* (c) 2004, Martin Mares <mj@ucw.cz>
*/
+#undef LOCAL_DEBUG
+
#include "lib/lib.h"
#include "lib/unaligned.h"
#include "lib/mempool.h"
#include "lib/bucket.h"
#include "lib/lizard.h"
#include "lib/bbuf.h"
+#include "lib/ff-utf8.h"
#include <stdlib.h>
#include <errno.h>
struct lizard_buffer *lizard;
};
+static uns get_attr_type;
+
+void
+get_attr_set_type(uns type)
+{
+ if (type < BUCKET_TYPE_PLAIN || type > BUCKET_TYPE_V33_LIZARD)
+ die("Unknown buckettype %x", type);
+ get_attr_type = type;
+}
+
+int
+get_attr(byte **pos, byte *end, struct parsed_attr *attr)
+{
+ byte *ptr = *pos;
+ if (ptr >= end)
+ return -1;
+ if (get_attr_type < BUCKET_TYPE_V33)
+ {
+ if (get_attr_type == BUCKET_TYPE_PLAIN)
+ {
+ while (ptr < end && *ptr == '\n')
+ ptr++;
+ *pos = ptr;
+ if (ptr >= end)
+ return -1;
+ }
+ else if (*ptr == '\n')
+ {
+ *pos = ++ptr;
+ attr->attr = 0;
+ return 0;
+ }
+ attr->attr = *ptr++;
+ attr->val = ptr;
+ while (ptr < end && *ptr != '\n')
+ ptr++;
+ attr->len = ptr++ - attr->val;
+ }
+ else
+ {
+ uns len;
+ GET_UTF8_32(ptr, len);
+ if (!len--)
+ {
+ *pos = ptr;
+ attr->attr = 0;
+ return 0;
+ }
+ attr->attr = ptr[len];
+ attr->val = ptr;
+ attr->len = len;
+ ptr += len+1;
+ }
+ if (ptr > end)
+ die("Incomplete attribute %c", attr->attr);
+ *pos = ptr;
+ return attr->attr;
+}
+
+int
+bget_attr(struct fastbuf *b, struct parsed_attr *attr)
+{
+ static bb_t buf;
+ if (get_attr_type < BUCKET_TYPE_V33)
+ {
+ int c = bgetc(b);
+ if (c < 0)
+ return -1;
+ if (get_attr_type == BUCKET_TYPE_PLAIN)
+ {
+ while (c == '\n')
+ c = bgetc(b);
+ if (c < 0)
+ return -1;
+ }
+ else if (c == '\n')
+ {
+ attr->attr = 0;
+ return 0;
+ }
+ attr->attr = c;
+
+ byte *ptr, *end;
+ uns len = bdirect_read_prepare(b, &ptr);
+ end = ptr + len;
+ attr->val = ptr;
+ while (ptr < end && *ptr != '\n')
+ ptr++;
+ if (ptr < end)
+ {
+ bdirect_read_commit(b, ptr+1);
+ attr->len = ptr - attr->val;
+ return attr->attr;
+ }
+
+ len = 0;
+ c = bgetc(b);
+ while (c >= 0 && c != '\n')
+ {
+ bb_grow(&buf, len+1);
+ buf.ptr[len++] = c;
+ c = bgetc(b);
+ }
+ if (c < 0)
+ die("Incomplete attribute %c", attr->attr);
+ attr->val = buf.ptr;
+ attr->len = len;
+ }
+ else
+ {
+ int len = bget_utf8_32(b);
+ if (len < 0)
+ return -1;
+ if (!len)
+ {
+ attr->attr = 0;
+ return 0;
+ }
+ attr->len = len-1;
+
+ byte *ptr;
+ int avail = bdirect_read_prepare(b, &ptr);
+ if (avail >= len)
+ {
+ attr->val = ptr;
+ attr->attr = ptr[len-1];
+ bdirect_read_commit(b, ptr + len);
+ return attr->attr;
+ }
+ bb_grow(&buf, --len);
+ breadb(b, buf.ptr, len);
+ attr->val = buf.ptr;
+ attr->len = len;
+ attr->attr = bgetc(b);
+ if (attr->attr < 0)
+ die("Incomplete attribute %c", attr->attr);
+ }
+ return attr->attr;
+}
+
struct buck2obj_buf *
buck2obj_alloc(void)
{
while (ptr < end)
{
uns len;
- GET_UTF8(ptr, len);
+ GET_UTF8_32(ptr, len);
if (!len--)
break;
byte type = ptr[len];
while (ptr < end)
{
uns len;
- GET_UTF8(ptr, len);
+ GET_UTF8_32(ptr, len);
if (!len--)
break;
byte type = ptr[len];
{
while (btell(body) < end && bgets(body, buf, sizeof(buf)))
if (buf[0])
- obj_add_attr(o_hdr, buf[0], buf+1);
+ obj_add_attr(o_body, buf[0], buf+1);
ASSERT(btell(body) == end);
}
}
else if (buck_type == BUCKET_TYPE_V33 || buck_type == BUCKET_TYPE_V33_LIZARD)
{
+ /* Avoid reading the whole bucket if only its header is needed. */
+ if (body_start)
+ {
+ sh_off_t start = btell(body);
+ sh_off_t end = start + buck_len;
+ while (btell(body) < end)
+ {
+ uns len = bget_utf8_32(body);
+ if (!len)
+ break;
+ byte *buf = mp_alloc_fast_noalign(o_hdr->pool, len);
+ bread(body, buf, len);
+ uns type = buf[--len];
+ buf[len] = 0;
+ obj_add_attr_ref(o_hdr, type, buf);
+ }
+ *body_start = btell(body) - start;
+ return 0;
+ }
+
/* Read all the bucket into 1 buffer, 0-copy if possible. */
byte *ptr, *end;
uns len = bdirect_read_prepare(body, &ptr);
{
/* Copy if the original buffer is too small.
* If it is write-protected, copy it also if it is uncompressed. */
+ DBG("NO ZC: %d < %d, %d %08x", len, buck_len, body->can_overwrite_buffer, buck_type);
bb_grow(&buf->bb, buck_len);
len = bread(body, buf->bb.ptr, buck_len);
ptr = buf->bb.ptr;
copied = 1;
}
- end = ptr + len;
+ else
+ DBG("ZC (%d >= %d, %d %08x)", len, buck_len, body->can_overwrite_buffer, buck_type);
+ end = ptr + buck_len;
- byte *start = ptr;
ptr = decode_attributes(ptr, end, o_hdr, 0); // header
- if (body_start)
- {
- *body_start = ptr - start;
- if (!copied)
- bdirect_read_commit(body, ptr);
- return 0;
- }
if (buck_type == BUCKET_TYPE_V33_LIZARD) // decompression
{
if (ptr + 8 > end)
- RET_ERR(EINVAL);
+ {
+ if (ptr == end) // truncated bucket
+ goto commit;
+ RET_ERR(EINVAL);
+ }
len = GET_U32(ptr);
ptr += 4;
uns adler = GET_U32(ptr);
ptr = decode_attributes(ptr, end, o_body, 2); // body
if (ptr != end)
RET_ERR(EINVAL);
+ commit:
if (!copied)
bdirect_read_commit_modified(body, ptr);
}
else
- RET_ERR(EINVAL);
+ {
+ bskip(body, buck_len);
+ RET_ERR(EINVAL);
+ }
return 0;
}
projects / libucw.git / blobdiff