Implemented update action : adelete.

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

diff --git a/index/extract.c b/index/extract.c
index 48dd978..be000c2 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.271 2007-12-07 14:09:09 adam Exp $
+/* $Id: extract.c,v 1.275 2007-12-20 11:15:42 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);
+        }
     }
 }
 
@@ -560,7 +585,7 @@ ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
 
 
 ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname, 
-                            int deleteFlag)
+                             enum zebra_recctrl_action_t action)
 {
     ZEBRA_RES r = ZEBRA_OK;
     int i, fd;
@@ -630,7 +655,7 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
     default:
        yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
     }
-    if (sysno && deleteFlag)
+    if (sysno && (action == action_delete || action == action_a_delete))
     {
         streamp = 0;
         fi = 0;
@@ -658,8 +683,7 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
         zebra_create_stream_fd(streamp, fd, 0);
     }
     r = zebra_extract_records_stream(zh, streamp,
-                                     deleteFlag ? 
-                                     action_delete : action_update,
+                                     action,
                                      0, /* tst_mode */
                                      zh->m_record_type,
                                      sysno,
@@ -944,14 +968,21 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
 
     if (! *sysno)
     {
-        /* new record */
+        /* new record AKA does not exist already */
         if (action == action_delete)
         {
-           yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
-                        pr_fname, (zint) start_offset);
+                yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+                        pr_fname, (zint) start_offset);
             yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
             return ZEBRA_FAIL;
         }
+        else if (action == action_a_delete)
+        {
+            if (show_progress)
+                yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+                        pr_fname, (zint) start_offset);
+            return ZEBRA_OK;
+        }
        else if (action == action_replace)
        {
            yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
@@ -1044,7 +1075,7 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
         extract_flush_record_keys(zh, *sysno, 0, delkeys,
                                   recordAttr->staticrank);
 #endif
-        if (action == action_delete)
+        if (action == action_delete || action == action_a_delete)
         {
             /* record going to be deleted */
 #if FLUSH2
@@ -1456,7 +1487,7 @@ 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)
 {
@@ -1725,32 +1756,13 @@ static void extract_add_complete_field(RecWord *p, zebra_map_t zm)
 
 static void extract_add_icu(RecWord *p, zebra_map_t zm)
 {
-    struct it_key key;
     const char *res_buf = 0;
     size_t res_len = 0;
-    ZebraHandle zh = p->extractCtrl->handle;
-    
-    int cat = zinfo_index_category_index;
-    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);
+
     zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
-    while (zebra_map_tokenize_next(zm, &res_buf, &res_len))
+    while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
     {
-        int i = 0;
-        key.mem[i++] = ch;
-        key.mem[i++] = p->record_id;
-        key.mem[i++] = p->section_id;
-        
-        if (zh->m_segment_indexing)
-            key.mem[i++] = p->segment;
-        key.mem[i++] = p->seqno;
-        key.len = i;
-
-        zebra_rec_keys_write(zh->reg->keys, res_buf, res_len, &key);
-        
+        extract_add_string(p, zm, res_buf, res_len);
         p->seqno++;
     }
 }
@@ -1760,8 +1772,8 @@ 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_icu
     extract_add_string
     
     extract_add_index_string