Some minor corrections to the LZ compression. In fact I wanted to

Jan Wieck · Jan Wieck · commit 211ed36635e3 · 1999-11-17T22:18:46.000Z
HAVE the required OID's first.

Jan
diff --git a/src/backend/utils/adt/lztext.c b/src/backend/utils/adt/lztext.c
@@ -1,10 +1,13 @@
 /* ----------
  * lztext.c -
  *
- * $Header: /cvsroot/pgsql/src/backend/utils/adt/Attic/lztext.c,v 1.1 1999/11/17 21:21:50 wieck Exp $
+ * $Header: /cvsroot/pgsql/src/backend/utils/adt/Attic/lztext.c,v 1.2 1999/11/17 22:18:45 wieck Exp $
  *
  *	Text type with internal LZ compressed representation. Uses the
  *	standard PostgreSQL compression method.
+ *
+ *	This code requires that the LZ compressor found in pg_lzcompress
+ *	codes a usable VARSIZE word at the beginning of the output buffer.
  * ----------
  */
 
@@ -42,7 +45,7 @@ lztextin(char *str)
 		return NULL;
 
 	/* ----------
-	 * Determine input size and eventually tuple size
+	 * Determine input size and maximum output Datum size
 	 * ----------
 	 */
 	rawsize = strlen(str);
@@ -56,8 +59,9 @@ lztextin(char *str)
 	pglz_compress(str, rawsize, tmp, NULL);
 
 	/* ----------
-	 * If we miss less than x% bytes at the end of the temp value,
-	 * so be it. Therefore we save a memcpy().
+	 * If we miss less than 25% bytes at the end of the temp value,
+	 * so be it. Therefore we save a palloc()/memcpy()/pfree()
+	 * sequence.
 	 * ----------
 	 */
 	if (tmp_size - tmp->varsize < 256 || 
@@ -141,7 +145,7 @@ lztextlen(lztext *lz)
 	 * without multibyte support, it's the remembered rawsize
 	 * ----------
 	 */
-	return lz->rawsize;
+	return PGLZ_RAW_SIZE(lz);
 }
 
 
@@ -166,7 +170,7 @@ lztextoctetlen(lztext *lz)
 	 * Return the varsize minus the VARSIZE field itself.
 	 * ----------
 	 */
-	return lz->varsize - sizeof(int32);
+	return VARSIZE(lz) - VARHDRSZ;
 }
 
 
@@ -208,8 +212,9 @@ text_lztext(text *txt)
 	pglz_compress(str, rawsize, tmp, NULL);
 
 	/* ----------
-	 * If we miss less than x% bytes at the end of the temp value,
-	 * so be it. Therefore we save a memcpy().
+	 * If we miss less than 25% bytes at the end of the temp value,
+	 * so be it. Therefore we save a palloc()/memcpy()/pfree()
+	 * sequence.
 	 * ----------
 	 */
 	if (tmp_size - tmp->varsize < 256 || 
@@ -250,15 +255,15 @@ lztext_text(lztext *lz)
 	 * Allocate and initialize the text result
 	 * ----------
 	 */
-	result = (text *) palloc(lz->rawsize + VARHDRSZ + 1);
+	result = (text *) palloc(PGLZ_RAW_SIZE(lz) + VARHDRSZ + 1);
 	VARSIZE(result) = lz->rawsize + VARHDRSZ;
 
 	/* ----------
 	 * Decompress directly into the text data area.
 	 * ----------
 	 */
-	pglz_decompress(lz, VARDATA(result));
 	VARDATA(result)[lz->rawsize] = 0;
+	pglz_decompress(lz, VARDATA(result));
 
 	return result;
 }
diff --git a/src/backend/utils/adt/pg_lzcompress.c b/src/backend/utils/adt/pg_lzcompress.c
@@ -1,7 +1,7 @@
 /* ----------
  * pg_lzcompress.c -
  *
- * $Header: /cvsroot/pgsql/src/backend/utils/adt/pg_lzcompress.c,v 1.1 1999/11/17 21:21:50 wieck Exp $
+ * $Header: /cvsroot/pgsql/src/backend/utils/adt/pg_lzcompress.c,v 1.2 1999/11/17 22:18:45 wieck Exp $
  *
  *		This is an implementation of LZ compression for PostgreSQL.
  *		It uses a simple history table and generates 2-3 byte tags
@@ -385,6 +385,8 @@ pglz_find_match (PGLZ_HistEntry **hstart, char *input, char *end,
 
 /* ----------
  * pglz_compress -
+ *
+ *		Compresses source into dest using strategy.
  * ----------
  */
 int
@@ -580,6 +582,8 @@ pglz_compress (char *source, int slen, PGLZ_Header *dest, PGLZ_Strategy *strateg
 
 /* ----------
  * pglz_decompress -
+ *
+ *		Decompresses source into dest.
  * ----------
  */
 int
diff --git a/src/include/utils/pg_lzcompress.h b/src/include/utils/pg_lzcompress.h
@@ -1,7 +1,7 @@
 /* ----------
  * pg_lzcompress.h -
  *
- * $Header: /cvsroot/pgsql/src/include/utils/pg_lzcompress.h,v 1.1 1999/11/17 21:21:51 wieck Exp $
+ * $Header: /cvsroot/pgsql/src/include/utils/pg_lzcompress.h,v 1.2 1999/11/17 22:18:46 wieck Exp $
  *
  *	Definitions for the builtin LZ compressor
  * ----------
@@ -37,9 +37,36 @@ typedef struct PGLZ_Header {
  */
 #define PGLZ_MAX_OUTPUT(_dlen)			((_dlen) + (((_dlen) | 0x07) >> 3)	\
 													 + sizeof(PGLZ_Header))
+
+/* ----------
+ * PGLZ_RAW_SIZE -
+ *
+ *		Macro to determine the uncompressed data size contained
+ *		in the entry.
+ * ----------
+ */
 #define PGLZ_RAW_SIZE(_lzdata)			(_lzdata->rawsize)
+
+/* ----------
+ * PGLZ_IS_COMPRESSED -
+ *
+ *		Macro to determine if the data itself is stored as raw
+ *		uncompressed data.
+ * ----------
+ */
 #define PGLZ_IS_COMPRESSED(_lzdata)		(_lzdata->varsize != 				\
-										 _lzdata->rawsize + sizeof(PGLZ_Header))
+										 _lzdata->rawsize + 				\
+										 				sizeof(PGLZ_Header))
+
+/* ----------
+ * PGLZ_RAW_DATA -
+ *
+ *		Macro to get access to the plain compressed or uncompressed
+ *		data. Useful if PGLZ_IS_COMPRESSED returns false.
+ * ----------
+ */
+#define PGLZ_RAW_DATA(_lzdata)			(((char *)(_lzdata)) + 				\
+														sizeof(PGLZ_Header))
 
 /* ----------
  * PGLZ_Strategy -
