X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=2f8e0fea8527f2ed023bd3cab3dce00cad3a9e48;hb=7f1be056d02d78487e9589f44177d683dbfa9fee;hp=094764b1e4479dba4d989980d0efc7191c07eb6b;hpb=a34c87ac6c08ec62534e4068a5bd1fac93df8abe;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 094764b..2f8e0fe 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,6 +1,6 @@
-/* $Id: extract.c,v 1.169 2004-12-02 14:05:03 adam Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
-   Index Data Aps
+/* $Id: extract.c,v 1.177 2005-03-17 09:48:46 adam Exp $
+   Copyright (C) 1995-2005
+   Index Data ApS
 
 This file is part of the Zebra server.
 
@@ -85,6 +85,7 @@ static void extract_init (struct recExtractCtrl *p, RecWord *w)
     w->seqno = 1;
     w->attrSet = VAL_BIB1;
     w->attrUse = 1016;
+    w->attrStr = 0;
     w->reg_type = 'w';
     w->extractCtrl = p;
     w->record_id = 0;
@@ -490,18 +491,6 @@ static int file_extract_record(ZebraHandle zh,
             }
             return 0;
         }
-        if (zh->reg->keys.buf_used == 0)
-        {
-            /* the extraction process returned no information - the record
-               is probably empty - unless flagShowRecords is in use */
-            if (!zh->m_flag_rw)
-                return 1;
-	    
-	    if (zh->records_processed < zh->m_file_verbose_limit)
-	        yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
-		    fname, recordOffset);
-            return 1;
-        }
         if (extractCtrl.match_criteria[0])
             matchStr = extractCtrl.match_criteria;
     }
@@ -533,6 +522,18 @@ static int file_extract_record(ZebraHandle zh,
 	    }
 	}
     }
+    if (! *sysno && zh->reg->keys.buf_used == 0)
+    {
+         /* the extraction process returned no information - the record
+            is probably empty - unless flagShowRecords is in use */
+         if (!zh->m_flag_rw)
+             return 1;
+  
+         if (zh->records_processed < zh->m_file_verbose_limit)
+	     yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
+	    fname, recordOffset);
+         return 1;
+    }
 
     if (! *sysno)
     {
@@ -576,7 +577,8 @@ static int file_extract_record(ZebraHandle zh,
 	if (!force_update && recordAttr->runNumber ==
             zebraExplain_runNumberIncrement (zh->reg->zei, 0))
 	{
-            yaz_log (YLOG_LOG, "run number = %d", recordAttr->runNumber);
+            yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
+			    recordAttr->runNumber);
 	    yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
                      zh->m_record_type, fname, recordOffset);
 	    extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
@@ -831,6 +833,7 @@ int buffer_extract_record (ZebraHandle zh,
     long recordOffset = 0;
     struct zebra_fetch_control fc;
     const char *pr_fname = fname;  /* filename to print .. */
+    int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
 
     if (!pr_fname)
 	pr_fname = "<no file>";  /* make it printable if file is omitted */
@@ -912,16 +915,6 @@ int buffer_extract_record (ZebraHandle zh,
 	yaz_log (YLOG_WARN, "extract error: no such filter");
 	return 0;
     }
-    if (zh->reg->keys.buf_used == 0)
-    {
-	/* the extraction process returned no information - the record
-	   is probably empty - unless flagShowRecords is in use */
-	if (test_mode)
-	    return 1;
-	yaz_log (YLOG_WARN, "No keys generated for record");
-	yaz_log (YLOG_WARN, " The file is probably empty");
-	return 1;
-    }
     /* match criteria */
     matchStr = NULL;
 
@@ -952,19 +945,28 @@ int buffer_extract_record (ZebraHandle zh,
 	    }
         }
     }
+    if (zh->reg->keys.buf_used == 0)
+    {
+	/* the extraction process returned no information - the record
+	   is probably empty - unless flagShowRecords is in use */
+	if (test_mode)
+	    return 1;
+    }
 
     if (! *sysno)
     {
         /* new record */
         if (delete_flag)
         {
-	    yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
-		  pr_fname, (long) recordOffset);
+	    if (show_progress)
+		yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
+			 pr_fname, (long) recordOffset);
             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
             return 1;
         }
-	yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
-	      (long) recordOffset);
+	if (show_progress)
+	    yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
+		     (long) recordOffset);
         rec = rec_new (zh->reg->records);
 
         *sysno = rec->sysno;
@@ -987,11 +989,13 @@ int buffer_extract_record (ZebraHandle zh,
         struct recKeys delkeys;
         struct sortKeys sortKeys;
 
-	if (!allow_update) {
-	      yaz_log (YLOG_LOG, "skipped %s %s %ld", 
-		    recordType, pr_fname, (long) recordOffset);
-	      logRecord(zh);
-	      return -1;
+	if (!allow_update)
+	{
+	    if (show_progress)
+		yaz_log (YLOG_LOG, "skipped %s %s %ld", 
+			 recordType, pr_fname, (long) recordOffset);
+	    logRecord(zh);
+	    return -1;
 	}
 
         rec = rec_get (zh->reg->records, *sysno);
@@ -1003,8 +1007,9 @@ int buffer_extract_record (ZebraHandle zh,
 	    if (recordAttr->runNumber ==
 		zebraExplain_runNumberIncrement (zh->reg->zei, 0))
 	    {
-		yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
-		      pr_fname, (long) recordOffset);
+		if (show_progress)
+		    yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
+			     pr_fname, (long) recordOffset);
 		extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
 		rec_rm (&rec);
 		logRecord(zh);
@@ -1025,14 +1030,19 @@ int buffer_extract_record (ZebraHandle zh,
             /* record going to be deleted */
             if (!delkeys.buf_used)
             {
-                yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
-                      pr_fname, (long) recordOffset);
-                yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
-            }
+		if (show_progress)
+		{
+		    yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
+			     pr_fname, (long) recordOffset);
+		    yaz_log (YLOG_WARN, "cannot delete file above, "
+			     "storeKeys false");
+		}
+	    }
             else
             {
-		yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
-		      pr_fname, (long) recordOffset);
+		if (show_progress)
+		    yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
+			     pr_fname, (long) recordOffset);
                 zh->records_deleted++;
                 if (matchStr)
                     dict_delete (zh->reg->matchDict, matchStr);
@@ -1047,14 +1057,18 @@ int buffer_extract_record (ZebraHandle zh,
             /* record going to be updated */
             if (!delkeys.buf_used)
             {
-                yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
-                      pr_fname, (long) recordOffset);
-                yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
-            }
+		if (show_progress)
+		{
+		    yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
+			     pr_fname, (long) recordOffset);
+		    yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
+		}
+	    }
             else
             {
-		yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
-		      pr_fname, (long) recordOffset);
+		if (show_progress)
+		    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);
                 zh->records_updated++;
@@ -1235,14 +1249,10 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         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);
 
-	attrSet = (int) key.mem[0] >> 16;
-	attrUse = (int) key.mem[0] & 65535;
-
         if (zh->reg->key_buf_used + 1024 > 
             (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
             extract_flushWriteKeys (zh,0);
@@ -1251,11 +1261,8 @@ 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 = zebraExplain_lookupSU (zei, attrSet, attrUse);
-        if (ch < 0)
-            ch = zebraExplain_addSU (zei, attrSet, attrUse);
+	ch = (int) key.mem[0];  /* ordinal for field/use/attribute */
 
-        assert (ch > 0);
 	zh->reg->key_buf_used +=
 	    key_SU_encode (ch,((char*)zh->reg->key_buf) +
                            zh->reg->key_buf_used);
@@ -1266,7 +1273,7 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
         ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
 
         key.len = 3;
-	if (key.mem[1]) /* filter specify record ID */
+	if (key.mem[1]) /* filter specified record ID */
 	    key.mem[0] = key.mem[1];
 	else
 	    key.mem[0] = sysno;
@@ -1426,9 +1433,7 @@ void extract_add_it_key (ZebraHandle zh,
     
     if (keys->buf_used+1024 > keys->buf_max)
     {
-        char *b;
-
-        b = (char *) xmalloc (keys->buf_max += 128000);
+        char *b = (char *) xmalloc (keys->buf_max += 128000);
         if (keys->buf_used > 0)
             memcpy (b, keys->buf, keys->buf_used);
         xfree (keys->buf);
@@ -1448,13 +1453,30 @@ void extract_add_it_key (ZebraHandle zh,
 void extract_add_index_string (RecWord *p, const char *str, int length)
 {
     struct it_key key;
+
+    ZebraHandle zh = p->extractCtrl->handle;
+    ZebraExplainInfo zei = zh->reg->zei;
+    int ch;
+
+    if (p->attrStr)
+    {
+	ch = zebraExplain_lookup_attr_str(zei, p->attrStr);
+	if (ch < 0)
+	    ch = zebraExplain_add_attr_str(zei, p->attrStr);
+    }
+    else
+    {
+	ch = zebraExplain_lookup_attr_su(zei, p->attrSet, p->attrUse);
+	if (ch < 0)
+	    ch = zebraExplain_add_attr_su(zei, p->attrSet, p->attrUse);
+    }
     key.len = 4;
-    key.mem[0] = p->attrSet * 65536 + p->attrUse;
+    key.mem[0] = ch;
     key.mem[1] = p->record_id;
     key.mem[2] = p->section_id;
     key.mem[3] = p->seqno;
 
-#if 1
+#if 0
     /* just for debugging .. */
     yaz_log(YLOG_LOG, "add: set=%d use=%d "
 	    "record_id=%lld section_id=%lld seqno=%lld",
@@ -1513,11 +1535,11 @@ void extract_add_string (RecWord *p, const char *string, int length)
 
 static void extract_add_incomplete_field (RecWord *p)
 {
-    const char *b = p->string;
-    int remain = p->length;
+    const char *b = p->term_buf;
+    int remain = p->term_len;
     const char **map = 0;
     
-    yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->length, p->string);
+    yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
 
     if (remain > 0)
 	map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
@@ -1530,7 +1552,7 @@ static void extract_add_incomplete_field (RecWord *p)
 	/* Skip spaces */
 	while (map && *map && **map == *CHR_SPACE)
 	{
-	    remain = p->length - (b - p->string);
+	    remain = p->term_len - (b - p->term_buf);
 	    if (remain > 0)
 		map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
 	    else
@@ -1545,7 +1567,7 @@ static void extract_add_incomplete_field (RecWord *p)
 
 	    while (i < IT_MAX_WORD && *cp)
 		buf[i++] = *(cp++);
-	    remain = p->length - (b - p->string);
+	    remain = p->term_len - (b - p->term_buf);
 	    if (remain > 0)
 		map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
 	    else
@@ -1560,13 +1582,13 @@ static void extract_add_incomplete_field (RecWord *p)
 
 static void extract_add_complete_field (RecWord *p)
 {
-    const char *b = p->string;
+    const char *b = p->term_buf;
     char buf[IT_MAX_WORD+1];
     const char **map = 0;
-    int i = 0, remain = p->length;
-    int first; /* first position */
+    int i = 0, remain = p->term_len;
 
-    yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'", p->length, p->string);
+    yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
+	    p->term_len, p->term_buf);
 
     if (remain > 0)
 	map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1);
@@ -1575,11 +1597,11 @@ static void extract_add_complete_field (RecWord *p)
     {
 	while (map && *map && **map == *CHR_SPACE)
 	{
-	    remain = p->length - (b - p->string);
+	    remain = p->term_len - (b - p->term_buf);
 
 	    if (remain > 0)
 	    {
-		first = i ? 0 : 1;
+		int first = i ? 0 : 1;  /* first position */
 		map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first);
 	    }
 	    else
@@ -1606,7 +1628,7 @@ static void extract_add_complete_field (RecWord *p)
 		while (i < IT_MAX_WORD && *cp)
 		    buf[i++] = *(cp++);
 	    }
-	    remain = p->length  - (b - p->string);
+	    remain = p->term_len  - (b - p->term_buf);
 	    if (remain > 0)
 	    {
 		map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
@@ -1631,10 +1653,10 @@ void extract_token_add (RecWord *p)
              p->string);
 #endif
     if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0,
-			       p->string, p->length)))
+			       p->term_buf, p->term_len)))
     {
-	p->string = wrbuf_buf(wrbuf);
-	p->length = wrbuf_len(wrbuf);
+	p->term_buf = wrbuf_buf(wrbuf);
+	p->term_len = wrbuf_len(wrbuf);
     }
     if (zebra_maps_is_complete (p->zebra_maps, p->reg_type))
 	extract_add_complete_field (p);