Added support for static ranking. Bug #174. The staticrank is enabled

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

diff --git a/index/extract.c b/index/extract.c
index 59822b9..5cc475e 100644 (file)
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.187 2005-06-23 06:45:46 adam Exp $
+/* $Id: extract.c,v 1.189 2005-08-18 12:50:17 adam Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -98,42 +98,45 @@ static void extract_init (struct recExtractCtrl *p, RecWord *w)
     w->section_id = 0;
 }
 
-static const char **searchRecordKey (ZebraHandle zh,
-                                     struct recKeys *reckeys,
-                                    int attrSetS, int attrUseS)
+static void searchRecordKey(ZebraHandle zh,
+                           const struct recKeys *reckeys,
+                           int attrSetS, int attrUseS,
+                           const char **ws, int ws_length)
 {
-    static const char *ws[32];
     void *decode_handle = iscz1_start();
     int off = 0;
     int startSeq = -1;
     int seqno = 0;
     int i;
+    int ch;
 
-    for (i = 0; i<32; i++)
+    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)
+       return ;
+
     while (off < reckeys->buf_used)
     {
         const char *src = reckeys->buf + off;
         struct it_key key;
        char *dst = (char*) &key;
-       int attrSet, attrUse;
 
        iscz1_decode(decode_handle, &dst, &src);
        assert(key.len <= 4 && key.len > 2);
 
-       attrSet = (int) key.mem[0] >> 16;
-       attrUse = (int) key.mem[0] & 65535;
        seqno = (int) key.mem[key.len-1];
 
-       if (attrUseS == attrUse && attrSetS == attrSet)
+       if (key.mem[0] == ch)
         {
             int woff;
 
             if (startSeq == -1)
                 startSeq = seqno;
             woff = seqno - startSeq;
-            if (woff >= 0 && woff < 31)
+            if (woff >= 0 && woff < ws_length)
                 ws[woff] = src;
         }
 
@@ -143,7 +146,6 @@ static const char **searchRecordKey (ZebraHandle zh,
     }
     iscz1_stop(decode_handle);
     assert (off == reckeys->buf_used);
-    return ws;
 }
 
 struct file_read_info {
@@ -218,7 +220,6 @@ static char *fileMatchStr (ZebraHandle zh,
     static char dstBuf[2048];      /* static here ??? */
     char *dst = dstBuf;
     const char *s = spec;
-    static const char **w;
 
     while (1)
     {
@@ -228,6 +229,7 @@ static char *fileMatchStr (ZebraHandle zh,
             break;
         if (*s == '(')
         {
+           const char *ws[32];
            char attset_str[64], attname_str[64];
            data1_attset *attset;
            int i;
@@ -260,8 +262,7 @@ static char *fileMatchStr (ZebraHandle zh,
                else
                    attUse = atoi (attname_str);
            }
-            w = searchRecordKey (zh, reckeys, attSet, attUse);
-            assert (w);
+            searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
 
             if (*s == ')')
             {
@@ -277,15 +278,15 @@ static char *fileMatchStr (ZebraHandle zh,
             s++;
 
             for (i = 0; i<32; i++)
-                if (matchFlag[i] && w[i])
+                if (matchFlag[i] && ws[i])
                 {
                     if (first)
                     {
                         *dst++ = ' ';
                         first = 0;
                     }
-                    strcpy (dst, w[i]);
-                    dst += strlen(w[i]);
+                    strcpy (dst, ws[i]);
+                    dst += strlen(ws[i]);
                 }
             if (first)
             {
@@ -402,6 +403,7 @@ static int file_extract_record(ZebraHandle zh,
     SYSNO sysnotmp;
     Record rec;
     off_t recordOffset = 0;
+    struct recExtractCtrl extractCtrl;
     
     /* announce database */
     if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
@@ -413,8 +415,6 @@ static int file_extract_record(ZebraHandle zh,
 
     if (fi->fd != -1)
     {
-       struct recExtractCtrl extractCtrl;
-
         /* we are going to read from a file, so prepare the extraction */
        create_rec_keys_codec(&zh->reg->keys);
 #if NATTR
@@ -435,6 +435,8 @@ static int file_extract_record(ZebraHandle zh,
        extractCtrl.schemaAdd = extract_schema_add;
        extractCtrl.dh = zh->reg->dh;
        extractCtrl.match_criteria[0] = '\0';
+       extractCtrl.staticrank = 0;
+       
        extractCtrl.first_record = fi->file_offset ? 0 : 1;
 
        extract_set_store_data_prepare(&extractCtrl);
@@ -479,7 +481,7 @@ static int file_extract_record(ZebraHandle zh,
             return 0;
         }
         if (extractCtrl.match_criteria[0])
-            matchStr = extractCtrl.match_criteria;
+            matchStr = extractCtrl.match_criteria;     
     }
 
     /* perform match if sysno not known and if match criteria is specified */
@@ -540,13 +542,15 @@ static int file_extract_record(ZebraHandle zh,
         *sysno = rec->sysno;
 
        recordAttr = rec_init_attr (zh->reg->zei, rec);
+       recordAttr->staticrank = extractCtrl.staticrank;
 
         if (matchStr)
         {
             dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
         }
        extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
+        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
+                                recordAttr->staticrank);
 
         zh->records_inserted++;
     }
@@ -577,6 +581,7 @@ static int file_extract_record(ZebraHandle zh,
            logRecord (zh);
            return 1;
        }
+       /* flush old keys for sort&search etc. */
         delkeys.buf_used = rec->size[recInfo_delKeys];
        delkeys.buf = rec->info[recInfo_delKeys];
 
@@ -584,7 +589,8 @@ static int file_extract_record(ZebraHandle zh,
         sortKeys.buf = rec->info[recInfo_sortKeys];
 
        extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
-        extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
+        extract_flushRecordKeys (zh, *sysno, 0, &delkeys,
+                                recordAttr->staticrank); /* old values */  
         if (deleteFlag)
         {
             /* record going to be deleted */
@@ -619,11 +625,14 @@ static int file_extract_record(ZebraHandle zh,
             }
             else
             {
+               /* flush new keys for sort&search etc */
                 if (zh->records_processed < zh->m_file_verbose_limit)
                     yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
                         zh->m_record_type, fname, recordOffset);
+               recordAttr->staticrank = extractCtrl.staticrank;
                 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
+                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
+                                        recordAttr->staticrank);
                 zh->records_updated++;
             }
         }
@@ -919,6 +928,7 @@ ZEBRA_RES buffer_extract_record (ZebraHandle zh,
     extractCtrl.dh = zh->reg->dh;
     extractCtrl.handle = zh;
     extractCtrl.match_criteria[0] = '\0';
+    extractCtrl.staticrank = 0;
     
     init_extractCtrl(zh, &extractCtrl);
 
@@ -1004,7 +1014,7 @@ ZEBRA_RES buffer_extract_record (ZebraHandle zh,
                          sizeof(*sysno), sysno);
         }
        extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
+        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys, 0);
 
         zh->records_inserted++;
     } 
@@ -1053,7 +1063,7 @@ ZEBRA_RES buffer_extract_record (ZebraHandle zh,
         sortKeys.buf = rec->info[recInfo_sortKeys];
 
        extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
-        extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
+        extract_flushRecordKeys (zh, *sysno, 0, &delkeys, 0);
         if (delete_flag)
         {
             /* record going to be deleted */
@@ -1099,7 +1109,7 @@ ZEBRA_RES buffer_extract_record (ZebraHandle zh,
                    yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
                             pr_fname, (long) recordOffset);
                 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
+                extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys, 0);
                 zh->records_updated++;
             }
         }
@@ -1219,6 +1229,7 @@ int explain_extract (void *handle, Record rec, data1_node *n)
 
     extractCtrl.flagShowRecords = 0;
     extractCtrl.match_criteria[0] = '\0';
+    extractCtrl.staticrank = 0;
     extractCtrl.handle = handle;
     extractCtrl.first_record = 1;
     
@@ -1243,9 +1254,9 @@ int explain_extract (void *handle, Record rec, data1_node *n)
        sortkeys.buf = rec->info[recInfo_sortKeys];
 
        extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
-       extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
+       extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys, 0);
     }
-    extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
+    extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys, 0);
     extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
 
     xfree (rec->info[recInfo_delKeys]);
@@ -1264,7 +1275,8 @@ int explain_extract (void *handle, Record rec, data1_node *n)
 }
 
 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
-                              int cmd, struct recKeys *reckeys)
+                              int cmd, struct recKeys *reckeys,
+                             zint staticrank)
 {
     void *decode_handle = iscz1_start();
     int off = 0;
@@ -1292,11 +1304,14 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
     while (off < reckeys->buf_used)
     {
         const char *src = reckeys->buf + off;
-        struct it_key key;
-       char *dst = (char*) &key;
+        struct it_key key_in;
+       struct it_key key_out;
+       char *dst = (char*) &key_in;
+       zint *keyp = key_out.mem;
+       size_t key_idx = 0;
 
        iscz1_decode(decode_handle, &dst, &src);
-       assert(key.len == 4);
+       assert(key_in.len == 4);
 
         if (zh->reg->key_buf_used + 1024 > 
             (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
@@ -1306,7 +1321,7 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
            (char*)zh->reg->key_buf + zh->reg->key_buf_used;
 
-       ch = (int) key.mem[0];  /* ordinal for field/use/attribute */
+       ch = (int) key_in.mem[0];  /* ordinal for field/use/attribute */
 
        zh->reg->key_buf_used +=
            key_SU_encode (ch,((char*)zh->reg->key_buf) +
@@ -1317,17 +1332,24 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
         ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
 
-        key.len = 3;
-       if (key.mem[1]) /* filter specified record ID */
-           key.mem[0] = key.mem[1];
+       if (zh->m_staticrank) /* rank config enabled ? */
+       {
+           *keyp++ = staticrank;
+           key_out.len = 4;
+       }
+       else
+           key_out.len = 3;
+
+       if (key_in.mem[1]) /* filter specified record ID */
+           *keyp++ = key_in.mem[1];
        else
-           key.mem[0] = sysno;
-       key.mem[1] = key.mem[2];  /* section_id */
-       key.mem[2] = key.mem[3];  /* sequence .. */
+           *keyp++ = sysno;
+       *keyp++ = key_in.mem[2];  /* section_id */
+       *keyp++ = key_in.mem[3];  /* sequence .. */
 
         memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
-               &key, sizeof(key));
-        (zh->reg->key_buf_used) += sizeof(key);
+               &key_out, sizeof(key_out));
+        (zh->reg->key_buf_used) += sizeof(key_out);
         off = src - reckeys->buf;
     }
     assert (off == reckeys->buf_used);