Refactor read_data:

 * New read_data_block is both sparse-file aware and uses zero-copy semantics
 * Push read_data_block down into specific formats (opens door to
   various encoded entry bodies, such as zip or gtar -S)
 * Reimplement read_data, read_data_skip, read_data_into_fd in terms
   of new read_data_block.
 * Update documentation
It's unfortunate that I couldn't just call the new interface
archive_read_data, but didn't want to upset the API that much.

1 files changed, 8 insertions, 0 deletions

diff --git a/lib/libarchive/archive.h.in b/lib/libarchive/archive.h.in
index 4fc70cf..fed02c0 100644
--- a/lib/libarchive/archive.h.in
+++ b/lib/libarchive/archive.h.in
@@ -166,6 +166,14 @@ int64_t		 archive_read_header_position(struct archive *);
 
 /* Read data from the body of an entry.  Similar to read(2). */
 ssize_t		 archive_read_data(struct archive *, void *, size_t);
+/*
+ * A zero-copy version of archive_read_data that also exposes the file offset
+ * of each returned block.  Note that the client has no way to specify
+ * the desired size of the block.  The API does gaurantee that offsets will
+ * be strictly increasing and that returned blocks will not overlap.
+ */
+int		 archive_read_data_block(struct archive *a,
+		    const void **buff, size_t *size, off_t *offset);
 
 /*-
  * Some convenience functions that are built on archive_read_data: