X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=be70f55e7ed43cf86f521e63b25737a781097315;hb=3d4f0f1ddbc8b1ec6df244bb400184f086e211c8;hp=a14b7f50f4724fb781a5cfe248b8bf769ec669f0;hpb=d57bb162b65a585e0ed3ad319d9673d66bc6d90c;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index a14b7f5..be70f55 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -1,10 +1,26 @@
-/*
- * Copyright (C) 1994-2002, Index Data 
- * All rights reserved.
- * Sebastian Hammer, Adam Dickmeiss
- *
- * $Id: extract.c,v 1.119 2002-05-07 11:05:19 adam Exp $
- */
+/* $Id: extract.c,v 1.128 2002-10-24 13:07:02 heikki Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+   Index Data Aps
+
+This file is part of the Zebra server.
+
+Zebra is free software; you can redistribute it and/or modify it under
+the terms of the GNU General Public License as published by the Free
+Software Foundation; either version 2, or (at your option) any later
+version.
+
+Zebra is distributed in the hope that it will be useful, but WITHOUT ANY
+WARRANTY; without even the implied warranty of MERCHANTABILITY or
+FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+for more details.
+
+You should have received a copy of the GNU General Public License
+along with Zebra; see the file LICENSE.zebra.  If not, write to the
+Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
+02111-1307, USA.
+*/
+
+
 #include <stdio.h>
 #include <assert.h>
 #ifdef WIN32
@@ -24,6 +40,9 @@
 #define PRINTF_OFF_T "%ld"
 #endif
 
+#define USE_SHELLSORT 0
+
+#if USE_SHELLSORT
 static void shellsort(void *ar, int r, size_t s,
                       int (*cmp)(const void *a, const void *b))
 {
@@ -46,7 +65,7 @@ static void shellsort(void *ar, int r, size_t s,
             memcpy (a+s*j, v, s);
         } 
 }
-
+#endif
 
 static void logRecord (ZebraHandle zh)
 {
@@ -395,19 +414,6 @@ struct recordLogInfo {
     struct recordGroup *rGroup;
 };
      
-static void recordLogPreamble (int level, const char *msg, void *info)
-{
-    struct recordLogInfo *p = (struct recordLogInfo *) info;
-    FILE *outf = yaz_log_file ();
-
-    if (level & LOG_LOG)
-        return ;
-    fprintf (outf, "File %s, offset %d, type %s\n",
-             p->rGroup->recordType, p->recordOffset, p->fname);
-    log_event_start (NULL, NULL);
-}
-
-
 static int recordExtract (ZebraHandle zh,
                           SYSNO *sysno, const char *fname,
                           struct recordGroup *rGroup, int deleteFlag,
@@ -419,7 +425,6 @@ static int recordExtract (ZebraHandle zh,
     char *matchStr;
     SYSNO sysnotmp;
     Record rec;
-    struct recordLogInfo logInfo;
     off_t recordOffset = 0;
 
     if (fi->fd != -1)
@@ -433,7 +438,9 @@ static int recordExtract (ZebraHandle zh,
 	zh->reg->keys.prevAttrUse = -1;
 	zh->reg->keys.prevAttrSet = -1;
 	zh->reg->keys.prevSeqNo = 0;
-	zh->reg->sortKeys = 0;
+	zh->reg->sortKeys.buf_used = 0;
+	zh->reg->sortKeys.buf_max = 0;
+	zh->reg->sortKeys.buf = 0;
 	
 	recordOffset = fi->file_moffset;
 	extractCtrl.offset = fi->file_moffset;
@@ -460,19 +467,19 @@ static int recordExtract (ZebraHandle zh,
 
         if (!rGroup->flagRw)
             printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset);
-
-        logInfo.fname = fname;
-        logInfo.recordOffset = recordOffset;
-        logInfo.rGroup = rGroup;
-        log_event_start (recordLogPreamble, &logInfo);
+        if (rGroup->flagRw)
+        {
+            char msg[512];
+            sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset);
+            yaz_log_init_prefix2 (msg);
+        }
 
         r = (*recType->extract)(clientData, &extractCtrl);
 
-        log_event_start (NULL, NULL);
-
+        yaz_log_init_prefix2 (0);
 	if (r == RECCTRL_EXTRACT_EOF)
 	    return 0;
-	else if (r == RECCTRL_EXTRACT_ERROR)
+	else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
 	{
             /* error occured during extraction ... */
             if (rGroup->flagRw &&
@@ -483,6 +490,18 @@ static int recordExtract (ZebraHandle zh,
             }
             return 0;
         }
+	else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
+	{
+            /* error occured during extraction ... */
+            if (rGroup->flagRw &&
+		zh->records_processed < rGroup->fileVerboseLimit)
+            {
+                logf (LOG_WARN, "no filter for %s %s " 
+                      PRINTF_OFF_T, rGroup->recordType,
+                      fname, recordOffset);
+            }
+            return 0;
+        }
         if (zh->reg->keys.buf_used == 0)
         {
             /* the extraction process returned no information - the record
@@ -546,8 +565,8 @@ static int recordExtract (ZebraHandle zh,
         {
             dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
         }
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
 	extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
 
         zh->records_inserted++;
     }
@@ -555,6 +574,7 @@ static int recordExtract (ZebraHandle zh,
     {
         /* record already exists */
         struct recKeys delkeys;
+        struct sortKeys sortKeys;
 
         rec = rec_get (zh->reg->records, *sysno);
         assert (rec);
@@ -574,7 +594,11 @@ static int recordExtract (ZebraHandle zh,
 	}
         delkeys.buf_used = rec->size[recInfo_delKeys];
 	delkeys.buf = rec->info[recInfo_delKeys];
-	extract_flushSortKeys (zh, *sysno, 0, &zh->reg->sortKeys);
+
+        sortKeys.buf_used = rec->size[recInfo_sortKeys];
+        sortKeys.buf = rec->info[recInfo_sortKeys];
+
+	extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
         extract_flushRecordKeys (zh, *sysno, 0, &delkeys);
         if (deleteFlag)
         {
@@ -613,6 +637,7 @@ static int recordExtract (ZebraHandle zh,
                 if (zh->records_processed < rGroup->fileVerboseLimit)
                     logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
                         rGroup->recordType, fname, recordOffset);
+                extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
                 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
                 zh->records_updated++;
             }
@@ -632,17 +657,10 @@ static int recordExtract (ZebraHandle zh,
     xfree (rec->info[recInfo_delKeys]);
     if (zh->reg->keys.buf_used > 0 && rGroup->flagStoreKeys == 1)
     {
-#if 1
         rec->size[recInfo_delKeys] = zh->reg->keys.buf_used;
         rec->info[recInfo_delKeys] = zh->reg->keys.buf;
         zh->reg->keys.buf = NULL;
         zh->reg->keys.buf_max = 0;
-#else
-        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]);
-#endif
     }
     else
     {
@@ -650,6 +668,14 @@ static int recordExtract (ZebraHandle zh,
         rec->size[recInfo_delKeys] = 0;
     }
 
+    /* update sort keys */
+    xfree (rec->info[recInfo_sortKeys]);
+
+    rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
+    rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
+    zh->reg->sortKeys.buf = NULL;
+    zh->reg->sortKeys.buf_max = 0;
+
     /* save file size of original record */
     zebraExplain_recordBytesIncrement (zh->reg->zei,
                                        - recordAttr->recordSize);
@@ -917,7 +943,7 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
     zh->reg->keys.prevAttrUse = -1;
     zh->reg->keys.prevAttrSet = -1;
     zh->reg->keys.prevSeqNo = 0;
-    zh->reg->sortKeys = 0;
+    zh->reg->sortKeys.buf_used = 0;
 
     extractCtrl.subType = subType;
     extractCtrl.init = extract_init;
@@ -939,19 +965,16 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
 
     if (r == RECCTRL_EXTRACT_EOF)
 	return 0;
-    else if (r == RECCTRL_EXTRACT_ERROR)
+    else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
     {
 	/* error occured during extraction ... */
-#if 1
-	yaz_log (LOG_WARN, "extract error");
-#else
-	if (rGroup->flagRw &&
-	    zh->records_processed < rGroup->fileVerboseLimit)
-	{
-	    logf (LOG_WARN, "fail %s %s %ld", rGroup->recordType,
-		  fname, (long) recordOffset);
-	}
-#endif
+	yaz_log (LOG_WARN, "extract error: generic");
+	return 0;
+    }
+    else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
+    {
+	/* error occured during extraction ... */
+	yaz_log (LOG_WARN, "extract error: no such filter");
 	return 0;
     }
     if (zh->reg->keys.buf_used == 0)
@@ -989,8 +1012,8 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
             dict_insert (zh->reg->matchDict, matchStr,
                          sizeof(*sysno), sysno);
         }
-        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
 	extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+        extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
     }
     else
     {
@@ -1049,6 +1072,7 @@ int extract_rec_in_mem (ZebraHandle zh, const char *recordType,
             {
 		logf (LOG_LOG, "update %s %s %ld", recordType,
 		      fname, (long) recordOffset);
+                extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
                 extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
             }
         }
@@ -1158,7 +1182,7 @@ int explain_extract (void *handle, Record rec, data1_node *n)
     zh->reg->keys.prevAttrUse = -1;
     zh->reg->keys.prevAttrSet = -1;
     zh->reg->keys.prevSeqNo = 0;
-    zh->reg->sortKeys = 0;
+    zh->reg->sortKeys.buf_used = 0;
     
     extractCtrl.init = extract_init;
     extractCtrl.tokenAdd = extract_token_add;
@@ -1175,11 +1199,15 @@ int explain_extract (void *handle, Record rec, data1_node *n)
     if (rec->size[recInfo_delKeys])
     {
 	struct recKeys delkeys;
-	struct sortKey *sortKeys = 0;
+	struct sortKeys sortkeys;
 
 	delkeys.buf_used = rec->size[recInfo_delKeys];
 	delkeys.buf = rec->info[recInfo_delKeys];
-	extract_flushSortKeys (zh, rec->sysno, 0, &sortKeys);
+
+	sortkeys.buf_used = rec->size[recInfo_sortKeys];
+	sortkeys.buf = rec->info[recInfo_sortKeys];
+
+	extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
 	extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys);
     }
     extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys);
@@ -1190,6 +1218,13 @@ int explain_extract (void *handle, Record rec, data1_node *n)
     rec->info[recInfo_delKeys] = zh->reg->keys.buf;
     zh->reg->keys.buf = NULL;
     zh->reg->keys.buf_max = 0;
+
+    xfree (rec->info[recInfo_sortKeys]);
+    rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
+    rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf;
+    zh->reg->sortKeys.buf = NULL;
+    zh->reg->sortKeys.buf_max = 0;
+
     return 0;
 }
 
@@ -1208,7 +1243,14 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
 
     if (!zh->reg->key_buf)
     {
-	int mem = 8*1024*1024;
+	int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
+	if (mem <= 0)
+	{
+	    logf(LOG_WARN, "Invalid memory setting, using default 8 MB");
+	    mem= 1024*1024*8;
+	}
+	/* FIXME: That "8" should be in a default settings include */
+	/* not hard-coded here! -H */
 	zh->reg->key_buf = (char**) xmalloc (mem);
 	zh->reg->ptr_top = mem/sizeof(char*);
 	zh->reg->ptr_i = 0;
@@ -1317,6 +1359,7 @@ void extract_flushWriteKeys (ZebraHandle zh)
         cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
         if (strcmp (cp, prevcp))
         {
+            encode_key_flush ( &encode_info, outf);
             encode_key_init (&encode_info);
             encode_key_write (cp, &encode_info, outf);
             prevcp = cp;
@@ -1324,6 +1367,7 @@ void extract_flushWriteKeys (ZebraHandle zh)
         else
             encode_key_write (cp + strlen(cp), &encode_info, outf);
     }
+    encode_key_flush ( &encode_info, outf);
 #else
     qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare);
     extract_get_fname_tmp (out_fname, key_file_no);
@@ -1355,6 +1399,7 @@ void extract_flushWriteKeys (ZebraHandle zh)
                 cp = key_buf[ptr_top-ptr_i];
                 encode_key_write (cp+key_y_len, &encode_info, outf);
             }
+            encode_key_flush ( &encode_info, outf);
             if (!i)
                 break;
             prevcp = key_buf[ptr_top-ptr_i];
@@ -1458,9 +1503,48 @@ void extract_add_index_string (RecWord *p, const char *string,
 static void extract_add_sort_string (RecWord *p, const char *string,
 				     int length)
 {
+#if 1
+    ZebraHandle zh = p->extractCtrl->handle;
+    struct sortKeys *sk = &zh->reg->sortKeys;
+    size_t off = 0;
+    int slen;
+
+    while (off < sk->buf_used)
+    {
+        int set, use, l;
+
+        l = key_SU_decode(&set, sk->buf + off);
+        off += l;
+        l = key_SU_decode(&use, sk->buf + off);
+        off += l;
+        l = key_SU_decode(&slen, sk->buf + off);
+        off += l + slen;
+        if (p->attrSet == set && p->attrUse == use)
+            return;
+    }
+    assert (off == sk->buf_used);
+    
+    if (sk->buf_used + IT_MAX_WORD > sk->buf_max)
+    {
+        char *b;
+        
+        b = (char *) xmalloc (sk->buf_max += 128000);
+        if (sk->buf_used > 0)
+            memcpy (b, sk->buf, sk->buf_used);
+        xfree (sk->buf);
+        sk->buf = b;
+    }
+    off += key_SU_encode(p->attrSet, sk->buf + off);
+    off += key_SU_encode(p->attrUse, sk->buf + off);
+    slen = strlen(string);
+    off += key_SU_encode(slen, sk->buf + off);
+    memcpy (sk->buf + off, string, slen);
+    sk->buf_used = off + slen;
+#else
     struct sortKey *sk;
     ZebraHandle zh = p->extractCtrl->handle;
 
+
     for (sk = zh->reg->sortKeys; sk; sk = sk->next)
 	if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse)
 	    return;
@@ -1475,6 +1559,7 @@ static void extract_add_sort_string (RecWord *p, const char *string,
 
     sk->attrSet = p->attrSet;
     sk->attrUse = p->attrUse;
+#endif
 }
 
 void extract_add_string (RecWord *p, const char *string, int length)
@@ -1599,22 +1684,28 @@ void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
 }
 
 void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
-                            int cmd, struct sortKey **skp)
+                            int cmd, struct sortKeys *sk)
 {
-    struct sortKey *sk = *skp;
     SortIdx sortIdx = zh->reg->sortIdx;
+    size_t off = 0;
 
     sortIdx_sysno (sortIdx, sysno);
-    while (sk)
+
+    while (off < sk->buf_used)
     {
-	struct sortKey *sk_next = sk->next;
-	sortIdx_type (sortIdx, sk->attrUse);
-	sortIdx_add (sortIdx, sk->string, sk->length);
-	xfree (sk->string);
-	xfree (sk);
-	sk = sk_next;
+        int set, use, slen, l;
+        
+        off += key_SU_decode(&set, sk->buf + off);
+        off += key_SU_decode(&use, sk->buf + off);
+        off += key_SU_decode(&slen, sk->buf + off);
+        
+        sortIdx_type(sortIdx, use);
+        if (cmd == 1)
+            sortIdx_add(sortIdx, sk->buf + off, slen);
+        else
+            sortIdx_add(sortIdx, "", 1);
+        off += slen;
     }
-    *skp = 0;
 }
 
 void encode_key_init (struct encode_info *i)
@@ -1622,6 +1713,10 @@ void encode_key_init (struct encode_info *i)
     i->sysno = 0;
     i->seqno = 0;
     i->cmd = -1;
+    i->prevsys=0;
+    i->prevseq=0;
+    i->prevcmd=-1;
+    i->keylen=0;
 }
 
 char *encode_key_int (int d, char *bp)
@@ -1649,6 +1744,11 @@ char *encode_key_int (int d, char *bp)
     return bp;
 }
 
+#ifdef OLDENCODE
+/* this is the old encode_key_write 
+ * may be deleted once we are confident that the new works
+ * HL 15-oct-2002
+ */
 void encode_key_write (char *k, struct encode_info *i, FILE *outf)
 {
     struct it_key key;
@@ -1675,3 +1775,108 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf)
     }
 }
 
+void encode_key_flush (struct encode_info *i, FILE *outf)
+{ /* dummy routine */
+}
+
+#else
+
+/* new encode_key_write
+ * The idea is to buffer one more key, and compare them
+ * If we are going to delete and insert the same key, 
+ * we may as well not bother. Should make a difference in 
+ * updates with small modifications (appending to a mbox)
+ */
+void encode_key_write (char *k, struct encode_info *i, FILE *outf)
+{
+    struct it_key key;
+    char *bp; 
+
+    if (*k)  /* first time for new key */
+    {
+        bp = i->buf;
+        while ((*bp++ = *k++))
+            ;
+	i->keylen= bp - i->buf -1;    
+	assert(i->keylen+1+sizeof(struct it_key) < ENCODE_BUFLEN);
+    }
+    else
+    {
+	bp=i->buf + i->keylen;
+	*bp++=0;
+	k++;
+    }
+
+    memcpy (&key, k+1, sizeof(struct it_key));
+    if (0==i->prevsys) /* no previous filter, fill up */
+    {
+        i->prevsys=key.sysno;
+	i->prevseq=key.seqno;
+	i->prevcmd=*k;
+    }
+    else if ( (i->prevsys==key.sysno) &&
+              (i->prevseq==key.seqno) &&
+	      (i->prevcmd!=*k) )
+    { /* same numbers, diff cmd, they cancel out */
+        i->prevsys=0;
+    }
+    else 
+    { /* different stuff, write previous, move buf */
+        bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
+	if (i->sysno != i->prevsys)
+	{
+	    i->sysno = i->prevsys;
+	    i->seqno = 0;
+        }
+        else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
+	{
+	    return; /* ??? Filters some sort of duplicates away */
+	            /* ??? Can this ever happen   -H 15oct02 */
+	}
+        bp = encode_key_int (i->prevseq - i->seqno, bp);
+        i->seqno = i->prevseq;
+        i->cmd = i->prevcmd;
+        if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+        {
+            logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+            exit (1);
+        }
+        i->keylen=0; /* ok, it's written, forget it */
+	i->prevsys=key.sysno;
+	i->prevseq=key.seqno;
+	i->prevcmd=*k;
+    }
+}
+
+void encode_key_flush (struct encode_info *i, FILE *outf)
+{ /* flush the last key from i */
+    char *bp =i->buf + i->keylen;
+    if (0==i->prevsys)
+    {
+        return; /* nothing to flush */
+    }
+    *bp++=0;
+    bp = encode_key_int ( (i->prevsys - i->sysno) * 2 + i->prevcmd, bp);
+    if (i->sysno != i->prevsys)
+    {
+        i->sysno = i->prevsys;
+        i->seqno = 0;
+    }
+    else if (!i->seqno && !i->prevseq && i->cmd == i->prevcmd)
+    {
+        return; /* ??? Filters some sort of duplicates away */
+                /* ??? Can this ever happen   -H 15oct02 */
+    }
+    bp = encode_key_int (i->prevseq - i->seqno, bp);
+    i->seqno = i->prevseq;
+    i->cmd = i->prevcmd;
+    if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+        exit (1);
+    }
+    i->keylen=0; /* ok, it's written, forget it */
+    i->prevsys=0; /* forget the values too */
+    i->prevseq=0;
+}
+#endif