X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=e06b9ae9287bbfb3d7abe85457195095ce22569e;hb=3042550d9340f8f9e3058702240e2cd376aa33fa;hp=f92ca7f932421d38a6640e8b9a72aad576a1d06f;hpb=c6959870998f868e6a0e9201739fb54aef696bc6;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index f92ca7f..e06b9ae 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,5 +1,5 @@
-/* $Id: extract.c,v 1.200 2005-12-09 10:45:04 adam Exp $
-   Copyright (C) 1995-2005
+/* $Id: extract.c,v 1.212 2006-05-10 14:13:45 adam Exp $
+   Copyright (C) 1995-2006
    Index Data ApS
 
 This file is part of the Zebra server.
@@ -32,15 +32,23 @@ 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>
 
-#if _FILE_OFFSET_BITS == 64
-#define PRINTF_OFF_T "%Ld"
+#ifdef WIN32
+#define PRINTF_OFF_T "%I64d"
+#else
+/* !WIN32 */
+#if SIZEOF_OFF_T == SIZEOF_LONG_LONG
+#define PRINTF_OFF_T "%lld"
 #else
 #define PRINTF_OFF_T "%ld"
 #endif
 
+#endif
+
+
 #define USE_SHELLSORT 0
 
 #if USE_SHELLSORT
@@ -206,6 +214,8 @@ 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 +226,8 @@ 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 +236,26 @@ 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 +272,7 @@ 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 +281,7 @@ static char *fileMatchStr (ZebraHandle zh,
             s++;
 
             for (i = 0; i<32; i++)
-                if (matchFlag[i] && ws[i])
+                if (ws[i])
                 {
                     if (first)
                     {
@@ -294,12 +304,12 @@ 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 +339,7 @@ static char *fileMatchStr (ZebraHandle zh,
 
             while (*s && *s != stopMarker)
             {
-                if (i < 63)
+                if (i+1 < sizeof(tmpString))
                     tmpString[i++] = *s++;
             }
             if (*s)
@@ -376,14 +386,15 @@ 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)
 {
+    const char *match_str_to_print = "";
     RecordAttr *recordAttr;
     int r;
     const char *matchStr = 0;
@@ -397,7 +408,7 @@ 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 +455,7 @@ 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 +465,7 @@ 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)
 	{
@@ -466,13 +477,19 @@ 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;	
     }
 
-    /* perform match if sysno not known and if match criteria is specified */
+    /* if matchStr is set now - we assume it's printable .
+       For internal matchStr (see below) we don't print */
+    if (matchStr)
+        match_str_to_print = matchStr;
+
+    /* perform internal match if sysno not known and if match criteria is
+       specified already */
     if (!sysno) 
     {
         sysnotmp = 0;
@@ -485,12 +502,14 @@ 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 +522,12 @@ 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 +538,31 @@ 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)
+        {
+            yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T 
+                    " " ZINT_FORMAT " %s" ,
+                    zh->m_record_type,
+                    fname, recordOffset, *sysno, match_str_to_print);
+        }
 	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 +612,42 @@ 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);
+                {
+                    yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T 
+                            " " ZINT_FORMAT " %s" ,
+                            zh->m_record_type,
+                            fname, recordOffset, *sysno, match_str_to_print);
+                }
                 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);
+            {
+                yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T 
+                        " " ZINT_FORMAT " %s" ,
+                        zh->m_record_type,
+                        fname, recordOffset, *sysno, match_str_to_print);
+            }
 	    recordAttr->staticrank = extractCtrl.staticrank;
 #if NATTR
             extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
@@ -717,13 +757,14 @@ 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 +815,7 @@ 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 +840,15 @@ 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 +859,15 @@ 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);
@@ -911,7 +958,7 @@ 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;
     }
     
@@ -964,8 +1011,11 @@ 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));
@@ -1003,9 +1053,12 @@ 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
@@ -1080,7 +1133,10 @@ 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);
@@ -1291,6 +1347,58 @@ int explain_extract (void *handle, Record rec, data1_node *n)
     return 0;
 }
 
+void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
+                             zebra_rec_keys_t reckeys)
+{
+    ZebraExplainInfo zei = zh->reg->zei;
+    struct ord_stat {
+        int no;
+        int ord;
+        struct ord_stat *next;
+    };
+
+    if (zebra_rec_keys_rewind(reckeys))
+    {
+        struct ord_stat *ord_list = 0;
+        struct ord_stat *p;
+	size_t slen;
+	const char *str;
+	struct it_key key_in;
+	while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
+        {
+            int ord = key_in.mem[0]; 
+
+            for (p = ord_list; p ; p = p->next)
+                if (p->ord == ord)
+                {
+                    p->no++;
+                    break;
+                }
+            if (!p)
+            {
+                p = xmalloc(sizeof(*p));
+                p->no = 1;
+                p->ord = ord;
+                p->next = ord_list;
+                ord_list = p;
+            }
+        }
+
+        p = ord_list;
+        while (p)
+        {
+            struct ord_stat *p1 = p;
+
+            if (is_insert)
+                zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
+            else
+                zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
+            p = p->next;
+            xfree(p1);
+        }
+    }
+}
+
 void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
                               int cmd,
 			      zebra_rec_keys_t reckeys,
@@ -1298,6 +1406,8 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
 {
     ZebraExplainInfo zei = zh->reg->zei;
 
+    extract_rec_keys_adjust(zh, cmd, reckeys);
+
     if (!zh->reg->key_buf)
     {
 	int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
@@ -1344,7 +1454,7 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
 	    zh->reg->key_buf_used +=
 		key_SU_encode(ch, (char*)zh->reg->key_buf +
 			      zh->reg->key_buf_used);
-	    
+
 	    /* copy the 0-terminated stuff from str to output */
 	    memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
 	    zh->reg->key_buf_used += slen;
@@ -1355,6 +1465,12 @@ 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;
 	    }
@@ -1530,7 +1646,8 @@ 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);
@@ -1559,7 +1676,7 @@ 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];
 	    
@@ -1932,6 +2049,7 @@ 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 +2071,42 @@ 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
@@ -2069,3 +2210,11 @@ void encode_key_flush (struct encode_info *i, FILE *outf)
     i->prevseq=0;
 }
 #endif
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+