Re: [PATCH 14/23] pack v4: object data copy

[Junio C Hamano <gitster@pobox.com>]

Nicolas Pitre <nico@fluxnic.net> writes:

> Blob and tag objects have no particular changes except for their object
> header.
>
> Delta objects are also copied as is, except for their delta base reference
> which is converted to the new way as used elsewhere in pack v4 encoding
> i.e. an index into the SHA1 table or a literal SHA1 prefixed by 0 if not
> found in the table (see add_sha1_ref).  This is true for both REF_DELTA
> as well as OFS_DELTA.
>
> Object payload is validated against the recorded CRC32 in the source
> pack index file when possible.
>
> Signed-off-by: Nicolas Pitre <nico@fluxnic.net>
> ---

The title somewhat confused me until I realized that this series is
building a program that would convert existing data from a single
pack into packv4 format, not a "pack-objects --pack-verison=4".

>  packv4-create.c | 66 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 66 insertions(+)
>
> diff --git a/packv4-create.c b/packv4-create.c
> index 6e0bb1d..a6dc818 100644
> --- a/packv4-create.c
> +++ b/packv4-create.c
> @@ -12,6 +12,7 @@
>  #include "object.h"
>  #include "tree-walk.h"
>  #include "pack.h"
> +#include "pack-revindex.h"
>  
>  
>  static int pack_compression_level = Z_DEFAULT_COMPRESSION;
> @@ -673,6 +674,71 @@ static unsigned int write_object_header(struct sha1file *f, enum object_type typ
>  	return end - buf;
>  }
>  
> +static unsigned long copy_object_data(struct sha1file *f, struct packed_git *p,
> +				      off_t offset)
> +{
> +	struct pack_window *w_curs = NULL;
> +	struct revindex_entry *revidx;
> +	enum object_type type;
> +	unsigned long avail, size, datalen, written;
> +	int hdrlen, idx_nr;
> +	unsigned char *src, *end, buf[24];
> +
> +	revidx = find_pack_revindex(p, offset);
> +	idx_nr = revidx->nr;
> +	datalen = revidx[1].offset - offset;
> +
> +	src = use_pack(p, &w_curs, offset, &avail);
> +	hdrlen = unpack_object_header_buffer(src, avail, &type, &size);
> +
> +	written = write_object_header(f, type, size);
> +
> +	if (type == OBJ_OFS_DELTA) {
> +		unsigned char c = src[hdrlen++];
> +		off_t base_offset = c & 127;
> +		while (c & 128) {
> +			base_offset += 1;
> +			if (!base_offset || MSB(base_offset, 7))
> +				die("delta offset overflow");
> +			c = src[hdrlen++];
> +			base_offset = (base_offset << 7) + (c & 127);
> +		}
> +		base_offset = offset - base_offset;
> +		if (base_offset <= 0 || base_offset >= offset)
> +			die("delta offset out of bound");
> +		revidx = find_pack_revindex(p, base_offset);
> +		end = add_sha1_ref(buf, nth_packed_object_sha1(p, revidx->nr));
> +		sha1write(f, buf, end - buf);
> +		written += end - buf;
> +	} else if (type == OBJ_REF_DELTA) {
> +		end = add_sha1_ref(buf, src + hdrlen);
> +		hdrlen += 20;
> +		sha1write(f, buf, end - buf);
> +		written += end - buf;
> +	}
> +
> +	if (p->index_version > 1 &&
> +	    check_pack_crc(p, &w_curs, offset, datalen, idx_nr))
> +		die("bad CRC for object at offset %"PRIuMAX" in %s",
> +		    (uintmax_t)offset, p->pack_name);
> +
> +	offset += hdrlen;
> +	datalen -= hdrlen;
> +
> +	while (datalen) {
> +		src = use_pack(p, &w_curs, offset, &avail);
> +		if (avail > datalen)
> +			avail = datalen;
> +		sha1write(f, src, avail);
> +		written += avail;
> +		offset += avail;
> +		datalen -= avail;
> +	}
> +	unuse_pack(&w_curs);
> +
> +	return written;
> +}
> +
>  static struct packed_git *open_pack(const char *path)
>  {
>  	char arg[PATH_MAX];