X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=4426e4afd4678ce3728bb16e8aa50b06048ff1fb;hb=e39158d4c6147865c129d2524f1b910e4e7921ed;hp=a7169f1d18400193a962160af100a29fd09131e2;hpb=d8728143be0a28eb22a93d91c4ba7d3ab6c3f6cb;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index a7169f1..4426e4a 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,109 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.11  1995-09-28 12:10:31  adam
+ * Revision 1.41  1995-12-06 16:06:42  adam
+ * Better diagnostics. Work on 'real' dictionary deletion.
+ *
+ * Revision 1.40  1995/12/05  16:57:40  adam
+ * More work on regular patterns.
+ *
+ * Revision 1.39  1995/12/05  13:20:18  adam
+ * Bug fix: file_read sometimes returned early EOF.
+ *
+ * Revision 1.38  1995/12/04  17:59:21  adam
+ * More work on regular expression conversion.
+ *
+ * Revision 1.37  1995/12/04  14:22:27  adam
+ * Extra arg to recType_byName.
+ * Started work on new regular expression parsed input to
+ * structured records.
+ *
+ * Revision 1.36  1995/11/30  08:34:29  adam
+ * Started work on commit facility.
+ * Changed a few malloc/free to xmalloc/xfree.
+ *
+ * Revision 1.35  1995/11/28  14:26:21  adam
+ * Bug fix: recordId with constant wasn't right.
+ * Bug fix: recordId dictionary entry wasn't deleted when needed.
+ *
+ * Revision 1.34  1995/11/28  09:09:38  adam
+ * Zebra config renamed.
+ * Use setting 'recordId' to identify record now.
+ * Bug fix in recindex.c: rec_release_blocks was invokeded even
+ * though the blocks were already released.
+ * File traversal properly deletes records when needed.
+ *
+ * Revision 1.33  1995/11/27  09:56:20  adam
+ * Record info elements better enumerated. Internal store of records.
+ *
+ * Revision 1.32  1995/11/25  10:24:05  adam
+ * More record fields - they are enumerated now.
+ * New options: flagStoreData flagStoreKey.
+ *
+ * Revision 1.31  1995/11/24  11:31:35  adam
+ * Commands add & del read filenames from stdin if source directory is
+ * empty.
+ * Match criteria supports 'constant' strings.
+ *
+ * Revision 1.30  1995/11/22  17:19:16  adam
+ * Record management uses the bfile system.
+ *
+ * Revision 1.29  1995/11/21  15:01:14  adam
+ * New general match criteria implemented.
+ * New feature: document groups.
+ *
+ * Revision 1.28  1995/11/21  09:20:30  adam
+ * Yet more work on record match.
+ *
+ * Revision 1.27  1995/11/20  16:59:45  adam
+ * New update method: the 'old' keys are saved for each records.
+ *
+ * Revision 1.26  1995/11/20  11:56:24  adam
+ * Work on new traversal.
+ *
+ * Revision 1.25  1995/11/16  15:34:54  adam
+ * Uses new record management system in both indexer and server.
+ *
+ * Revision 1.24  1995/11/15  19:13:08  adam
+ * Work on record management.
+ *
+ * Revision 1.23  1995/10/27  14:00:10  adam
+ * Implemented detection of database availability.
+ *
+ * Revision 1.22  1995/10/17  18:02:07  adam
+ * New feature: databases. Implemented as prefix to words in dictionary.
+ *
+ * Revision 1.21  1995/10/10  12:24:38  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.20  1995/10/06  13:52:05  adam
+ * Bug fixes. Handler may abort further scanning.
+ *
+ * Revision 1.19  1995/10/04  12:55:16  adam
+ * Bug fix in ranked search. Use=Any keys inserted.
+ *
+ * Revision 1.18  1995/10/04  09:37:08  quinn
+ * Fixed bug.
+ *
+ * Revision 1.17  1995/10/03  14:28:57  adam
+ * Buffered read in extract works.
+ *
+ * Revision 1.16  1995/10/03  14:28:45  adam
+ * Work on more effecient read handler in extract.
+ *
+ * Revision 1.15  1995/10/02  15:42:53  adam
+ * Extract uses file descriptors instead of FILE pointers.
+ *
+ * Revision 1.14  1995/10/02  15:29:13  adam
+ * More logging in file_extract.
+ *
+ * Revision 1.13  1995/09/29  14:01:39  adam
+ * Bug fixes.
+ *
+ * Revision 1.12  1995/09/28  14:22:56  adam
+ * Sort uses smaller temporary files.
+ *
+ * Revision 1.11  1995/09/28  12:10:31  adam
  * Bug fixes. Field prefix used in queries.
  *
  * Revision 1.10  1995/09/28  09:19:41  adam
@@ -52,80 +154,162 @@
 #include <recctrl.h>
 #include "index.h"
 
-static Dict file_idx;
-static SYSNO sysno_next;
-static int key_fd = -1;
-static int sys_idx_fd = -1;
-static char *key_buf;
-static int key_offset, key_buf_size;
-static int key_cmd;
-static int key_sysno;
+#include "recindex.h"
+
+static Dict matchDict;
+
+static Records records = NULL;
 
-void key_open (const char *fname)
+static char **key_buf;
+static size_t ptr_top;
+static size_t ptr_i;
+static size_t key_buf_used;
+static int key_file_no;
+
+static int records_inserted = 0;
+static int records_updated = 0;
+static int records_deleted = 0;
+
+#define MATCH_DICT "match"
+
+void key_open (int mem)
 {
-    void *file_key;
-    if (key_fd != -1)
-        return;
-    if ((key_fd = open (fname, O_RDWR|O_CREAT, 0666)) == -1)
+    if (mem < 50000)
+        mem = 50000;
+    key_buf = xmalloc (mem);
+    ptr_top = mem/sizeof(char*);
+    ptr_i = 0;
+
+    key_buf_used = 0;
+    key_file_no = 0;
+
+    if (!(matchDict = dict_open (MATCH_DICT, 20, 1)))
     {
-        logf (LOG_FATAL|LOG_ERRNO, "open %s", fname);
+        logf (LOG_FATAL, "dict_open fail of %s", MATCH_DICT);
         exit (1);
     }
-    logf (LOG_DEBUG, "key_open of %s", fname);
-    key_buf_size = 49100;
-    key_buf = xmalloc (key_buf_size);
-    key_offset = 0;
-    if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
+    assert (!records);
+    records = rec_open (1);
+}
+
+struct encode_info {
+    int  sysno;
+    int  seqno;
+    char buf[512];
+};
+
+void encode_key_init (struct encode_info *i)
+{
+    i->sysno = 0;
+    i->seqno = 0;
+}
+
+char *encode_key_int (int d, char *bp)
+{
+    if (d <= 63)
+        *bp++ = d;
+    else if (d <= 16383)
+    {
+        *bp++ = 64 + (d>>8);
+        *bp++ = d  & 255;
+    }
+    else if (d <= 4194303)
     {
-        logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
-        exit (1);
+        *bp++ = 128 + (d>>16);
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
     }
-    file_key = dict_lookup (file_idx, ".");
-    if (file_key)
-        memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
     else
-        sysno_next = 1;
-    if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
     {
-        logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
-        exit (1);
+        *bp++ = 192 + (d>>24);
+        *bp++ = (d>>16) & 255;
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
     }
+    return bp;
 }
 
-int key_close (void)
+void encode_key_write (char *k, struct encode_info *i, FILE *outf)
 {
-    if (key_fd == -1)
+    struct it_key key;
+    char *bp = i->buf;
+
+    while ((*bp++ = *k++))
+        ;
+    memcpy (&key, k+1, sizeof(struct it_key));
+    bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
+    if (i->sysno != key.sysno)
     {
-        logf (LOG_DEBUG, "key_close - but no file");
-        return 0;
+        i->sysno = key.sysno;
+        i->seqno = 0;
+    }
+    bp = encode_key_int (key.seqno - i->seqno, bp);
+    i->seqno = key.seqno;
+    if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+        exit (1);
     }
-    close (key_fd);
-    close (sys_idx_fd);
-    dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
-    dict_close (file_idx);
-    key_fd = -1;
-    xfree (key_buf);
-    return 1;
 }
 
-void wordFlush (int sysno)
+void key_flush (void)
 {
-    size_t i = 0;
-    int w;
+    FILE *outf;
+    char out_fname[200];
+    char *prevcp, *cp;
+    struct encode_info encode_info;
+    
+    if (ptr_i <= 0)
+        return;
+
+    key_file_no++;
+    logf (LOG_LOG, "sorting section %d", key_file_no);
+    qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
+    sprintf (out_fname, TEMP_FNAME, key_file_no);
 
-    if (key_fd == -1)
-	return; 
-    while (i < key_offset)
+    if (!(outf = fopen (out_fname, "w")))
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
+        exit (1);
+    }
+    logf (LOG_LOG, "writing section %d", key_file_no);
+    prevcp = cp = key_buf[ptr_top-ptr_i];
+    
+    encode_key_init (&encode_info);
+    encode_key_write (cp, &encode_info, outf);
+    while (--ptr_i > 0)
     {
-        w = write (key_fd, key_buf + i, key_offset - i);
-        if (w == -1)
+        cp = key_buf[ptr_top-ptr_i];
+        if (strcmp (cp, prevcp))
         {
-            logf (LOG_FATAL|LOG_ERRNO, "Write key fail");
-            exit (1);
+            encode_key_init (&encode_info);
+            encode_key_write (cp, &encode_info, outf);
+            prevcp = cp;
         }
-        i += w;
+        else
+            encode_key_write (cp + strlen(cp), &encode_info, outf);
+    }
+    if (fclose (outf))
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
+        exit (1);
     }
-    key_offset = 0;
+    logf (LOG_LOG, "finished section %d", key_file_no);
+    ptr_i = 0;
+    key_buf_used = 0;
+}
+
+int key_close (void)
+{
+    key_flush ();
+    xfree (key_buf);
+    rec_close (&records);
+    dict_close (matchDict);
+
+    logf (LOG_LOG, "Records inserted %6d", records_inserted);
+    logf (LOG_LOG, "Records updated  %6d", records_updated);
+    logf (LOG_LOG, "Records deleted  %6d", records_deleted);
+    return key_file_no;
 }
 
 static void wordInit (RecWord *p)
@@ -135,57 +319,591 @@ static void wordInit (RecWord *p)
     p->which = Word_String;
 }
 
-static void wordAdd (const RecWord *p)
+struct recKeys {
+    int buf_used;
+    int buf_max;
+    char *buf;
+} reckeys;
+
+static void addRecordKey (const RecWord *p)
 {
-    struct it_key key;
-    char x;
+    char *dst;
+    char attrSet;
+    short attrUse;
     size_t i;
 
-    if (key_offset + 1000 > key_buf_size)
+    if (reckeys.buf_used+1024 > reckeys.buf_max)
     {
-        char *new_key_buf;
+        char *b;
 
-        key_buf_size *= 2;
-        new_key_buf = xmalloc (2*key_buf_size);
-        memcpy (new_key_buf, key_buf, key_offset);
-        xfree (key_buf);
-        key_buf = new_key_buf;
+        b = xmalloc (reckeys.buf_max += 65000);
+        if (reckeys.buf_used > 0)
+            memcpy (b, reckeys.buf, reckeys.buf_used);
+        xfree (reckeys.buf);
+        reckeys.buf = b;
     }
-    key_offset += index_word_prefix (key_buf + key_offset,
-                                     p->attrSet, p->attrUse);
+    dst = reckeys.buf + reckeys.buf_used;
     switch (p->which)
     {
     case Word_String:
+        attrSet = p->attrSet;
+        memcpy (dst, &attrSet, sizeof(attrSet));
+        dst += sizeof(attrSet);
+
+        attrUse = p->attrUse;
+        memcpy (dst, &attrUse, sizeof(attrUse));
+        dst += sizeof(attrUse);
+        
         for (i = 0; p->u.string[i]; i++)
-            key_buf[key_offset++] = index_char_cvt (p->u.string[i]);
-        key_buf[key_offset++] = '\0';
+            *dst++ = p->u.string[i];
+        *dst++ = '\0';
+
+        memcpy (dst, &p->seqno, sizeof(p->seqno));
+        dst += sizeof(p->seqno);
+
         break;
     default:
-        return ;
+        return;
+    }
+    reckeys.buf_used = dst - reckeys.buf;
+}
+
+static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys, 
+                             const char *databaseName)
+{
+    int off = 0;
+    while (off < reckeys->buf_used)
+    {
+        const char *src = reckeys->buf + off;
+        char attrSet;
+        short attrUse;
+        struct it_key key;
+        
+        memcpy (&attrSet, src, sizeof(attrSet));
+        src += sizeof(attrSet);
+
+        memcpy (&attrUse, src, sizeof(attrUse));
+        src += sizeof(attrUse);
+
+        if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
+            key_flush ();
+        ++ptr_i;
+        key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used;
+        key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used,
+                                           attrSet, attrUse, databaseName);
+        while (*src)
+            ((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++);
+        src++;
+        ((char*)key_buf) [key_buf_used++] = '\0';
+        
+        ((char*) key_buf)[key_buf_used++] = cmd;
+
+        memcpy (&key.seqno, src, sizeof(key.seqno));
+        src += sizeof(key.seqno);
+        key.sysno = sysno;
+        memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
+        key_buf_used += sizeof(key);
+        off = src - reckeys->buf;
+    }
+    assert (off == reckeys->buf_used);
+}
+
+static const char **searchRecordKey (struct recKeys *reckeys,
+                               int attrSetS, int attrUseS)
+{
+    static const char *ws[32];
+    int off = 0;
+    int startSeq = -1;
+    int i;
+
+    for (i = 0; i<32; i++)
+        ws[i] = NULL;
+    
+    while (off < reckeys->buf_used)
+    {
+        const char *src = reckeys->buf + off;
+        char attrSet;
+        short attrUse;
+        int seqno;
+        const char *wstart;
+        
+        memcpy (&attrSet, src, sizeof(attrSet));
+        src += sizeof(attrSet);
+
+        memcpy (&attrUse, src, sizeof(attrUse));
+        src += sizeof(attrUse);
+
+        wstart = src;
+        while (*src++)
+            ;
+
+        memcpy (&seqno, src, sizeof(seqno));
+        src += sizeof(seqno);
+
+#if 0
+        logf (LOG_LOG, "(%d,%d) %d %s", attrSet, attrUse, seqno, wstart);
+#endif
+        if (attrUseS == attrUse && attrSetS == attrSet)
+        {
+            int woff;
+
+
+            if (startSeq == -1)
+                startSeq = seqno;
+            woff = seqno - startSeq;
+            if (woff >= 0 && woff < 31)
+                ws[woff] = wstart;
+        }
+
+        off = src - reckeys->buf;
+    }
+    assert (off == reckeys->buf_used);
+    return ws;
+}
+
+static void addRecordKeyAny (const RecWord *p)
+{
+    if (p->attrSet != 1 || p->attrUse != 1016)
+    {
+        RecWord w;
+
+        memcpy (&w, p, sizeof(w));
+        w.attrSet = 1;
+        w.attrUse = 1016;
+        addRecordKey (&w);
+    }
+    addRecordKey (p);
+}
+
+
+#define FILE_READ_BUFSIZE 4096
+
+static int file_noread;
+#if FILE_READ_BUFSIZE
+static char *file_buf;
+static int file_offset;
+static int file_bufsize;
+#endif
+
+static void file_read_start (int fd)
+{
+    file_noread = 0;
+#if FILE_READ_BUFSIZE
+    file_offset = 0;
+    file_buf = xmalloc (FILE_READ_BUFSIZE);
+    file_bufsize = read (fd, file_buf, FILE_READ_BUFSIZE);
+#endif
+}
+
+static void file_read_stop (int fd)
+{
+#if FILE_READ_BUFSIZE
+    xfree (file_buf);
+    file_buf = NULL;
+#endif
+}
+
+static int file_read (int fd, char *buf, size_t count)
+{
+#if FILE_READ_BUFSIZE
+    int l = file_bufsize - file_offset;
+
+    if (count > l)
+    {
+        int r;
+        if (l > 0)
+            memcpy (buf, file_buf + file_offset, l);
+        count = count-l;
+        if (count > FILE_READ_BUFSIZE)
+        {
+            if ((r = read (fd, buf + l, count)) == -1)
+            {
+                logf (LOG_FATAL|LOG_ERRNO, "read");
+                exit (1);
+            }
+            file_bufsize = 0;
+            file_offset = 0;
+            file_noread += l+r;
+            return l+r;
+        }
+        file_bufsize = r = read (fd, file_buf, FILE_READ_BUFSIZE);
+        if (r == -1)
+        {
+            logf (LOG_FATAL|LOG_ERRNO, "read");
+            exit (1);
+        }
+        else if (r <= count)
+        {
+            file_offset = r;
+            memcpy (buf + l, file_buf, r);
+            file_noread += l+r;
+            return l+r;
+        }
+        else
+        {
+            file_offset = count;
+            memcpy (buf + l, file_buf, count - l);
+            file_noread += count;
+            return count;
+        }
     }
-    x = (key_cmd == 'a') ? 1 : 0;
-    memcpy (key_buf + key_offset, &x, 1);
-    key_offset++;
+    memcpy (buf, file_buf + file_offset, count);
+    file_offset += count;
+    file_noread += count;
+    return count;
+#else
+    int r;
+    r = read (fd, buf, count);
+    if (r > 0)
+        file_noread += r;
+    return r;
+#endif
+}
 
-    key.sysno = key_sysno;
-    key.seqno = p->seqno;
-    memcpy (key_buf + key_offset, &key, sizeof(key));
-    key_offset += sizeof(key);
+static int atois (const char **s)
+{
+    int val = 0, c;
+    while ( (c=**s) >= '0' && c <= '9')
+    {
+        val = val*10 + c - '0';
+        ++(*s);
+    }
+    return val;
 }
 
-void file_extract (int cmd, const char *fname, const char *kname)
+static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
+                           const char *fname,
+                           const char *spec)
 {
+    static char dstBuf[2048];
+    char *dst = dstBuf;
+    const char *s = spec;
+    static const char **w;
     int i;
+
+    while (1)
+    {
+        while (*s == ' ' || *s == '\t')
+            s++;
+        if (!*s)
+            break;
+        if (*s == '(')
+        {
+            char matchFlag[32];
+            int attrSet, attrUse;
+            int first = 1;
+
+            s++;
+            attrSet = atois (&s);
+            if (*s != ',')
+            {
+                logf (LOG_WARN, "Missing , in match criteria %s in group %s",
+                      spec, rGroup->groupName ? rGroup->groupName : "none");
+                return NULL;
+            }
+            s++;
+            attrUse = atois (&s);
+            w = searchRecordKey (reckeys, attrSet, attrUse);
+            assert (w);
+
+            if (*s == ')')
+            {
+                for (i = 0; i<32; i++)
+                    matchFlag[i] = 1;
+            }
+            else
+            {
+                logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
+                      spec, rGroup->groupName ? rGroup->groupName : "none");
+                return NULL;
+            }
+            s++;
+
+            for (i = 0; i<32; i++)
+                if (matchFlag[i] && w[i])
+                {
+                    if (first)
+                    {
+                        *dst++ = ' ';
+                        first = 0;
+                    }
+                    strcpy (dst, w[i]);
+                    dst += strlen(w[i]);
+                }
+            if (first)
+            {
+                logf (LOG_WARN, "Record in file %s didn't contain match"
+                      " fields in (%d,%d)", fname, attrSet, attrUse);
+                return NULL;
+            }
+        }
+        else if (*s == '$')
+        {
+            int spec_len;
+            char special[64];
+            const char *spec_src = NULL;
+            const char *s1 = ++s;
+            while (*s1 && *s1 != ' ' && *s1 != '\t')
+                s1++;
+
+            spec_len = s1 - s;
+            if (spec_len > 63)
+                spec_len = 63;
+            memcpy (special, s, spec_len);
+            special[spec_len] = '\0';
+            s = s1;
+
+            if (!strcmp (special, "group"))
+                spec_src = rGroup->groupName;
+            else if (!strcmp (special, "database"))
+                spec_src = rGroup->databaseName;
+            else if (!strcmp (special, "filename"))
+                spec_src = fname;
+            else if (!strcmp (special, "type"))
+                spec_src = rGroup->recordType;
+            else 
+                spec_src = NULL;
+            if (spec_src)
+            {
+                strcpy (dst, spec_src);
+                dst += strlen (spec_src);
+            }
+        }
+        else if (*s == '\"' || *s == '\'')
+        {
+            int stopMarker = *s++;
+            char tmpString[64];
+            int i = 0;
+
+            while (*s && *s != stopMarker)
+            {
+                if (i < 63)
+                    tmpString[i++] = *s++;
+            }
+            if (*s)
+                s++;
+            tmpString[i] = '\0';
+            strcpy (dst, tmpString);
+            dst += strlen (tmpString);
+        }
+        else
+        {
+            logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
+                  spec, rGroup->groupName ? rGroup->groupName : "none");
+            return NULL;
+        }
+        *dst++ = 1;
+    }
+    if (dst == dstBuf)
+    {
+        logf (LOG_WARN, "No match criteria for record %s in group %s",
+              fname, rGroup->groupName ? rGroup->groupName : "none");
+        return NULL;
+    }
+    return dstBuf;
+}
+
+static int recordExtract (SYSNO *sysno, const char *fname,
+                          struct recordGroup *rGroup, int deleteFlag,
+                          int fd, RecType recType, char *subType)
+{
+    struct recExtractCtrl extractCtrl;
+    int r;
+    char *matchStr;
+    SYSNO sysnotmp;
+    Record rec;
+
+    if (fd != -1)
+    {
+        extractCtrl.fd = fd;
+        /* extract keys */
+        extractCtrl.subType = subType;
+        extractCtrl.init = wordInit;
+        extractCtrl.add = addRecordKeyAny;
+
+        reckeys.buf_used = 0;
+        extractCtrl.readf = file_read;
+        r = (*recType->extract)(&extractCtrl);
+  
+        if (r)      
+        {
+            logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
+            return 0;
+        }
+    }
+
+    /* perform match if sysno not known and if match criteria is specified */
+       
+    matchStr = NULL;
+    if (!sysno) 
+    {
+        sysnotmp = 0;
+        sysno = &sysnotmp;
+        if (rGroup->recordId && *rGroup->recordId)
+        {
+            char *rinfo;
+        
+            matchStr = fileMatchStr (&reckeys, rGroup, fname, 
+                                     rGroup->recordId);
+            if (matchStr)
+            {
+                rinfo = dict_lookup (matchDict, matchStr);
+                if (rinfo)
+                    memcpy (sysno, rinfo+1, sizeof(*sysno));
+            }
+            else
+            {
+                logf (LOG_WARN, "Record not inserted");
+                return 0;
+            }
+        }
+    }
+
+    /* new record ? */
+    if (! *sysno)
+    {
+        if (deleteFlag)
+        {
+            logf (LOG_LOG, "? %s", fname);
+            return 1;
+        }
+        logf (LOG_LOG, "add %s %s", rGroup->recordType, fname);
+        rec = rec_new (records);
+        *sysno = rec->sysno;
+
+        if (matchStr)
+        {
+            dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
+        }
+        flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
+
+        records_inserted++;
+    }
+    else
+    {
+        struct recKeys delkeys;
+
+        rec = rec_get (records, *sysno);
+
+        delkeys.buf_used = rec->size[recInfo_delKeys];
+	delkeys.buf = rec->info[recInfo_delKeys];
+        flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]);
+        if (deleteFlag)
+        {
+            if (!delkeys.buf_used)
+            {
+                logf (LOG_WARN, "cannot delete %s: storeKeys false",
+                      fname);
+            }
+            else
+            {
+                SYSNO sysnoz = 0;
+                logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname);
+                records_deleted++;
+                if (matchStr)
+                    dict_insert (matchDict, matchStr, sizeof(sysnoz), &sysnoz);
+                rec_del (records, &rec);
+            }
+            return 1;
+        }
+        else
+        {
+            if (!delkeys.buf_used)
+            {
+                logf (LOG_WARN, "cannot update %s: storeKeys false",
+                      fname);
+            }
+            else
+            {
+                logf (LOG_LOG, "update %s %s", rGroup->recordType,
+                      fname);
+                flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); 
+                records_updated++;
+            }
+        }
+    }
+    xfree (rec->info[recInfo_fileType]);
+    rec->info[recInfo_fileType] =
+        rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
+
+    xfree (rec->info[recInfo_filename]);
+    rec->info[recInfo_filename] =
+        rec_strdup (fname, &rec->size[recInfo_filename]);
+
+    xfree (rec->info[recInfo_delKeys]);
+    if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
+    {
+        rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
+        rec->size[recInfo_delKeys] = reckeys.buf_used;
+        memcpy (rec->info[recInfo_delKeys], reckeys.buf,
+                rec->size[recInfo_delKeys]);
+    }
+    else
+    {
+        rec->info[recInfo_delKeys] = NULL;
+        rec->size[recInfo_delKeys] = 0;
+    }
+
+    xfree (rec->info[recInfo_storeData]);
+    if (rGroup->flagStoreData == 1)
+    {
+        rec->size[recInfo_storeData] = file_noread;
+        rec->info[recInfo_storeData] = xmalloc (file_noread);
+#if FILE_READ_BUFSIZE
+        if (file_noread < FILE_READ_BUFSIZE)
+	    memcpy (rec->info[recInfo_storeData], file_buf, file_noread);
+        else
+#endif
+        {
+            if (lseek (fd, 0L, SEEK_SET) < 0)
+            {
+                logf (LOG_ERRNO|LOG_FATAL, "seek to 0 in %s", fname);
+                exit (1);
+            }
+            if (read (fd, rec->info[recInfo_storeData], file_noread) 
+                < file_noread)
+            {
+                logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
+                      file_noread, fname);
+                exit (1);
+            }
+        }
+    }
+    else
+    {
+        rec->info[recInfo_storeData] = NULL;
+        rec->size[recInfo_storeData] = 0;
+    }
+    xfree (rec->info[recInfo_databaseName]);
+    rec->info[recInfo_databaseName] =
+        rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]); 
+
+    rec_put (records, &rec);
+    return 1;
+}
+
+int fileExtract (SYSNO *sysno, const char *fname, 
+                 const struct recordGroup *rGroupP, int deleteFlag)
+{
+    int i, fd;
+    char gprefix[128];
     char ext[128];
-    SYSNO sysno;
     char ext_res[128];
-    const char *file_type;
-    void *file_info;
-    FILE *inf;
-    struct recExtractCtrl extractCtrl;
-    RecType rt;
+    char subType[128];
+    RecType recType;
+    struct recordGroup rGroupM;
+    struct recordGroup *rGroup = &rGroupM;
 
-    logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
+    memcpy (rGroup, rGroupP, sizeof(*rGroupP));
+   
+    if (!rGroup->groupName || !*rGroup->groupName)
+        *gprefix = '\0';
+    else
+        sprintf (gprefix, "%s.", rGroup->groupName);
+
+    logf (LOG_DEBUG, "fileExtract %s", fname);
+
+    /* determine file extension */
     for (i = strlen(fname); --i >= 0; )
         if (fname[i] == '/')
         {
@@ -197,37 +915,97 @@ void file_extract (int cmd, const char *fname, const char *kname)
             strcpy (ext, fname+i+1);
             break;
         }
-    sprintf (ext_res, "fileExtension.%s", ext);
-    if (!(file_type = res_get (common_resource, ext_res)))
-        return;
-    if (!(rt = recType_byName (file_type)))
-        return;
-    file_info = dict_lookup (file_idx, kname);
-    if (!file_info)
+    /* determine file type - depending on extension */
+    if (!rGroup->recordType)
     {
-        sysno = sysno_next++;
-        dict_insert (file_idx, kname, sizeof(sysno), &sysno);
-        lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
-        write (sys_idx_fd, file_type, strlen (file_type)+1);
-        write (sys_idx_fd, kname, strlen(kname)+1);
+        sprintf (ext_res, "%srecordType.%s", gprefix, ext);
+        if (!(rGroup->recordType = res_get (common_resource, ext_res)))
+        {
+            sprintf (ext_res, "%srecordType", gprefix);
+            if (!(rGroup->recordType = res_get (common_resource, ext_res)))
+            {
+                logf (LOG_LOG, "? %s", fname);
+                return 0;
+            }
+        }
+    }
+    if (!rGroup->recordType)
+    {
+        logf (LOG_LOG, "? record %s", fname);
+        return 0;
+    }
+    if (!(recType = recType_byName (rGroup->recordType, subType)))
+    {
+        logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
+        return 0;
     }
-    else
-        memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
 
-    if (!(inf = fopen (fname, "r")))
+    /* determine match criteria */
+    if (!rGroup->recordId)
     {
-        logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
-        return;
+        sprintf (ext_res, "%srecordId.%s", gprefix, ext);
+        rGroup->recordId = res_get (common_resource, ext_res);
     }
-    extractCtrl.inf = inf;
-    extractCtrl.subType = "";
-    extractCtrl.init = wordInit;
-    extractCtrl.add = wordAdd;
-    key_sysno = sysno;
-    key_cmd = cmd;
-    (*rt->extract)(&extractCtrl);
-    fclose (inf);
-    wordFlush (sysno);
-}
 
+    /* determine database name */
+    if (!rGroup->databaseName)
+    {
+        sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
+        if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
+        {
+            sprintf (ext_res, "%sdatabase", gprefix);
+            rGroup->databaseName = res_get (common_resource, ext_res);
+        }
+    }
+    if (!rGroup->databaseName)
+        rGroup->databaseName = "Default";
+
+    if (rGroup->flagStoreData == -1)
+    {
+        const char *sval;
+        sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
+        if (!(sval = res_get (common_resource, ext_res)))
+        {
+            sprintf (ext_res, "%sstoreData", gprefix);
+            sval = res_get (common_resource, ext_res);
+        }
+        if (sval)
+            rGroup->flagStoreData = atoi (sval);
+    }
+    if (rGroup->flagStoreData == -1)
+        rGroup->flagStoreData = 0;
+
+    if (rGroup->flagStoreKeys == -1)
+    {
+        const char *sval;
+
+        sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
+        if (!(sval = res_get (common_resource, ext_res)))
+        {
+            sprintf (ext_res, "%sstoreKeys", gprefix);
+            sval = res_get (common_resource, ext_res);
+        }
+        if (sval)
+            rGroup->flagStoreKeys = atoi (sval);
+    }
+    if (rGroup->flagStoreKeys == -1)
+        rGroup->flagStoreKeys = 0;
+
+    if (sysno && deleteFlag)
+        fd = -1;
+    else
+    {
+        if ((fd = open (fname, O_RDONLY)) == -1)
+        {
+            logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
+            return 0;
+        }
+    }
+    file_read_start (fd);
+    recordExtract (sysno, fname, rGroup, deleteFlag, fd, recType, subType);
+    file_read_stop (fd);
+    if (fd != -1)
+        close (fd);
+    return 1;
+}