Code

Dump all refs and marks during a checkpoint in fast-import.
[git.git] / fast-import.c
index c0cadc4279dc0eebf476a831b6cf8ed767eadd6d..d9ed3e2f18ccba41da216c9a94031b2ae44025c8 100644 (file)
@@ -17,8 +17,8 @@ Format of STDIN stream:
 
   new_commit ::= 'commit' sp ref_str lf
     mark?
-    ('author' sp name '<' email '>' ts tz lf)?
-    'committer' sp name '<' email '>' ts tz lf
+    ('author' sp name '<' email '>' when lf)?
+    'committer' sp name '<' email '>' when lf
     commit_msg
     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
@@ -26,7 +26,8 @@ Format of STDIN stream:
     lf;
   commit_msg ::= data;
 
-  file_change ::= file_del | file_obm | file_inm;
+  file_change ::= file_clr | file_del | file_obm | file_inm;
+  file_clr ::= 'deleteall' lf;
   file_del ::= 'D' sp path_str lf;
   file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
   file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
@@ -34,7 +35,7 @@ Format of STDIN stream:
 
   new_tag ::= 'tag' sp tag_str lf
     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
-       'tagger' sp name '<' email '>' ts tz lf
+       'tagger' sp name '<' email '>' when lf
     tag_msg;
   tag_msg ::= data;
 
@@ -81,13 +82,17 @@ Format of STDIN stream:
   path_str    ::= path    | '"' quoted(path)    '"' ;
   mode        ::= '100644' | '644'
                 | '100755' | '755'
-                | '140000'
+                | '120000'
                 ;
 
   declen ::= # unsigned 32 bit value, ascii base10 notation;
   bigint ::= # unsigned integer value, ascii base10 notation;
   binary_data ::= # file content, not interpreted;
 
+  when         ::= raw_when | rfc2822_when;
+  raw_when     ::= ts sp tz;
+  rfc2822_when ::= # Valid RFC 2822 date and time;
+
   sp ::= # ASCII space character;
   lf ::= # ASCII newline (LF) character;
 
@@ -117,6 +122,7 @@ Format of STDIN stream:
 #include "object.h"
 #include "blob.h"
 #include "tree.h"
+#include "commit.h"
 #include "delta.h"
 #include "pack.h"
 #include "refs.h"
@@ -234,9 +240,16 @@ struct hash_list
        unsigned char sha1[20];
 };
 
+typedef enum {
+       WHENSPEC_RAW = 1,
+       WHENSPEC_RFC2822,
+       WHENSPEC_NOW,
+} whenspec_type;
+
 /* Configured limits on output */
 static unsigned long max_depth = 10;
 static unsigned long max_packsize = (1LL << 32) - 1;
+static int force_update;
 
 /* Stats and misc. counters */
 static uintmax_t alloc_count;
@@ -247,6 +260,7 @@ static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 static unsigned long object_count;
 static unsigned long branch_count;
 static unsigned long branch_load_count;
+static int failure;
 
 /* Memory pools */
 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
@@ -294,6 +308,7 @@ static struct tag *first_tag;
 static struct tag *last_tag;
 
 /* Input stream parsing */
+static whenspec_type whenspec = WHENSPEC_RAW;
 static struct strbuf command_buf;
 static uintmax_t next_mark;
 static struct dbuf new_data;
@@ -823,7 +838,7 @@ static void end_packfile(void)
        last_blob.depth = 0;
 }
 
-static void checkpoint(void)
+static void cycle_packfile(void)
 {
        end_packfile();
        start_packfile();
@@ -916,7 +931,7 @@ static int store_object(
 
                /* This new object needs to *not* have the current pack_id. */
                e->pack_id = pack_id + 1;
-               checkpoint();
+               cycle_packfile();
 
                /* We cannot carry a delta into the new pack. */
                if (delta) {
@@ -969,10 +984,9 @@ static int store_object(
        pack_size += s.total_out;
 
        free(out);
-       if (delta)
-               free(delta);
+       free(delta);
        if (last) {
-               if (last->data && !last->no_free)
+               if (!last->no_free)
                        free(last->data);
                last->data = dat;
                last->offset = e->offset;
@@ -1268,19 +1282,48 @@ del_entry:
        return 1;
 }
 
-static void dump_branches(void)
+static int update_branch(struct branch *b)
 {
        static const char *msg = "fast-import";
+       struct ref_lock *lock;
+       unsigned char old_sha1[20];
+
+       if (read_ref(b->name, old_sha1))
+               hashclr(old_sha1);
+       lock = lock_any_ref_for_update(b->name, old_sha1);
+       if (!lock)
+               return error("Unable to lock %s", b->name);
+       if (!force_update && !is_null_sha1(old_sha1)) {
+               struct commit *old_cmit, *new_cmit;
+
+               old_cmit = lookup_commit_reference_gently(old_sha1, 0);
+               new_cmit = lookup_commit_reference_gently(b->sha1, 0);
+               if (!old_cmit || !new_cmit) {
+                       unlock_ref(lock);
+                       return error("Branch %s is missing commits.", b->name);
+               }
+
+               if (!in_merge_bases(old_cmit, new_cmit)) {
+                       unlock_ref(lock);
+                       warn("Not updating %s"
+                               " (new tip %s does not contain %s)",
+                               b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
+                       return -1;
+               }
+       }
+       if (write_ref_sha1(lock, b->sha1, msg) < 0)
+               return error("Unable to update %s", b->name);
+       return 0;
+}
+
+static void dump_branches(void)
+{
        unsigned int i;
        struct branch *b;
-       struct ref_lock *lock;
 
        for (i = 0; i < branch_table_sz; i++) {
-               for (b = branch_table[i]; b; b = b->table_next_branch) {
-                       lock = lock_any_ref_for_update(b->name, NULL);
-                       if (!lock || write_ref_sha1(lock, b->sha1, msg) < 0)
-                               die("Can't write %s", b->name);
-               }
+               for (b = branch_table[i]; b; b = b->table_next_branch)
+                       failure |= update_branch(b);
        }
 }
 
@@ -1289,13 +1332,13 @@ static void dump_tags(void)
        static const char *msg = "fast-import";
        struct tag *t;
        struct ref_lock *lock;
-       char path[PATH_MAX];
+       char ref_name[PATH_MAX];
 
        for (t = first_tag; t; t = t->next_tag) {
-               sprintf(path, "refs/tags/%s", t->name);
-               lock = lock_any_ref_for_update(path, NULL);
+               sprintf(ref_name, "tags/%s", t->name);
+               lock = lock_ref_sha1(ref_name, NULL);
                if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
-                       die("Can't write %s", path);
+                       failure |= error("Unable to update %s", ref_name);
        }
 }
 
@@ -1397,6 +1440,64 @@ static void *cmd_data (size_t *size)
        return buffer;
 }
 
+static int validate_raw_date(const char *src, char *result, int maxlen)
+{
+       const char *orig_src = src;
+       char *endp, sign;
+
+       strtoul(src, &endp, 10);
+       if (endp == src || *endp != ' ')
+               return -1;
+
+       src = endp + 1;
+       if (*src != '-' && *src != '+')
+               return -1;
+       sign = *src;
+
+       strtoul(src + 1, &endp, 10);
+       if (endp == src || *endp || (endp - orig_src) >= maxlen)
+               return -1;
+
+       strcpy(result, orig_src);
+       return 0;
+}
+
+static char *parse_ident(const char *buf)
+{
+       const char *gt;
+       size_t name_len;
+       char *ident;
+
+       gt = strrchr(buf, '>');
+       if (!gt)
+               die("Missing > in ident string: %s", buf);
+       gt++;
+       if (*gt != ' ')
+               die("Missing space after > in ident string: %s", buf);
+       gt++;
+       name_len = gt - buf;
+       ident = xmalloc(name_len + 24);
+       strncpy(ident, buf, name_len);
+
+       switch (whenspec) {
+       case WHENSPEC_RAW:
+               if (validate_raw_date(gt, ident + name_len, 24) < 0)
+                       die("Invalid raw date \"%s\" in ident: %s", gt, buf);
+               break;
+       case WHENSPEC_RFC2822:
+               if (parse_date(gt, ident + name_len, 24) < 0)
+                       die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
+               break;
+       case WHENSPEC_NOW:
+               if (strcmp("now", gt))
+                       die("Date in ident must be 'now': %s", buf);
+               datestamp(ident + name_len, 24);
+               break;
+       }
+
+       return ident;
+}
+
 static void cmd_new_blob(void)
 {
        size_t l;
@@ -1521,9 +1622,7 @@ static void file_change_m(struct branch *b)
        }
 
        tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode);
-
-       if (p_uq)
-               free(p_uq);
+       free(p_uq);
 }
 
 static void file_change_d(struct branch *b)
@@ -1539,8 +1638,15 @@ static void file_change_d(struct branch *b)
                p = p_uq;
        }
        tree_content_remove(&b->branch_tree, p);
-       if (p_uq)
-               free(p_uq);
+       free(p_uq);
+}
+
+static void file_change_deleteall(struct branch *b)
+{
+       release_tree_content_recursive(b->branch_tree.tree);
+       hashclr(b->branch_tree.versions[0].sha1);
+       hashclr(b->branch_tree.versions[1].sha1);
+       load_tree(&b->branch_tree);
 }
 
 static void cmd_from(struct branch *b)
@@ -1659,11 +1765,11 @@ static void cmd_new_commit(void)
        read_next_command();
        cmd_mark();
        if (!strncmp("author ", command_buf.buf, 7)) {
-               author = strdup(command_buf.buf);
+               author = parse_ident(command_buf.buf + 7);
                read_next_command();
        }
        if (!strncmp("committer ", command_buf.buf, 10)) {
-               committer = strdup(command_buf.buf);
+               committer = parse_ident(command_buf.buf + 10);
                read_next_command();
        }
        if (!committer)
@@ -1687,6 +1793,8 @@ static void cmd_new_commit(void)
                        file_change_m(b);
                else if (!strncmp("D ", command_buf.buf, 2))
                        file_change_d(b);
+               else if (!strcmp("deleteall", command_buf.buf))
+                       file_change_deleteall(b);
                else
                        die("Unsupported file_change: %s", command_buf.buf);
                read_next_command();
@@ -1696,7 +1804,7 @@ static void cmd_new_commit(void)
        store_tree(&b->branch_tree);
        hashcpy(b->branch_tree.versions[0].sha1,
                b->branch_tree.versions[1].sha1);
-       size_dbuf(&new_data, 97 + msglen
+       size_dbuf(&new_data, 114 + msglen
                + merge_count * 49
                + (author
                        ? strlen(author) + strlen(committer)
@@ -1712,15 +1820,12 @@ static void cmd_new_commit(void)
                free(merge_list);
                merge_list = next;
        }
-       if (author)
-               sp += sprintf(sp, "%s\n", author);
-       else
-               sp += sprintf(sp, "author %s\n", committer + 10);
-       sp += sprintf(sp, "%s\n\n", committer);
+       sp += sprintf(sp, "author %s\n", author ? author : committer);
+       sp += sprintf(sp, "committer %s\n", committer);
+       *sp++ = '\n';
        memcpy(sp, msg, msglen);
        sp += msglen;
-       if (author)
-               free(author);
+       free(author);
        free(committer);
        free(msg);
 
@@ -1785,7 +1890,7 @@ static void cmd_new_tag(void)
        /* tagger ... */
        if (strncmp("tagger ", command_buf.buf, 7))
                die("Expected tagger command, got %s", command_buf.buf);
-       tagger = strdup(command_buf.buf);
+       tagger = parse_ident(command_buf.buf + 7);
 
        /* tag payload/message */
        read_next_command();
@@ -1797,7 +1902,8 @@ static void cmd_new_tag(void)
        sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
        sp += sprintf(sp, "type %s\n", type_names[OBJ_COMMIT]);
        sp += sprintf(sp, "tag %s\n", t->name);
-       sp += sprintf(sp, "%s\n\n", tagger);
+       sp += sprintf(sp, "tagger %s\n", tagger);
+       *sp++ = '\n';
        memcpy(sp, msg, msglen);
        sp += msglen;
        free(tagger);
@@ -1834,18 +1940,21 @@ static void cmd_reset_branch(void)
 
 static void cmd_checkpoint(void)
 {
-       if (object_count)
-               checkpoint();
+       if (object_count) {
+               cycle_packfile();
+               dump_branches();
+               dump_tags();
+               dump_marks();
+       }
        read_next_command();
 }
 
 static const char fast_import_usage[] =
-"git-fast-import [--depth=n] [--active-branches=n] [--export-marks=marks.file] [--branch-log=log]";
+"git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
 
 int main(int argc, const char **argv)
 {
-       int i;
-       uintmax_t total_count, duplicate_count;
+       int i, show_stats = 1;
 
        git_config(git_default_config);
 
@@ -1854,6 +1963,17 @@ int main(int argc, const char **argv)
 
                if (*a != '-' || !strcmp(a, "--"))
                        break;
+               else if (!strncmp(a, "--date-format=", 14)) {
+                       const char *fmt = a + 14;
+                       if (!strcmp(fmt, "raw"))
+                               whenspec = WHENSPEC_RAW;
+                       else if (!strcmp(fmt, "rfc2822"))
+                               whenspec = WHENSPEC_RFC2822;
+                       else if (!strcmp(fmt, "now"))
+                               whenspec = WHENSPEC_NOW;
+                       else
+                               die("unknown --date-format argument %s", fmt);
+               }
                else if (!strncmp(a, "--max-pack-size=", 16))
                        max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
                else if (!strncmp(a, "--depth=", 8))
@@ -1862,6 +1982,12 @@ int main(int argc, const char **argv)
                        max_active_branches = strtoul(a + 18, NULL, 0);
                else if (!strncmp(a, "--export-marks=", 15))
                        mark_file = a + 15;
+               else if (!strcmp(a, "--force"))
+                       force_update = 1;
+               else if (!strcmp(a, "--quiet"))
+                       show_stats = 0;
+               else if (!strcmp(a, "--stats"))
+                       show_stats = 1;
                else
                        die("unknown option %s", a);
        }
@@ -1901,31 +2027,32 @@ int main(int argc, const char **argv)
        unkeep_all_packs();
        dump_marks();
 
-       total_count = 0;
-       for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
-               total_count += object_count_by_type[i];
-       duplicate_count = 0;
-       for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
-               duplicate_count += duplicate_count_by_type[i];
-
-       fprintf(stderr, "%s statistics:\n", argv[0]);
-       fprintf(stderr, "---------------------------------------------------------------------\n");
-       fprintf(stderr, "Alloc'd objects: %10ju\n", alloc_count);
-       fprintf(stderr, "Total objects:   %10ju (%10ju duplicates                  )\n", total_count, duplicate_count);
-       fprintf(stderr, "      blobs  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
-       fprintf(stderr, "      trees  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
-       fprintf(stderr, "      commits:   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
-       fprintf(stderr, "      tags   :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
-       fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
-       fprintf(stderr, "      marks:     %10ju (%10ju unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
-       fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
-       fprintf(stderr, "Memory total:    %10ju KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
-       fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
-       fprintf(stderr, "     objects:    %10ju KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
-       fprintf(stderr, "---------------------------------------------------------------------\n");
-       pack_report();
-       fprintf(stderr, "---------------------------------------------------------------------\n");
-       fprintf(stderr, "\n");
+       if (show_stats) {
+               uintmax_t total_count = 0, duplicate_count = 0;
+               for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
+                       total_count += object_count_by_type[i];
+               for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
+                       duplicate_count += duplicate_count_by_type[i];
+
+               fprintf(stderr, "%s statistics:\n", argv[0]);
+               fprintf(stderr, "---------------------------------------------------------------------\n");
+               fprintf(stderr, "Alloc'd objects: %10ju\n", alloc_count);
+               fprintf(stderr, "Total objects:   %10ju (%10ju duplicates                  )\n", total_count, duplicate_count);
+               fprintf(stderr, "      blobs  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
+               fprintf(stderr, "      trees  :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
+               fprintf(stderr, "      commits:   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
+               fprintf(stderr, "      tags   :   %10ju (%10ju duplicates %10ju deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
+               fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
+               fprintf(stderr, "      marks:     %10ju (%10ju unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
+               fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
+               fprintf(stderr, "Memory total:    %10ju KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
+               fprintf(stderr, "       pools:    %10lu KiB\n", total_allocd/1024);
+               fprintf(stderr, "     objects:    %10ju KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
+               fprintf(stderr, "---------------------------------------------------------------------\n");
+               pack_report();
+               fprintf(stderr, "---------------------------------------------------------------------\n");
+               fprintf(stderr, "\n");
+       }
 
-       return 0;
+       return failure ? 1 : 0;
 }