X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=9d2742c928eb35ce7195ba94fe2bd4742a26bff1;hb=3c37091eb3e508bff58b297eaa0feab5e92ff53b;hp=f92ca7f932421d38a6640e8b9a72aad576a1d06f;hpb=c6959870998f868e6a0e9201739fb54aef696bc6;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index f92ca7f..9d2742c 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.200 2005-12-09 10:45:04 adam Exp $
+/* $Id: extract.c,v 1.204 2006-03-20 15:17:30 mike Exp $
    Copyright (C) 1995-2005
    Index Data ApS
 
@@ -32,6 +32,7 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 #include <fcntl.h>
 
 #include "index.h"
+#include "orddict.h"
 #include <direntz.h>
 #include <charmap.h>
 
@@ -206,6 +207,8 @@ static void file_end (void *handle, off_t offset)
     }
 }
 
+#define FILE_MATCH_BLANK "\t "
+
 static char *fileMatchStr (ZebraHandle zh,
 			   zebra_rec_keys_t reckeys,
                            const char *fname, const char *spec)
@@ -216,8 +219,8 @@ static char *fileMatchStr (ZebraHandle zh,
 
     while (1)
     {
-        while (*s == ' ' || *s == '\t')
-            s++;
+	for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
+	    ;
         if (!*s)
             break;
         if (*s == '(')
@@ -226,21 +229,26 @@ static char *fileMatchStr (ZebraHandle zh,
 	    char attset_str[64], attname_str[64];
 	    data1_attset *attset;
 	    int i;
-            char matchFlag[32];
             int attSet = 1, attUse = 1;
             int first = 1;
-
-            s++;
-	    for (i = 0; *s && *s != ',' && *s != ')'; s++)
-		if (i < 63)
+	    
+	    for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
+		;
+	    for (i = 0; *s && *s != ',' && *s != ')' && 
+		     !strchr(FILE_MATCH_BLANK, *s); s++)
+		if (i+1 < sizeof(attset_str))
 		    attset_str[i++] = *s;
 	    attset_str[i] = '\0';
-
+	    
+	    for (; strchr(FILE_MATCH_BLANK, *s); s++)
+		;
 	    if (*s == ',')
 	    {
-		s++;
-		for (i = 0; *s && *s != ')'; s++)
-		    if (i < 63)
+		for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
+		    ;
+		for (i = 0; *s && *s != ')' && 
+			 !strchr(FILE_MATCH_BLANK, *s); s++)
+		    if (i+1 < sizeof(attname_str))
 			attname_str[i++] = *s;
 		attname_str[i] = '\0';
 	    }
@@ -257,12 +265,7 @@ static char *fileMatchStr (ZebraHandle zh,
 	    }
             searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
 
-            if (*s == ')')
-            {
-                for (i = 0; i<32; i++)
-                    matchFlag[i] = 1;
-            }
-            else
+            if (*s != ')')
             {
                 yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
                       spec, zh->m_group ? zh->m_group : "none");
@@ -271,7 +274,7 @@ static char *fileMatchStr (ZebraHandle zh,
             s++;
 
             for (i = 0; i<32; i++)
-                if (matchFlag[i] && ws[i])
+                if (ws[i])
                 {
                     if (first)
                     {
@@ -294,12 +297,12 @@ static char *fileMatchStr (ZebraHandle zh,
             char special[64];
             const char *spec_src = NULL;
             const char *s1 = ++s;
-            while (*s1 && *s1 != ' ' && *s1 != '\t')
+            while (*s1 && !strchr(FILE_MATCH_BLANK, *s1))
                 s1++;
 
             spec_len = s1 - s;
-            if (spec_len > 63)
-                spec_len = 63;
+            if (spec_len > sizeof(special)-1)
+                spec_len = sizeof(special)-1;
             memcpy (special, s, spec_len);
             special[spec_len] = '\0';
             s = s1;
@@ -329,7 +332,7 @@ static char *fileMatchStr (ZebraHandle zh,
 
             while (*s && *s != stopMarker)
             {
-                if (i < 63)
+                if (i+1 < sizeof(tmpString))
                     tmpString[i++] = *s++;
             }
             if (*s)
@@ -490,7 +493,9 @@ static int file_extract_record(ZebraHandle zh,
 	}
 	if (matchStr)
 	{
-            char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+            char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
+					  matchStr);
 	    if (rinfo)
 	    {
 		assert(*rinfo == sizeof(*sysno));
@@ -533,7 +538,9 @@ static int file_extract_record(ZebraHandle zh,
 
         if (matchStr)
         {
-            dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+            dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
+			    sizeof(*sysno), sysno);
         }
 #if NATTR
 	extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
@@ -595,7 +602,10 @@ static int file_extract_record(ZebraHandle zh,
 			     zh->m_record_type, fname, recordOffset);
                 zh->records_deleted++;
                 if (matchStr)
-                    dict_delete (zh->reg->matchDict, matchStr);
+		{
+		    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+                    dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
+		}
                 rec_del (zh->reg->records, &rec);
             }
 	    rec_rm (&rec);
@@ -964,8 +974,11 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh,
                 }
             }
         }
-        if (matchStr) {
-	    char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+        if (matchStr) 
+	{
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+	    char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
+					  matchStr);
             if (rinfo)
 	    {
 		assert(*rinfo == sizeof(*sysno));
@@ -1003,8 +1016,9 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh,
 
         if (matchStr)
         {
-            dict_insert (zh->reg->matchDict, matchStr,
-                         sizeof(*sysno), sysno);
+	    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+            dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
+			    sizeof(*sysno), sysno);
         }
 #if NATTR
 	extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
@@ -1080,7 +1094,10 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh,
 			     pr_fname, (long) recordOffset);
                 zh->records_deleted++;
                 if (matchStr)
-                    dict_delete (zh->reg->matchDict, matchStr);
+		{
+		    int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+                    dict_delete_ord(zh->reg->matchDict, db_ord, matchStr);
+		}
                 rec_del (zh->reg->records, &rec);
             }
 	    rec_rm (&rec);
@@ -1355,6 +1372,12 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
 
 	    if (zh->m_staticrank) /* rank config enabled ? */
 	    {
+		if (staticrank < 0)
+		{
+		    yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0",
+			    (long) staticrank);
+		    staticrank = 0;
+		}
 		*keyp++ = staticrank;
 		key_out.len = 4;
 	    }
@@ -1932,6 +1955,7 @@ void encode_key_init (struct encode_info *i)
     i->prevcmd=-1;
     i->keylen=0;
     i->encode_handle = iscz1_start();
+    i->decode_handle = iscz1_start();
 }
 
 #define OLDENCODE 1
@@ -1953,19 +1977,42 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf)
     /* and copy & align key so we can mangle */
     memcpy (&key, k+1, sizeof(struct it_key));  /* *k is insert/delete */
 
+#if 0
+    /* debugging */
+    key_logdump_txt(YLOG_LOG, &key, *k ? "i" : "d");
+#endif
+    assert(key.mem[0] >= 0);
+
     bp0 = bp++;
     iscz1_encode(i->encode_handle, &bp, &src);
+
     *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */
     if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
     {
         yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
         exit (1);
     }
+
+#if 0
+    /* debugging */
+    if (1)
+    {
+	struct it_key key2;
+	const char *src = bp0+1;
+	char *dst = (char*) &key2;
+	iscz1_decode(i->decode_handle, &dst, &src);
+
+	key_logdump_txt(YLOG_LOG, &key2, *k ? "i" : "d");
+
+	assert(key2.mem[1]);
+    }
+#endif
 }
 
 void encode_key_flush (struct encode_info *i, FILE *outf)
-{ /* dummy routine */
+{ 
     iscz1_stop(i->encode_handle);
+    iscz1_stop(i->decode_handle);
 }
 
 #else