-/* $Id: extract.c,v 1.214 2006-05-18 12:03:05 adam Exp $
+/* $Id: extract.c,v 1.220 2006-06-13 12:02:06 adam Exp $
Copyright (C) 1995-2006
Index Data ApS
#endif
+#define ENCODE_BUFLEN 768
+struct encode_info {
+ void *encode_handle;
+ void *decode_handle;
+ char buf[ENCODE_BUFLEN];
+};
+
+static int log_level = 0;
+static int log_level_initialized = 1;
+
+static void zebra_init_log_level()
+{
+ if (!log_level_initialized)
+ {
+ log_level = yaz_log_module_level("extract");
+ log_level_initialized = 1;
+ }
+}
+
+static void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
+ int cmd, zebra_rec_keys_t reckeys,
+ zint staticrank);
+static void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno,
+ int cmd, zebra_rec_keys_t skp);
+static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
+static void extract_token_add (RecWord *p);
+
+static void encode_key_init (struct encode_info *i);
+static void encode_key_write (char *k, struct encode_info *i, FILE *outf);
+static void encode_key_flush (struct encode_info *i, FILE *outf);
#define USE_SHELLSORT 0
++zh->records_processed;
if (!(zh->records_processed % 1000))
{
- 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);
+ 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->zebra_maps = p->zebra_maps;
w->seqno = 1;
-#if NATTR
-#else
- w->attrSet = VAL_BIB1;
- w->attrUse = 1016;
-#endif
- w->index_name = 0;
+ w->index_name = "any";
w->index_type = 'w';
w->extractCtrl = p;
w->record_id = 0;
static void searchRecordKey(ZebraHandle zh,
zebra_rec_keys_t reckeys,
- int attrSetS, int attrUseS,
+ const char *index_name,
const char **ws, int ws_length)
{
int i;
- int ch;
+ int ch = -1;
for (i = 0; i<ws_length; i++)
ws[i] = NULL;
- ch = zebraExplain_lookup_attr_su_any_index(zh->reg->zei,
- attrSetS, attrUseS);
+ if (ch < 0)
+ ch = zebraExplain_lookup_attr_str(zh->reg->zei, '0', index_name);
+ if (ch < 0)
+ ch = zebraExplain_lookup_attr_str(zh->reg->zei, 'p', index_name);
+ if (ch < 0)
+ ch = zebraExplain_lookup_attr_str(zh->reg->zei, 'w', index_name);
+
if (ch < 0)
return ;
if (zebra_rec_keys_rewind(reckeys))
{
- int startSeq = -1;
+ zint startSeq = -1;
const char *str;
size_t slen;
struct it_key key;
if (key.mem[0] == ch)
{
- int woff;
+ zint woff;
if (startSeq == -1)
startSeq = seqno;
{
const char *ws[32];
char attset_str[64], attname_str[64];
- data1_attset *attset;
int i;
- int attSet = 1, attUse = 1;
int first = 1;
for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
for (; strchr(FILE_MATCH_BLANK, *s); s++)
;
- if (*s == ',')
+ if (*s != ',')
+ strcpy(attname_str, attset_str);
+ else
{
for (s++; strchr(FILE_MATCH_BLANK, *s); s++)
;
attname_str[i++] = *s;
attname_str[i] = '\0';
}
-
- if ((attset = data1_get_attset (zh->reg->dh, attset_str)))
- {
- data1_att *att;
- attSet = attset->reference;
- att = data1_getattbyname(zh->reg->dh, attset, attname_str);
- if (att)
- attUse = att->value;
- else
- attUse = atoi (attname_str);
- }
- searchRecordKey (zh, reckeys, attSet, attUse, ws, 32);
+
+ searchRecordKey (zh, reckeys, attname_str, ws, 32);
if (*s != ')')
{
RecType recType;
void *recTypeClientData;
+ zebra_init_log_level();
+
if (!zh->m_group || !*zh->m_group)
*gprefix = '\0';
else
sprintf (gprefix, "%s.", zh->m_group);
- yaz_log (YLOG_DEBUG, "fileExtract %s", fname);
+ yaz_log(log_level, "zebra_extract_file %s", fname);
/* determine file extension */
*ext = '\0';
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 zebra_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;
const char *pr_fname = fname; /* filename to print .. */
int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
+ zebra_init_log_level();
+
if (!pr_fname)
pr_fname = "<no file>"; /* make it printable if file is omitted */
if (recordType && *recordType)
{
- yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType);
+ yaz_log(log_level, "Record type explicitly specified: %s", recordType);
recType = recType_byName (zh->reg->recTypes, zh->res, recordType,
&clientData);
}
yaz_log (YLOG_WARN, "No such record type defined");
return ZEBRA_FAIL;
}
- yaz_log (YLOG_DEBUG, "Get record type from rgroup: %s",zh->m_record_type);
+ yaz_log(log_level, "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;
/* record going to be deleted */
if (zebra_rec_keys_empty(delkeys))
{
- yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
- pr_fname, (long) recordOffset);
- yaz_log (YLOG_WARN, "cannot delete file above, "
- "storeKeys false (3)");
+ 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
{
if (show_progress)
- yaz_log (YLOG_LOG, "delete %s %s %ld", recordType,
- pr_fname, (long) recordOffset);
+ yaz_log(YLOG_LOG, "delete %s %s %ld", recordType,
+ pr_fname, (long) recordOffset);
zh->records_deleted++;
if (matchStr)
{
else
{
if (show_progress)
- yaz_log (YLOG_LOG, "update %s %s %ld", recordType,
- pr_fname, (long) recordOffset);
+ yaz_log(YLOG_LOG, "update %s %s %ld", recordType,
+ pr_fname, (long) recordOffset);
recordAttr->staticrank = extractCtrl.staticrank;
extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys,
return ZEBRA_OK;
}
-int explain_extract (void *handle, Record rec, data1_node *n)
+ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n)
{
ZebraHandle zh = (ZebraHandle) handle;
struct recExtractCtrl extractCtrl;
zebra_rec_keys_get_buf(zh->reg->sortKeys,
&rec->info[recInfo_sortKeys],
&rec->size[recInfo_sortKeys]);
-
- return 0;
+ return ZEBRA_OK;
}
void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
struct it_key key_in;
while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
{
- int ord = key_in.mem[0];
+ int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
for (p = ord_list; p ; p = p->next)
if (p->ord == ord)
(char*)zh->reg->key_buf + zh->reg->key_buf_used;
/* encode the ordinal value (field/use/attribute) .. */
- ch = (int) key_in.mem[0];
+ ch = CAST_ZINT_TO_INT(key_in.mem[0]);
zh->reg->key_buf_used +=
key_SU_encode(ch, (char*)zh->reg->key_buf +
zh->reg->key_buf_used);
#endif
if (!zh->reg->key_buf || ptr_i <= 0)
{
- yaz_log (YLOG_DEBUG, " nothing to flush section=%d buf=%p i=%d",
+ yaz_log(log_level, " 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 ",
+ yaz_log(log_level, " buf=%p ",
zh->reg->key_buf);
- yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i);
- yaz_log (YLOG_DEBUG, " reg=%p ",zh->reg);
+ yaz_log(log_level, " ptr=%d ",zh->reg->ptr_i);
+ yaz_log(log_level, " reg=%p ",zh->reg);
return;
}
(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",
+ yaz_log(log_level, " 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,
{
char dst_buf[IT_MAX_WORD];
char *dst_term = dst_buf;
- int ord, seqno;
+ int ord;
+ zint seqno;
int index_type;
+
assert(key.len <= 4 && key.len > 2);
- seqno = (int) key.mem[key.len-1];
- ord = key.mem[0];
+ seqno = key.mem[key.len-1];
+ ord = CAST_ZINT_TO_INT(key.mem[0]);
zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type,
- 0/* db */, 0/* set */, 0/* use */,
- 0 /* string_index */);
+ 0/* db */, 0 /* string_index */);
assert(index_type);
zebra_term_untrans_iconv(zh, nmem, index_type,
&dst_term, str);
while (zebra_rec_keys_read(reckeys, &str, &slen, &key))
{
char dst_buf[IT_MAX_WORD];
- int seqno;
+ zint seqno;
int index_type;
+ int ord = CAST_ZINT_TO_INT(key.mem[0]);
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);
+ zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, &db, 0);
- seqno = (int) key.mem[key.len-1];
+ seqno = 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);
+ yaz_log(YLOG_LOG, "ord=%d seqno=" ZINT_FORMAT
+ " term=%s", ord, seqno, dst_buf);
}
}
}
-void extract_add_index_string(RecWord *p, const char *str, int length)
+static void extract_add_index_string(RecWord *p, const char *str, int length)
{
struct it_key key;
ZebraExplainInfo zei = zh->reg->zei;
int ch;
- if (p->index_name)
- {
- ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
- if (ch < 0)
- ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
- }
- else
- {
-#if NATTR
- return;
-#else
- ch = zebraExplain_lookup_attr_su(zei, p->index_type,
- p->attrSet, p->attrUse);
- if (ch < 0)
- ch = zebraExplain_add_attr_su(zei, p->index_type,
- p->attrSet, p->attrUse);
-#endif
- }
+ if (!p->index_name)
+ return;
+
+ ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
+ if (ch < 0)
+ ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
+
key.len = 4;
key.mem[0] = ch;
key.mem[1] = p->record_id;
ZebraExplainInfo zei = zh->reg->zei;
int ch;
- if (p->index_name)
- {
- ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
- if (ch < 0)
- ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
- }
- else
- {
-#if NATTR
- return;
-#else
- ch = zebraExplain_lookup_attr_su(zei, p->index_type,
- VAL_IDXPATH, p->attrUse);
- if (ch < 0)
- ch = zebraExplain_add_attr_su(zei, p->index_type,
- VAL_IDXPATH, p->attrUse);
-#endif
- }
+ if (!p->index_name)
+ return;
+
+ ch = zebraExplain_lookup_attr_str(zei, p->index_type, p->index_name);
+ if (ch < 0)
+ ch = zebraExplain_add_attr_str(zei, p->index_type, p->index_name);
key.len = 4;
key.mem[0] = ch;
key.mem[1] = p->record_id;
zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
}
-void extract_add_string (RecWord *p, const char *string, int length)
+static void extract_add_string (RecWord *p, const char *string, int length)
{
assert (length > 0);
if (zebra_maps_is_sort (p->zebra_maps, p->index_type))
int remain = p->term_len;
const char **map = 0;
- yaz_log(YLOG_DEBUG, "Incomplete field, w='%.*s'", p->term_len, p->term_buf);
-
if (remain > 0)
map = zebra_maps_input(p->zebra_maps, p->index_type, &b, remain, 0);
const char **map = 0;
int i = 0, remain = p->term_len;
- yaz_log(YLOG_DEBUG, "Complete field, w='%.*s'",
- p->term_len, p->term_buf);
-
if (remain > 0)
map = zebra_maps_input (p->zebra_maps, p->index_type, &b, remain, 1);
{
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++);
}
extract_add_string (p, buf, i);
}
-void extract_token_add (RecWord *p)
+static void extract_token_add(RecWord *p)
{
WRBUF wrbuf;
-#if 0
- 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);
-#endif
+ if (log_level)
+ yaz_log(log_level, "extract_token_add "
+ "type=%c index=%s seqno=" ZINT_FORMAT " s=%.*s",
+ p->index_type, p->index_name,
+ p->seqno, p->term_len, p->term_buf);
if ((wrbuf = zebra_replace(p->zebra_maps, p->index_type, 0,
p->term_buf, p->term_len)))
{
p->setStoreData = extract_set_store_data_cb;
}
-void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
+static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid)
{
ZebraHandle zh = (ZebraHandle) p->handle;
zebraExplain_addSchema (zh->reg->zei, oid);
while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
{
- int ord = (int) key_in.mem[0];
+ int ord = CAST_ZINT_TO_INT(key_in.mem[0]);
sortIdx_type(sortIdx, ord);
if (cmd == 1)
}
}
-void encode_key_init (struct encode_info *i)
+static 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;
i->encode_handle = iscz1_start();
i->decode_handle = iscz1_start();
}
-#define OLDENCODE 1
-
-#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)
+static void encode_key_write (char *k, struct encode_info *i, FILE *outf)
{
struct it_key key;
char *bp = i->buf, *bp0;
#endif
}
-void encode_key_flush (struct encode_info *i, FILE *outf)
+static void encode_key_flush (struct encode_info *i, FILE *outf)
{
iscz1_stop(i->encode_handle);
iscz1_stop(i->decode_handle);
}
-#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)
- {
- yaz_log (YLOG_FATAL|YLOG_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)
- {
- yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite");
- exit (1);
- }
- i->keylen=0; /* ok, it's written, forget it */
- i->prevsys=0; /* forget the values too */
- i->prevseq=0;
-}
-#endif
/*
* Local variables:
* c-basic-offset: 4
projects / idzebra-moved-to-github.git / blobdiff