-/* $Id: extract.c,v 1.189 2005-08-18 12:50:17 adam Exp $
+/* $Id: extract.c,v 1.206 2006-03-30 09:52:15 adam Exp $
Copyright (C) 1995-2005
Index Data ApS
#include <fcntl.h>
#include "index.h"
+#include "orddict.h"
#include <direntz.h>
#include <charmap.h>
}
static void searchRecordKey(ZebraHandle zh,
- const struct recKeys *reckeys,
+ zebra_rec_keys_t reckeys,
int attrSetS, int attrUseS,
const char **ws, int ws_length)
{
- void *decode_handle = iscz1_start();
- int off = 0;
- int startSeq = -1;
- int seqno = 0;
int i;
int ch;
if (ch < 0)
return ;
- while (off < reckeys->buf_used)
+ if (zebra_rec_keys_rewind(reckeys))
{
- const char *src = reckeys->buf + off;
- struct it_key key;
- char *dst = (char*) &key;
-
- iscz1_decode(decode_handle, &dst, &src);
- assert(key.len <= 4 && key.len > 2);
-
- seqno = (int) key.mem[key.len-1];
-
- if (key.mem[0] == ch)
- {
- int woff;
-
- if (startSeq == -1)
- startSeq = seqno;
- woff = seqno - startSeq;
- if (woff >= 0 && woff < ws_length)
- ws[woff] = src;
- }
+ int startSeq = -1;
+ const char *str;
+ size_t slen;
+ struct it_key key;
+ zint seqno;
+ while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+ {
+ assert(key.len <= 4 && key.len > 2);
- while (*src++)
- ;
- off = src - reckeys->buf;
+ seqno = key.mem[key.len-1];
+
+ if (key.mem[0] == ch)
+ {
+ int woff;
+
+ if (startSeq == -1)
+ startSeq = seqno;
+ woff = seqno - startSeq;
+ if (woff >= 0 && woff < ws_length)
+ ws[woff] = str;
+ }
+ }
}
- iscz1_stop(decode_handle);
- assert (off == reckeys->buf_used);
}
struct file_read_info {
}
}
+#define FILE_MATCH_BLANK "\t "
+
static char *fileMatchStr (ZebraHandle zh,
- struct recKeys *reckeys,
+ zebra_rec_keys_t reckeys,
const char *fname, const char *spec)
{
static char dstBuf[2048]; /* static here ??? */
while (1)
{
- while (*s == ' ' || *s == '\t')
- s++;
+ for (; *s && strchr(FILE_MATCH_BLANK, *s); s++)
+ ;
if (!*s)
break;
if (*s == '(')
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';
}
}
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");
s++;
for (i = 0; i<32; i++)
- if (matchFlag[i] && ws[i])
+ if (ws[i])
{
if (first)
{
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;
while (*s && *s != stopMarker)
{
- if (i < 63)
+ if (i+1 < sizeof(tmpString))
tmpString[i++] = *s++;
}
if (*s)
struct recordGroup *rGroup;
};
-void create_rec_keys_codec(struct recKeys *keys)
-{
- keys->buf_used = 0;
- iscz1_reset(keys->codec_handle);
-}
-
static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl)
{
int i;
if (fi->fd != -1)
{
/* we are going to read from a file, so prepare the extraction */
- create_rec_keys_codec(&zh->reg->keys);
+ zebra_rec_keys_reset(zh->reg->keys);
+
#if NATTR
- create_rec_keys_codec(&zh->reg->sortKeys);
+ zebra_rec_keys_reset(zh->reg->sortKeys);
#else
zh->reg->sortKeys.buf_used = 0;
#endif
if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
{
-
- matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
+ matchStr = fileMatchStr (zh, zh->reg->keys, fname,
zh->m_record_id);
if (!matchStr)
{
}
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));
}
}
}
- if (! *sysno && zh->reg->keys.buf_used == 0)
+ if (! *sysno && zebra_rec_keys_empty(zh->reg->keys) )
{
/* the extraction process returned no information - the record
is probably empty - unless flagShowRecords is in use */
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
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
- extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
+#endif
+ extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
recordAttr->staticrank);
-
zh->records_inserted++;
}
else
{
/* record already exists */
- struct recKeys delkeys;
+ zebra_rec_keys_t delkeys = zebra_rec_keys_open();
+
#if NATTR
- struct recKeys sortKeys;
+ zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
#else
struct sortKeys sortKeys;
#endif
recordAttr = rec_init_attr (zh->reg->zei, rec);
- if (!force_update && recordAttr->runNumber ==
- zebraExplain_runNumberIncrement (zh->reg->zei, 0))
- {
- yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT,
- 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);
- logRecord (zh);
- return 1;
- }
- /* flush old keys for sort&search etc. */
- delkeys.buf_used = rec->size[recInfo_delKeys];
- delkeys.buf = rec->info[recInfo_delKeys];
+ zebra_rec_keys_set_buf(delkeys,
+ rec->info[recInfo_delKeys],
+ rec->size[recInfo_delKeys],
+ 0);
+#if NATTR
+ zebra_rec_keys_set_buf(sortKeys,
+ rec->info[recInfo_sortKeys],
+ rec->size[recInfo_sortKeys],
+ 0);
+ extract_flushSortKeys (zh, *sysno, 0, sortKeys);
+#else
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,
+#endif
+
+ extract_flushRecordKeys (zh, *sysno, 0, delkeys,
recordAttr->staticrank); /* old values */
if (deleteFlag)
{
/* record going to be deleted */
- if (!delkeys.buf_used)
+ 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_WARN, "cannot delete file above, storeKeys false");
+ zh->m_record_type, fname, recordOffset);
+ 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);
+ zh->m_record_type, fname, 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);
}
else
{
- /* record going to be updated */
- if (!delkeys.buf_used)
- {
+ /* 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_WARN, "cannot update file above, storeKeys false");
- }
- 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);
- recordAttr->staticrank = extractCtrl.staticrank;
- extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
- extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
+ recordAttr->staticrank = extractCtrl.staticrank;
+#if NATTR
+ extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+#else
+ extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+#endif
+ extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
recordAttr->staticrank);
- zh->records_updated++;
- }
+ zh->records_updated++;
}
+ zebra_rec_keys_close(delkeys);
+#if NATTR
+ zebra_rec_keys_close(sortKeys);
+#endif
}
/* update file type */
xfree (rec->info[recInfo_fileType]);
/* update delete keys */
xfree (rec->info[recInfo_delKeys]);
- if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
+ if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 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;
+ zebra_rec_keys_get_buf(zh->reg->keys,
+ &rec->info[recInfo_delKeys],
+ &rec->size[recInfo_delKeys]);
}
else
{
/* update sort keys */
xfree (rec->info[recInfo_sortKeys]);
+#if NATTR
+ zebra_rec_keys_get_buf(zh->reg->sortKeys,
+ &rec->info[recInfo_sortKeys],
+ &rec->size[recInfo_sortKeys]);
+#else
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;
+#endif
/* save file size of original record */
zebraExplain_recordBytesIncrement (zh->reg->zei,
If not, and a record is provided, then sysno is got from there
*/
-ZEBRA_RES buffer_extract_record (ZebraHandle zh,
- const char *buf, size_t buf_size,
- int delete_flag,
- int test_mode,
- const char *recordType,
- SYSNO *sysno,
- const char *match_criteria,
- const char *fname,
- int force_update,
- int allow_update)
+ZEBRA_RES buffer_extract_record(ZebraHandle zh,
+ const char *buf, size_t buf_size,
+ int delete_flag,
+ int test_mode,
+ const char *recordType,
+ SYSNO *sysno,
+ const char *match_criteria,
+ const char *fname,
+ int force_update,
+ int allow_update)
{
+ SYSNO sysno0 = 0;
RecordAttr *recordAttr;
struct recExtractCtrl extractCtrl;
int r;
extractCtrl.first_record = 1;
extractCtrl.fh = &fc;
- create_rec_keys_codec(&zh->reg->keys);
+ zebra_rec_keys_reset(zh->reg->keys);
+
#if NATTR
- create_rec_keys_codec(&zh->reg->sortKeys);
+ zebra_rec_keys_reset(zh->reg->sortKeys);
#else
zh->reg->sortKeys.buf_used = 0;
#endif
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;
}
yaz_log (YLOG_WARN, "extract error: no such filter");
return ZEBRA_FAIL;
}
- /* match criteria */
- matchStr = NULL;
if (extractCtrl.match_criteria[0])
match_criteria = extractCtrl.match_criteria;
- if (! *sysno) {
- char *rinfo;
+ if (!sysno) {
+
+ sysno = &sysno0;
+
if (match_criteria && *match_criteria) {
matchStr = match_criteria;
} else {
if (zh->m_record_id && *zh->m_record_id) {
- matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
+ matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
zh->m_record_id);
if (!matchStr)
{
}
}
}
- if (matchStr) {
- rinfo = dict_lookup (zh->reg->matchDict, matchStr);
+ if (matchStr)
+ {
+ int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+ char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
+ matchStr);
if (rinfo)
{
assert(*rinfo == sizeof(*sysno));
}
}
}
- if (zh->reg->keys.buf_used == 0)
+ if (zebra_rec_keys_empty(zh->reg->keys))
{
/* the extraction process returned no information - the record
is probably empty - unless flagShowRecords is in use */
/* new record */
if (delete_flag)
{
- if (show_progress)
- yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
+ yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
pr_fname, (long) recordOffset);
yaz_log (YLOG_WARN, "cannot delete record above (seems new)");
return ZEBRA_FAIL;
*sysno = rec->sysno;
recordAttr = rec_init_attr (zh->reg->zei, rec);
+ recordAttr->staticrank = extractCtrl.staticrank;
if (matchStr)
{
- dict_insert (zh->reg->matchDict, matchStr,
- sizeof(*sysno), sysno);
+ int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
+ dict_insert_ord(zh->reg->matchDict, db_ord, matchStr,
+ sizeof(*sysno), sysno);
}
+#if NATTR
+ extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+#else
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
- extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys, 0);
-
+#endif
+
+#if 0
+ print_rec_keys(zh, zh->reg->keys);
+#endif
+ extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
+ recordAttr->staticrank);
zh->records_inserted++;
}
else
{
/* record already exists */
- struct recKeys delkeys;
+ zebra_rec_keys_t delkeys = zebra_rec_keys_open();
#if NATTR
- struct recKeys sortKeys;
+ zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
#else
struct sortKeys sortKeys;
#endif
if (!allow_update)
{
- if (show_progress)
- yaz_log (YLOG_LOG, "skipped %s %s %ld",
+ yaz_log (YLOG_LOG, "skipped %s %s %ld",
recordType, pr_fname, (long) recordOffset);
logRecord(zh);
return ZEBRA_FAIL;
assert (rec);
recordAttr = rec_init_attr (zh->reg->zei, rec);
-
- if (!force_update) {
- if (recordAttr->runNumber ==
- zebraExplain_runNumberIncrement (zh->reg->zei, 0))
- {
- if (show_progress)
- 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 ZEBRA_FAIL;
- }
- }
-
- delkeys.buf_used = rec->size[recInfo_delKeys];
- delkeys.buf = rec->info[recInfo_delKeys];
+ zebra_rec_keys_set_buf(delkeys,
+ rec->info[recInfo_delKeys],
+ rec->size[recInfo_delKeys],
+ 0);
+#if NATTR
+ zebra_rec_keys_set_buf(sortKeys,
+ rec->info[recInfo_sortKeys],
+ rec->size[recInfo_sortKeys],
+ 0);
+#else
sortKeys.buf_used = rec->size[recInfo_sortKeys];
sortKeys.buf = rec->info[recInfo_sortKeys];
+#endif
+#if NATTR
+ extract_flushSortKeys (zh, *sysno, 0, sortKeys);
+#else
extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
- extract_flushRecordKeys (zh, *sysno, 0, &delkeys, 0);
+#endif
+ extract_flushRecordKeys (zh, *sysno, 0, delkeys,
+ recordAttr->staticrank);
if (delete_flag)
{
/* record going to be deleted */
- if (!delkeys.buf_used)
+ if (zebra_rec_keys_empty(delkeys))
{
- if (show_progress)
- {
- yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
- pr_fname, (long) recordOffset);
- yaz_log (YLOG_WARN, "cannot delete file above, "
- "storeKeys false");
- }
+ yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
+ pr_fname, (long) recordOffset);
+ yaz_log (YLOG_WARN, "cannot delete file above, "
+ "storeKeys false (3)");
}
else
{
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);
}
else
{
- /* record going to be updated */
- if (!delkeys.buf_used)
- {
- if (show_progress)
- {
- yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
- pr_fname, (long) recordOffset);
- yaz_log (YLOG_WARN, "cannot update file above, storeKeys false");
- }
- }
- else
- {
- if (show_progress)
+ if (show_progress)
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, 0);
- zh->records_updated++;
- }
+ recordAttr->staticrank = extractCtrl.staticrank;
+#if NATTR
+ extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
+#else
+ extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
+#endif
+ extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
+ recordAttr->staticrank);
+ zh->records_updated++;
}
+ zebra_rec_keys_close(delkeys);
+#if NATTR
+ zebra_rec_keys_close(sortKeys);
+#endif
}
/* update file type */
xfree (rec->info[recInfo_fileType]);
/* update delete keys */
xfree (rec->info[recInfo_delKeys]);
- if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1)
+ if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 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;
+ zebra_rec_keys_get_buf(zh->reg->keys,
+ &rec->info[recInfo_delKeys],
+ &rec->size[recInfo_delKeys]);
}
else
{
rec->info[recInfo_delKeys] = NULL;
rec->size[recInfo_delKeys] = 0;
}
-
/* update sort keys */
xfree (rec->info[recInfo_sortKeys]);
+#if NATTR
+ zebra_rec_keys_get_buf(zh->reg->sortKeys,
+ &rec->info[recInfo_sortKeys],
+ &rec->size[recInfo_sortKeys]);
+#else
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;
+#endif
/* save file size of original record */
zebraExplain_recordBytesIncrement (zh->reg->zei,
abort ();
}
- create_rec_keys_codec(&zh->reg->keys);
+ zebra_rec_keys_reset(zh->reg->keys);
+
#if NATTR
- create_rec_keys_codec(&zh->reg->sortKeys);
+ zebra_rec_keys_reset(zh->reg->sortKeys);
#else
zh->reg->sortKeys.buf_used = 0;
#endif
if (rec->size[recInfo_delKeys])
{
- struct recKeys delkeys;
+ zebra_rec_keys_t delkeys = zebra_rec_keys_open();
+
#if NATTR
- struct recKeys sortkeys;
+ zebra_rec_keys_t sortkeys = zebra_rec_keys_open();
#else
struct sortKeys sortkeys;
#endif
- delkeys.buf_used = rec->size[recInfo_delKeys];
- delkeys.buf = rec->info[recInfo_delKeys];
+ zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
+ rec->size[recInfo_delKeys],
+ 0);
+ extract_flushRecordKeys (zh, rec->sysno, 0, delkeys, 0);
+ zebra_rec_keys_close(delkeys);
+#if NATTR
+ zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
+ rec->size[recInfo_sortKeys],
+ 0);
+ extract_flushSortKeys (zh, rec->sysno, 0, sortkeys);
+ zebra_rec_keys_close(sortkeys);
+#else
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, 0);
+#endif
}
- extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys, 0);
+ extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
+#if NATTR
+ extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys);
+#else
extract_flushSortKeys (zh, rec->sysno, 1, &zh->reg->sortKeys);
+#endif
xfree (rec->info[recInfo_delKeys]);
- 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;
+ zebra_rec_keys_get_buf(zh->reg->keys,
+ &rec->info[recInfo_delKeys],
+ &rec->size[recInfo_delKeys]);
xfree (rec->info[recInfo_sortKeys]);
+#if NATTR
+ zebra_rec_keys_get_buf(zh->reg->sortKeys,
+ &rec->info[recInfo_sortKeys],
+ &rec->size[recInfo_sortKeys]);
+#else
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;
+#endif
return 0;
}
void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
- int cmd, struct recKeys *reckeys,
+ int cmd,
+ zebra_rec_keys_t reckeys,
zint staticrank)
{
- void *decode_handle = iscz1_start();
- int off = 0;
- int ch = 0;
ZebraExplainInfo zei = zh->reg->zei;
if (!zh->reg->key_buf)
}
zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
- while (off < reckeys->buf_used)
+ if (zebra_rec_keys_rewind(reckeys))
{
- const char *src = reckeys->buf + off;
- struct it_key key_in;
- struct it_key key_out;
- char *dst = (char*) &key_in;
- zint *keyp = key_out.mem;
- size_t key_idx = 0;
-
- iscz1_decode(decode_handle, &dst, &src);
- assert(key_in.len == 4);
-
- if (zh->reg->key_buf_used + 1024 >
- (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
- 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;
-
- ch = (int) key_in.mem[0]; /* ordinal for field/use/attribute */
-
- 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 (zh->m_staticrank) /* rank config enabled ? */
+ size_t slen;
+ const char *str;
+ struct it_key key_in;
+ while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
{
- *keyp++ = staticrank;
- key_out.len = 4;
- }
- else
- key_out.len = 3;
+ int ch = 0;
+ struct it_key key_out;
+ zint *keyp = key_out.mem;
- if (key_in.mem[1]) /* filter specified record ID */
- *keyp++ = key_in.mem[1];
- else
- *keyp++ = sysno;
- *keyp++ = key_in.mem[2]; /* section_id */
- *keyp++ = key_in.mem[3]; /* sequence .. */
+ assert(key_in.len == 4);
+
+ /* check for buffer overflow */
+ if (zh->reg->key_buf_used + 1024 >
+ (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*))
+ 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;
+
+ /* encode the ordinal value (field/use/attribute) .. */
+ ch = (int) key_in.mem[0];
+ 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;
+ ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = '\0';
- memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used,
- &key_out, sizeof(key_out));
- (zh->reg->key_buf_used) += sizeof(key_out);
- off = src - reckeys->buf;
+ /* the delete/insert indicator */
+ ((char*)zh->reg->key_buf)[(zh->reg->key_buf_used)++] = cmd;
+
+ 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;
+ }
+ else
+ key_out.len = 3;
+
+ if (key_in.mem[1]) /* filter specified record ID */
+ *keyp++ = key_in.mem[1];
+ else
+ *keyp++ = sysno;
+ *keyp++ = key_in.mem[2]; /* section_id */
+ *keyp++ = key_in.mem[3]; /* sequence .. */
+
+ memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used,
+ &key_out, sizeof(key_out));
+ (zh->reg->key_buf_used) += sizeof(key_out);
+ }
}
- assert (off == reckeys->buf_used);
- iscz1_stop(decode_handle);
}
void extract_flushWriteKeys (ZebraHandle zh, int final)
zh->reg->key_buf_used = 0;
}
-void extract_add_it_key (ZebraHandle zh,
- struct recKeys *keys,
- int reg_type,
- const char *str, int slen, struct it_key *key)
-{
- char *dst;
- const char *src = (char*) key;
-
- if (keys->buf_used+1024 > keys->buf_max)
- {
- char *b = (char *) xmalloc (keys->buf_max += 128000);
- if (keys->buf_used > 0)
- memcpy (b, keys->buf, keys->buf_used);
- xfree (keys->buf);
- keys->buf = b;
- }
- dst = keys->buf + keys->buf_used;
-
- iscz1_encode(keys->codec_handle, &dst, &src);
-
-#if REG_TYPE_PREFIX
- *dst++ = reg_type;
-#endif
- memcpy (dst, str, slen);
- dst += slen;
- *dst++ = '\0';
- keys->buf_used = dst - keys->buf;
-}
-
-ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, struct recKeys *reckeys,
+ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
+ zebra_rec_keys_t reckeys,
zebra_snippets *snippets)
{
- void *decode_handle = iscz1_start();
- int off = 0;
- int seqno = 0;
NMEM nmem = nmem_create();
-
- yaz_log(YLOG_LOG, "zebra_rec_keys_snippets buf=%p sz=%d", reckeys->buf,
- reckeys->buf_used);
- assert(reckeys->buf);
- while (off < reckeys->buf_used)
+ if (zebra_rec_keys_rewind(reckeys))
{
- const char *src = reckeys->buf + off;
- struct it_key key;
- char *dst = (char*) &key;
- char dst_buf[IT_MAX_WORD];
- char *dst_term = dst_buf;
-
- iscz1_decode(decode_handle, &dst, &src);
- assert(key.len <= 4 && key.len > 2);
-
- seqno = (int) key.mem[key.len-1];
-
- zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
- zebra_snippets_append(snippets, seqno, src[0], key.mem[0], dst_term);
- while (*src++)
- ;
- off = src - reckeys->buf;
- nmem_reset(nmem);
+ const char *str;
+ size_t slen;
+ struct it_key key;
+ while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+ {
+ char dst_buf[IT_MAX_WORD];
+ char *dst_term = dst_buf;
+ int ord, seqno;
+ int index_type;
+ assert(key.len <= 4 && key.len > 2);
+ seqno = (int) key.mem[key.len-1];
+ ord = key.mem[0];
+
+ zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
+ 0/* db */, 0/* set */, 0/* use */,
+ 0 /* string index */);
+ assert(index_type);
+ zebra_term_untrans_iconv(zh, nmem, index_type,
+ &dst_term, str);
+ zebra_snippets_append(snippets, seqno, ord, dst_term);
+ nmem_reset(nmem);
+ }
}
nmem_destroy(nmem);
- iscz1_stop(decode_handle);
return ZEBRA_OK;
}
-void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys)
+void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
{
- void *decode_handle = iscz1_start();
- int off = 0;
- int seqno = 0;
- NMEM nmem = nmem_create();
-
- yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
- reckeys->buf_used);
- assert(reckeys->buf);
- while (off < reckeys->buf_used)
+ yaz_log(YLOG_LOG, "print_rec_keys");
+ if (zebra_rec_keys_rewind(reckeys))
{
- const char *src = reckeys->buf + off;
- struct it_key key;
- char *dst = (char*) &key;
- int attrSet, attrUse;
- char dst_buf[IT_MAX_WORD];
- char *dst_term = dst_buf;
-
- 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];
-
- zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1);
-
- yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
- key.mem[0], seqno, dst_term);
- while (*src++)
- ;
- off = src - reckeys->buf;
- nmem_reset(nmem);
+ const char *str;
+ size_t slen;
+ struct it_key key;
+ while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
+ {
+ char dst_buf[IT_MAX_WORD];
+ int seqno;
+ int index_type;
+ const char *db = 0;
+ assert(key.len <= 4 && key.len > 2);
+
+ zebraExplain_lookup_ord(zh->reg->zei,
+ key.mem[0], &index_type, &db, 0, 0, 0);
+
+ seqno = (int) key.mem[key.len-1];
+
+ zebra_term_untrans(zh, index_type, dst_buf, str);
+
+ yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
+ key.mem[0], seqno, dst_buf);
+ }
}
- nmem_destroy(nmem);
- iscz1_stop(decode_handle);
}
void extract_add_index_string (RecWord *p, const char *str, int length)
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);
+ if (1)
+ {
+ char strz[80];
+ int i;
+
+ strz[0] = 0;
+ for (i = 0; i<length && i < 20; i++)
+ sprintf(strz+strlen(strz), "%02X", str[i] & 0xff);
+ /* just for debugging .. */
+ yaz_log(YLOG_LOG, "add: set=%d use=%d "
+ "record_id=%lld section_id=%lld seqno=%lld %s",
+ p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno,
+ strz);
+ }
#endif
- extract_add_it_key(p->extractCtrl->handle,
- &zh->reg->keys,
- p->index_type, str,
- length, &key);
+ zebra_rec_keys_write(zh->reg->keys, str, length, &key);
}
#if NATTR
key.mem[2] = p->section_id;
key.mem[3] = p->seqno;
- extract_add_it_key(p->extractCtrl->handle,
- &zh->reg->sortKeys,
- p->index_type, str,
- length, &key);
+ zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
}
#else
static void extract_add_sort_string (RecWord *p, const char *str, int length)
{
int set, use, slen;
- off += key_SU_decode(&set, sk->buf + off);
- off += key_SU_decode(&use, sk->buf + off);
- off += key_SU_decode(&slen, sk->buf + off);
+ off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
+ off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
+ off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
off += slen;
if (p->attrSet == set && p->attrUse == use)
return;
}
#if NATTR
+#error not done yet with zebra_rec_keys_t
void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
int cmd, struct recKeys *reckeys)
{
{
int set, use, slen;
- off += key_SU_decode(&set, sk->buf + off);
- off += key_SU_decode(&use, sk->buf + off);
- off += key_SU_decode(&slen, sk->buf + off);
+ off += key_SU_decode(&set, (unsigned char *) sk->buf + off);
+ off += key_SU_decode(&use, (unsigned char *) sk->buf + off);
+ off += key_SU_decode(&slen, (unsigned char *) sk->buf + off);
sortIdx_type(sortIdx, use);
if (cmd == 1)
i->prevcmd=-1;
i->keylen=0;
i->encode_handle = iscz1_start();
+ i->decode_handle = iscz1_start();
}
#define OLDENCODE 1
/* 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