X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=a980bcf26b0fede3d9ba9f7da960bb4100fd773c;hb=9ceac3e96fe083b93b6fd4a676003dd10482e465;hp=86f868b3a9e7c355a1b6c92e29ed1bf4ee6448d4;hpb=f7a3769dede0071696bdcc13ae2ee1efe6d52d96;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 86f868b..a980bcf 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.214 2006-05-18 12:03:05 adam Exp $
+/* $Id: extract.c,v 1.218 2006-05-30 13:44:44 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
 
@@ -96,12 +115,7 @@ static void extract_init (struct recExtractCtrl *p, RecWord *w)
 {
     w->zebra_maps = p->zebra_maps;
     w->seqno = 1;
-#if NATTR
-#else
-    w->attrSet = VAL_BIB1;
-    w->attrUse = 1016;
-#endif
-    w->index_name = 0;
+    w->index_name = "any";
     w->index_type = 'w';
     w->extractCtrl = p;
     w->record_id = 0;
@@ -110,17 +124,22 @@ static void extract_init (struct recExtractCtrl *p, RecWord *w)
 
 static void searchRecordKey(ZebraHandle zh,
 			    zebra_rec_keys_t reckeys,
-			    int attrSetS, int attrUseS,
+                            const char *index_name,
 			    const char **ws, int ws_length)
 {
     int i;
-    int ch;
+    int ch = -1;
 
     for (i = 0; i<ws_length; i++)
         ws[i] = NULL;
 
-    ch = zebraExplain_lookup_attr_su_any_index(zh->reg->zei,
-					       attrSetS, attrUseS);
+    if (ch < 0)
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, '0', index_name);
+    if (ch < 0)
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, 'p', index_name);
+    if (ch < 0)
+        ch = zebraExplain_lookup_attr_str(zh->reg->zei, 'w', index_name);
+
     if (ch < 0)
 	return ;
 
@@ -236,9 +255,7 @@ static char *fileMatchStr (ZebraHandle zh,
         {
 	    const char *ws[32];
 	    char attset_str[64], attname_str[64];
-	    data1_attset *attset;
 	    int i;
-            int attSet = 1, attUse = 1;
             int first = 1;
 	    
 	    for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
@@ -251,7 +268,9 @@ static char *fileMatchStr (ZebraHandle zh,
 	    
 	    for (; strchr(FILE_MATCH_BLANK, *s); s++)
 		;
-	    if (*s == ',')
+	    if (*s != ',')
+                strcpy(attname_str, attset_str);
+            else
 	    {
 		for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
 		    ;
@@ -261,18 +280,8 @@ static char *fileMatchStr (ZebraHandle zh,
 			attname_str[i++] = *s;
 		attname_str[i] = '\0';
 	    }
-	    
-	    if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
-	    {
-		data1_att *att;
-		attSet = attset->reference;
-		att = data1_getattbyname(zh->reg->dh, attset, attname_str);
-		if (att)
-		    attUse = att->value;
-		else
-		    attUse = atoi (attname_str);
-	    }
-            searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
+
+            searchRecordKey (zh, reckeys, attname_str, ws, 32);
 
             if (*s != ')')
             {
@@ -1571,8 +1580,7 @@ ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
 	    ord = 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);
@@ -1601,7 +1609,7 @@ void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
 	    assert(key.len <= 4 && key.len > 2);
 
 	    zebraExplain_lookup_ord(zh->reg->zei,
-				    key.mem[0], &index_type, &db, 0, 0, 0);
+				    key.mem[0], &index_type, &db, 0);
 	    
 	    seqno = (int) key.mem[key.len-1];
 	    
@@ -1621,24 +1629,13 @@ void extract_add_index_string(RecWord *p, const char *str, int length)
     ZebraExplainInfo zei = zh->reg->zei;
     int ch;
 
-    if (p->index_name)
-    {
-	ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
-	if (ch < 0)
-	    ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
-    }
-    else
-    {
-#if NATTR
-	return;
-#else
-	ch = zebraExplain_lookup_attr_su(zei, p->index_type, 
-					 p->attrSet, p->attrUse);
-	if (ch < 0)
-	    ch = zebraExplain_add_attr_su(zei, p->index_type,
-					  p->attrSet, p->attrUse);
-#endif
-    }
+    if (!p->index_name)
+        return;
+
+    ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
+    if (ch < 0)
+        ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
+
     key.len = 4;
     key.mem[0] = ch;
     key.mem[1] = p->record_id;
@@ -1673,24 +1670,12 @@ static void extract_add_sort_string(RecWord *p, const char *str, int length)
     ZebraExplainInfo zei = zh->reg->zei;
     int ch;
 
-    if (p->index_name)
-    {
-	ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
-	if (ch < 0)
-	    ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
-    }
-    else
-    {
-#if NATTR
-	return;
-#else
-	ch = zebraExplain_lookup_attr_su(zei, p->index_type, 
-                                         VAL_IDXPATH, p->attrUse);
-	if (ch < 0)
-	    ch = zebraExplain_add_attr_su(zei, p->index_type,
-					  VAL_IDXPATH, p->attrUse);
-#endif
-    }
+    if (!p->index_name)
+        return;
+
+    ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
+    if (ch < 0)
+        ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
     key.len = 4;
     key.mem[0] = ch;
     key.mem[1] = p->record_id;
@@ -1915,24 +1900,10 @@ 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;
@@ -1983,107 +1954,6 @@ 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