X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=edf6e372cd846e30de182d305f3d43cdafc52b7b;hb=52551bf774d771a0422a401946fd0c5ee3788f34;hp=ee771ac6587a8f0d91f68177d3cffbacf58ec1ea;hpb=00d69171180b9edc123986794b572be57caa5bb2;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index ee771ac..edf6e37 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,5 +1,5 @@
-/* $Id: extract.c,v 1.133 2002-12-16 22:59:34 adam Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+/* $Id: extract.c,v 1.143 2003-03-13 04:25:17 pop Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003
    Index Data Aps
 
 This file is part of the Zebra server.
@@ -440,8 +440,6 @@ static int recordExtract (ZebraHandle zh,
 	zh->reg->keys.prevAttrSet = -1;
 	zh->reg->keys.prevSeqNo = 0;
 	zh->reg->sortKeys.buf_used = 0;
-	zh->reg->sortKeys.buf_max = 0;
-	zh->reg->sortKeys.buf = 0;
 	
 	recordOffset = fi->file_moffset;
 	extractCtrl.offset = fi->file_moffset;
@@ -916,9 +914,11 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
 				 &rGroup,
 				 delete_flag,
 				 test_mode,
+				 recordType,
 				 sysno,
 				 match_criteria,
-				 "<no file>"));
+				 "<no file>",
+				 0,1));
 }
 /*
   If sysno is provided, then it's used to identify the reocord.
@@ -931,19 +931,21 @@ int bufferExtractRecord (ZebraHandle zh,
 			 struct recordGroup *rGroup, 
 			 int delete_flag,
 			 int test_mode, 
+			 const char *recordType,
 			 int *sysno,
 			 const char *match_criteria,
-			 const char *fname)
+			 const char *fname,
+			 int force_update,
+			 int allow_update)
 
 {
     RecordAttr *recordAttr;
     struct recExtractCtrl extractCtrl;
     int i, r;
     char *matchStr = 0;
-    RecType recType;
+    RecType recType = NULL;
     char subType[1024];
     void *clientData;
-    SYSNO sysnotmp;
     Record rec;
     long recordOffset = 0;
     struct zebra_fetch_control fc;
@@ -962,33 +964,45 @@ int bufferExtractRecord (ZebraHandle zh,
     extractCtrl.endf = zebra_record_int_end;
     extractCtrl.fh = &fc;
 
+    zh->reg->keys.buf_used = 0;
+    zh->reg->keys.prevAttrUse = -1;
+    zh->reg->keys.prevAttrSet = -1;
+    zh->reg->keys.prevSeqNo = 0;
+    zh->reg->sortKeys.buf_used = 0;
+
     /* announce database */
+
+    if (!(rGroup->databaseName)) {
+        logf (LOG_WARN, "Invalid record group, no database name given");
+	return 0;
+    }
+    
     if (zebraExplain_curDatabase (zh->reg->zei, rGroup->databaseName))
     {
         if (zebraExplain_newDatabase (zh->reg->zei, rGroup->databaseName, 0))
-	    return 0;
+            return 0;
     }
-
-    if (!(rGroup->recordType)) {
-        logf (LOG_WARN, "No such record type defined");
-        return 0;
+    
+    if (*recordType) {
+        logf (LOG_DEBUG, "Record type explicitly specified: %s", recordType);
+        recType = recType_byName (zh->reg->recTypes, recordType, subType,
+                                  &clientData);
+    } else {
+        if (!(rGroup->recordType)) {
+            logf (LOG_WARN, "No such record type defined");
+            return 0;
+        }
+        logf (LOG_DEBUG, "Get record type from rgroup: %s",rGroup->recordType);
+        recType = recType_byName (zh->reg->recTypes, rGroup->recordType, subType,
+                                  &clientData);
+        recordType = rGroup->recordType;
     }
-
-    if (!(recType =
-	  recType_byName (zh->reg->recTypes, rGroup->recordType, subType,
-			  &clientData)))
-    {
+    
+    if (!recType) {
         logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
         return 0;
     }
-    zh->reg->keys.buf_used = 0;
-    zh->reg->keys.prevAttrUse = -1;
-    zh->reg->keys.prevAttrSet = -1;
-    zh->reg->keys.prevSeqNo = 0;
-    zh->reg->sortKeys.buf_used = 0;
-    zh->reg->sortKeys.buf_max = 0;
-    zh->reg->sortKeys.buf = 0;
-
+    
     extractCtrl.subType = subType;
     extractCtrl.init = extract_init;
     extractCtrl.tokenAdd = extract_token_add;
@@ -1034,25 +1048,24 @@ int bufferExtractRecord (ZebraHandle zh,
     /* match criteria */
     matchStr = NULL;
 
-    if (! *sysno) {
-      char *rinfo;
-      if (strlen(match_criteria) > 0) {
-	matchStr = (char *)match_criteria;
-      } else {
-	if (rGroup->recordId && *rGroup->recordId) {
-	  matchStr = fileMatchStr (zh, &zh->reg->keys, rGroup, fname, 
-				   rGroup->recordId);
-	}
-      }
-      if (matchStr) {
-	rinfo = dict_lookup (zh->reg->matchDict, matchStr);
-	if (rinfo)
-	  memcpy (sysno, rinfo+1, sizeof(*sysno));
-      } else {
-	logf (LOG_WARN, "Bad match criteria (recordID)");
-	return 0;
-      }
-
+    if (! *sysno && match_criteria) {
+        char *rinfo;
+        if (*match_criteria) {
+            matchStr = (char *)match_criteria;
+        } else {
+            if (rGroup->recordId && *rGroup->recordId) {
+                matchStr = fileMatchStr (zh, &zh->reg->keys, rGroup, fname, 
+                                         rGroup->recordId);
+            }
+        }
+        if (matchStr) {
+            rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+            if (rinfo)
+                memcpy (sysno, rinfo+1, sizeof(*sysno));
+        } else {
+            logf (LOG_WARN, "Bad match criteria (recordID)");
+            return 0;
+        }
     }
 
     if (! *sysno)
@@ -1060,12 +1073,12 @@ int bufferExtractRecord (ZebraHandle zh,
         /* new record */
         if (delete_flag)
         {
-	    logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
+	    logf (LOG_LOG, "delete %s %s %ld", recordType,
 		  fname, (long) recordOffset);
             logf (LOG_WARN, "cannot delete record above (seems new)");
             return 1;
         }
-	logf (LOG_LOG, "add %s %s %ld", rGroup->recordType, fname,
+	logf (LOG_LOG, "add %s %s %ld", recordType, fname,
 	      (long) recordOffset);
         rec = rec_new (zh->reg->records);
 
@@ -1082,28 +1095,38 @@ int bufferExtractRecord (ZebraHandle zh,
         extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
 
         zh->records_inserted++;
-    }
+    } 
     else
     {
         /* record already exists */
         struct recKeys delkeys;
         struct sortKeys sortKeys;
 
+	if (!allow_update) {
+	      logf (LOG_LOG, "skipped %s %s %ld", 
+		    recordType, fname, (long) recordOffset);
+	      logRecord(zh);
+	      return -1;
+	}
+
         rec = rec_get (zh->reg->records, *sysno);
         assert (rec);
 	
 	recordAttr = rec_init_attr (zh->reg->zei, rec);
 
-	if (recordAttr->runNumber ==
-	    zebraExplain_runNumberIncrement (zh->reg->zei, 0))
-	{
-	    logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType,
-		  fname, (long) recordOffset);
-	    extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
-	    rec_rm (&rec);
-            logRecord(zh);
-	    return 1;
+	if (!force_update) {
+	  if (recordAttr->runNumber ==
+	      zebraExplain_runNumberIncrement (zh->reg->zei, 0))
+	    {
+	      logf (LOG_LOG, "skipped %s %s %ld", recordType,
+		    fname, (long) recordOffset);
+	      extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
+	      rec_rm (&rec);
+	      logRecord(zh);
+	      return 1;
+	    }
 	}
+
         delkeys.buf_used = rec->size[recInfo_delKeys];
 	delkeys.buf = rec->info[recInfo_delKeys];
 
@@ -1117,13 +1140,13 @@ int bufferExtractRecord (ZebraHandle zh,
             /* record going to be deleted */
             if (!delkeys.buf_used)
             {
-                logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
+                logf (LOG_LOG, "delete %s %s %ld", recordType,
                       fname, (long) recordOffset);
                 logf (LOG_WARN, "cannot delete file above, storeKeys false");
             }
             else
             {
-		logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType,
+		logf (LOG_LOG, "delete %s %s %ld", recordType,
 		      fname, (long) recordOffset);
                 zh->records_deleted++;
                 if (matchStr)
@@ -1139,13 +1162,13 @@ int bufferExtractRecord (ZebraHandle zh,
             /* record going to be updated */
             if (!delkeys.buf_used)
             {
-                logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
+                logf (LOG_LOG, "update %s %s %ld", recordType,
                       fname, (long) recordOffset);
                 logf (LOG_WARN, "cannot update file above, storeKeys false");
             }
             else
             {
-		logf (LOG_LOG, "update %s %s %ld", rGroup->recordType,
+		logf (LOG_LOG, "update %s %s %ld", recordType,
 		      fname, (long) recordOffset);
                 extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
                 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
@@ -1156,7 +1179,7 @@ int bufferExtractRecord (ZebraHandle zh,
     /* update file type */
     xfree (rec->info[recInfo_fileType]);
     rec->info[recInfo_fileType] =
-        rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
+        rec_strdup (recordType, &rec->size[recInfo_fileType]);
 
     /* update filename */
     xfree (rec->info[recInfo_filename]);
@@ -1591,7 +1614,7 @@ static void extract_add_sort_string (RecWord *p, const char *string,
 {
     ZebraHandle zh = p->extractCtrl->handle;
     struct sortKeys *sk = &zh->reg->sortKeys;
-    size_t off = 0;
+    int off = 0;
 
     while (off < sk->buf_used)
     {
@@ -1727,10 +1750,11 @@ void extract_token_add (RecWord *p)
 {
     WRBUF wrbuf;
 
+#if 0
     yaz_log (LOG_LOG, "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
              p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
              p->string);
-
+#endif
     if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
 			       p->string, p->length)))
     {
@@ -1753,13 +1777,13 @@ void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
                             int cmd, struct sortKeys *sk)
 {
     SortIdx sortIdx = zh->reg->sortIdx;
-    size_t off = 0;
+    int off = 0;
 
     sortIdx_sysno (sortIdx, sysno);
 
     while (off < sk->buf_used)
     {
-        int set, use, slen, l;
+        int set, use, slen;
         
         off += key_SU_decode(&set, sk->buf + off);
         off += key_SU_decode(&use, sk->buf + off);