-/* $Id: extract.c,v 1.268 2007-11-06 10:29:59 adam Exp $
+/* $Id: extract.c,v 1.276 2007-12-20 11:21:29 adam Exp $
Copyright (C) 1995-2007
Index Data ApS
}
+static void snippet_add_icu(RecWord *p, int ord, zebra_map_t zm)
+{
+ struct snip_rec_info *h = p->extractCtrl->handle;
+
+ const char *res_buf = 0;
+ size_t res_len = 0;
+
+ const char *display_buf = 0;
+ size_t display_len = 0;
+
+ zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+ while (zebra_map_tokenize_next(zm, &res_buf, &res_len,
+ &display_buf, &display_len))
+ {
+ zebra_snippets_appendn(h->snippets, p->seqno, 0, ord,
+ display_buf, display_len);
+ p->seqno++;
+ }
+}
+
static void snippet_token_add(RecWord *p)
{
struct snip_rec_info *h = p->extractCtrl->handle;
int ch = zebraExplain_lookup_attr_str(
zei, zinfo_index_category_index, p->index_type, p->index_name);
- if (zebra_maps_is_complete(zm))
- snippet_add_complete_field(p, ch, zm);
+ if (zebra_maps_is_icu(zm))
+ snippet_add_icu(p, ch, zm);
else
- snippet_add_incomplete_field(p, ch, zm);
+ {
+ if (zebra_maps_is_complete(zm))
+ snippet_add_complete_field(p, ch, zm);
+ else
+ snippet_add_incomplete_field(p, ch, zm);
+ }
}
}
struct recordGroup *rGroup;
};
-static void all_matches_add(struct recExtractCtrl *ctrl)
+/** \brief add the always-matches index entry and map to real record ID
+ \param ctrl record control
+ \param record_id custom record ID
+ \param sysno system record ID
+
+ This function serves two purposes.. It adds the always matches
+ entry and makes a pointer from the custom record ID (if defined)
+ back to the system record ID (sysno)
+ See zebra_recid_to_sysno .
+ */
+static void all_matches_add(struct recExtractCtrl *ctrl, zint record_id,
+ zint sysno)
{
RecWord word;
extract_init(ctrl, &word);
+ word.record_id = record_id;
+ /* we use the seqno as placeholder for a way to get back to
+ record database from _ALLRECORDS.. This is used if a custom
+ RECORD was defined */
+ word.seqno = sysno;
word.index_name = "_ALLRECORDS";
word.index_type = "w";
- word.seqno = 1;
+
extract_add_index_string(&word, zinfo_index_category_alwaysmatches,
"", 0);
}
ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
- int deleteFlag)
+ enum zebra_recctrl_action_t action)
{
ZEBRA_RES r = ZEBRA_OK;
int i, fd;
default:
yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type);
}
- if (sysno && deleteFlag)
+ if (sysno && (action == action_delete || action == action_a_delete))
{
streamp = 0;
fi = 0;
zebra_create_stream_fd(streamp, fd, 0);
}
r = zebra_extract_records_stream(zh, streamp,
- deleteFlag ?
- action_delete : action_update,
+ action,
0, /* tst_mode */
zh->m_record_type,
sysno,
else
end_offset = stream->tellf(stream);
- all_matches_add(&extractCtrl);
-
if (extractCtrl.match_criteria[0])
match_criteria = extractCtrl.match_criteria;
}
}
}
}
+
if (zebra_rec_keys_empty(zh->reg->keys))
{
/* the extraction process returned no information - the record
if (! *sysno)
{
- /* new record */
+ /* new record AKA does not exist already */
if (action == action_delete)
{
- yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
- pr_fname, (zint) start_offset);
+ yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
yaz_log(YLOG_WARN, "cannot delete record above (seems new)");
return ZEBRA_FAIL;
}
+ else if (action == action_a_delete)
+ {
+ if (show_progress)
+ yaz_log(YLOG_LOG, "adelete %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ return ZEBRA_OK;
+ }
else if (action == action_replace)
{
yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
*sysno = rec->sysno;
+
+ if (stream)
+ {
+ all_matches_add(&extractCtrl,
+ zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+ *sysno);
+ }
+
+
recordAttr = rec_init_attr(zh->reg->zei, rec);
if (extractCtrl.staticrank < 0)
{
rec = rec_get(zh->reg->records, *sysno);
assert(rec);
+
+ if (stream)
+ {
+ all_matches_add(&extractCtrl,
+ zebra_rec_keys_get_custom_record_id(zh->reg->keys),
+ *sysno);
+ }
recordAttr = rec_init_attr(zh->reg->zei, rec);
extract_flush_record_keys(zh, *sysno, 0, delkeys,
recordAttr->staticrank);
#endif
- if (action == action_delete)
+ if (action == action_delete || action == action_a_delete)
{
/* record going to be deleted */
#if FLUSH2
}
-ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
+ZEBRA_RES zebra_rec_keys_to_snippets1(ZebraHandle zh,
zebra_rec_keys_t reckeys,
zebra_snippets *snippets)
{
static void extract_add_icu(RecWord *p, zebra_map_t zm)
{
- struct it_key key;
const char *res_buf = 0;
size_t res_len = 0;
- ZebraHandle zh = p->extractCtrl->handle;
- int r = zebra_map_tokenize(zm, p->term_buf, p->term_len,
- &res_buf, &res_len);
- int cat = zinfo_index_category_index;
- int ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
- if (ch < 0)
- ch = zebraExplain_add_attr_str(zh->reg->zei, cat, p->index_type, p->index_name);
- while (r)
- {
- int i = 0;
- key.mem[i++] = ch;
- key.mem[i++] = p->record_id;
- key.mem[i++] = p->section_id;
-
- if (zh->m_segment_indexing)
- key.mem[i++] = p->segment;
- key.mem[i++] = p->seqno;
- key.len = i;
- yaz_log(YLOG_LOG, "keys_write %.*s", (int) res_len, res_buf);
- zebra_rec_keys_write(zh->reg->keys, res_buf, res_len, &key);
-
+ zebra_map_tokenize_start(zm, p->term_buf, p->term_len);
+ while (zebra_map_tokenize_next(zm, &res_buf, &res_len, 0, 0))
+ {
+ extract_add_string(p, zm, res_buf, res_len);
p->seqno++;
- r = zebra_map_tokenize(zm, 0, 0, &res_buf, &res_len);
}
}
\param p token data to be indexed
Call sequence:
- extract_token
- zebra_add_{in}_complete
+ extract_token_add
+ extract_add_{in}_complete / extract_add_icu
extract_add_string
extract_add_index_string