don't need to even do the readdir */
        if (show_others || show_killed) {
-               read_directory(".", "", 0);
+               const char *path = ".", *base = "";
+               int baselen = prefix_len;
+
+               if (baselen)
+                       path = base = prefix;
+               read_directory(path, base, baselen);
                qsort(dir, nr_dir, sizeof(struct nond_on_fs *), cmp_name);
                if (show_others)
-                       for (i = 0; i < nr_dir; i++)
-                               printf("%s%.*s%c", tag_other,
-                                      dir[i]->len, dir[i]->name,
-                                      line_terminator);
+                       show_other_files();
                if (show_killed)
                        show_killed_files();
        }
@@ -256,95 +474,216 @@ static void show_files(void)
                                continue;
                        if (show_unmerged && !ce_stage(ce))
                                continue;
-                       if (!show_stage)
-                               printf("%s%s%c",
-                                      ce_stage(ce) ? tag_unmerged :
-                                      tag_cached,
-                                      ce->name, line_terminator);
-                       else
-                               printf("%s%06o %s %d\t%s%c",
-                                      ce_stage(ce) ? tag_unmerged :
-                                      tag_cached,
-                                      ntohl(ce->ce_mode),
-                                      sha1_to_hex(ce->sha1),
-                                      ce_stage(ce),
-                                      ce->name, line_terminator); 
+                       show_ce_entry(ce_stage(ce) ? tag_unmerged : tag_cached, ce);
                }
        }
-       if (show_deleted) {
+       if (show_deleted | show_modified) {
                for (i = 0; i < active_nr; i++) {
                        struct cache_entry *ce = active_cache[i];
                        struct stat st;
+                       int err;
                        if (excluded(ce->name) != show_ignored)
                                continue;
-                       if (!lstat(ce->name, &st))
-                               continue;
-                       printf("%s%s%c", tag_removed, ce->name,
-                              line_terminator);
+                       err = lstat(ce->name, &st);
+                       if (show_deleted && err)
+                               show_ce_entry(tag_removed, ce);
+                       if (show_modified && ce_modified(ce, &st))
+                               show_ce_entry(tag_modified, ce);
+               }
+       }
+}
+
+/*
+ * Prune the index to only contain stuff starting with "prefix"
+ */
+static void prune_cache(void)
+{
+       int pos = cache_name_pos(prefix, prefix_len);
+       unsigned int first, last;
+
+       if (pos < 0)
+               pos = -pos-1;
+       active_cache += pos;
+       active_nr -= pos;
+       first = 0;
+       last = active_nr;
+       while (last > first) {
+               int next = (last + first) >> 1;
+               struct cache_entry *ce = active_cache[next];
+               if (!strncmp(ce->name, prefix, prefix_len)) {
+                       first = next+1;
+                       continue;
+               }
+               last = next;
+       }
+       active_nr = last;
+}
+
+static void verify_pathspec(void)
+{
+       const char **p, *n, *prev;
+       char *real_prefix;
+       unsigned long max;
+
+       prev = NULL;
+       max = PATH_MAX;
+       for (p = pathspec; (n = *p) != NULL; p++) {
+               int i, len = 0;
+               for (i = 0; i < max; i++) {
+                       char c = n[i];
+                       if (prev && prev[i] != c)
+                               break;
+                       if (!c || c == '*' || c == '?')
+                               break;
+                       if (c == '/')
+                               len = i+1;
+               }
+               prev = n;
+               if (len < max) {
+                       max = len;
+                       if (!max)
+                               break;
                }
        }
+
+       if (prefix_offset > max || memcmp(prev, prefix, prefix_offset))
+               die("git-ls-files: cannot generate relative filenames containing '..'");
+
+       real_prefix = NULL;
+       prefix_len = max;
+       if (max) {
+               real_prefix = xmalloc(max + 1);
+               memcpy(real_prefix, prev, max);
+               real_prefix[max] = 0;
+       }
+       prefix = real_prefix;
 }
 
-static const char *ls_files_usage =
-       "git-ls-files [-z] [-t] (--[cached|deleted|others|stage|unmerged|killed])* "
-       "[ --ignored [--exclude=<pattern>] [--exclude-from=<file>) ]";
+static const char ls_files_usage[] =
+       "git-ls-files [-z] [-t] (--[cached|deleted|others|stage|unmerged|killed|modified])* "
+       "[ --ignored ] [--exclude=<pattern>] [--exclude-from=<file>] "
+       "[ --exclude-per-directory=<filename> ] [--] [<file>]*";
 
-int main(int argc, char **argv)
+int main(int argc, const char **argv)
 {
        int i;
+       int exc_given = 0;
+
+       prefix = setup_git_directory();
+       if (prefix)
+               prefix_offset = strlen(prefix);
+       git_config(git_default_config);
 
        for (i = 1; i < argc; i++) {
-               char *arg = argv[i];
+               const char *arg = argv[i];
 
+               if (!strcmp(arg, "--")) {
+                       i++;
+                       break;
+               }
                if (!strcmp(arg, "-z")) {
                        line_terminator = 0;
-               } else if (!strcmp(arg, "-t")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-t")) {
                        tag_cached = "H ";
                        tag_unmerged = "M ";
                        tag_removed = "R ";
+                       tag_modified = "C ";
                        tag_other = "? ";
                        tag_killed = "K ";
-               } else if (!strcmp(arg, "-c") || !strcmp(arg, "--cached")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-c") || !strcmp(arg, "--cached")) {
                        show_cached = 1;
-               } else if (!strcmp(arg, "-d") || !strcmp(arg, "--deleted")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-d") || !strcmp(arg, "--deleted")) {
                        show_deleted = 1;
-               } else if (!strcmp(arg, "-o") || !strcmp(arg, "--others")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-m") || !strcmp(arg, "--modified")) {
+                       show_modified = 1;
+                       continue;
+               }
+               if (!strcmp(arg, "-o") || !strcmp(arg, "--others")) {
                        show_others = 1;
-               } else if (!strcmp(arg, "-i") || !strcmp(arg, "--ignored")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-i") || !strcmp(arg, "--ignored")) {
                        show_ignored = 1;
-               } else if (!strcmp(arg, "-s") || !strcmp(arg, "--stage")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-s") || !strcmp(arg, "--stage")) {
                        show_stage = 1;
-               } else if (!strcmp(arg, "-k") || !strcmp(arg, "--killed")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-k") || !strcmp(arg, "--killed")) {
                        show_killed = 1;
-               } else if (!strcmp(arg, "-u") || !strcmp(arg, "--unmerged")) {
+                       continue;
+               }
+               if (!strcmp(arg, "-u") || !strcmp(arg, "--unmerged")) {
                        /* There's no point in showing unmerged unless
                         * you also show the stage information.
                         */
                        show_stage = 1;
                        show_unmerged = 1;
-               } else if (!strcmp(arg, "-x") && i+1 < argc) {
-                       add_exclude(argv[++i]);
-               } else if (!strncmp(arg, "--exclude=", 10)) {
-                       add_exclude(arg+10);
-               } else if (!strcmp(arg, "-X") && i+1 < argc) {
+                       continue;
+               }
+               if (!strcmp(arg, "-x") && i+1 < argc) {
+                       exc_given = 1;
+                       add_exclude(argv[++i], "", 0, &exclude_list[EXC_CMDL]);
+                       continue;
+               }
+               if (!strncmp(arg, "--exclude=", 10)) {
+                       exc_given = 1;
+                       add_exclude(arg+10, "", 0, &exclude_list[EXC_CMDL]);
+                       continue;
+               }
+               if (!strcmp(arg, "-X") && i+1 < argc) {
+                       exc_given = 1;
                        add_excludes_from_file(argv[++i]);
-               } else if (!strncmp(arg, "--exclude-from=", 15)) {
+                       continue;
+               }
+               if (!strncmp(arg, "--exclude-from=", 15)) {
+                       exc_given = 1;
                        add_excludes_from_file(arg+15);
-               } else
+                       continue;
+               }
+               if (!strncmp(arg, "--exclude-per-directory=", 24)) {
+                       exc_given = 1;
+                       exclude_per_dir = arg + 24;
+                       continue;
+               }
+               if (!strcmp(arg, "--full-name")) {
+                       prefix_offset = 0;
+                       continue;
+               }
+               if (*arg == '-')
                        usage(ls_files_usage);
+               break;
        }
 
-       if (show_ignored && !nr_excludes) {
+       pathspec = get_pathspec(prefix, argv + i);
+
+       /* Verify that the pathspec matches the prefix */
+       if (pathspec)
+               verify_pathspec();
+
+       if (show_ignored && !exc_given) {
                fprintf(stderr, "%s: --ignored needs some exclude pattern\n",
                        argv[0]);
                exit(1);
        }
 
        /* With no flags, we default to showing the cached files */
-       if (!(show_stage | show_deleted | show_others | show_unmerged | show_killed))
+       if (!(show_stage | show_deleted | show_others | show_unmerged |
+             show_killed | show_modified))
                show_cached = 1;
 
        read_cache();
+       if (prefix)
+               prune_cache();
        show_files();
        return 0;
 }