fast-import: prevent producing bad delta
[gitweb.git] / fast-import.c
index b3ba778924908946a0e20ebfdae9bc2058e8a55e..6dad9ff4db81bbdda0e44f2c59bb3dcc9f9fb585 100644 (file)
@@ -24,10 +24,12 @@ Format of STDIN stream:
     commit_msg
     ('from' sp committish lf)?
     ('merge' sp committish lf)*
-    file_change*
+    (file_change | ls)*
     lf?;
   commit_msg ::= data;
 
+  ls ::= 'ls' sp '"' quoted(path) '"' lf;
+
   file_change ::= file_clr
     | file_del
     | file_rnm
@@ -132,7 +134,7 @@ Format of STDIN stream:
   ts    ::= # time since the epoch in seconds, ascii base10 notation;
   tz    ::= # GIT style timezone;
 
-     # note: comments and cat requests may appear anywhere
+     # note: comments, ls and cat requests may appear anywhere
      # in the input, except within a data command.  Any form
      # of the data command always escapes the related input
      # from comment processing.
@@ -141,7 +143,9 @@ Format of STDIN stream:
      # must be the first character on that line (an lf
      # preceded it).
      #
+
   cat_blob ::= 'cat-blob' sp (hexsha1 | idnum) lf;
+  ls_tree  ::= 'ls' sp (hexsha1 | idnum) sp path_str lf;
 
   comment ::= '#' not_lf* lf;
   not_lf  ::= # Any byte that is not ASCII newline (LF);
@@ -166,6 +170,11 @@ Format of STDIN stream:
 #define DEPTH_BITS 13
 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
 
+/*
+ * We abuse the setuid bit on directories to mean "do not delta".
+ */
+#define NO_DELTA S_ISUID
+
 struct object_entry {
        struct pack_idx_entry idx;
        struct object_entry *next;
@@ -270,7 +279,6 @@ struct recent_command {
 /* Configured limits on output */
 static unsigned long max_depth = 10;
 static off_t max_packsize;
-static uintmax_t big_file_threshold = 512 * 1024 * 1024;
 static int force_update;
 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
 static int pack_compression_seen;
@@ -315,6 +323,7 @@ static struct mark_set *marks;
 static const char *export_marks_file;
 static const char *import_marks_file;
 static int import_marks_file_from_stream;
+static int import_marks_file_ignore_missing;
 static int relative_marks_paths;
 
 /* Our last blob */
@@ -359,6 +368,7 @@ static int cat_blob_fd = STDOUT_FILENO;
 
 static void parse_argv(void);
 static void parse_cat_blob(void);
+static void parse_ls(struct branch *b);
 
 static void write_branch_report(FILE *rpt, struct branch *b)
 {
@@ -1409,8 +1419,9 @@ static void mktree(struct tree_content *t, int v, struct strbuf *b)
                struct tree_entry *e = t->entries[i];
                if (!e->versions[v].mode)
                        continue;
-               strbuf_addf(b, "%o %s%c", (unsigned int)e->versions[v].mode,
-                                       e->name->str_dat, '\0');
+               strbuf_addf(b, "%o %s%c",
+                       (unsigned int)(e->versions[v].mode & ~NO_DELTA),
+                       e->name->str_dat, '\0');
                strbuf_add(b, e->versions[v].sha1, 20);
        }
 }
@@ -1420,7 +1431,7 @@ static void store_tree(struct tree_entry *root)
        struct tree_content *t = root->tree;
        unsigned int i, j, del;
        struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
-       struct object_entry *le;
+       struct object_entry *le = NULL;
 
        if (!is_null_sha1(root->versions[1].sha1))
                return;
@@ -1430,7 +1441,8 @@ static void store_tree(struct tree_entry *root)
                        store_tree(t->entries[i]);
        }
 
-       le = find_object(root->versions[0].sha1);
+       if (!(root->versions[0].mode & NO_DELTA))
+               le = find_object(root->versions[0].sha1);
        if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
                mktree(t, 0, &old_tree);
                lo.data = old_tree;
@@ -1464,6 +1476,7 @@ static void tree_content_replace(
 {
        if (!S_ISDIR(mode))
                die("Root cannot be a non-directory");
+       hashclr(root->versions[0].sha1);
        hashcpy(root->versions[1].sha1, sha1);
        if (root->tree)
                release_tree_content_recursive(root->tree);
@@ -1508,6 +1521,23 @@ static int tree_content_set(
                                if (e->tree)
                                        release_tree_content_recursive(e->tree);
                                e->tree = subtree;
+
+                               /*
+                                * We need to leave e->versions[0].sha1 alone
+                                * to avoid modifying the preimage tree used
+                                * when writing out the parent directory.
+                                * But after replacing the subdir with a
+                                * completely different one, it's not a good
+                                * delta base any more, and besides, we've
+                                * thrown away the tree entries needed to
+                                * make a delta against it.
+                                *
+                                * So let's just explicitly disable deltas
+                                * for the subtree.
+                                */
+                               if (S_ISDIR(e->versions[0].mode))
+                                       e->versions[0].mode |= NO_DELTA;
+
                                hashclr(root->versions[1].sha1);
                                return 1;
                        }
@@ -1782,7 +1812,11 @@ static void read_marks(void)
 {
        char line[512];
        FILE *f = fopen(import_marks_file, "r");
-       if (!f)
+       if (f)
+               ;
+       else if (import_marks_file_ignore_missing && errno == ENOENT)
+               return; /* Marks file does not exist */
+       else
                die_errno("cannot read '%s'", import_marks_file);
        while (fgets(line, sizeof(line), f)) {
                uintmax_t mark;
@@ -2596,6 +2630,8 @@ static void parse_new_commit(void)
                        note_change_n(b, prev_fanout);
                else if (!strcmp("deleteall", command_buf.buf))
                        file_change_deleteall(b);
+               else if (!prefixcmp(command_buf.buf, "ls "))
+                       parse_ls(b);
                else {
                        unread_command_buf = 1;
                        break;
@@ -2819,6 +2855,153 @@ static void parse_cat_blob(void)
        cat_blob(oe, sha1);
 }
 
+static struct object_entry *dereference(struct object_entry *oe,
+                                       unsigned char sha1[20])
+{
+       unsigned long size;
+       char *buf = NULL;
+       if (!oe) {
+               enum object_type type = sha1_object_info(sha1, NULL);
+               if (type < 0)
+                       die("object not found: %s", sha1_to_hex(sha1));
+               /* cache it! */
+               oe = insert_object(sha1);
+               oe->type = type;
+               oe->pack_id = MAX_PACK_ID;
+               oe->idx.offset = 1;
+       }
+       switch (oe->type) {
+       case OBJ_TREE:  /* easy case. */
+               return oe;
+       case OBJ_COMMIT:
+       case OBJ_TAG:
+               break;
+       default:
+               die("Not a treeish: %s", command_buf.buf);
+       }
+
+       if (oe->pack_id != MAX_PACK_ID) {       /* in a pack being written */
+               buf = gfi_unpack_entry(oe, &size);
+       } else {
+               enum object_type unused;
+               buf = read_sha1_file(sha1, &unused, &size);
+       }
+       if (!buf)
+               die("Can't load object %s", sha1_to_hex(sha1));
+
+       /* Peel one layer. */
+       switch (oe->type) {
+       case OBJ_TAG:
+               if (size < 40 + strlen("object ") ||
+                   get_sha1_hex(buf + strlen("object "), sha1))
+                       die("Invalid SHA1 in tag: %s", command_buf.buf);
+               break;
+       case OBJ_COMMIT:
+               if (size < 40 + strlen("tree ") ||
+                   get_sha1_hex(buf + strlen("tree "), sha1))
+                       die("Invalid SHA1 in commit: %s", command_buf.buf);
+       }
+
+       free(buf);
+       return find_object(sha1);
+}
+
+static struct object_entry *parse_treeish_dataref(const char **p)
+{
+       unsigned char sha1[20];
+       struct object_entry *e;
+
+       if (**p == ':') {       /* <mark> */
+               char *endptr;
+               e = find_mark(strtoumax(*p + 1, &endptr, 10));
+               if (endptr == *p + 1)
+                       die("Invalid mark: %s", command_buf.buf);
+               if (!e)
+                       die("Unknown mark: %s", command_buf.buf);
+               *p = endptr;
+               hashcpy(sha1, e->idx.sha1);
+       } else {        /* <sha1> */
+               if (get_sha1_hex(*p, sha1))
+                       die("Invalid SHA1: %s", command_buf.buf);
+               e = find_object(sha1);
+               *p += 40;
+       }
+
+       while (!e || e->type != OBJ_TREE)
+               e = dereference(e, sha1);
+       return e;
+}
+
+static void print_ls(int mode, const unsigned char *sha1, const char *path)
+{
+       static struct strbuf line = STRBUF_INIT;
+
+       /* See show_tree(). */
+       const char *type =
+               S_ISGITLINK(mode) ? commit_type :
+               S_ISDIR(mode) ? tree_type :
+               blob_type;
+
+       if (!mode) {
+               /* missing SP path LF */
+               strbuf_reset(&line);
+               strbuf_addstr(&line, "missing ");
+               quote_c_style(path, &line, NULL, 0);
+               strbuf_addch(&line, '\n');
+       } else {
+               /* mode SP type SP object_name TAB path LF */
+               strbuf_reset(&line);
+               strbuf_addf(&line, "%06o %s %s\t",
+                               mode & ~NO_DELTA, type, sha1_to_hex(sha1));
+               quote_c_style(path, &line, NULL, 0);
+               strbuf_addch(&line, '\n');
+       }
+       cat_blob_write(line.buf, line.len);
+}
+
+static void parse_ls(struct branch *b)
+{
+       const char *p;
+       struct tree_entry *root = NULL;
+       struct tree_entry leaf = {NULL};
+
+       /* ls SP (<treeish> SP)? <path> */
+       p = command_buf.buf + strlen("ls ");
+       if (*p == '"') {
+               if (!b)
+                       die("Not in a commit: %s", command_buf.buf);
+               root = &b->branch_tree;
+       } else {
+               struct object_entry *e = parse_treeish_dataref(&p);
+               root = new_tree_entry();
+               hashcpy(root->versions[1].sha1, e->idx.sha1);
+               load_tree(root);
+               if (*p++ != ' ')
+                       die("Missing space after tree-ish: %s", command_buf.buf);
+       }
+       if (*p == '"') {
+               static struct strbuf uq = STRBUF_INIT;
+               const char *endp;
+               strbuf_reset(&uq);
+               if (unquote_c_style(&uq, p, &endp))
+                       die("Invalid path: %s", command_buf.buf);
+               if (*endp)
+                       die("Garbage after path in: %s", command_buf.buf);
+               p = uq.buf;
+       }
+       tree_content_get(root, p, &leaf);
+       /*
+        * A directory in preparation would have a sha1 of zero
+        * until it is saved.  Save, for simplicity.
+        */
+       if (S_ISDIR(leaf.versions[1].mode))
+               store_tree(&leaf);
+
+       print_ls(leaf.versions[1].mode, leaf.versions[1].sha1, p);
+       if (!b || root != &b->branch_tree)
+               release_tree_entry(root);
+}
+
 static void checkpoint(void)
 {
        checkpoint_requested = 0;
@@ -2854,7 +3037,8 @@ static char* make_fast_import_path(const char *path)
        return strbuf_detach(&abs_path, NULL);
 }
 
-static void option_import_marks(const char *marks, int from_stream)
+static void option_import_marks(const char *marks,
+                                       int from_stream, int ignore_missing)
 {
        if (import_marks_file) {
                if (from_stream)
@@ -2868,6 +3052,7 @@ static void option_import_marks(const char *marks, int from_stream)
        import_marks_file = make_fast_import_path(marks);
        safe_create_leading_directories_const(import_marks_file);
        import_marks_file_from_stream = from_stream;
+       import_marks_file_ignore_missing = ignore_missing;
 }
 
 static void option_date_format(const char *fmt)
@@ -2967,18 +3152,21 @@ static int parse_one_feature(const char *feature, int from_stream)
        if (!prefixcmp(feature, "date-format=")) {
                option_date_format(feature + 12);
        } else if (!prefixcmp(feature, "import-marks=")) {
-               option_import_marks(feature + 13, from_stream);
+               option_import_marks(feature + 13, from_stream, 0);
+       } else if (!prefixcmp(feature, "import-marks-if-exists=")) {
+               option_import_marks(feature + strlen("import-marks-if-exists="),
+                                       from_stream, 1);
        } else if (!prefixcmp(feature, "export-marks=")) {
                option_export_marks(feature + 13);
        } else if (!strcmp(feature, "cat-blob")) {
                ; /* Don't die - this feature is supported */
-       } else if (!prefixcmp(feature, "relative-marks")) {
+       } else if (!strcmp(feature, "relative-marks")) {
                relative_marks_paths = 1;
-       } else if (!prefixcmp(feature, "no-relative-marks")) {
+       } else if (!strcmp(feature, "no-relative-marks")) {
                relative_marks_paths = 0;
-       } else if (!prefixcmp(feature, "force")) {
+       } else if (!strcmp(feature, "force")) {
                force_update = 1;
-       } else if (!strcmp(feature, "notes")) {
+       } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
                ; /* do nothing; we have the feature */
        } else {
                return 0;
@@ -3042,10 +3230,6 @@ static int git_pack_config(const char *k, const char *v, void *cb)
                max_packsize = git_config_ulong(k, v);
                return 0;
        }
-       if (!strcmp(k, "core.bigfilethreshold")) {
-               long n = git_config_int(k, v);
-               big_file_threshold = 0 < n ? n : 0;
-       }
        return git_default_config(k, v, cb);
 }
 
@@ -3119,6 +3303,8 @@ int main(int argc, const char **argv)
        while (read_next_command() != EOF) {
                if (!strcmp("blob", command_buf.buf))
                        parse_new_blob();
+               else if (!prefixcmp(command_buf.buf, "ls "))
+                       parse_ls(NULL);
                else if (!prefixcmp(command_buf.buf, "commit "))
                        parse_new_commit();
                else if (!prefixcmp(command_buf.buf, "tag "))