X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=8d4cd9ca7035e4895b06ba089d811efbf89b3ea4;hb=53f91ef0ea7d588a26e90bf87ad46bdb84938d6f;hp=75bc9e004db2c39b8365680409d553089b215c0c;hpb=6ba9698e88c0283e40fa5980a1a6b551fff2d597;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 75bc9e0..8d4cd9c 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.215 2006-05-19 13:49:34 adam Exp $
+/* $Id: extract.c,v 1.219 2006-06-07 10:14:40 adam Exp $
    Copyright (C) 1995-2006
    Index Data ApS
 
@@ -48,6 +48,25 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 
 #endif
 
+#define ENCODE_BUFLEN 768
+struct encode_info {
+#if 0
+    int  sysno;  /* previously written values for delta-compress */
+    int  seqno;
+    int  cmd;
+    int prevsys; /* buffer for skipping insert/delete pairs */
+    int prevseq;
+    int prevcmd;
+    int keylen; /* tells if we have an unwritten key in buf, and how long*/
+#endif
+    void *encode_handle;
+    void *decode_handle;
+    char buf[ENCODE_BUFLEN];
+};
+
+static void encode_key_init (struct encode_info *i);
+static void encode_key_write (char *k, struct encode_info *i, FILE *outf);
+static void encode_key_flush (struct encode_info *i, FILE *outf);
 
 #define USE_SHELLSORT 0
 
@@ -126,7 +145,7 @@ static void searchRecordKey(ZebraHandle zh,
 
     if (zebra_rec_keys_rewind(reckeys))
     {
-	int startSeq = -1;
+	zint startSeq = -1;
 	const char *str;
 	size_t slen;
 	struct it_key key;
@@ -139,7 +158,7 @@ static void searchRecordKey(ZebraHandle zh,
 	    
 	    if (key.mem[0] == ch)
 	    {
-		int woff;
+		zint woff;
 		
 		if (startSeq == -1)
 		    startSeq = seqno;
@@ -1281,7 +1300,7 @@ void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
 	struct it_key key_in;
 	while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
         {
-            int ord = key_in.mem[0]; 
+            int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
 
             for (p = ord_list; p ; p = p->next)
                 if (p->ord == ord)
@@ -1365,7 +1384,7 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
 		(char*)zh->reg->key_buf + zh->reg->key_buf_used;
 
 	    /* encode the ordinal value (field/use/attribute) .. */
-	    ch = (int) key_in.mem[0];
+	    ch = CAST_ZINT_TO_INT(key_in.mem[0]);
 	    zh->reg->key_buf_used +=
 		key_SU_encode(ch, (char*)zh->reg->key_buf +
 			      zh->reg->key_buf_used);
@@ -1554,15 +1573,16 @@ ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
 	{
 	    char dst_buf[IT_MAX_WORD];
 	    char *dst_term = dst_buf;
-	    int ord, seqno;
+	    int ord;
+            zint seqno;
 	    int index_type;
+
 	    assert(key.len <= 4 && key.len > 2);
-	    seqno = (int) key.mem[key.len-1];
-	    ord = key.mem[0];
+	    seqno = key.mem[key.len-1];
+	    ord = CAST_ZINT_TO_INT(key.mem[0]);
 	    
 	    zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
-				    0/* db */, 0/* set */, 0/* use */,
-				    0 /* string_index */);
+				    0/* db */, 0 /* string_index */);
 	    assert(index_type);
 	    zebra_term_untrans_iconv(zh, nmem, index_type,
 				     &dst_term, str);
@@ -1585,20 +1605,20 @@ void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
 	while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
 	{
 	    char dst_buf[IT_MAX_WORD];
-	    int seqno;
+	    zint seqno;
 	    int index_type;
+            int ord = CAST_ZINT_TO_INT(key.mem[0]);
 	    const char *db = 0;
 	    assert(key.len <= 4 && key.len > 2);
 
-	    zebraExplain_lookup_ord(zh->reg->zei,
-				    key.mem[0], &index_type, &db, 0, 0, 0);
+	    zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
 	    
-	    seqno = (int) key.mem[key.len-1];
+	    seqno = key.mem[key.len-1];
 	    
 	    zebra_term_untrans(zh, index_type, dst_buf, str);
 	    
-	    yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
-		    key.mem[0], seqno, dst_buf); 
+	    yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT 
+                    " term=%s", ord, seqno, dst_buf); 
 	}
     }
 }
@@ -1869,7 +1889,7 @@ void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
 
 	while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
         {
-            int ord = (int) key_in.mem[0];
+            int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
             
             sortIdx_type(sortIdx, ord);
             if (cmd == 1)
@@ -1882,24 +1902,10 @@ void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
 
 void encode_key_init (struct encode_info *i)
 {
-    i->sysno = 0;
-    i->seqno = 0;
-    i->cmd = -1;
-    i->prevsys=0;
-    i->prevseq=0;
-    i->prevcmd=-1;
-    i->keylen=0;
     i->encode_handle = iscz1_start();
     i->decode_handle = iscz1_start();
 }
 
-#define OLDENCODE 1
-
-#ifdef OLDENCODE
-/* this is the old encode_key_write 
- * may be deleted once we are confident that the new works
- * HL 15-oct-2002
- */
 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
 {
     struct it_key key;
@@ -1950,107 +1956,6 @@ void encode_key_flush (struct encode_info *i, FILE *outf)
     iscz1_stop(i->decode_handle);
 }
 
-#else
-
-/* new encode_key_write
- * The idea is to buffer one more key, and compare them
- * If we are going to delete and insert the same key, 
- * we may as well not bother. Should make a difference in 
- * updates with small modifications (appending to a mbox)
- */
-void encode_key_write (char *k, struct encode_info *i, FILE *outf)
-{
-    struct it_key key;
-    char *bp; 
-
-    if (*k)  /* first time for new key */
-    {
-        bp = i->buf;
-        while ((*bp++ = *k++))
-            ;
-	i->keylen= bp - i->buf -1;    
-	assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
-    }
-    else
-    {
-	bp=i->buf + i->keylen;
-	*bp++=0;
-	k++;
-    }
-
-    memcpy (&key, k+1, sizeof(struct it_key));
-    if (0==i->prevsys) /* no previous filter, fill up */
-    {
-        i->prevsys=key.sysno;
-	i->prevseq=key.seqno;
-	i->prevcmd=*k;
-    }
-    else if ( (i->prevsys==key.sysno) &&
-              (i->prevseq==key.seqno) &&
-	      (i->prevcmd!=*k) )
-    { /* same numbers, diff cmd, they cancel out */
-        i->prevsys=0;
-    }
-    else 
-    { /* different stuff, write previous, move buf */
-        bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
-	if (i->sysno != i->prevsys)
-	{
-	    i->sysno = i->prevsys;
-	    i->seqno = 0;
-        }
-        else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
-	{
-	    return; /* ??? Filters some sort of duplicates away */
-	            /* ??? Can this ever happen   -H 15oct02 */
-	}
-        bp = encode_key_int (i->prevseq - i->seqno, bp);
-        i->seqno = i->prevseq;
-        i->cmd = i->prevcmd;
-        if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
-        {
-            yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
-            exit (1);
-        }
-        i->keylen=0; /* ok, it's written, forget it */
-	i->prevsys=key.sysno;
-	i->prevseq=key.seqno;
-	i->prevcmd=*k;
-    }
-}
-
-void encode_key_flush (struct encode_info *i, FILE *outf)
-{ /* flush the last key from i */
-    char *bp =i->buf + i->keylen;
-    if (0==i->prevsys)
-    {
-        return; /* nothing to flush */
-    }
-    *bp++=0;
-    bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
-    if (i->sysno != i->prevsys)
-    {
-        i->sysno = i->prevsys;
-        i->seqno = 0;
-    }
-    else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
-    {
-        return; /* ??? Filters some sort of duplicates away */
-                /* ??? Can this ever happen   -H 15oct02 */
-    }
-    bp = encode_key_int (i->prevseq - i->seqno, bp);
-    i->seqno = i->prevseq;
-    i->cmd = i->prevcmd;
-    if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
-    {
-        yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
-        exit (1);
-    }
-    i->keylen=0; /* ok, it's written, forget it */
-    i->prevsys=0; /* forget the values too */
-    i->prevseq=0;
-}
-#endif
 /*
  * Local variables:
  * c-basic-offset: 4