Use a hashtable for objects instead of a sorted list
[git.git] / pack-objects.c
index fc969e3..2135e9a 100644 (file)
@@ -1,11 +1,11 @@
-#include <ctype.h>
 #include "cache.h"
 #include "object.h"
 #include "delta.h"
 #include "pack.h"
 #include "csum-file.h"
+#include <sys/time.h>
 
-static const char pack_usage[] = "git-pack-objects [--window=N] [--depth=N] {--stdout | base-name} < object-list";
+static const char pack_usage[] = "git-pack-objects [--non-empty] [--local] [--incremental] [--window=N] [--depth=N] {--stdout | base-name} < object-list";
 
 struct object_entry {
        unsigned char sha1[20];
@@ -18,11 +18,16 @@ struct object_entry {
        struct object_entry *delta;
 };
 
+static unsigned char object_list_sha1[20];
+static int non_empty = 0;
+static int local = 0;
+static int incremental = 0;
 static struct object_entry **sorted_by_sha, **sorted_by_type;
 static struct object_entry *objects = NULL;
 static int nr_objects = 0, nr_alloc = 0;
 static const char *base_name;
 static unsigned char pack_file_sha1[20];
+static int progress = 0;
 
 static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
 {
@@ -34,7 +39,7 @@ static void *delta_against(void *buf, unsigned long size, struct object_entry *e
        if (!otherbuf)
                die("unable to read %s", sha1_to_hex(entry->delta->sha1));
         delta_buf = diff_delta(otherbuf, othersize,
-                              buf, size, &delta_size, ~0UL);
+                              buf, size, &delta_size, 0);
         if (!delta_buf || delta_size != entry->delta_size)
                die("delta size changed");
         free(buf);
@@ -116,7 +121,7 @@ static unsigned long write_one(struct sha1file *f,
                return offset;
        e->offset = offset;
        offset += write_object(f, e);
-       /* if we are delitified, write out its base object. */
+       /* if we are deltified, write out its base object. */
        if (e->delta)
                offset = write_one(f, e->delta, offset);
        return offset;
@@ -133,7 +138,7 @@ static void write_pack_file(void)
        if (!base_name)
                f = sha1fd(1, "<stdout>");
        else
-               f = sha1create("%s.%s", base_name, "pack");
+               f = sha1create("%s-%s.%s", base_name, sha1_to_hex(object_list_sha1), "pack");
        hdr.hdr_signature = htonl(PACK_SIGNATURE);
        hdr.hdr_version = htonl(PACK_VERSION);
        hdr.hdr_entries = htonl(nr_objects);
@@ -150,7 +155,7 @@ static void write_pack_file(void)
 static void write_index_file(void)
 {
        int i;
-       struct sha1file *f = sha1create("%s.%s", base_name, "idx");
+       struct sha1file *f = sha1create("%s-%s.%s", base_name, sha1_to_hex(object_list_sha1), "idx");
        struct object_entry **list = sorted_by_sha;
        struct object_entry **last = list + nr_objects;
        unsigned int array[256];
@@ -187,11 +192,26 @@ static void write_index_file(void)
        sha1close(f, NULL, 1);
 }
 
-static void add_object_entry(unsigned char *sha1, unsigned int hash)
+static int add_object_entry(unsigned char *sha1, unsigned int hash)
 {
        unsigned int idx = nr_objects;
        struct object_entry *entry;
 
+       if (incremental || local) {
+               struct packed_git *p;
+
+               for (p = packed_git; p; p = p->next) {
+                       struct pack_entry e;
+
+                       if (find_pack_entry_one(sha1, &e, p)) {
+                               if (incremental)
+                                       return 0;
+                               if (local && !p->pack_local)
+                                       return 0;
+                       }
+               }
+       }
+
        if (idx >= nr_alloc) {
                unsigned int needed = (idx + 1024) * 3 / 2;
                objects = xrealloc(objects, needed * sizeof(*entry));
@@ -202,6 +222,7 @@ static void add_object_entry(unsigned char *sha1, unsigned int hash)
        memcpy(entry->sha1, sha1, 20);
        entry->hash = hash;
        nr_objects = idx+1;
+       return 1;
 }
 
 static void check_object(struct object_entry *entry)
@@ -343,10 +364,13 @@ static void find_deltas(struct object_entry **list, int window, int depth)
        int i, idx;
        unsigned int array_size = window * sizeof(struct unpacked);
        struct unpacked *array = xmalloc(array_size);
+       int eye_candy;
 
        memset(array, 0, array_size);
        i = nr_objects;
        idx = 0;
+       eye_candy = i - (nr_objects / 20);
+
        while (--i >= 0) {
                struct object_entry *entry = list[i];
                struct unpacked *n = array + idx;
@@ -354,6 +378,10 @@ static void find_deltas(struct object_entry **list, int window, int depth)
                char type[10];
                int j;
 
+               if (progress && i <= eye_candy) {
+                       eye_candy -= nr_objects / 20;
+                       fputc('.', stderr);
+               }
                free(n->data);
                n->entry = entry;
                n->data = read_sha1_file(entry->sha1, type, &size);
@@ -375,18 +403,109 @@ static void find_deltas(struct object_entry **list, int window, int depth)
                if (idx >= window)
                        idx = 0;
        }
+
+       for (i = 0; i < window; ++i)
+               free(array[i].data);
+       free(array);
+}
+
+static void prepare_pack(int window, int depth)
+{
+       get_object_details();
+
+       if (progress)
+               fprintf(stderr, "Packing %d objects", nr_objects);
+       sorted_by_type = create_sorted_list(type_size_sort);
+       if (window && depth)
+               find_deltas(sorted_by_type, window+1, depth);
+       if (progress)
+               fputc('\n', stderr);
+       write_pack_file();
+}
+
+static int reuse_cached_pack(unsigned char *sha1, int pack_to_stdout)
+{
+       static const char cache[] = "pack-cache/pack-%s.%s";
+       char *cached_pack, *cached_idx;
+       int ifd, ofd, ifd_ix = -1;
+
+       cached_pack = git_path(cache, sha1_to_hex(sha1), "pack");
+       ifd = open(cached_pack, O_RDONLY);
+       if (ifd < 0)
+               return 0;
+
+       if (!pack_to_stdout) {
+               cached_idx = git_path(cache, sha1_to_hex(sha1), "idx");
+               ifd_ix = open(cached_idx, O_RDONLY);
+               if (ifd_ix < 0) {
+                       close(ifd);
+                       return 0;
+               }
+       }
+
+       fprintf(stderr, "Reusing %d objects pack %s\n", nr_objects,
+               sha1_to_hex(sha1));
+
+       if (pack_to_stdout) {
+               if (copy_fd(ifd, 1))
+                       exit(1);
+               close(ifd);
+       }
+       else {
+               char name[PATH_MAX];
+               snprintf(name, sizeof(name),
+                        "%s-%s.%s", base_name, sha1_to_hex(sha1), "pack");
+               ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
+               if (ofd < 0)
+                       die("unable to open %s (%s)", name, strerror(errno));
+               if (copy_fd(ifd, ofd))
+                       exit(1);
+               close(ifd);
+
+               snprintf(name, sizeof(name),
+                        "%s-%s.%s", base_name, sha1_to_hex(sha1), "idx");
+               ofd = open(name, O_CREAT | O_EXCL | O_WRONLY, 0666);
+               if (ofd < 0)
+                       die("unable to open %s (%s)", name, strerror(errno));
+               if (copy_fd(ifd_ix, ofd))
+                       exit(1);
+               close(ifd_ix);
+               puts(sha1_to_hex(sha1));
+       }
+
+       return 1;
 }
 
 int main(int argc, char **argv)
 {
+       SHA_CTX ctx;
        char line[PATH_MAX + 20];
        int window = 10, depth = 10, pack_to_stdout = 0;
+       struct object_entry **list;
        int i;
+       struct timeval prev_tv;
+       int eye_candy = 0;
+       int eye_candy_incr = 500;
+
+
+       setup_git_directory();
 
        for (i = 1; i < argc; i++) {
                const char *arg = argv[i];
 
                if (*arg == '-') {
+                       if (!strcmp("--non-empty", arg)) {
+                               non_empty = 1;
+                               continue;
+                       }
+                       if (!strcmp("--local", arg)) {
+                               local = 1;
+                               continue;
+                       }
+                       if (!strcmp("--incremental", arg)) {
+                               incremental = 1;
+                               continue;
+                       }
                        if (!strncmp("--window=", arg, 9)) {
                                char *end;
                                window = strtoul(arg+9, &end, 0);
@@ -415,13 +534,36 @@ int main(int argc, char **argv)
        if (pack_to_stdout != !base_name)
                usage(pack_usage);
 
+       progress = isatty(2);
+
+       prepare_packed_git();
+       if (progress) {
+               fprintf(stderr, "Generating pack...\n");
+               gettimeofday(&prev_tv, NULL);
+       }
        while (fgets(line, sizeof(line), stdin) != NULL) {
                unsigned int hash;
                char *p;
                unsigned char sha1[20];
 
+               if (progress && (eye_candy <= nr_objects)) {
+                       fprintf(stderr, "Counting objects...%d\r", nr_objects);
+                       if (eye_candy && (50 <= eye_candy_incr)) {
+                               struct timeval tv;
+                               int time_diff;
+                               gettimeofday(&tv, NULL);
+                               time_diff = (tv.tv_sec - prev_tv.tv_sec);
+                               time_diff <<= 10;
+                               time_diff += (tv.tv_usec - prev_tv.tv_usec);
+                               if ((1 << 9) < time_diff)
+                                       eye_candy_incr += 50;
+                               else if (50 < eye_candy_incr)
+                                       eye_candy_incr -= 50;
+                       }
+                       eye_candy += eye_candy_incr;
+               }
                if (get_sha1_hex(line, sha1))
-                       die("expected sha1, got garbage");
+                       die("expected sha1, got garbage:\n %s", line);
                hash = 0;
                p = line+40;
                while (*p) {
@@ -432,17 +574,28 @@ int main(int argc, char **argv)
                }
                add_object_entry(sha1, hash);
        }
-       get_object_details();
-
-       fprintf(stderr, "Packing %d objects\n", nr_objects);
+       if (progress)
+               fprintf(stderr, "Done counting %d objects.\n", nr_objects);
+       if (non_empty && !nr_objects)
+               return 0;
 
        sorted_by_sha = create_sorted_list(sha1_sort);
-       sorted_by_type = create_sorted_list(type_size_sort);
-       if (window && depth)
-               find_deltas(sorted_by_type, window+1, depth);
-
-       write_pack_file();
-       if (!pack_to_stdout)
-               write_index_file();
+       SHA1_Init(&ctx);
+       list = sorted_by_sha;
+       for (i = 0; i < nr_objects; i++) {
+               struct object_entry *entry = *list++;
+               SHA1_Update(&ctx, entry->sha1, 20);
+       }
+       SHA1_Final(object_list_sha1, &ctx);
+
+       if (reuse_cached_pack(object_list_sha1, pack_to_stdout))
+               ;
+       else {
+               prepare_pack(window, depth);
+               if (!pack_to_stdout) {
+                       write_index_file();
+                       puts(sha1_to_hex(object_list_sha1));
+               }
+       }
        return 0;
 }