+#include <ctype.h>
#include "cache.h"
#include "object.h"
#include "delta.h"
+#include "csum-file.h"
-static const char pack_usage[] = "git-pack-objects [--window=N] base-name < object-list";
+static const char pack_usage[] = "git-pack-objects [--window=N] [--depth=N] base-name < object-list";
enum object_type {
OBJ_NONE,
OBJ_COMMIT,
OBJ_TREE,
OBJ_BLOB,
- OBJ_DELTA // NOTE! This is _not_ the same as a "delta" object in the filesystem
+ OBJ_DELTA
};
struct object_entry {
unsigned long size;
unsigned long offset;
unsigned int depth;
- unsigned int flags;
+ unsigned int hash;
enum object_type type;
unsigned long delta_size;
struct object_entry *delta;
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 int delta_window = 10;
static const char *base_name;
-
-struct myfile {
- int fd;
- unsigned long chars;
- unsigned char buffer[8192];
-};
-
-static FILE *create_file(const char *suffix)
-{
- static char filename[PATH_MAX];
- unsigned len;
-
- len = snprintf(filename, PATH_MAX, "%s.%s", base_name, suffix);
- if (len >= PATH_MAX)
- die("you wascally wabbit, you");
- return fopen(filename, "w");
-}
-
-static unsigned long fwrite_compressed(void *in, unsigned long size, FILE *f)
-{
- z_stream stream;
- unsigned long maxsize;
- void *out;
-
- memset(&stream, 0, sizeof(stream));
- deflateInit(&stream, Z_DEFAULT_COMPRESSION);
- maxsize = deflateBound(&stream, size);
- out = xmalloc(maxsize);
-
- /* Compress it */
- stream.next_in = in;
- stream.avail_in = size;
-
- stream.next_out = out;
- stream.avail_out = maxsize;
-
- while (deflate(&stream, Z_FINISH) == Z_OK)
- /* nothing */;
- deflateEnd(&stream);
-
- size = stream.total_out;
- fwrite(out, size, 1, f);
- free(out);
- return size;
-}
+static unsigned char pack_file_sha1[20];
static void *delta_against(void *buf, unsigned long size, struct object_entry *entry)
{
if (!otherbuf)
die("unable to read %s", sha1_to_hex(entry->delta->sha1));
- delta_buf = diff_delta(buf, size, otherbuf, othersize, &delta_size);
- if (delta_size != entry->delta_size)
+ delta_buf = diff_delta(otherbuf, othersize,
+ buf, size, &delta_size, ~0UL);
+ if (!delta_buf || delta_size != entry->delta_size)
die("delta size changed");
free(buf);
free(otherbuf);
return delta_buf;
}
-static unsigned long write_object(FILE *f, struct object_entry *entry)
+static unsigned long write_object(struct sha1file *f, struct object_entry *entry)
{
unsigned long size;
char type[10];
void *buf = read_sha1_file(entry->sha1, type, &size);
- char header[21];
+ char header[25];
unsigned hdrlen, datalen;
if (!buf)
* instead.
*/
header[0] = ".CTB"[entry->type];
- datalen = htonl(size);
- memcpy(header+1, &datalen, 4);
hdrlen = 5;
if (entry->delta) {
- memcpy(header+1, entry->delta, 20);
+ header[0] = 'D';
+ memcpy(header+5, entry->delta, 20);
buf = delta_against(buf, size, entry);
size = entry->delta_size;
- hdrlen = 21;
+ hdrlen = 25;
}
- fwrite(header, hdrlen, 1, f);
- datalen = fwrite_compressed(buf, size, f);
+ datalen = htonl(size);
+ memcpy(header+1, &datalen, 4);
+ sha1write(f, header, hdrlen);
+ datalen = sha1write_compressed(f, buf, size);
free(buf);
return hdrlen + datalen;
}
static void write_pack_file(void)
{
int i;
- FILE *f = create_file("pack");
+ struct sha1file *f = sha1create("%s.%s", base_name, "pack");
unsigned long offset = 0;
unsigned long mb;
entry->offset = offset;
offset += write_object(f, entry);
}
- fclose(f);
+ sha1close(f, pack_file_sha1, 1);
mb = offset >> 20;
offset &= 0xfffff;
}
static void write_index_file(void)
{
int i;
- FILE *f = create_file("idx");
+ struct sha1file *f = sha1create("%s.%s", base_name, "idx");
struct object_entry **list = sorted_by_sha;
struct object_entry **last = list + nr_objects;
unsigned int array[256];
/*
* Write the first-level table (the list is sorted,
* but we use a 256-entry lookup to be able to avoid
- * having to do eight extra binary search iterations)
+ * having to do eight extra binary search iterations).
*/
for (i = 0; i < 256; i++) {
struct object_entry **next = list;
array[i] = htonl(next - sorted_by_sha);
list = next;
}
- fwrite(array, 256, sizeof(int), f);
+ sha1write(f, array, 256 * sizeof(int));
/*
* Write the actual SHA1 entries..
for (i = 0; i < nr_objects; i++) {
struct object_entry *entry = *list++;
unsigned int offset = htonl(entry->offset);
- fwrite(&offset, 4, 1, f);
- fwrite(entry->sha1, 20, 1, f);
+ sha1write(f, &offset, 4);
+ sha1write(f, entry->sha1, 20);
}
- fclose(f);
+ sha1write(f, pack_file_sha1, 20);
+ sha1close(f, NULL, 1);
}
-static void add_object_entry(unsigned char *sha1)
+static void add_object_entry(unsigned char *sha1, unsigned int hash)
{
unsigned int idx = nr_objects;
struct object_entry *entry;
entry = objects + idx;
memset(entry, 0, sizeof(*entry));
memcpy(entry->sha1, sha1, 20);
+ entry->hash = hash;
nr_objects = idx+1;
}
return -1;
if (a->type > b->type)
return 1;
+ if (a->hash < b->hash)
+ return -1;
+ if (a->hash > b->hash)
+ return 1;
if (a->size < b->size)
return -1;
if (a->size > b->size)
};
/*
- * We search for deltas in a list sorted by type and by size, and
- * walk the "old" chain backwards in the list, so if the type
- * has changed or the size difference is too big, there's no point
- * in even continuing the walk, since the other old objects are
- * going to be even smaller or of a different type. So return -1
- * once we determine that there's no point even trying.
+ * We search for deltas _backwards_ in a list sorted by type and
+ * by size, so that we see progressively smaller and smaller files.
+ * That's because we prefer deltas to be from the bigger file
+ * to the smaller - deletes are potentially cheaper, but perhaps
+ * more importantly, the bigger file is likely the more recent
+ * one.
*/
-static int try_delta(struct unpacked *cur, struct unpacked *old)
+static int try_delta(struct unpacked *cur, struct unpacked *old, unsigned max_depth)
{
struct object_entry *cur_entry = cur->entry;
struct object_entry *old_entry = old->entry;
- unsigned long size, oldsize, delta_size;
+ unsigned long size, oldsize, delta_size, sizediff;
+ long max_size;
void *delta_buf;
/* Don't bother doing diffs between different types */
if (cur_entry->type != old_entry->type)
return -1;
- /* Size is guaranteed to be larger than or equal to oldsize */
size = cur_entry->size;
+ if (size < 50)
+ return -1;
oldsize = old_entry->size;
- if (size - oldsize > oldsize / 4)
+ sizediff = oldsize > size ? oldsize - size : size - oldsize;
+ if (sizediff > size / 8)
return -1;
+ if (old_entry->depth >= max_depth)
+ return 0;
/*
* NOTE!
* more space-efficient (deletes don't have to say _what_ they
* delete).
*/
- delta_buf = diff_delta(cur->data, size, old->data, oldsize, &delta_size);
+ max_size = size / 2 - 20;
+ if (cur_entry->delta)
+ max_size = cur_entry->delta_size-1;
+ if (sizediff >= max_size)
+ return -1;
+ delta_buf = diff_delta(old->data, oldsize,
+ cur->data, size, &delta_size, max_size);
if (!delta_buf)
- die("unable to create delta");
- if (delta_size + 20 < size / 2) {
- if (!cur_entry->delta || cur_entry->delta_size > delta_size) {
- cur_entry->delta = old_entry;
- cur_entry->delta_size = delta_size;
- }
- }
+ return 0;
+ cur_entry->delta = old_entry;
+ cur_entry->delta_size = delta_size;
+ cur_entry->depth = old_entry->depth + 1;
free(delta_buf);
+ return 0;
}
-static void find_deltas(struct object_entry **list, int window)
+static void find_deltas(struct object_entry **list, int window, int depth)
{
- unsigned int i;
+ int i, idx;
unsigned int array_size = window * sizeof(struct unpacked);
struct unpacked *array = xmalloc(array_size);
memset(array, 0, array_size);
- for (i = 0; i < nr_objects; i++) {
- unsigned int idx = i % window;
+ i = nr_objects;
+ idx = 0;
+ while (--i >= 0) {
struct object_entry *entry = list[i];
struct unpacked *n = array + idx;
unsigned long size;
n->data = read_sha1_file(entry->sha1, type, &size);
if (size != entry->size)
die("object %s inconsistent object length (%lu vs %lu)", sha1_to_hex(entry->sha1), size, entry->size);
- for (j = 0; j < window; j++) {
- unsigned int other_idx = idx - 1 - j;
+ j = window;
+ while (--j > 0) {
+ unsigned int other_idx = idx + j;
struct unpacked *m;
- if (other_idx < 0)
- other_idx += window;
+ if (other_idx >= window)
+ other_idx -= window;
m = array + other_idx;
if (!m->entry)
break;
- if (try_delta(n, m) < 0)
+ if (try_delta(n, m, depth) < 0)
break;
}
+ idx++;
+ if (idx >= window)
+ idx = 0;
}
}
int main(int argc, char **argv)
{
- char line[128];
+ char line[PATH_MAX + 20];
+ int window = 10, depth = 10;
int i;
for (i = 1; i < argc; i++) {
if (*arg == '-') {
if (!strncmp("--window=", arg, 9)) {
char *end;
- delta_window = strtoul(arg+9, &end, 0);
+ window = strtoul(arg+9, &end, 0);
if (!arg[9] || *end)
usage(pack_usage);
continue;
}
+ if (!strncmp("--depth=", arg, 8)) {
+ char *end;
+ depth = strtoul(arg+8, &end, 0);
+ if (!arg[8] || *end)
+ usage(pack_usage);
+ continue;
+ }
usage(pack_usage);
}
if (base_name)
usage(pack_usage);
while (fgets(line, sizeof(line), stdin) != NULL) {
+ unsigned int hash;
+ char *p;
unsigned char sha1[20];
+
if (get_sha1_hex(line, sha1))
die("expected sha1, got garbage");
- add_object_entry(sha1);
+ hash = 0;
+ p = line+40;
+ while (*p) {
+ unsigned char c = *p++;
+ if (isspace(c))
+ continue;
+ hash = hash * 11 + c;
+ }
+ add_object_entry(sha1, hash);
}
get_object_details();
sorted_by_sha = create_sorted_list(sha1_sort);
sorted_by_type = create_sorted_list(type_size_sort);
- if (delta_window)
- find_deltas(sorted_by_type, delta_window);
+ if (window && depth)
+ find_deltas(sorted_by_type, window+1, depth);
write_pack_file();
write_index_file();
projects / git.git / blobdiff