ICU functional for scan and snippets.

[idzebra-moved-to-github.git] / index / extract.c

diff --git a/index/extract.c b/index/extract.c
index 6ac93de..aed0650 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.268 2007-11-06 10:29:59 adam Exp $
+/* $Id: extract.c,v 1.273 2007-12-13 11:09:20 adam Exp $
    Copyright (C) 1995-2007
    Index Data ApS
 
@@ -265,6 +265,26 @@ static void snippet_add_incomplete_field(RecWord *p, int ord, zebra_map_t zm)
 
 }
 
+static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm)
+{
+    struct snip_rec_info *h = p->extractCtrl->handle;
+
+    const char *res_buf = 0;
+    size_t res_len = 0;
+
+    const char *display_buf = 0;
+    size_t display_len = 0;
+
+    zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+    while (zebra_map_tokenize_next(zm, &res_buf, &res_len,
+                                   &display_buf, &display_len))
+    {
+        zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+                               display_buf, display_len);
+        p->seqno++;
+    }
+}
+
 static void snippet_token_add(RecWord *p)
 {
     struct snip_rec_info *h = p->extractCtrl->handle;
@@ -277,10 +297,15 @@ static void snippet_token_add(RecWord *p)
         int ch = zebraExplain_lookup_attr_str(
             zei, zinfo_index_category_index, p->index_type, p->index_name);
 
-        if (zebra_maps_is_complete(zm))
-            snippet_add_complete_field(p, ch, zm);
+        if (zebra_maps_is_icu(zm))
+            snippet_add_icu(p, ch, zm);
         else
-            snippet_add_incomplete_field(p, ch, zm);
+        {
+            if (zebra_maps_is_complete(zm))
+                snippet_add_complete_field(p, ch, zm);
+            else
+                snippet_add_incomplete_field(p, ch, zm);
+        }
     }
 }
 
@@ -520,13 +545,29 @@ struct recordLogInfo {
     struct recordGroup *rGroup;
 };
 
-static void all_matches_add(struct recExtractCtrl *ctrl)
+/** \brief add the always-matches index entry and map to real record ID
+    \param ctrl record control
+    \param record_id custom record ID
+    \param sysno system record ID
+    
+    This function serves two purposes.. It adds the always matches
+    entry and makes a pointer from the custom record ID (if defined)
+    back to the system record ID (sysno)
+    See zebra_recid_to_sysno .
+  */
+static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
+                            zint sysno)
 {
     RecWord word;
     extract_init(ctrl, &word);
+    word.record_id = record_id;
+    /* we use the seqno as placeholder for a way to get back to
+       record database from _ALLRECORDS.. This is used if a custom
+       RECORD was defined */
+    word.seqno = sysno;
     word.index_name = "_ALLRECORDS";
     word.index_type = "w";
-    word.seqno = 1;
+
     extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
                               "", 0);
 }
@@ -874,8 +915,6 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
         else
             end_offset = stream->tellf(stream);
 
-        all_matches_add(&extractCtrl);
-        
         if (extractCtrl.match_criteria[0])
             match_criteria = extractCtrl.match_criteria;
     }
@@ -919,6 +958,7 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
            }
        }
     }
+
     if (zebra_rec_keys_empty(zh->reg->keys))
     {
        /* the extraction process returned no information - the record
@@ -951,6 +991,15 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 
         *sysno = rec->sysno;
 
+
+        if (stream)
+        {
+            all_matches_add(&extractCtrl,
+                            zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+                            *sysno);
+        }
+
+
        recordAttr = rec_init_attr(zh->reg->zei, rec);
        if (extractCtrl.staticrank < 0)
         {
@@ -992,6 +1041,13 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 
         rec = rec_get(zh->reg->records, *sysno);
         assert(rec);
+
+        if (stream)
+        {
+            all_matches_add(&extractCtrl,
+                            zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+                            *sysno);
+        }
        
        recordAttr = rec_init_attr(zh->reg->zei, rec);
 
@@ -1425,7 +1481,7 @@ void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
 }
 
 
-ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
+ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh,
                                      zebra_rec_keys_t reckeys,
                                      zebra_snippets *snippets)
 {
@@ -1698,13 +1754,15 @@ static void extract_add_icu(RecWord *p, zebra_map_t zm)
     const char *res_buf = 0;
     size_t res_len = 0;
     ZebraHandle zh = p->extractCtrl->handle;
-    int r = zebra_map_tokenize(zm, p->term_buf, p->term_len,
-                               &res_buf, &res_len);
+    
     int cat = zinfo_index_category_index;
-    int ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
+    int ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 
+                                          p->index_type, p->index_name);
     if (ch < 0)
-        ch = zebraExplain_add_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
-    while (r)
+        ch = zebraExplain_add_attr_str(zh->reg->zei, cat, 
+                                       p->index_type, p->index_name);
+    zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+    while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
     {
         int i = 0;
         key.mem[i++] = ch;
@@ -1716,11 +1774,9 @@ static void extract_add_icu(RecWord *p, zebra_map_t zm)
         key.mem[i++] = p->seqno;
         key.len = i;
 
-        yaz_log(YLOG_LOG, "keys_write %.*s", (int) res_len, res_buf);
         zebra_rec_keys_write(zh->reg->keys, res_buf, res_len, &key);
         
         p->seqno++;
-        r = zebra_map_tokenize(zm, 0, 0, &res_buf, &res_len);
     }
 }
 
@@ -1729,8 +1785,8 @@ static void extract_add_icu(RecWord *p, zebra_map_t zm)
     \param p token data to be indexed
 
     Call sequence:
-    extract_token
-    zebra_add_{in}_complete
+    extract_token_add
+    extract_add_{in}_complete
     extract_add_string
     
     extract_add_index_string