X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=a93373ce8f259656e1c6d37299c33ae9bb483729;hb=b3f079a571e404665620e8eb3bb1dd7db1ff6669;hp=7e81792087d202566ab39ceeee290a95ec184e42;hpb=13551199cc4f367c3699fd079d3f7021075a6059;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 7e81792..a93373c 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -170,8 +170,7 @@ static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
 
     while (map)
     {
-	char buf[IT_MAX_WORD+1];
-	int i, remain;
+	int remain;
 
 	/* Skip spaces */
 	while (map && *map && **map == *CHR_SPACE)
@@ -189,17 +188,10 @@ static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
         {
             zebra_snippets_appendn(h->snippets, p->seqno, 1, ord,
                                    start, last - start);
-
         }
         start = last;
-
-	i = 0;
 	while (map && *map && **map != *CHR_SPACE)
 	{
-	    const char *cp = *map;
-
-	    while (i < IT_MAX_WORD && *cp)
-		buf[i++] = *(cp++);
 	    remain = p->term_len - (b - p->term_buf);
             last = b;
 	    if (remain > 0)
@@ -207,8 +199,8 @@ static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
 	    else
 		map = 0;
 	}
-	if (!i)
-	    return;
+        if (start == last)
+            return ;
 
         if (first)
         {   
@@ -285,7 +277,6 @@ void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 {
     struct recExtractCtrl extractCtrl;
     struct snip_rec_info info;
-    int r;
 
     extractCtrl.stream = stream;
     extractCtrl.first_record = 1;
@@ -308,8 +299,7 @@ void extract_snippet(ZebraHandle zh, zebra_snippets *sn,
 
     extractCtrl.setStoreData = 0;
 
-    r = (*rt->extract)(recTypeClientData, &extractCtrl);
-
+    (*rt->extract)(recTypeClientData, &extractCtrl);
 }
 
 static void searchRecordKey(ZebraHandle zh,
@@ -575,7 +565,6 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
     char gprefix[128];
     char ext[128];
     char ext_res[128];
-    struct file_read_info *fi = 0;
     const char *original_record_type = 0;
     RecType recType;
     void *recTypeClientData;
@@ -641,7 +630,6 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
     if (sysno && (action == action_delete || action == action_a_delete))
     {
         streamp = 0;
-        fi = 0;
     }
     else
     {
@@ -1810,6 +1798,11 @@ static void extract_add_icu(RecWord *p, zebra_map_t zm)
     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
     while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
     {
+        if (res_len > IT_MAX_WORD)
+        {
+            yaz_log(YLOG_LOG, "Truncating long term %ld", (long) res_len);
+            res_len = IT_MAX_WORD;
+        }
         extract_add_string(p, zm, res_buf, res_len);
         p->seqno++;
     }
@@ -1835,7 +1828,6 @@ static void extract_token_add(RecWord *p)
 {
     ZebraHandle zh = p->extractCtrl->handle;
     zebra_map_t zm = zebra_map_get_or_add(zh->reg->zebra_maps, p->index_type);
-    WRBUF wrbuf;
 
     if (log_level_details)
     {
@@ -1844,11 +1836,6 @@ static void extract_token_add(RecWord *p)
                 p->index_type, p->index_name, 
                 p->seqno, p->term_len, p->term_buf);
     }
-    if ((wrbuf = zebra_replace(zm, 0, p->term_buf, p->term_len)))
-    {
-        p->term_buf = wrbuf_buf(wrbuf);
-        p->term_len = wrbuf_len(wrbuf);
-    }
     if (zebra_maps_is_icu(zm))
     {
         extract_add_icu(p, zm);