X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=10ff2bfd21c07fa568311b1d92ac136068cacadf;hb=3168ab325caf86d3bac38c33f7d3b701d1110539;hp=45fde5d7c362d8c7af3b93eb8e59a592033c7274;hpb=52824e358151328181178232866be5766dcfce6c;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 45fde5d..10ff2bf 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,4 +1,4 @@
-/* $Id: extract.c,v 1.199 2005-11-10 11:25:47 adam Exp $
+/* $Id: extract.c,v 1.208 2006-05-09 11:31:37 marc 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)
@@ -376,13 +379,13 @@ static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
     ctrl->flagShowRecords = !zh->m_flag_rw;
 }
 
-static int file_extract_record(ZebraHandle zh,
-			       SYSNO *sysno, const char *fname,
-			       int deleteFlag,
-			       struct file_read_info *fi,
-			       int force_update,
-			       RecType recType,
-			       void *recTypeClientData)
+static ZEBRA_RES file_extract_record(ZebraHandle zh,
+				     SYSNO *sysno, const char *fname,
+				     int deleteFlag,
+				     struct file_read_info *fi,
+				     int force_update,
+				     RecType recType,
+				     void *recTypeClientData)
 {
     RecordAttr *recordAttr;
     int r;
@@ -397,7 +400,7 @@ static int file_extract_record(ZebraHandle zh,
     {
         if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0],
 				      zh->m_explain_database))
-	    return 0;
+	    return ZEBRA_FAIL;
     }
 
     if (fi->fd != -1)
@@ -444,7 +447,7 @@ static int file_extract_record(ZebraHandle zh,
 
         yaz_log_init_prefix2 (0);
 	if (r == RECCTRL_EXTRACT_EOF)
-	    return 0;
+	    return ZEBRA_FAIL;
 	else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
 	{
             /* error occured during extraction ... */
@@ -454,7 +457,7 @@ static int file_extract_record(ZebraHandle zh,
                 yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
                       fname, recordOffset);
             }
-            return 0;
+            return ZEBRA_FAIL;
         }
 	else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
 	{
@@ -466,7 +469,7 @@ static int file_extract_record(ZebraHandle zh,
                       PRINTF_OFF_T, zh->m_record_type,
                       fname, recordOffset);
             }
-            return 0;
+            return ZEBRA_FAIL;
         }
         if (extractCtrl.match_criteria[0])
             matchStr = extractCtrl.match_criteria;	
@@ -485,12 +488,14 @@ static int file_extract_record(ZebraHandle zh,
 	    if (!matchStr)
 	    {
 		yaz_log(YLOG_WARN, "Bad match criteria");
-		return 0;
+		return ZEBRA_FAIL;
 	    }
 	}
 	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));
@@ -503,12 +508,12 @@ static int file_extract_record(ZebraHandle zh,
          /* the extraction process returned no information - the record
             is probably empty - unless flagShowRecords is in use */
          if (!zh->m_flag_rw)
-             return 1;
+             return ZEBRA_OK;
   
          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;
+         return ZEBRA_OK;
     }
 
     if (! *sysno)
@@ -519,22 +524,36 @@ static int file_extract_record(ZebraHandle zh,
 	    yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
 		  fname, recordOffset);
             yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
-            return 1;
+            return ZEBRA_OK;
         }
-        if (zh->records_processed < zh->m_file_verbose_limit)
-            yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
-                  fname, recordOffset);
-        rec = rec_new (zh->reg->records);
 
+        rec = rec_new (zh->reg->records);
+        
         *sysno = rec->sysno;
-
+        
+        if (zh->records_processed < zh->m_file_verbose_limit)
+          if (matchStr)
+            yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T 
+                    " " ZINT_FORMAT " %s" ,
+                    zh->m_record_type,
+                    fname, recordOffset, *sysno, matchStr);
+          else
+            yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T 
+                    " " ZINT_FORMAT , 
+                    zh->m_record_type,
+                    fname, recordOffset, *sysno);
+        
 	recordAttr = rec_init_attr (zh->reg->zei, rec);
 	recordAttr->staticrank = extractCtrl.staticrank;
 
         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);
 #else
@@ -584,30 +603,54 @@ static int file_extract_record(ZebraHandle zh,
             /* record going to be deleted */
             if (zebra_rec_keys_empty(delkeys))
             {
-                yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
-			 zh->m_record_type, fname, recordOffset);
+                yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T 
+                         " " ZINT_FORMAT,
+			 zh->m_record_type, fname, recordOffset, *sysno);
                 yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false (1)");
             }
             else
             {
                 if (zh->records_processed < zh->m_file_verbose_limit)
-                    yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
-			     zh->m_record_type, fname, recordOffset);
+                  if (matchStr)
+                    yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T 
+                            " " ZINT_FORMAT " %s" ,
+                            zh->m_record_type,
+                            fname, recordOffset, *sysno, matchStr);
+                  else
+                    yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T 
+                            " " ZINT_FORMAT , 
+                            zh->m_record_type,
+                            fname, recordOffset, *sysno);
+
+
+
                 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);
             logRecord (zh);
-            return 1;
+            return ZEBRA_OK;
         }
         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);
+                  if (matchStr)
+                    yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T 
+                            " " ZINT_FORMAT " %s" ,
+                            zh->m_record_type,
+                            fname, recordOffset, *sysno, matchStr);
+                  else
+                    yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T 
+                            " " ZINT_FORMAT , 
+                            zh->m_record_type,
+                            fname, recordOffset, *sysno);
+
 	    recordAttr->staticrank = extractCtrl.staticrank;
 #if NATTR
             extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
@@ -717,13 +760,14 @@ static int file_extract_record(ZebraHandle zh,
     /* commit this record */
     rec_put (zh->reg->records, &rec);
     logRecord (zh);
-    return 1;
+    return ZEBRA_OK;
 }
 
-int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, 
-		 int deleteFlag)
+ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, 
+			     int deleteFlag)
 {
-    int r, i, fd;
+    ZEBRA_RES r = ZEBRA_OK;
+    int i, fd;
     char gprefix[128];
     char ext[128];
     char ext_res[128];
@@ -774,7 +818,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
 			  &recTypeClientData)))
     {
         yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type);
-        return 0;
+        return ZEBRA_FAIL;
     }
 
     switch(recType->version)
@@ -799,16 +843,15 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
         else
             strcpy (full_rep, fname);
         
-
         if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
         {
             yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
 	    zh->m_record_type = original_record_type;
-            return 0;
+            return ZEBRA_FAIL;
         }
     }
     fi = file_read_start (fd);
-    do
+    while(1)
     {
 	fi->file_moffset = fi->file_offset;
 	fi->file_more = 0;  /* file_end not called (yet) */
@@ -819,8 +862,15 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname,
 	    fi->file_offset = fi->file_moffset;
 	    lseek(fi->fd, fi->file_moffset, SEEK_SET);
 	}
+	if (r != ZEBRA_OK)
+	{
+	    break;
+	}
+	if (sysno)
+	{
+	    break;
+	}
     }
-    while (r && !sysno);
     file_read_stop (fi);
     if (fd != -1)
         close (fd);
@@ -845,6 +895,7 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh,
 				int force_update,
 				int allow_update)
 {
+    SYSNO sysno0 = 0;
     RecordAttr *recordAttr;
     struct recExtractCtrl extractCtrl;
     int r;
@@ -910,7 +961,7 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh,
     
     if (!recType)
     {
-        yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
+        yaz_log (YLOG_WARN, "No such record type: %s", recordType);
         return ZEBRA_FAIL;
     }
     
@@ -942,14 +993,14 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh,
 	yaz_log (YLOG_WARN, "extract error: no such filter");
 	return ZEBRA_FAIL;
     }
-    /* match criteria */
-    matchStr = NULL;
 
     if (extractCtrl.match_criteria[0])
 	match_criteria = extractCtrl.match_criteria;
 
-    if (! *sysno) {
-        char *rinfo;
+    if (!sysno) {
+
+	sysno = &sysno0;
+
         if (match_criteria && *match_criteria) {
             matchStr = match_criteria;
         } else {
@@ -963,8 +1014,11 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh,
                 }
             }
         }
-        if (matchStr) {
-            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));
@@ -1002,9 +1056,12 @@ 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);
 #else
@@ -1079,7 +1136,10 @@ 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);
@@ -1354,6 +1414,12 @@ 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;
 	    }
@@ -1529,7 +1595,8 @@ ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
 	    ord = key.mem[0];
 	    
 	    zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
-				    0/* db */, 0/* set */, 0/* use */);
+				    0/* db */, 0/* set */, 0/* use */,
+				    0 /* string_index */);
 	    assert(index_type);
 	    zebra_term_untrans_iconv(zh, nmem, index_type,
 				     &dst_term, str);
@@ -1558,7 +1625,7 @@ void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
 	    assert(key.len <= 4 && key.len > 2);
 
 	    zebraExplain_lookup_ord(zh->reg->zei,
-				    key.mem[0], &index_type, &db, 0, 0);
+				    key.mem[0], &index_type, &db, 0, 0, 0);
 	    
 	    seqno = (int) key.mem[key.len-1];
 	    
@@ -1931,6 +1998,7 @@ 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
@@ -1952,19 +2020,42 @@ 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