pack-objects: pass fullname down to add_object_entry()
[gitweb.git] / builtin-pack-objects.c
index c72e07a2bb73b74e6be9a038fae4eafd0b88e2b2..631fb8c9a6d136ad31ffc99731d3b54c26b7bda5 100644 (file)
 #include "diff.h"
 #include "revision.h"
 #include "list-objects.h"
+#include "progress.h"
 
 static const char pack_usage[] = "\
 git-pack-objects [{ -q | --progress | --all-progress }] \n\
        [--local] [--incremental] [--window=N] [--depth=N] \n\
-       [--no-reuse-delta] [--delta-base-offset] [--non-empty] \n\
-       [--revs [--unpacked | --all]*] [--reflog] [--stdout | base-name] \n\
-       [<ref-list | <object-list]";
+       [--no-reuse-delta] [--no-reuse-object] [--delta-base-offset] \n\
+       [--non-empty] [--revs [--unpacked | --all]*] [--reflog] \n\
+       [--stdout | base-name] [<ref-list | <object-list]";
 
 struct object_entry {
        unsigned char sha1[20];
@@ -54,7 +55,7 @@ static struct object_entry *objects;
 static uint32_t nr_objects, nr_alloc, nr_result;
 
 static int non_empty;
-static int no_reuse_delta;
+static int no_reuse_delta, no_reuse_object;
 static int local;
 static int incremental;
 static int allow_ofs_delta;
@@ -62,10 +63,13 @@ static const char *pack_tmp_name, *idx_tmp_name;
 static char tmpname[PATH_MAX];
 static unsigned char pack_file_sha1[20];
 static int progress = 1;
-static volatile sig_atomic_t progress_update;
 static int window = 10;
+static int depth = 50;
 static int pack_to_stdout;
 static int num_preferred_base;
+static struct progress progress_state;
+static int pack_compression_level = Z_DEFAULT_COMPRESSION;
+static int pack_compression_seen;
 
 /*
  * The object names in objects array are hashed with this hashtable,
@@ -344,56 +348,6 @@ static void copy_pack_data(struct sha1file *f,
        }
 }
 
-static int check_loose_inflate(unsigned char *data, unsigned long len, unsigned long expect)
-{
-       z_stream stream;
-       unsigned char fakebuf[4096];
-       int st;
-
-       memset(&stream, 0, sizeof(stream));
-       stream.next_in = data;
-       stream.avail_in = len;
-       stream.next_out = fakebuf;
-       stream.avail_out = sizeof(fakebuf);
-       inflateInit(&stream);
-
-       while (1) {
-               st = inflate(&stream, Z_FINISH);
-               if (st == Z_STREAM_END || st == Z_OK) {
-                       st = (stream.total_out == expect &&
-                             stream.total_in == len) ? 0 : -1;
-                       break;
-               }
-               if (st != Z_BUF_ERROR) {
-                       st = -1;
-                       break;
-               }
-               stream.next_out = fakebuf;
-               stream.avail_out = sizeof(fakebuf);
-       }
-       inflateEnd(&stream);
-       return st;
-}
-
-static int revalidate_loose_object(struct object_entry *entry,
-                                  unsigned char *map,
-                                  unsigned long mapsize)
-{
-       /* we already know this is a loose object with new type header. */
-       enum object_type type;
-       unsigned long size, used;
-
-       if (pack_to_stdout)
-               return 0;
-
-       used = unpack_object_header_gently(map, mapsize, &type, &size);
-       if (!used)
-               return -1;
-       map += used;
-       mapsize -= used;
-       return check_loose_inflate(map, mapsize, size);
-}
-
 static unsigned long write_object(struct sha1file *f,
                                  struct object_entry *entry)
 {
@@ -410,7 +364,9 @@ static unsigned long write_object(struct sha1file *f,
                crc32_begin(f);
 
        obj_type = entry->type;
-       if (! entry->in_pack)
+       if (no_reuse_object)
+               to_reuse = 0;   /* explicit */
+       else if (!entry->in_pack)
                to_reuse = 0;   /* can't reuse what we don't have */
        else if (obj_type == OBJ_REF_DELTA || obj_type == OBJ_OFS_DELTA)
                to_reuse = 1;   /* check_object() decided it for us */
@@ -423,25 +379,6 @@ static unsigned long write_object(struct sha1file *f,
                                 * and we do not need to deltify it.
                                 */
 
-       if (!entry->in_pack && !entry->delta) {
-               unsigned char *map;
-               unsigned long mapsize;
-               map = map_sha1_file(entry->sha1, &mapsize);
-               if (map && !legacy_loose_object(map)) {
-                       /* We can copy straight into the pack file */
-                       if (revalidate_loose_object(entry, map, mapsize))
-                               die("corrupt loose object %s",
-                                   sha1_to_hex(entry->sha1));
-                       sha1write(f, map, mapsize);
-                       munmap(map, mapsize);
-                       written++;
-                       reused++;
-                       return mapsize;
-               }
-               if (map)
-                       munmap(map, mapsize);
-       }
-
        if (!to_reuse) {
                buf = read_sha1_file(entry->sha1, &type, &size);
                if (!buf)
@@ -483,7 +420,7 @@ static unsigned long write_object(struct sha1file *f,
                        sha1write(f, entry->delta->sha1, 20);
                        hdrlen += 20;
                }
-               datalen = sha1write_compressed(f, buf, size);
+               datalen = sha1write_compressed(f, buf, size, pack_compression_level);
                free(buf);
        }
        else {
@@ -558,22 +495,25 @@ static off_t write_one(struct sha1file *f,
        return offset + size;
 }
 
+static int open_object_dir_tmp(const char *path)
+{
+    snprintf(tmpname, sizeof(tmpname), "%s/%s", get_object_directory(), path);
+    return mkstemp(tmpname);
+}
+
 static off_t write_pack_file(void)
 {
        uint32_t i;
        struct sha1file *f;
        off_t offset, last_obj_offset = 0;
        struct pack_header hdr;
-       unsigned last_percent = 999;
        int do_progress = progress;
 
        if (pack_to_stdout) {
                f = sha1fd(1, "<stdout>");
                do_progress >>= 1;
        } else {
-               int fd;
-               snprintf(tmpname, sizeof(tmpname), "tmp_pack_XXXXXX");
-               fd = mkstemp(tmpname);
+               int fd = open_object_dir_tmp("tmp_pack_XXXXXX");
                if (fd < 0)
                        die("unable to create %s: %s\n", tmpname, strerror(errno));
                pack_tmp_name = xstrdup(tmpname);
@@ -581,7 +521,7 @@ static off_t write_pack_file(void)
        }
 
        if (do_progress)
-               fprintf(stderr, "Writing %u objects.\n", nr_result);
+               start_progress(&progress_state, "Writing %u objects...", "", nr_result);
 
        hdr.hdr_signature = htonl(PACK_SIGNATURE);
        hdr.hdr_version = htonl(PACK_VERSION);
@@ -593,18 +533,11 @@ static off_t write_pack_file(void)
        for (i = 0; i < nr_objects; i++) {
                last_obj_offset = offset;
                offset = write_one(f, objects + i, offset);
-               if (do_progress) {
-                       unsigned percent = written * 100 / nr_result;
-                       if (progress_update || percent != last_percent) {
-                               fprintf(stderr, "%4u%% (%u/%u) done\r",
-                                       percent, written, nr_result);
-                               progress_update = 0;
-                               last_percent = percent;
-                       }
-               }
+               if (do_progress)
+                       display_progress(&progress_state, written);
        }
        if (do_progress)
-               fputc('\n', stderr);
+               stop_progress(&progress_state);
  done:
        if (written != nr_result)
                die("wrote %u objects while expecting %u", written, nr_result);
@@ -630,10 +563,8 @@ static void write_index_file(off_t last_obj_offset, unsigned char *sha1)
        uint32_t array[256];
        uint32_t i, index_version;
        SHA_CTX ctx;
-       int fd;
 
-       snprintf(tmpname, sizeof(tmpname), "tmp_idx_XXXXXX");
-       fd = mkstemp(tmpname);
+       int fd = open_object_dir_tmp("tmp_idx_XXXXXX");
        if (fd < 0)
                die("unable to create %s: %s\n", tmpname, strerror(errno));
        idx_tmp_name = xstrdup(tmpname);
@@ -792,6 +723,9 @@ static unsigned name_hash(const char *name)
        unsigned char c;
        unsigned hash = 0;
 
+       if (!name)
+               return 0;
+
        /*
         * This effectively just creates a sortable number from the
         * last sixteen non-whitespace characters. Last characters
@@ -806,12 +740,13 @@ static unsigned name_hash(const char *name)
 }
 
 static int add_object_entry(const unsigned char *sha1, enum object_type type,
-                           unsigned hash, int exclude)
+                           const char *name, int exclude)
 {
        struct object_entry *entry;
        struct packed_git *p, *found_pack = NULL;
        off_t found_offset = 0;
        int ix;
+       unsigned hash = name_hash(name);
 
        ix = nr_objects ? locate_object_entry_hash(sha1) : -1;
        if (ix >= 0) {
@@ -865,10 +800,8 @@ static int add_object_entry(const unsigned char *sha1, enum object_type type,
        else
                object_ix[-1 - ix] = nr_objects;
 
-       if (progress_update) {
-               fprintf(stderr, "Counting objects...%u\r", nr_objects);
-               progress_update = 0;
-       }
+       if (progress)
+               display_progress(&progress_state, nr_objects);
 
        return 1;
 }
@@ -1002,10 +935,9 @@ static void add_pbase_object(struct tree_desc *tree,
                if (cmp < 0)
                        return;
                if (name[cmplen] != '/') {
-                       unsigned hash = name_hash(fullname);
                        add_object_entry(entry.sha1,
                                         S_ISDIR(entry.mode) ? OBJ_TREE : OBJ_BLOB,
-                                        hash, 1);
+                                        fullname, 1);
                        return;
                }
                if (S_ISDIR(entry.mode)) {
@@ -1065,10 +997,11 @@ static int check_pbase_path(unsigned hash)
        return 0;
 }
 
-static void add_preferred_base_object(const char *name, unsigned hash)
+static void add_preferred_base_object(const char *name)
 {
        struct pbase_tree *it;
        int cmplen;
+       unsigned hash = name_hash(name);
 
        if (!num_preferred_base || check_pbase_path(hash))
                return;
@@ -1076,7 +1009,7 @@ static void add_preferred_base_object(const char *name, unsigned hash)
        cmplen = name_cmp_len(name);
        for (it = pbase_tree; it; it = it->next) {
                if (cmplen == 0) {
-                       add_object_entry(it->pcache.sha1, OBJ_TREE, 0, 1);
+                       add_object_entry(it->pcache.sha1, OBJ_TREE, NULL, 1);
                }
                else {
                        struct tree_desc tree;
@@ -1131,8 +1064,8 @@ static void check_object(struct object_entry *entry)
                buf = use_pack(p, &w_curs, entry->in_pack_offset, &avail);
 
                /*
-                * We want in_pack_type even if we do not reuse delta.
-                * There is no point not reusing non-delta representations.
+                * We want in_pack_type even if we do not reuse delta
+                * since non-delta representations could still be reused.
                 */
                used = unpack_object_header_gently(buf, avail,
                                                   &entry->in_pack_type,
@@ -1390,7 +1323,6 @@ static void find_deltas(struct object_entry **list, int window, int depth)
        uint32_t i = nr_objects, idx = 0, processed = 0;
        unsigned int array_size = window * sizeof(struct unpacked);
        struct unpacked *array;
-       unsigned last_percent = 999;
        int max_depth;
 
        if (!nr_objects)
@@ -1398,7 +1330,7 @@ static void find_deltas(struct object_entry **list, int window, int depth)
        array = xmalloc(array_size);
        memset(array, 0, array_size);
        if (progress)
-               fprintf(stderr, "Deltifying %u objects.\n", nr_result);
+               start_progress(&progress_state, "Deltifying %u objects...", "", nr_result);
 
        do {
                struct object_entry *entry = list[--i];
@@ -1408,15 +1340,8 @@ static void find_deltas(struct object_entry **list, int window, int depth)
                if (!entry->preferred_base)
                        processed++;
 
-               if (progress) {
-                       unsigned percent = processed * 100 / nr_result;
-                       if (percent != last_percent || progress_update) {
-                               fprintf(stderr, "%4u%% (%u/%u) done\r",
-                                       percent, processed, nr_result);
-                               progress_update = 0;
-                               last_percent = percent;
-                       }
-               }
+               if (progress)
+                       display_progress(&progress_state, processed);
 
                if (entry->delta)
                        /* This happens if we decided to reuse existing
@@ -1471,7 +1396,7 @@ static void find_deltas(struct object_entry **list, int window, int depth)
        } while (i > 0);
 
        if (progress)
-               fputc('\n', stderr);
+               stop_progress(&progress_state);
 
        for (i = 0; i < window; ++i) {
                free_delta_index(array[i].index);
@@ -1498,34 +1423,26 @@ static void prepare_pack(int window, int depth)
        free(delta_list);
 }
 
-static void progress_interval(int signum)
-{
-       progress_update = 1;
-}
-
-static void setup_progress_signal(void)
-{
-       struct sigaction sa;
-       struct itimerval v;
-
-       memset(&sa, 0, sizeof(sa));
-       sa.sa_handler = progress_interval;
-       sigemptyset(&sa.sa_mask);
-       sa.sa_flags = SA_RESTART;
-       sigaction(SIGALRM, &sa, NULL);
-
-       v.it_interval.tv_sec = 1;
-       v.it_interval.tv_usec = 0;
-       v.it_value = v.it_interval;
-       setitimer(ITIMER_REAL, &v, NULL);
-}
-
 static int git_pack_config(const char *k, const char *v)
 {
        if(!strcmp(k, "pack.window")) {
                window = git_config_int(k, v);
                return 0;
        }
+       if(!strcmp(k, "pack.depth")) {
+               depth = git_config_int(k, v);
+               return 0;
+       }
+       if (!strcmp(k, "pack.compression")) {
+               int level = git_config_int(k, v);
+               if (level == -1)
+                       level = Z_DEFAULT_COMPRESSION;
+               else if (level < 0 || level > Z_BEST_COMPRESSION)
+                       die("bad pack compression level %d", level);
+               pack_compression_level = level;
+               pack_compression_seen = 1;
+               return 0;
+       }
        return git_default_config(k, v);
 }
 
@@ -1533,7 +1450,6 @@ static void read_object_list_from_stdin(void)
 {
        char line[40 + 1 + PATH_MAX + 2];
        unsigned char sha1[20];
-       unsigned hash;
 
        for (;;) {
                if (!fgets(line, sizeof(line), stdin)) {
@@ -1556,22 +1472,20 @@ static void read_object_list_from_stdin(void)
                if (get_sha1_hex(line, sha1))
                        die("expected sha1, got garbage:\n %s", line);
 
-               hash = name_hash(line+41);
-               add_preferred_base_object(line+41, hash);
-               add_object_entry(sha1, 0, hash, 0);
+               add_preferred_base_object(line+41);
+               add_object_entry(sha1, 0, line+41, 0);
        }
 }
 
 static void show_commit(struct commit *commit)
 {
-       add_object_entry(commit->object.sha1, OBJ_COMMIT, 0, 0);
+       add_object_entry(commit->object.sha1, OBJ_COMMIT, NULL, 0);
 }
 
 static void show_object(struct object_array_entry *p)
 {
-       unsigned hash = name_hash(p->name);
-       add_preferred_base_object(p->name, hash);
-       add_object_entry(p->item->sha1, p->item->type, hash, 0);
+       add_preferred_base_object(p->name);
+       add_object_entry(p->item->sha1, p->item->type, p->name, 0);
 }
 
 static void show_edge(struct commit *commit)
@@ -1612,9 +1526,15 @@ static void get_object_list(int ac, const char **av)
        traverse_commit_list(&revs, show_commit, show_object);
 }
 
+static int adjust_perm(const char *path, mode_t mode)
+{
+       if (chmod(path, mode))
+               return -1;
+       return adjust_shared_perm(path);
+}
+
 int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 {
-       int depth = 10;
        int use_internal_rev_list = 0;
        int thin = 0;
        uint32_t i;
@@ -1631,6 +1551,8 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
        rp_ac = 2;
 
        git_config(git_pack_config);
+       if (!pack_compression_seen && core_compression_seen)
+               pack_compression_level = core_compression_level;
 
        progress = isatty(2);
        for (i = 1; i < argc; i++) {
@@ -1651,6 +1573,18 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
                        incremental = 1;
                        continue;
                }
+               if (!prefixcmp(arg, "--compression=")) {
+                       char *end;
+                       int level = strtoul(arg+14, &end, 0);
+                       if (!arg[14] || *end)
+                               usage(pack_usage);
+                       if (level == -1)
+                               level = Z_DEFAULT_COMPRESSION;
+                       else if (level < 0 || level > Z_BEST_COMPRESSION)
+                               die("bad pack compression level %d", level);
+                       pack_compression_level = level;
+                       continue;
+               }
                if (!prefixcmp(arg, "--window=")) {
                        char *end;
                        window = strtoul(arg+9, &end, 0);
@@ -1681,6 +1615,10 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
                        no_reuse_delta = 1;
                        continue;
                }
+               if (!strcmp("--no-reuse-object", arg)) {
+                       no_reuse_object = no_reuse_delta = 1;
+                       continue;
+               }
                if (!strcmp("--delta-base-offset", arg)) {
                        allow_ofs_delta = 1;
                        continue;
@@ -1750,44 +1688,48 @@ int cmd_pack_objects(int argc, const char **argv, const char *prefix)
 
        prepare_packed_git();
 
-       if (progress) {
-               fprintf(stderr, "Generating pack...\n");
-               setup_progress_signal();
-       }
-
+       if (progress)
+               start_progress(&progress_state, "Generating pack...",
+                              "Counting objects: ", 0);
        if (!use_internal_rev_list)
                read_object_list_from_stdin();
        else {
                rp_av[rp_ac] = NULL;
                get_object_list(rp_ac, rp_av);
        }
-
-       if (progress)
+       if (progress) {
+               stop_progress(&progress_state);
                fprintf(stderr, "Done counting %u objects.\n", nr_objects);
+       }
+
        if (non_empty && !nr_result)
                return 0;
        if (progress && (nr_objects != nr_result))
                fprintf(stderr, "Result has %u objects.\n", nr_result);
        if (nr_result)
                prepare_pack(window, depth);
-       if (progress == 1 && pack_to_stdout) {
-               /* the other end usually displays progress itself */
-               struct itimerval v = {{0,},};
-               setitimer(ITIMER_REAL, &v, NULL);
-               signal(SIGALRM, SIG_IGN );
-               progress_update = 0;
-       }
        last_obj_offset = write_pack_file();
        if (!pack_to_stdout) {
                unsigned char object_list_sha1[20];
+               mode_t mode = umask(0);
+
+               umask(mode);
+               mode = 0444 & ~mode;
+
                write_index_file(last_obj_offset, object_list_sha1);
                snprintf(tmpname, sizeof(tmpname), "%s-%s.pack",
                         base_name, sha1_to_hex(object_list_sha1));
+               if (adjust_perm(pack_tmp_name, mode))
+                       die("unable to make temporary pack file readable: %s",
+                           strerror(errno));
                if (rename(pack_tmp_name, tmpname))
                        die("unable to rename temporary pack file: %s",
                            strerror(errno));
                snprintf(tmpname, sizeof(tmpname), "%s-%s.idx",
                         base_name, sha1_to_hex(object_list_sha1));
+               if (adjust_perm(idx_tmp_name, mode))
+                       die("unable to make temporary index file readable: %s",
+                           strerror(errno));
                if (rename(idx_tmp_name, tmpname))
                        die("unable to rename temporary index file: %s",
                            strerror(errno));