-/* $Id: extract.c,v 1.196 2005-10-28 07:25:30 adam Exp $
+/* $Id: extract.c,v 1.197 2005-10-28 09:22:50 adam Exp $
Copyright (C) 1995-2005
Index Data ApS
}
static void searchRecordKey(ZebraHandle zh,
-#if NEW_REC_KEYS
zebra_rec_keys_t reckeys,
-#else
- const struct recKeys *reckeys,
-#endif
int attrSetS, int attrUseS,
const char **ws, int ws_length)
{
-#if NEW_REC_KEYS
-#else
- void *decode_handle = iscz1_start();
- int off = 0;
- int startSeq = -1;
- int seqno = 0;
-#endif
int i;
int ch;
if (ch < 0)
return ;
-#if NEW_REC_KEYS
if (zebra_rec_keys_rewind(reckeys))
{
int startSeq = -1;
}
}
}
-#else
- while (off < reckeys->buf_used)
- {
- 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;
- }
-
- while (*src++)
- ;
- off = src - reckeys->buf;
- }
- iscz1_stop(decode_handle);
- assert (off == reckeys->buf_used);
-#endif
}
struct file_read_info {
}
static char *fileMatchStr (ZebraHandle zh,
-#if NEW_REC_KEYS
zebra_rec_keys_t reckeys,
-#else
- struct recKeys *reckeys,
-#endif
const char *fname, const char *spec)
{
static char dstBuf[2048]; /* static here ??? */
struct recordGroup *rGroup;
};
-#if NEW_REC_KEYS
-#else
-void create_rec_keys_codec(struct recKeys *keys)
-{
- keys->buf_used = 0;
- iscz1_reset(keys->codec_handle);
-}
-#endif
-
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 */
-#if NEW_REC_KEYS
zebra_rec_keys_reset(zh->reg->keys);
-#else
- create_rec_keys_codec(&zh->reg->keys);
-#endif
#if NATTR
-#if NEW_REC_KEYS
zebra_rec_keys_reset(zh->reg->sortKeys);
#else
- create_rec_keys_codec(&zh->reg->sortKeys);
-#endif
-#else
zh->reg->sortKeys.buf_used = 0;
#endif
recordOffset = fi->file_moffset;
if (matchStr == 0 && zh->m_record_id && *zh->m_record_id)
{
-#if NEW_REC_KEYS
matchStr = fileMatchStr (zh, zh->reg->keys, fname,
zh->m_record_id);
-#else
- matchStr = fileMatchStr (zh, &zh->reg->keys, fname,
- zh->m_record_id);
-#endif
if (!matchStr)
{
yaz_log(YLOG_WARN, "Bad match criteria");
}
}
}
- if (! *sysno
-#if NEW_REC_KEYS
- && zebra_rec_keys_empty(zh->reg->keys)
-#else
- && zh->reg->keys.buf_used == 0
-#endif
- )
+ 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 */
dict_insert (zh->reg->matchDict, matchStr, sizeof(*sysno), sysno);
}
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-#if NEW_REC_KEYS
extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
recordAttr->staticrank);
-#else
- extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
- recordAttr->staticrank);
-#endif
zh->records_inserted++;
}
else
{
/* record already exists */
-#if NEW_REC_KEYS
zebra_rec_keys_t delkeys = zebra_rec_keys_open();
-#else
- struct recKeys delkeys;
-#endif
#if NATTR
-#if NEW_REC_KEYS
zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
#else
- struct recKeys sortKeys;
-#endif
-#else
struct sortKeys sortKeys;
#endif
recordAttr = rec_init_attr (zh->reg->zei, rec);
-#if NEW_REC_KEYS
zebra_rec_keys_set_buf(delkeys,
rec->info[recInfo_delKeys],
rec->size[recInfo_delKeys],
0);
-#else
- /* flush old keys for sort&search etc. */
- delkeys.buf_used = rec->size[recInfo_delKeys];
- delkeys.buf = rec->info[recInfo_delKeys];
-#endif
-#if NEW_REC_KEYS && NATTR
+#if NATTR
zebra_rec_keys_set_buf(sortKeys,
rec->info[recInfo_sortKeys],
rec->size[recInfo_sortKeys],
extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
#endif
-#if NEW_REC_KEYS
extract_flushRecordKeys (zh, *sysno, 0, delkeys,
recordAttr->staticrank); /* old values */
-#else
- extract_flushRecordKeys (zh, *sysno, 0, &delkeys,
- recordAttr->staticrank); /* old values */
-#endif
if (deleteFlag)
{
/* record going to be deleted */
- if (
-#if NEW_REC_KEYS
- zebra_rec_keys_empty(delkeys)
-#else
- !delkeys.buf_used
-#endif
- )
+ if (zebra_rec_keys_empty(delkeys))
{
yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T,
zh->m_record_type, fname, recordOffset);
else
{
/* record going to be updated */
- if (
-#if NEW_REC_KEYS
- zebra_rec_keys_empty(delkeys)
-#else
- !delkeys.buf_used
-#endif
- )
+ if (zebra_rec_keys_empty(delkeys))
{
yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T,
zh->m_record_type, fname, recordOffset);
zh->m_record_type, fname, recordOffset);
recordAttr->staticrank = extractCtrl.staticrank;
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-#if NEW_REC_KEYS
extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
recordAttr->staticrank);
-#else
- extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
- recordAttr->staticrank);
-#endif
zh->records_updated++;
}
}
-#if NEW_REC_KEYS
zebra_rec_keys_close(delkeys);
-#endif
-#if NATTR && NEW_REC_KWYS
+#if NATTR
zebra_rec_keys_close(sortKeys);
#endif
}
/* update delete keys */
xfree (rec->info[recInfo_delKeys]);
- if (
-#if NEW_REC_KEYS
- !zebra_rec_keys_empty(zh->reg->keys)
-#else
- zh->reg->keys.buf_used > 0
-#endif
- && zh->m_store_keys == 1)
+ if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
{
-#if NEW_REC_KEYS
zebra_rec_keys_get_buf(zh->reg->keys,
&rec->info[recInfo_delKeys],
&rec->size[recInfo_delKeys]);
-#else
-
- 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;
-#endif
}
else
{
extractCtrl.first_record = 1;
extractCtrl.fh = &fc;
-#if NEW_REC_KEYS
zebra_rec_keys_reset(zh->reg->keys);
-#else
- create_rec_keys_codec(&zh->reg->keys);
-#endif
+
#if NATTR
-#if NEW_REC_KEYS
zebra_rec_keys_reset(zh->reg->sortKeys);
#else
- create_rec_keys_codec(&zh->reg->sortKeys);
-#endif
-#else
zh->reg->sortKeys.buf_used = 0;
#endif
if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0]))
matchStr = match_criteria;
} else {
if (zh->m_record_id && *zh->m_record_id) {
-#if NEW_REC_KEYS
matchStr = fileMatchStr (zh, zh->reg->keys, pr_fname,
zh->m_record_id);
-#else
- matchStr = fileMatchStr (zh, &zh->reg->keys, pr_fname,
- zh->m_record_id);
-#endif
if (!matchStr)
{
yaz_log (YLOG_WARN, "Bad match criteria (recordID)");
}
}
}
- if (
-#if NEW_REC_KEYS
- zebra_rec_keys_empty(zh->reg->keys)
-#else
- zh->reg->keys.buf_used == 0
-#endif
-)
+ if (zebra_rec_keys_empty(zh->reg->keys))
{
/* the extraction process returned no information - the record
is probably empty - unless flagShowRecords is in use */
}
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-#if NEW_REC_KEYS
#if 0
print_rec_keys(zh, zh->reg->keys);
#endif
extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
recordAttr->staticrank);
-#else
-#if 0
- print_rec_keys(zh, &zh->reg->keys);
-#endif
- extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
- recordAttr->staticrank);
-#endif
-
zh->records_inserted++;
}
else
{
/* record already exists */
-#if NEW_REC_KEYS
zebra_rec_keys_t delkeys = zebra_rec_keys_open();
-#else
- struct recKeys delkeys;
-#endif
#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 NEW_REC_KEYS
zebra_rec_keys_set_buf(delkeys,
rec->info[recInfo_delKeys],
rec->size[recInfo_delKeys],
0);
-#else
- delkeys.buf_used = rec->size[recInfo_delKeys];
- delkeys.buf = rec->info[recInfo_delKeys];
-#endif
sortKeys.buf_used = rec->size[recInfo_sortKeys];
sortKeys.buf = rec->info[recInfo_sortKeys];
-#if NEW_REC_KEYS && NATTR
+#if NATTR
extract_flushSortKeys (zh, *sysno, 0, sortKeys);
#else
extract_flushSortKeys (zh, *sysno, 0, &sortKeys);
#endif
-#if NEW_REC_KEYS
extract_flushRecordKeys (zh, *sysno, 0, delkeys,
recordAttr->staticrank);
-#else
- extract_flushRecordKeys (zh, *sysno, 0, &delkeys,
- recordAttr->staticrank);
-#endif
if (delete_flag)
{
/* record going to be deleted */
- if (
-#if NEW_REC_KEYS
- zebra_rec_keys_empty(delkeys)
-#else
- !delkeys.buf_used
-#endif
- )
+ if (zebra_rec_keys_empty(delkeys))
{
if (show_progress)
{
else
{
/* record going to be updated */
- if (
-#if NEW_REC_KEYS
- zebra_rec_keys_empty(delkeys)
-#else
- !delkeys.buf_used
-#endif
- )
+ if (zebra_rec_keys_empty(delkeys))
{
if (show_progress)
{
pr_fname, (long) recordOffset);
recordAttr->staticrank = extractCtrl.staticrank;
extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys);
-#if NEW_REC_KEYS
extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
recordAttr->staticrank);
-#else
- extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys,
- recordAttr->staticrank);
-#endif
zh->records_updated++;
}
}
-#if NEW_REC_KEYS
zebra_rec_keys_close(delkeys);
-#endif
-#if NEW_REC_KEYS && NATTR
+#if NATTR
zebra_rec_keys_close(sortKeys);
#endif
}
/* update delete keys */
xfree (rec->info[recInfo_delKeys]);
- if (
-#if NEW_REC_KEYS
- !zebra_rec_keys_empty(zh->reg->keys)
-#else
- zh->reg->keys.buf_used > 0
-#endif
- && zh->m_store_keys == 1)
+ if (!zebra_rec_keys_empty(zh->reg->keys) && zh->m_store_keys == 1)
{
-#if NEW_REC_KEYS
zebra_rec_keys_get_buf(zh->reg->keys,
&rec->info[recInfo_delKeys],
&rec->size[recInfo_delKeys]);
-#else
-
- 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;
-#endif
}
else
{
abort ();
}
-#if NEW_REC_KEYS
zebra_rec_keys_reset(zh->reg->keys);
-#else
- create_rec_keys_codec(&zh->reg->keys);
-#endif
#if NATTR
-#if NEW_REC_KEYS
zebra_rec_keys_reset(zh->reg->sortKeys);
#else
- create_rec_keys_codec(&zh->reg->sortKeys);
-#endif
-#else
zh->reg->sortKeys.buf_used = 0;
#endif
extractCtrl.init = extract_init;
if (rec->size[recInfo_delKeys])
{
-#if NEW_REC_KEYS
zebra_rec_keys_t delkeys = zebra_rec_keys_open();
-#else
- struct recKeys delkeys;
-#endif
-
+
#if NATTR
-#if NEW_REC_KEYS
zebra_rec_keys_t sortkeys = zzebra_rec_keys_open();
#else
- struct recKeys sortkeys;
-#endif
-#else
struct sortKeys sortkeys;
#endif
-#if NEW_REC_KEYS
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);
-#else
- delkeys.buf_used = rec->size[recInfo_delKeys];
- delkeys.buf = rec->info[recInfo_delKeys];
- extract_flushRecordKeys (zh, rec->sysno, 0, &delkeys, 0);
-#endif
-#if NATTR && NEW_REC_KEYS
+#if NATTR
zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
rec->size[recInfo_sortKeys],
0);
extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys);
#endif
}
-#if NEW_REC_KEYS
extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0);
-#else
- extract_flushRecordKeys (zh, rec->sysno, 1, &zh->reg->keys, 0);
-#endif
-#if NATTR && NEW_REC_KEYS
+#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]);
-#if NEW_REC_KEYS
zebra_rec_keys_get_buf(zh->reg->keys,
&rec->info[recInfo_delKeys],
&rec->size[recInfo_delKeys]);
-#else
- 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;
-#endif
xfree (rec->info[recInfo_sortKeys]);
rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used;
void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
int cmd,
-#if NEW_REC_KEYS
zebra_rec_keys_t reckeys,
-#else
- struct recKeys *reckeys,
-#endif
zint staticrank)
{
-#if NEW_REC_KEYS
-#else
- void *decode_handle = iscz1_start();
- int off = 0;
- int ch = 0;
-#endif
ZebraExplainInfo zei = zh->reg->zei;
if (!zh->reg->key_buf)
}
zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
-#if NEW_REC_KEYS
if (zebra_rec_keys_rewind(reckeys))
{
size_t slen;
(zh->reg->key_buf_used) += sizeof(key_out);
}
}
-#else
- while (off < reckeys->buf_used)
- {
- 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;
-
- 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 ? */
- {
- *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);
- off = src - reckeys->buf;
- }
- assert (off == reckeys->buf_used);
- iscz1_stop(decode_handle);
-#endif
}
void extract_flushWriteKeys (ZebraHandle zh, int final)
zh->reg->key_buf_used = 0;
}
-#if NEW_REC_KEYS
void extract_add_it_key (ZebraHandle zh,
zebra_rec_keys_t *keys,
int reg_type,
{
zebra_rec_keys_write(*keys, reg_type, str, slen, key);
}
-#else
-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;
-}
-#endif
ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh,
-#if NEW_REC_KEYS
zebra_rec_keys_t reckeys,
-#else
- struct recKeys *reckeys,
-#endif
zebra_snippets *snippets)
{
NMEM nmem = nmem_create();
-
-#if NEW_REC_KEYS
if (zebra_rec_keys_rewind(reckeys))
{
const char *str;
nmem_reset(nmem);
}
}
-#else
- int off = 0;
- int seqno = 0;
- void *decode_handle = iscz1_start();
- assert(reckeys->buf);
- while (off < reckeys->buf_used)
- {
- 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;
- int index_type = 0, ord;
-
- iscz1_decode(decode_handle, &dst, &src);
- 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 */);
- assert(index_type);
- zebra_term_untrans_iconv(zh, nmem, index_type,
- &dst_term, src);
- zebra_snippets_append(snippets, seqno, ord, dst_term);
- while (*src++)
- ;
- off = src - reckeys->buf;
- nmem_reset(nmem);
- }
- iscz1_stop(decode_handle);
-#endif
nmem_destroy(nmem);
return ZEBRA_OK;
}
-void print_rec_keys(ZebraHandle zh,
-#if NEW_REC_KEYS
- zebra_rec_keys_t reckeys
-#else
- struct recKeys *reckeys
-#endif
-)
+void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys)
{
-#if NEW_REC_KEYS
yaz_log(YLOG_LOG, "print_rec_keys");
if (zebra_rec_keys_rewind(reckeys))
{
key.mem[0], seqno, dst_buf);
}
}
-#else
- int off = 0;
- int seqno = 0;
- void *decode_handle = iscz1_start();
- yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf,
- reckeys->buf_used);
- assert(reckeys->buf);
- while (off < reckeys->buf_used)
- {
- const char *src = reckeys->buf + off;
- struct it_key key;
- char *dst = (char*) &key;
- char dst_buf[IT_MAX_WORD];
- int index_type;
- const char *db = 0;
-
- iscz1_decode(decode_handle, &dst, &src);
- assert(key.len <= 4 && key.len > 2);
-
- seqno = (int) key.mem[key.len-1];
-
- zebraExplain_lookup_ord(zh->reg->zei,
- key.mem[0], &index_type, &db, 0, 0);
-
- zebra_term_untrans(zh, index_type, dst_buf, src);
-
- yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s",
- key.mem[0], seqno, dst_buf);
- while (*src++)
- ;
- off = src - reckeys->buf;
- }
- iscz1_stop(decode_handle);
-#endif
}
void extract_add_index_string (RecWord *p, const char *str, int length)
}
#if NATTR
+#error not done yet with zebra_rec_keys_t
void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
int cmd, struct recKeys *reckeys)
{