gitweb: Filter out commit ID from @difftree in git_commit and git_commitdiff
[gitweb.git] / index-pack.c
index 785fe71a6fb5e518f92d62bace43cdd13f3c37ae..e33f60524f240f288df7dc6d398486d1d6e7e5ee 100644 (file)
@@ -2,6 +2,10 @@
 #include "delta.h"
 #include "pack.h"
 #include "csum-file.h"
+#include "blob.h"
+#include "commit.h"
+#include "tag.h"
+#include "tree.h"
 
 static const char index_pack_usage[] =
 "git-index-pack [-o index-file] pack-file";
@@ -9,77 +13,102 @@ static const char index_pack_usage[] =
 struct object_entry
 {
        unsigned long offset;
+       unsigned long size;
+       unsigned int hdr_size;
        enum object_type type;
        enum object_type real_type;
        unsigned char sha1[20];
 };
 
+union delta_base {
+       unsigned char sha1[20];
+       unsigned long offset;
+};
+
+/*
+ * Even if sizeof(union delta_base) == 24 on 64-bit archs, we really want
+ * to memcmp() only the first 20 bytes.
+ */
+#define UNION_BASE_SZ  20
+
 struct delta_entry
 {
        struct object_entry *obj;
-       unsigned char base_sha1[20];
+       union delta_base base;
 };
 
 static const char *pack_name;
-static unsigned char *pack_base;
-static unsigned long pack_size;
 static struct object_entry *objects;
 static struct delta_entry *deltas;
 static int nr_objects;
 static int nr_deltas;
 
-static void open_pack_file(void)
+/* We always read in 4kB chunks. */
+static unsigned char input_buffer[4096];
+static unsigned long input_offset, input_len, consumed_bytes;
+static SHA_CTX input_ctx;
+static int input_fd;
+
+/*
+ * Make sure at least "min" bytes are available in the buffer, and
+ * return the pointer to the buffer.
+ */
+static void * fill(int min)
 {
-       int fd;
-       struct stat st;
+       if (min <= input_len)
+               return input_buffer + input_offset;
+       if (min > sizeof(input_buffer))
+               die("cannot fill %d bytes", min);
+       if (input_offset) {
+               SHA1_Update(&input_ctx, input_buffer, input_offset);
+               memcpy(input_buffer, input_buffer + input_offset, input_len);
+               input_offset = 0;
+       }
+       do {
+               int ret = xread(input_fd, input_buffer + input_len,
+                               sizeof(input_buffer) - input_len);
+               if (ret <= 0) {
+                       if (!ret)
+                               die("early EOF");
+                       die("read error on input: %s", strerror(errno));
+               }
+               input_len += ret;
+       } while (input_len < min);
+       return input_buffer;
+}
+
+static void use(int bytes)
+{
+       if (bytes > input_len)
+               die("used more bytes than were available");
+       input_len -= bytes;
+       input_offset += bytes;
+       consumed_bytes += bytes;
+}
 
-       fd = open(pack_name, O_RDONLY);
-       if (fd < 0)
+static void open_pack_file(void)
+{
+       input_fd = open(pack_name, O_RDONLY);
+       if (input_fd < 0)
                die("cannot open packfile '%s': %s", pack_name,
                    strerror(errno));
-       if (fstat(fd, &st)) {
-               int err = errno;
-               close(fd);
-               die("cannot fstat packfile '%s': %s", pack_name,
-                   strerror(err));
-       }
-       pack_size = st.st_size;
-       pack_base = mmap(NULL, pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
-       if (pack_base == MAP_FAILED) {
-               int err = errno;
-               close(fd);
-               die("cannot mmap packfile '%s': %s", pack_name,
-                   strerror(err));
-       }
-       close(fd);
+       SHA1_Init(&input_ctx);
 }
 
 static void parse_pack_header(void)
 {
-       const struct pack_header *hdr;
-       unsigned char sha1[20];
-       SHA_CTX ctx;
-
-       /* Ensure there are enough bytes for the header and final SHA1 */
-       if (pack_size < sizeof(struct pack_header) + 20)
-               die("packfile '%s' is too small", pack_name);
+       struct pack_header *hdr = fill(sizeof(struct pack_header));
 
        /* Header consistency check */
-       hdr = (void *)pack_base;
        if (hdr->hdr_signature != htonl(PACK_SIGNATURE))
                die("packfile '%s' signature mismatch", pack_name);
-       if (hdr->hdr_version != htonl(PACK_VERSION))
-               die("packfile '%s' version %d different from ours %d",
-                   pack_name, ntohl(hdr->hdr_version), PACK_VERSION);
+       if (!pack_version_ok(hdr->hdr_version))
+               die("packfile '%s' version %d unsupported",
+                   pack_name, ntohl(hdr->hdr_version));
 
        nr_objects = ntohl(hdr->hdr_entries);
-
-       /* Check packfile integrity */
-       SHA1_Init(&ctx);
-       SHA1_Update(&ctx, pack_base, pack_size - 20);
-       SHA1_Final(sha1, &ctx);
-       if (memcmp(sha1, pack_base + pack_size - 20, 20))
-               die("packfile '%s' SHA1 mismatch", pack_name);
+       use(sizeof(struct pack_header));
+       /*fprintf(stderr, "Indexing %d objects\n", nr_objects);*/
 }
 
 static void bad_object(unsigned long offset, const char *format,
@@ -97,86 +126,121 @@ static void bad_object(unsigned long offset, const char *format, ...)
            pack_name, offset, buf);
 }
 
-static void *unpack_entry_data(unsigned long offset,
-                              unsigned long *current_pos, unsigned long size)
+static void *unpack_entry_data(unsigned long offset, unsigned long size)
 {
-       unsigned long pack_limit = pack_size - 20;
-       unsigned long pos = *current_pos;
        z_stream stream;
        void *buf = xmalloc(size);
 
        memset(&stream, 0, sizeof(stream));
        stream.next_out = buf;
        stream.avail_out = size;
-       stream.next_in = pack_base + pos;
-       stream.avail_in = pack_limit - pos;
+       stream.next_in = fill(1);
+       stream.avail_in = input_len;
        inflateInit(&stream);
 
        for (;;) {
                int ret = inflate(&stream, 0);
-               if (ret == Z_STREAM_END)
+               use(input_len - stream.avail_in);
+               if (stream.total_out == size && ret == Z_STREAM_END)
                        break;
                if (ret != Z_OK)
                        bad_object(offset, "inflate returned %d", ret);
+               stream.next_in = fill(1);
+               stream.avail_in = input_len;
        }
        inflateEnd(&stream);
-       if (stream.total_out != size)
-               bad_object(offset, "size mismatch (expected %lu, got %lu)",
-                          size, stream.total_out);
-       *current_pos = pack_limit - stream.avail_in;
        return buf;
 }
 
-static void *unpack_raw_entry(unsigned long offset,
-                             enum object_type *obj_type,
-                             unsigned long *obj_size,
-                             unsigned char *delta_base,
-                             unsigned long *next_obj_offset)
+static void *unpack_raw_entry(struct object_entry *obj, union delta_base *delta_base)
 {
-       unsigned long pack_limit = pack_size - 20;
-       unsigned long pos = offset;
-       unsigned char c;
-       unsigned long size;
+       unsigned char *p, c;
+       unsigned long size, base_offset;
        unsigned shift;
-       enum object_type type;
-       void *data;
 
-       c = pack_base[pos++];
-       type = (c >> 4) & 7;
+       obj->offset = consumed_bytes;
+
+       p = fill(1);
+       c = *p;
+       use(1);
+       obj->type = (c >> 4) & 7;
        size = (c & 15);
        shift = 4;
        while (c & 0x80) {
-               if (pos >= pack_limit)
-                       bad_object(offset, "object extends past end of pack");
-               c = pack_base[pos++];
+               p = fill(1);
+               c = *p;
+               use(1);
                size += (c & 0x7fUL) << shift;
                shift += 7;
        }
+       obj->size = size;
 
-       switch (type) {
-       case OBJ_DELTA:
-               if (pos + 20 >= pack_limit)
-                       bad_object(offset, "object extends past end of pack");
-               memcpy(delta_base, pack_base + pos, 20);
-               pos += 20;
-               /* fallthru */
+       switch (obj->type) {
+       case OBJ_REF_DELTA:
+               hashcpy(delta_base->sha1, fill(20));
+               use(20);
+               break;
+       case OBJ_OFS_DELTA:
+               memset(delta_base, 0, sizeof(*delta_base));
+               p = fill(1);
+               c = *p;
+               use(1);
+               base_offset = c & 127;
+               while (c & 128) {
+                       base_offset += 1;
+                       if (!base_offset || base_offset & ~(~0UL >> 7))
+                               bad_object(obj->offset, "offset value overflow for delta base object");
+                       p = fill(1);
+                       c = *p;
+                       use(1);
+                       base_offset = (base_offset << 7) + (c & 127);
+               }
+               delta_base->offset = obj->offset - base_offset;
+               if (delta_base->offset >= obj->offset)
+                       bad_object(obj->offset, "delta base offset is out of bound");
+               break;
        case OBJ_COMMIT:
        case OBJ_TREE:
        case OBJ_BLOB:
        case OBJ_TAG:
-               data = unpack_entry_data(offset, &pos, size);
                break;
        default:
-               bad_object(offset, "bad object type %d", type);
+               bad_object(obj->offset, "bad object type %d", obj->type);
        }
+       obj->hdr_size = consumed_bytes - obj->offset;
 
-       *obj_type = type;
-       *obj_size = size;
-       *next_obj_offset = pos;
+       return unpack_entry_data(obj->offset, obj->size);
+}
+
+static void * get_data_from_pack(struct object_entry *obj)
+{
+       unsigned long from = obj[0].offset + obj[0].hdr_size;
+       unsigned long len = obj[1].offset - from;
+       unsigned pg_offset = from % getpagesize();
+       unsigned char *map, *data;
+       z_stream stream;
+       int st;
+
+       map = mmap(NULL, len + pg_offset, PROT_READ, MAP_PRIVATE,
+                  input_fd, from - pg_offset);
+       if (map == MAP_FAILED)
+               die("cannot mmap packfile '%s': %s", pack_name, strerror(errno));
+       data = xmalloc(obj->size);
+       memset(&stream, 0, sizeof(stream));
+       stream.next_out = data;
+       stream.avail_out = obj->size;
+       stream.next_in = map + pg_offset;
+       stream.avail_in = len;
+       inflateInit(&stream);
+       while ((st = inflate(&stream, Z_FINISH)) == Z_OK);
+       inflateEnd(&stream);
+       if (st != Z_STREAM_END || stream.total_out != obj->size)
+               die("serious inflate inconsistency");
+       munmap(map, len + pg_offset);
        return data;
 }
 
-static int find_delta(const unsigned char *base_sha1)
+static int find_delta(const union delta_base *base)
 {
        int first = 0, last = nr_deltas;
 
@@ -185,7 +249,7 @@ static int find_delta(const unsigned char *base_sha1)
                 struct delta_entry *delta = &deltas[next];
                 int cmp;
 
-                cmp = memcmp(base_sha1, delta->base_sha1, 20);
+                cmp = memcmp(base, &delta->base, UNION_BASE_SZ);
                 if (!cmp)
                         return next;
                 if (cmp < 0) {
@@ -197,18 +261,18 @@ static int find_delta(const unsigned char *base_sha1)
         return -first-1;
 }
 
-static int find_deltas_based_on_sha1(const unsigned char *base_sha1,
-                                    int *first_index, int *last_index)
+static int find_delta_childs(const union delta_base *base,
+                            int *first_index, int *last_index)
 {
-       int first = find_delta(base_sha1);
+       int first = find_delta(base);
        int last = first;
        int end = nr_deltas - 1;
 
        if (first < 0)
                return -1;
-       while (first > 0 && !memcmp(deltas[first-1].base_sha1, base_sha1, 20))
+       while (first > 0 && !memcmp(&deltas[first - 1].base, base, UNION_BASE_SZ))
                --first;
-       while (last < end && !memcmp(deltas[last+1].base_sha1, base_sha1, 20))
+       while (last < end && !memcmp(&deltas[last + 1].base, base, UNION_BASE_SZ))
                ++last;
        *first_index = first;
        *last_index = last;
@@ -224,10 +288,10 @@ static void sha1_object(const void *data, unsigned long size,
        const char *type_str;
 
        switch (type) {
-       case OBJ_COMMIT: type_str = "commit"; break;
-       case OBJ_TREE:   type_str = "tree"; break;
-       case OBJ_BLOB:   type_str = "blob"; break;
-       case OBJ_TAG:    type_str = "tag"; break;
+       case OBJ_COMMIT: type_str = commit_type; break;
+       case OBJ_TREE:   type_str = tree_type; break;
+       case OBJ_BLOB:   type_str = blob_type; break;
+       case OBJ_TAG:    type_str = tag_type; break;
        default:
                die("bad type %d", type);
        }
@@ -248,25 +312,34 @@ static void resolve_delta(struct delta_entry *delta, void *base_data,
        unsigned long delta_size;
        void *result;
        unsigned long result_size;
-       enum object_type delta_type;
-       unsigned char base_sha1[20];
-       unsigned long next_obj_offset;
+       union delta_base delta_base;
        int j, first, last;
 
        obj->real_type = type;
-       delta_data = unpack_raw_entry(obj->offset, &delta_type,
-                                     &delta_size, base_sha1,
-                                     &next_obj_offset);
+       delta_data = get_data_from_pack(obj);
+       delta_size = obj->size;
        result = patch_delta(base_data, base_size, delta_data, delta_size,
                             &result_size);
        free(delta_data);
        if (!result)
                bad_object(obj->offset, "failed to apply delta");
        sha1_object(result, result_size, type, obj->sha1);
-       if (!find_deltas_based_on_sha1(obj->sha1, &first, &last)) {
+
+       hashcpy(delta_base.sha1, obj->sha1);
+       if (!find_delta_childs(&delta_base, &first, &last)) {
                for (j = first; j <= last; j++)
-                       resolve_delta(&deltas[j], result, result_size, type);
+                       if (deltas[j].obj->type == OBJ_REF_DELTA)
+                               resolve_delta(&deltas[j], result, result_size, type);
        }
+
+       memset(&delta_base, 0, sizeof(delta_base));
+       delta_base.offset = obj->offset;
+       if (!find_delta_childs(&delta_base, &first, &last)) {
+               for (j = first; j <= last; j++)
+                       if (deltas[j].obj->type == OBJ_OFS_DELTA)
+                               resolve_delta(&deltas[j], result, result_size, type);
+       }
+
        free(result);
 }
 
@@ -274,16 +347,16 @@ static int compare_delta_entry(const void *a, const void *b)
 {
        const struct delta_entry *delta_a = a;
        const struct delta_entry *delta_b = b;
-       return memcmp(delta_a->base_sha1, delta_b->base_sha1, 20);
+       return memcmp(&delta_a->base, &delta_b->base, UNION_BASE_SZ);
 }
 
-static void parse_pack_objects(void)
+/* Parse all objects and return the pack content SHA1 hash */
+static void parse_pack_objects(unsigned char *sha1)
 {
        int i;
-       unsigned long offset = sizeof(struct pack_header);
-       unsigned char base_sha1[20];
+       struct delta_entry *delta = deltas;
        void *data;
-       unsigned long data_size;
+       struct stat st;
 
        /*
         * First pass:
@@ -293,22 +366,32 @@ static void parse_pack_objects(void)
         */
        for (i = 0; i < nr_objects; i++) {
                struct object_entry *obj = &objects[i];
-               obj->offset = offset;
-               data = unpack_raw_entry(offset, &obj->type, &data_size,
-                                       base_sha1, &offset);
+               data = unpack_raw_entry(obj, &delta->base);
                obj->real_type = obj->type;
-               if (obj->type == OBJ_DELTA) {
-                       struct delta_entry *delta = &deltas[nr_deltas++];
+               if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA) {
+                       nr_deltas++;
                        delta->obj = obj;
-                       memcpy(delta->base_sha1, base_sha1, 20);
+                       delta++;
                } else
-                       sha1_object(data, data_size, obj->type, obj->sha1);
+                       sha1_object(data, obj->size, obj->type, obj->sha1);
                free(data);
        }
-       if (offset != pack_size - 20)
+       objects[i].offset = consumed_bytes;
+
+       /* Check pack integrity */
+       SHA1_Update(&input_ctx, input_buffer, input_offset);
+       SHA1_Final(sha1, &input_ctx);
+       if (hashcmp(fill(20), sha1))
+               die("packfile '%s' SHA1 mismatch", pack_name);
+       use(20);
+
+       /* If input_fd is a file, we should have reached its end now. */
+       if (fstat(input_fd, &st))
+               die("cannot fstat packfile '%s': %s", pack_name, strerror(errno));
+       if (S_ISREG(st.st_mode) && st.st_size != consumed_bytes)
                die("packfile '%s' has junk at the end", pack_name);
 
-       /* Sort deltas by base SHA1 for fast searching */
+       /* Sort deltas by base SHA1/offset for fast searching */
        qsort(deltas, nr_deltas, sizeof(struct delta_entry),
              compare_delta_entry);
 
@@ -322,22 +405,36 @@ static void parse_pack_objects(void)
         */
        for (i = 0; i < nr_objects; i++) {
                struct object_entry *obj = &objects[i];
-               int j, first, last;
+               union delta_base base;
+               int j, ref, ref_first, ref_last, ofs, ofs_first, ofs_last;
 
-               if (obj->type == OBJ_DELTA)
+               if (obj->type == OBJ_REF_DELTA || obj->type == OBJ_OFS_DELTA)
                        continue;
-               if (find_deltas_based_on_sha1(obj->sha1, &first, &last))
+               hashcpy(base.sha1, obj->sha1);
+               ref = !find_delta_childs(&base, &ref_first, &ref_last);
+               memset(&base, 0, sizeof(base));
+               base.offset = obj->offset;
+               ofs = !find_delta_childs(&base, &ofs_first, &ofs_last);
+               if (!ref && !ofs)
                        continue;
-               data = unpack_raw_entry(obj->offset, &obj->type, &data_size,
-                                       base_sha1, &offset);
-               for (j = first; j <= last; j++)
-                       resolve_delta(&deltas[j], data, data_size, obj->type);
+               data = get_data_from_pack(obj);
+               if (ref)
+                       for (j = ref_first; j <= ref_last; j++)
+                               if (deltas[j].obj->type == OBJ_REF_DELTA)
+                                       resolve_delta(&deltas[j], data,
+                                                     obj->size, obj->type);
+               if (ofs)
+                       for (j = ofs_first; j <= ofs_last; j++)
+                               if (deltas[j].obj->type == OBJ_OFS_DELTA)
+                                       resolve_delta(&deltas[j], data,
+                                                     obj->size, obj->type);
                free(data);
        }
 
        /* Check for unresolved deltas */
        for (i = 0; i < nr_deltas; i++) {
-               if (deltas[i].obj->real_type == OBJ_DELTA)
+               if (deltas[i].obj->real_type == OBJ_REF_DELTA ||
+                   deltas[i].obj->real_type == OBJ_OFS_DELTA)
                        die("packfile '%s' has unresolved deltas",  pack_name);
        }
 }
@@ -346,24 +443,34 @@ static int sha1_compare(const void *_a, const void *_b)
 {
        struct object_entry *a = *(struct object_entry **)_a;
        struct object_entry *b = *(struct object_entry **)_b;
-       return memcmp(a->sha1, b->sha1, 20);
+       return hashcmp(a->sha1, b->sha1);
 }
 
+/*
+ * On entry *sha1 contains the pack content SHA1 hash, on exit it is
+ * the SHA1 hash of sorted object names.
+ */
 static void write_index_file(const char *index_name, unsigned char *sha1)
 {
        struct sha1file *f;
-       struct object_entry **sorted_by_sha =
-               xcalloc(nr_objects, sizeof(struct object_entry *));
-       struct object_entry **list = sorted_by_sha;
-       struct object_entry **last = sorted_by_sha + nr_objects;
+       struct object_entry **sorted_by_sha, **list, **last;
        unsigned int array[256];
        int i;
        SHA_CTX ctx;
 
-       for (i = 0; i < nr_objects; ++i)
-               sorted_by_sha[i] = &objects[i];
-       qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
-             sha1_compare);
+       if (nr_objects) {
+               sorted_by_sha =
+                       xcalloc(nr_objects, sizeof(struct object_entry *));
+               list = sorted_by_sha;
+               last = sorted_by_sha + nr_objects;
+               for (i = 0; i < nr_objects; ++i)
+                       sorted_by_sha[i] = &objects[i];
+               qsort(sorted_by_sha, nr_objects, sizeof(sorted_by_sha[0]),
+                     sha1_compare);
+
+       }
+       else
+               sorted_by_sha = list = last = NULL;
 
        unlink(index_name);
        f = sha1create("%s", index_name);
@@ -402,7 +509,7 @@ static void write_index_file(const char *index_name, unsigned char *sha1)
                sha1write(f, obj->sha1, 20);
                SHA1_Update(&ctx, obj->sha1, 20);
        }
-       sha1write(f, pack_base + pack_size - 20, 20);
+       sha1write(f, sha1, 20);
        sha1close(f, NULL, 1);
        free(sorted_by_sha);
        SHA1_Final(sha1, &ctx);
@@ -437,10 +544,10 @@ int main(int argc, char **argv)
                usage(index_pack_usage);
        if (!index_name) {
                int len = strlen(pack_name);
-               if (len < 5 || strcmp(pack_name + len - 5, ".pack"))
+               if (!has_extension(pack_name, ".pack"))
                        die("packfile name '%s' does not end with '.pack'",
                            pack_name);
-               index_name_buf = xmalloc(len - 1);
+               index_name_buf = xmalloc(len);
                memcpy(index_name_buf, pack_name, len - 5);
                strcpy(index_name_buf + len - 5, ".idx");
                index_name = index_name_buf;
@@ -448,9 +555,9 @@ int main(int argc, char **argv)
 
        open_pack_file();
        parse_pack_header();
-       objects = xcalloc(nr_objects, sizeof(struct object_entry));
+       objects = xcalloc(nr_objects + 1, sizeof(struct object_entry));
        deltas = xcalloc(nr_objects, sizeof(struct delta_entry));
-       parse_pack_objects();
+       parse_pack_objects(sha1);
        free(deltas);
        write_index_file(index_name, sha1);
        free(objects);