-/* $Id: extract.c,v 1.147 2004-01-22 11:27:21 adam Exp $
- Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
- Index Data Aps
+/* $Id: extract.c,v 1.171 2005-01-15 19:38:25 adam Exp $
+ Copyright (C) 1995-2005
+ Index Data ApS
This file is part of the Zebra server.
02111-1307, USA.
*/
-
#include <stdio.h>
#include <assert.h>
+#include <ctype.h>
#ifdef WIN32
#include <io.h>
#else
++zh->records_processed;
if (!(zh->records_processed % 1000))
{
- logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d",
+ yaz_log (YLOG_LOG, "Records: "ZINT_FORMAT" i/u/d "
+ ZINT_FORMAT"/"ZINT_FORMAT"/"ZINT_FORMAT,
zh->records_processed, zh->records_inserted, zh->records_updated,
zh->records_deleted);
}
w->attrUse = 1016;
w->reg_type = 'w';
w->extractCtrl = p;
+ w->record_id = 0;
+ w->section_id = 0;
}
static const char **searchRecordKey (ZebraHandle zh,
int attrSetS, int attrUseS)
{
static const char *ws[32];
+ void *decode_handle = iscz1_start();
int off = 0;
int startSeq = -1;
- int i;
int seqno = 0;
-#if SU_SCHEME
- int chS, ch;
-#else
- short attrUse;
- char attrSet;
-#endif
+ int i;
for (i = 0; i<32; i++)
ws[i] = NULL;
-#if SU_SCHEME
- chS = zebraExplain_lookupSU (zh->reg->zei, attrSetS, attrUseS);
- if (chS < 0)
- return ws;
-#endif
while (off < reckeys->buf_used)
{
-
const char *src = reckeys->buf + off;
- const char *wstart;
- int lead;
-
- lead = *src++;
-#if SU_SCHEME
- if ((lead & 3)<3)
- {
- memcpy (&ch, src, sizeof(ch));
- src += sizeof(ch);
- }
-#else
- if (!(lead & 1))
- {
- memcpy (&attrSet, src, sizeof(attrSet));
- src += sizeof(attrSet);
- }
- if (!(lead & 2))
- {
- memcpy (&attrUse, src, sizeof(attrUse));
- src += sizeof(attrUse);
- }
-#endif
- wstart = src;
- while (*src++)
- ;
- if (lead & 60)
- seqno += ((lead>>2) & 15)-1;
- else
- {
- memcpy (&seqno, src, sizeof(seqno));
- src += sizeof(seqno);
- }
- if (
-#if SU_SCHEME
- ch == chS
-#else
- attrUseS == attrUse && attrSetS == attrSet
-#endif
- )
+ struct it_key key;
+ char *dst = (char*) &key;
+ int attrSet, attrUse;
+
+ iscz1_decode(decode_handle, &dst, &src);
+ assert(key.len < 4 && key.len > 2);
+
+ attrSet = (int) key.mem[0] >> 16;
+ attrUse = (int) key.mem[0] & 65535;
+ seqno = (int) key.mem[key.len-1];
+
+ if (attrUseS == attrUse && attrSetS == attrSet)
{
int woff;
-
if (startSeq == -1)
startSeq = seqno;
woff = seqno - startSeq;
if (woff >= 0 && woff < 31)
- ws[woff] = wstart;
+ ws[woff] = src;
}
+ while (*src++)
+ ;
off = src - reckeys->buf;
}
+ iscz1_stop(decode_handle);
assert (off == reckeys->buf_used);
return ws;
}
}
else
{
- logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
+ yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s",
spec, zh->m_group ? zh->m_group : "none");
return NULL;
}
}
if (first)
{
- logf (LOG_WARN, "Record didn't contain match"
+ yaz_log (YLOG_WARN, "Record didn't contain match"
" fields in (%s,%s)", attset_str, attname_str);
return NULL;
}
}
else
{
- logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
+ yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s",
spec, zh->m_group ? zh->m_group : "none");
return NULL;
}
}
if (dst == dstBuf)
{
- logf (LOG_WARN, "No match criteria for record %s in group %s",
+ yaz_log (YLOG_WARN, "No match criteria for record %s in group %s",
fname, zh->m_group ? zh->m_group : "none");
return NULL;
}
int recordOffset;
struct recordGroup *rGroup;
};
+
+void create_rec_keys_codec(struct recKeys *keys)
+{
+ keys->buf_used = 0;
+ iscz1_reset(keys->codec_handle);
+}
static int file_extract_record(ZebraHandle zh,
SYSNO *sysno, const char *fname,
{
RecordAttr *recordAttr;
int r;
- char *matchStr;
+ const char *matchStr = 0;
SYSNO sysnotmp;
Record rec;
off_t recordOffset = 0;
RecType recType;
void *clientData;
- char subType[128];
if (!(recType =
- recType_byName (zh->reg->recTypes, zh->m_record_type, subType,
+ recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type,
&clientData)))
{
- logf (LOG_WARN, "No such record type: %s", zh->m_record_type);
+ yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
return 0;
}
/* we are going to read from a file, so prepare the extraction */
int i;
- zh->reg->keys.buf_used = 0;
- zh->reg->keys.prevAttrUse = -1;
- zh->reg->keys.prevAttrSet = -1;
- zh->reg->keys.prevSeqNo = 0;
+ create_rec_keys_codec(&zh->reg->keys);
+
zh->reg->sortKeys.buf_used = 0;
recordOffset = fi->file_moffset;
extractCtrl.tellf = file_tell;
extractCtrl.endf = file_end;
extractCtrl.fh = fi;
- extractCtrl.subType = subType;
extractCtrl.init = extract_init;
extractCtrl.tokenAdd = extract_token_add;
extractCtrl.schemaAdd = extract_schema_add;
extractCtrl.dh = zh->reg->dh;
+ extractCtrl.match_criteria[0] = '\0';
extractCtrl.handle = zh;
for (i = 0; i<256; i++)
{
if (zh->m_flag_rw &&
zh->records_processed < zh->m_file_verbose_limit)
{
- logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
+ yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type,
fname, recordOffset);
}
return 0;
if (zh->m_flag_rw &&
zh->records_processed < zh->m_file_verbose_limit)
{
- logf (LOG_WARN, "no filter for %s %s "
+ yaz_log (YLOG_WARN, "no filter for %s %s "
PRINTF_OFF_T, zh->m_record_type,
fname, recordOffset);
}
if (!zh->m_flag_rw)
return 1;
- logf (LOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type,
- fname, recordOffset);
+ 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;
}
+ if (extractCtrl.match_criteria[0])
+ matchStr = extractCtrl.match_criteria;
}
/* perform match if sysno not known and if match criteria is specified */
-
- matchStr = NULL;
if (!sysno)
{
sysnotmp = 0;
sysno = &sysnotmp;
- if (zh->m_record_id && *zh->m_record_id)
+
+ if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
{
- char *rinfo;
matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
zh->m_record_id);
- if (matchStr)
- {
- rinfo = dict_lookup (zh->reg->matchDict, matchStr);
- if (rinfo)
- memcpy (sysno, rinfo+1, sizeof(*sysno));
- }
- else
- {
- logf (LOG_WARN, "Bad match criteria");
- return 0;
- }
- }
+ if (!matchStr)
+ {
+ yaz_log(YLOG_WARN, "Bad match criteria");
+ return 0;
+ }
+ }
+ if (matchStr)
+ {
+ char *rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+ if (rinfo)
+ {
+ assert(*rinfo == sizeof(*sysno));
+ memcpy (sysno, rinfo+1, sizeof(*sysno));
+ }
+ }
}
if (! *sysno)
/* new record */
if (deleteFlag)
{
- logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
+ yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type,
fname, recordOffset);
- logf (LOG_WARN, "cannot delete record above (seems new)");
+ yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
return 1;
}
if (zh->records_processed < zh->m_file_verbose_limit)
- logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
+ yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type,
fname, recordOffset);
rec = rec_new (zh->reg->records);
if (!force_update && recordAttr->runNumber ==
zebraExplain_runNumberIncrement (zh->reg->zei, 0))
{
- yaz_log (LOG_LOG, "run number = %d", recordAttr->runNumber);
- yaz_log (LOG_LOG, "skipped %s %s " PRINTF_OFF_T,
+ yaz_log (YLOG_LOG, "run number = %d", recordAttr->runNumber);
+ yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T,
zh->m_record_type, fname, recordOffset);
extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
rec_rm (&rec);
/* record going to be deleted */
if (!delkeys.buf_used)
{
- logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
+ yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
zh->m_record_type, fname, recordOffset);
- logf (LOG_WARN, "cannot delete file above, storeKeys false");
+ yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
}
else
{
if (zh->records_processed < zh->m_file_verbose_limit)
- logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T,
+ yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
zh->m_record_type, fname, recordOffset);
zh->records_deleted++;
if (matchStr)
/* record going to be updated */
if (!delkeys.buf_used)
{
- logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
+ yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
zh->m_record_type, fname, recordOffset);
- logf (LOG_WARN, "cannot update file above, storeKeys false");
+ yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
}
else
{
if (zh->records_processed < zh->m_file_verbose_limit)
- logf (LOG_LOG, "update %s %s " PRINTF_OFF_T,
+ yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
zh->m_record_type, fname, recordOffset);
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
xmalloc (recordAttr->recordSize);
if (lseek (fi->fd, recordOffset, SEEK_SET) < 0)
{
- logf (LOG_ERRNO|LOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
+ yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s",
recordOffset, fname);
exit (1);
}
if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize)
< recordAttr->recordSize)
{
- logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
+ yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s",
recordAttr->recordSize, fname);
exit (1);
}
char ext[128];
char ext_res[128];
struct file_read_info *fi;
+ const char *original_record_type = 0;
if (!zh->m_group || !*zh->m_group)
*gprefix = '\0';
else
sprintf (gprefix, "%s.", zh->m_group);
- logf (LOG_DEBUG, "fileExtract %s", fname);
+ yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
/* determine file extension */
*ext = '\0';
break;
}
/* determine file type - depending on extension */
+ original_record_type = zh->m_record_type;
if (!zh->m_record_type)
{
sprintf (ext_res, "%srecordType.%s", gprefix, ext);
if (!zh->m_record_type)
{
if (zh->records_processed < zh->m_file_verbose_limit)
- logf (LOG_LOG, "? %s", fname);
+ yaz_log (YLOG_LOG, "? %s", fname);
return 0;
}
/* determine match criteria */
if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1)
{
- logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep);
+ yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep);
+ zh->m_record_type = original_record_type;
return 0;
}
}
file_read_stop (fi);
if (fd != -1)
close (fd);
+ zh->m_record_type = original_record_type;
return r;
}
int delete_flag,
int test_mode,
const char *recordType,
- int *sysno,
+ SYSNO *sysno,
const char *match_criteria,
const char *fname,
int force_update,
RecordAttr *recordAttr;
struct recExtractCtrl extractCtrl;
int i, r;
- char *matchStr = 0;
+ const char *matchStr = 0;
RecType recType = NULL;
- char subType[1024];
void *clientData;
Record rec;
long recordOffset = 0;
extractCtrl.endf = zebra_record_int_end;
extractCtrl.fh = &fc;
- zh->reg->keys.buf_used = 0;
- zh->reg->keys.prevAttrUse = -1;
- zh->reg->keys.prevAttrSet = -1;
- zh->reg->keys.prevSeqNo = 0;
+ create_rec_keys_codec(&zh->reg->keys);
+
zh->reg->sortKeys.buf_used = 0;
if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
}
if (recordType && *recordType) {
- logf (LOG_DEBUG, "Record type explicitly specified: %s", recordType);
- recType = recType_byName (zh->reg->recTypes, recordType, subType,
+ yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
+ recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
&clientData);
} else {
if (!(zh->m_record_type)) {
- logf (LOG_WARN, "No such record type defined");
+ yaz_log (YLOG_WARN, "No such record type defined");
return 0;
}
- logf (LOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
- recType = recType_byName (zh->reg->recTypes, zh->m_record_type, subType,
- &clientData);
+ yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
+ recType = recType_byName (zh->reg->recTypes, zh->res,
+ zh->m_record_type, &clientData);
recordType = zh->m_record_type;
}
if (!recType) {
- logf (LOG_WARN, "No such record type: %s", zh->m_record_type);
+ yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type);
return 0;
}
- extractCtrl.subType = subType;
extractCtrl.init = extract_init;
extractCtrl.tokenAdd = extract_token_add;
extractCtrl.schemaAdd = extract_schema_add;
extractCtrl.handle = zh;
extractCtrl.zebra_maps = zh->reg->zebra_maps;
extractCtrl.flagShowRecords = 0;
+ extractCtrl.match_criteria[0] = '\0';
for (i = 0; i<256; i++)
{
if (zebra_maps_is_positioned(zh->reg->zebra_maps, i))
else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
{
/* error occured during extraction ... */
- yaz_log (LOG_WARN, "extract error: generic");
+ yaz_log (YLOG_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");
+ yaz_log (YLOG_WARN, "extract error: no such filter");
return 0;
}
if (zh->reg->keys.buf_used == 0)
is probably empty - unless flagShowRecords is in use */
if (test_mode)
return 1;
- logf (LOG_WARN, "No keys generated for record");
- logf (LOG_WARN, " The file is probably empty");
+ yaz_log (YLOG_WARN, "No keys generated for record");
+ yaz_log (YLOG_WARN, " The file is probably empty");
return 1;
}
/* match criteria */
matchStr = NULL;
+ if (extractCtrl.match_criteria[0])
+ match_criteria = extractCtrl.match_criteria;
+
if (! *sysno) {
char *rinfo;
if (match_criteria && *match_criteria) {
- matchStr = (char *)match_criteria;
+ matchStr = match_criteria;
} else {
if (zh->m_record_id && *zh->m_record_id) {
matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
zh->m_record_id);
+ if (!matchStr)
+ {
+ yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
+ return 1;
+ }
}
}
if (matchStr) {
rinfo = dict_lookup (zh->reg->matchDict, matchStr);
if (rinfo)
+ {
+ assert(*rinfo == sizeof(*sysno));
memcpy (sysno, rinfo+1, sizeof(*sysno));
- } else {
- logf (LOG_WARN, "Bad match criteria (recordID)");
- return 0;
+ }
}
}
/* new record */
if (delete_flag)
{
- logf (LOG_LOG, "delete %s %s %ld", recordType,
+ yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
pr_fname, (long) recordOffset);
- logf (LOG_WARN, "cannot delete record above (seems new)");
+ yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
return 1;
}
- logf (LOG_LOG, "add %s %s %ld", recordType, pr_fname,
+ yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname,
(long) recordOffset);
rec = rec_new (zh->reg->records);
struct sortKeys sortKeys;
if (!allow_update) {
- logf (LOG_LOG, "skipped %s %s %ld",
+ yaz_log (YLOG_LOG, "skipped %s %s %ld",
recordType, pr_fname, (long) recordOffset);
logRecord(zh);
return -1;
if (recordAttr->runNumber ==
zebraExplain_runNumberIncrement (zh->reg->zei, 0))
{
- logf (LOG_LOG, "skipped %s %s %ld", recordType,
+ yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType,
pr_fname, (long) recordOffset);
extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys);
rec_rm (&rec);
logRecord(zh);
- return 0;
+ return -1;
}
}
/* record going to be deleted */
if (!delkeys.buf_used)
{
- logf (LOG_LOG, "delete %s %s %ld", recordType,
+ yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
pr_fname, (long) recordOffset);
- logf (LOG_WARN, "cannot delete file above, storeKeys false");
+ yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false");
}
else
{
- logf (LOG_LOG, "delete %s %s %ld", recordType,
+ yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
pr_fname, (long) recordOffset);
zh->records_deleted++;
if (matchStr)
/* record going to be updated */
if (!delkeys.buf_used)
{
- logf (LOG_LOG, "update %s %s %ld", recordType,
+ yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
pr_fname, (long) recordOffset);
- logf (LOG_WARN, "cannot update file above, storeKeys false");
+ yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
}
else
{
- logf (LOG_LOG, "update %s %s %ld", recordType,
+ yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
pr_fname, (long) recordOffset);
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys);
abort ();
}
- zh->reg->keys.buf_used = 0;
- zh->reg->keys.prevAttrUse = -1;
- zh->reg->keys.prevAttrSet = -1;
- zh->reg->keys.prevSeqNo = 0;
+ create_rec_keys_codec(&zh->reg->keys);
+
zh->reg->sortKeys.buf_used = 0;
extractCtrl.init = extract_init;
extractCtrl.seqno[i] = 0;
extractCtrl.zebra_maps = zh->reg->zebra_maps;
extractCtrl.flagShowRecords = 0;
+ extractCtrl.match_criteria[0] = '\0';
extractCtrl.handle = handle;
if (n)
void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
int cmd, struct recKeys *reckeys)
{
-#if SU_SCHEME
-#else
- unsigned char attrSet = (unsigned char) -1;
- unsigned short attrUse = (unsigned short) -1;
-#endif
- int seqno = 0;
+ void *decode_handle = iscz1_start();
int off = 0;
int ch = 0;
ZebraExplainInfo zei = zh->reg->zei;
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");
+ yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB");
mem= 1024*1024*8;
}
/* FIXME: That "8" should be in a default settings include */
zh->reg->key_file_no = 0;
}
zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
+
while (off < reckeys->buf_used)
{
const char *src = reckeys->buf + off;
struct it_key key;
- int lead;
-
- lead = *src++;
+ char *dst = (char*) &key;
+ int attrSet, attrUse;
+
+ iscz1_decode(decode_handle, &dst, &src);
+ assert(key.len == 4);
+
+ attrSet = (int) key.mem[0] >> 16;
+ attrUse = (int) key.mem[0] & 65535;
-#if SU_SCHEME
- if ((lead & 3) < 3)
- {
- memcpy (&ch, src, sizeof(ch));
- src += sizeof(ch);
- }
-#else
- if (!(lead & 1))
- {
- memcpy (&attrSet, src, sizeof(attrSet));
- src += sizeof(attrSet);
- }
- if (!(lead & 2))
- {
- memcpy (&attrUse, src, sizeof(attrUse));
- src += sizeof(attrUse);
- }
-#endif
if (zh->reg->key_buf_used + 1024 >
(zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
- extract_flushWriteKeys (zh);
+ extract_flushWriteKeys (zh,0);
++(zh->reg->ptr_i);
+ assert(zh->reg->ptr_i > 0);
(zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] =
(char*)zh->reg->key_buf + zh->reg->key_buf_used;
-#if SU_SCHEME
-#else
+
ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
if (ch < 0)
ch = zebraExplain_addSU (zei, attrSet, attrUse);
-#endif
+
assert (ch > 0);
zh->reg->key_buf_used +=
key_SU_encode (ch,((char*)zh->reg->key_buf) +
zh->reg->key_buf_used);
-
while (*src)
((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++;
src++;
((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0';
((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd;
- if (lead & 60)
- seqno += ((lead>>2) & 15)-1;
- else
- {
- memcpy (&seqno, src, sizeof(seqno));
- src += sizeof(seqno);
- }
- key.seqno = seqno;
- key.sysno = sysno;
- memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used, &key, sizeof(key));
+ key.len = 3;
+ if (key.mem[1]) /* filter specify record ID */
+ key.mem[0] = key.mem[1];
+ else
+ key.mem[0] = sysno;
+ key.mem[1] = key.mem[2]; /* section_id */
+ key.mem[2] = key.mem[3]; /* sequence .. */
+
+ memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
+ &key, sizeof(key));
(zh->reg->key_buf_used) += sizeof(key);
off = src - reckeys->buf;
}
assert (off == reckeys->buf_used);
+ iscz1_stop(decode_handle);
}
-void extract_flushWriteKeys (ZebraHandle zh)
+void extract_flushWriteKeys (ZebraHandle zh, int final)
+ /* optimizing: if final=1, and no files written yet */
+ /* push the keys directly to merge, sidestepping the */
+ /* temp file altogether. Speeds small updates */
{
FILE *outf;
char out_fname[200];
char *prevcp, *cp;
struct encode_info encode_info;
int ptr_i = zh->reg->ptr_i;
+ int temp_policy;
#if SORT_EXTRA
int i;
#endif
if (!zh->reg->key_buf || ptr_i <= 0)
+ {
+ yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
+ zh->reg->key_file_no, zh->reg->key_buf, ptr_i);
+ yaz_log (YLOG_DEBUG, " buf=%p ",
+ zh->reg->key_buf);
+ yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
+ yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
+
return;
+ }
(zh->reg->key_file_no)++;
- logf (LOG_LOG, "sorting section %d", (zh->reg->key_file_no));
+ yaz_log (YLOG_LOG, "sorting section %d", (zh->reg->key_file_no));
+ yaz_log (YLOG_DEBUG, " sort_buff at %p n=%d",
+ zh->reg->key_buf + zh->reg->ptr_top - ptr_i,ptr_i);
#if !SORT_EXTRA
qsort (zh->reg->key_buf + zh->reg->ptr_top - ptr_i, ptr_i,
sizeof(char*), key_qsort_compare);
+
+ /* zebra.cfg: tempfiles:
+ Y: always use temp files (old way)
+ A: use temp files, if more than one (auto)
+ = if this is both the last and the first
+ N: never bother with temp files (new) */
+
+ temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]);
+ if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') {
+ yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ",
+ temp_policy);
+ temp_policy='A';
+ }
+
+ if ( ( temp_policy =='N' ) || /* always from memory */
+ ( ( temp_policy =='A' ) && /* automatic */
+ (zh->reg->key_file_no == 1) && /* this is first time */
+ (final) ) ) /* and last (=only) time */
+ { /* go directly from memory */
+ zh->reg->key_file_no =0; /* signal not to read files */
+ zebra_index_merge(zh);
+ zh->reg->ptr_i = 0;
+ zh->reg->key_buf_used = 0;
+ return;
+ }
+
+ /* Not doing directly from memory, write into a temp file */
extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no);
if (!(outf = fopen (out_fname, "wb")))
{
- logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
exit (1);
}
- logf (LOG_LOG, "writing section %d", zh->reg->key_file_no);
+ yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no);
prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
encode_key_init (&encode_info);
if (!(outf = fopen (out_fname, "wb")))
{
- logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname);
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname);
exit (1);
}
- logf (LOG_LOG, "writing section %d", key_file_no);
+ yaz_log (YLOG_LOG, "writing section %d", key_file_no);
i = ptr_i;
prevcp = key_buf[ptr_top-i];
while (1)
{
key_y_len = strlen(prevcp)+1;
#if 0
- logf (LOG_LOG, "key_y_len: %2d %02x %02x %s",
+ yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s",
key_y_len, prevcp[0], prevcp[1], 2+prevcp);
#endif
qsort (key_buf + ptr_top-ptr_i, ptr_i - i,
#endif
if (fclose (outf))
{
- logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname);
exit (1);
}
- logf (LOG_LOG, "finished section %d", zh->reg->key_file_no);
+ yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no);
zh->reg->ptr_i = 0;
zh->reg->key_buf_used = 0;
}
-void extract_add_index_string (RecWord *p, const char *string,
- int length)
+void extract_add_it_key (ZebraHandle zh,
+ int reg_type,
+ const char *str, int slen, struct it_key *key)
{
char *dst;
- unsigned char attrSet;
- unsigned short attrUse;
- int lead = 0;
- int diff = 0;
- int *pseqno = &p->seqno;
- ZebraHandle zh = p->extractCtrl->handle;
- ZebraExplainInfo zei = zh->reg->zei;
struct recKeys *keys = &zh->reg->keys;
+ const char *src = (char*) key;
if (keys->buf_used+1024 > keys->buf_max)
{
}
dst = keys->buf + keys->buf_used;
- attrSet = p->attrSet;
- if (keys->buf_used > 0 && keys->prevAttrSet == attrSet)
- lead |= 1;
- else
- keys->prevAttrSet = attrSet;
- attrUse = p->attrUse;
- if (keys->buf_used > 0 && keys->prevAttrUse == attrUse)
- lead |= 2;
- else
- keys->prevAttrUse = attrUse;
-#if 1
- diff = 1 + *pseqno - keys->prevSeqNo;
- if (diff >= 1 && diff <= 15)
- lead |= (diff << 2);
- else
- diff = 0;
-#endif
- keys->prevSeqNo = *pseqno;
-
- *dst++ = lead;
+ iscz1_encode(keys->codec_handle, &dst, &src);
-#if SU_SCHEME
- if ((lead & 3) < 3)
- {
- int ch = zebraExplain_lookupSU (zei, attrSet, attrUse);
- if (ch < 0)
- {
- ch = zebraExplain_addSU (zei, attrSet, attrUse);
- yaz_log (LOG_DEBUG, "addSU set=%d use=%d SU=%d",
- attrSet, attrUse, ch);
- }
- assert (ch > 0);
- memcpy (dst, &ch, sizeof(ch));
- dst += sizeof(ch);
- }
-#else
- if (!(lead & 1))
- {
- memcpy (dst, &attrSet, sizeof(attrSet));
- dst += sizeof(attrSet);
- }
- if (!(lead & 2))
- {
- memcpy (dst, &attrUse, sizeof(attrUse));
- dst += sizeof(attrUse);
- }
-#endif
- *dst++ = p->reg_type;
- memcpy (dst, string, length);
- dst += length;
+ *dst++ = reg_type;
+ memcpy (dst, str, slen);
+ dst += slen;
*dst++ = '\0';
-
- if (!diff)
- {
- memcpy (dst, pseqno, sizeof(*pseqno));
- dst += sizeof(*pseqno);
- }
keys->buf_used = dst - keys->buf;
}
-static void extract_add_sort_string (RecWord *p, const char *string,
+void extract_add_index_string (RecWord *p, const char *str, int length)
+{
+ struct it_key key;
+ key.len = 4;
+ key.mem[0] = p->attrSet * 65536 + p->attrUse;
+ key.mem[1] = p->record_id;
+ key.mem[2] = p->section_id;
+ key.mem[3] = p->seqno;
+
+#if 0
+ /* just for debugging .. */
+ yaz_log(YLOG_LOG, "add: set=%d use=%d "
+ "record_id=%lld section_id=%lld seqno=%lld",
+ p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno);
+#endif
+
+ extract_add_it_key(p->extractCtrl->handle, p->reg_type, str,
+ length, &key);
+}
+
+static void extract_add_sort_string (RecWord *p, const char *str,
int length)
{
ZebraHandle zh = p->extractCtrl->handle;
off += key_SU_encode(p->attrSet, sk->buf + off);
off += key_SU_encode(p->attrUse, sk->buf + off);
off += key_SU_encode(length, sk->buf + off);
- memcpy (sk->buf + off, string, length);
+ memcpy (sk->buf + off, str, length);
sk->buf_used = off + length;
}
const char *b = p->string;
int remain = p->length;
const char **map = 0;
+
+ yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->length, p->string);
if (remain > 0)
- map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+ map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
while (map)
{
{
remain = p->length - (b - p->string);
if (remain > 0)
- map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+ map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
else
map = 0;
}
buf[i++] = *(cp++);
remain = p->length - (b - p->string);
if (remain > 0)
- map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+ map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0);
else
map = 0;
}
char buf[IT_MAX_WORD+1];
const char **map = 0;
int i = 0, remain = p->length;
+ int first; /* first position */
+
+ yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'", p->length, p->string);
if (remain > 0)
- map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain);
+ map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1);
while (remain > 0 && i < IT_MAX_WORD)
{
while (map && *map && **map == *CHR_SPACE)
{
remain = p->length - (b - p->string);
+
if (remain > 0)
- map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain);
+ {
+ first = i ? 0 : 1;
+ map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first);
+ }
else
map = 0;
}
{
const char *cp = *map;
- if (i >= IT_MAX_WORD)
- break;
- while (i < IT_MAX_WORD && *cp)
- buf[i++] = *(cp++);
+ if (**map == *CHR_CUT)
+ {
+ i = 0;
+ }
+ else
+ {
+ if (i >= IT_MAX_WORD)
+ break;
+ yaz_log(YLOG_DEBUG, "Adding string to index '%d'", **map);
+ while (i < IT_MAX_WORD && *cp)
+ buf[i++] = *(cp++);
+ }
remain = p->length - (b - p->string);
if (remain > 0)
+ {
map = zebra_maps_input (p->zebra_maps, p->reg_type, &b,
- remain);
+ remain, 0);
+ }
else
map = 0;
}
{
WRBUF wrbuf;
#if 0
- yaz_log (LOG_LOG, "token_add "
+ yaz_log (YLOG_LOG, "token_add "
"reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s",
p->reg_type, p->attrSet, p->attrUse, p->seqno, p->length,
p->string);
i->prevseq=0;
i->prevcmd=-1;
i->keylen=0;
+ i->encode_handle = iscz1_start();
}
-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)
- {
- *bp++ = 128 + (d>>16);
- *bp++ = (d>>8) & 255;
- *bp++ = d & 255;
- }
- else
- {
- *bp++ = 192 + (d>>24);
- *bp++ = (d>>16) & 255;
- *bp++ = (d>>8) & 255;
- *bp++ = d & 255;
- }
- return bp;
-}
#define OLDENCODE 1
#ifdef OLDENCODE
void encode_key_write (char *k, struct encode_info *i, FILE *outf)
{
struct it_key key;
- char *bp = i->buf;
+ char *bp = i->buf, *bp0;
+ const char *src = (char *) &key;
+ /* copy term to output 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)
- {
- i->sysno = key.sysno;
- i->seqno = 0;
- }
- else if (!i->seqno && !key.seqno && i->cmd == *k)
- return;
- bp = encode_key_int (key.seqno - i->seqno, bp);
- i->seqno = key.seqno;
- i->cmd = *k;
+ /* and copy & align key so we can mangle */
+ memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */
+
+ 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)
{
- logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
exit (1);
}
}
void encode_key_flush (struct encode_info *i, FILE *outf)
{ /* dummy routine */
+ iscz1_stop(i->encode_handle);
}
#else
i->cmd = i->prevcmd;
if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
{
- logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
exit (1);
}
i->keylen=0; /* ok, it's written, forget it */
i->cmd = i->prevcmd;
if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
{
- logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+ yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
exit (1);
}
i->keylen=0; /* ok, it's written, forget it */