X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=lib%2Fbuck2obj.c;h=d75b48ab19e212b2d65e8996981db3b6dda85325;hb=c673da72822d7704572716de80402829921efdc3;hp=b19963f7803284acdbaed5c5f53ef9d8d7db9bfa;hpb=271d2a02e7b798fd9781018c060d49e47f277cba;p=libucw.git

diff --git a/lib/buck2obj.c b/lib/buck2obj.c
index b19963f7..d75b48ab 100644
--- a/lib/buck2obj.c
+++ b/lib/buck2obj.c
@@ -1,5 +1,5 @@
 /*
- *	Bucket -> Object Converter
+ *	Generating Objects from Buckets
  *
  *	(c) 2004, Robert Spalek <robert@ucw.cz>
  *	(c) 2004, Martin Mares <mj@ucw.cz>
@@ -7,13 +7,14 @@
 
 #include "lib/lib.h"
 #include "lib/unaligned.h"
-#include "lib/pools.h"
+#include "lib/mempool.h"
 #include "lib/fastbuf.h"
-#include "charset/unicode.h"
+#include "lib/unicode.h"
 #include "lib/object.h"
 #include "lib/bucket.h"
 #include "lib/lizard.h"
-#include "lib/buck2obj.h"
+#include "lib/bbuf.h"
+#include "lib/ff-utf8.h"
 
 #include <stdlib.h>
 #include <errno.h>
@@ -21,10 +22,6 @@
 
 #define	RET_ERR(num)	({ errno = num; return -1; })
 
-#define	GBUF_TYPE	byte
-#define	GBUF_PREFIX(x)	bb_##x
-#include "lib/gbuf.h"
-
 struct buck2obj_buf
 {
   bb_t bb;
@@ -87,26 +84,57 @@ decode_attributes(byte *ptr, byte *end, struct odes *o, uns can_overwrite)
 int
 buck2obj_parse(struct buck2obj_buf *buf, uns buck_type, uns buck_len, struct fastbuf *body, struct odes *o_hdr, uns *body_start, struct odes *o_body)
 {
-  if (buck_type == BUCKET_TYPE_PLAIN)
+  if (buck_type <= BUCKET_TYPE_PLAIN)
   {
-    if (body_start)
+    if (body_start)			// there is no header part
       *body_start = 0;
-    obj_read_multi(body, o_hdr);	// ignore empty lines, read until EOF or NUL
+    // ignore empty lines and read until the end of the bucket
+    sh_off_t end = btell(body) + buck_len;
+    byte buf[MAX_ATTR_SIZE];
+    while (btell(body) < end && bgets(body, buf, sizeof(buf)))
+      if (buf[0])
+	obj_add_attr(o_hdr, buf[0], buf+1);
+    ASSERT(btell(body) == end);
   }
   else if (buck_type == BUCKET_TYPE_V30)
   {
     sh_off_t start = btell(body);
-    obj_read(body, o_hdr);		// end on EOF or the first empty line
+    sh_off_t end = start + buck_len;
+    byte buf[MAX_ATTR_SIZE];
+    while (btell(body) < end && bgets(body, buf, sizeof(buf)) && buf[0])
+      obj_add_attr(o_hdr, buf[0], buf+1);
     if (body_start)
       *body_start = btell(body) - start;
     else
     {
-      obj_read(body, o_body);
-      bgetc(body);
+      while (btell(body) < end && bgets(body, buf, sizeof(buf)))
+	if (buf[0])
+	  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(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);
@@ -121,26 +149,22 @@ buck2obj_parse(struct buck2obj_buf *buf, uns buck_type, uns buck_len, struct fas
       ptr = buf->bb.ptr;
       copied = 1;
     }
-    end = ptr + len;
+    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 + 4 > end)
+      if (ptr + 8 > end)
 	RET_ERR(EINVAL);
       len = GET_U32(ptr);
       ptr += 4;
+      uns adler = GET_U32(ptr);
+      ptr += 4;
       byte *new_ptr = lizard_decompress_safe(ptr, buf->lizard, len);
       if (!new_ptr)
 	return -1;
+      if (adler32(new_ptr, len) != adler)
+	RET_ERR(EINVAL);
       if (!copied)
 	bdirect_read_commit(body, end);
       ptr = new_ptr;
@@ -154,7 +178,10 @@ buck2obj_parse(struct buck2obj_buf *buf, uns buck_type, uns buck_len, struct fas
       bdirect_read_commit_modified(body, ptr);
   }
   else
-    RET_ERR(EINVAL);
+    {
+      bskip(body, buck_len);
+      RET_ERR(EINVAL);
+    }
   return 0;
 }
 
@@ -167,3 +194,17 @@ obj_read_bucket(struct buck2obj_buf *buf, struct mempool *pool, uns buck_type, u
   else
     return o;
 }
+
+int
+obj_read(struct fastbuf *f, struct odes *o)
+{
+  byte buf[MAX_ATTR_SIZE];
+
+  while (bgets(f, buf, sizeof(buf)))
+    {
+      if (!buf[0])
+	return 1;
+      obj_add_attr(o, buf[0], buf+1);
+    }
+  return 0;
+}