-/* $Id: extract.c,v 1.247 2007-01-15 15:10:16 adam Exp $
+/* $Id: extract.c,v 1.257 2007-05-08 12:50:04 adam Exp $
Copyright (C) 1995-2007
Index Data ApS
static int log_level_details = 0;
static int log_level_initialized = 0;
+/* 1 if we use eliminitate identical delete/insert keys */
+/* eventually this the 0-case code will be removed */
+#define FLUSH2 1
+
+void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
+ zebra_rec_keys_t ins_keys,
+ zint ins_rank,
+ zebra_rec_keys_t del_keys,
+ zint del_rank);
+
static void zebra_init_log_level(void)
{
if (!log_level_initialized)
static void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid);
static void extract_token_add (RecWord *p);
+static void check_log_limit(ZebraHandle zh)
+{
+ if (zh->records_processed + zh->records_skipped == zh->m_file_verbose_limit)
+ {
+ yaz_log(YLOG_LOG, "More than %d file log entries. Omitting rest",
+ zh->m_file_verbose_limit);
+ }
+}
+
static void logRecord (ZebraHandle zh)
{
+ check_log_limit(zh);
++zh->records_processed;
if (!(zh->records_processed % 1000))
{
"", 0);
}
+ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
+ struct ZebraRecStream *stream,
+ enum zebra_recctrl_action_t action,
+ int test_mode,
+ const char *recordType,
+ zint *sysno,
+ const char *match_criteria,
+ const char *fname,
+ RecType recType,
+ void *recTypeClientData);
+
+
ZEBRA_RES zebra_extract_file(ZebraHandle zh, zint *sysno, const char *fname,
int deleteFlag)
{
}
if (!zh->m_record_type)
{
- if (zh->records_processed < zh->m_file_verbose_limit)
+ check_log_limit(zh);
+ if (zh->records_processed + zh->records_skipped
+ < zh->m_file_verbose_limit)
yaz_log (YLOG_LOG, "? %s", fname);
+ zh->records_skipped++;
return 0;
}
/* determine match criteria */
streamp = &stream;
zebra_create_stream_fd(streamp, fd, 0);
}
- while(1)
- {
- int more = 0;
- r = zebra_extract_record_stream(zh, streamp,
- deleteFlag,
- 0, /* tst_mode */
- zh->m_record_type,
- sysno,
- 0, /*match_criteria */
- fname,
- 1, /* force_update */
- 1, /* allow_update */
- recType, recTypeClientData, &more);
- if (!more)
- break;
- if (sysno)
- {
- break;
- }
- }
+ r = zebra_extract_records_stream(zh, streamp,
+ deleteFlag ?
+ action_delete : action_update,
+ 0, /* tst_mode */
+ zh->m_record_type,
+ sysno,
+ 0, /*match_criteria */
+ fname,
+ recType, recTypeClientData);
if (streamp)
stream.destroy(streamp);
zh->m_record_type = original_record_type;
ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh,
const char *buf, size_t buf_size,
- int delete_flag,
+ enum zebra_recctrl_action_t action,
int test_mode,
const char *recordType,
zint *sysno,
const char *match_criteria,
- const char *fname,
- int force_update,
- int allow_update)
+ const char *fname)
{
struct ZebraRecStream stream;
ZEBRA_RES res;
void *clientData;
RecType recType = 0;
- int more = 0;
if (recordType && *recordType)
{
zebra_create_stream_mem(&stream, buf, buf_size);
- res = zebra_extract_record_stream(zh, &stream,
- delete_flag,
- test_mode,
- recordType,
- sysno,
- match_criteria,
- fname,
- force_update,
- allow_update,
- recType, clientData, &more);
+ res = zebra_extract_records_stream(zh, &stream,
+ action,
+ test_mode,
+ recordType,
+ sysno,
+ match_criteria,
+ fname,
+ recType, clientData);
stream.destroy(&stream);
return res;
}
+ZEBRA_RES zebra_extract_records_stream(ZebraHandle zh,
+ struct ZebraRecStream *stream,
+ enum zebra_recctrl_action_t action,
+ int test_mode,
+ const char *recordType,
+ zint *sysno,
+ const char *match_criteria,
+ const char *fname,
+ RecType recType,
+ void *recTypeClientData)
+{
+ ZEBRA_RES res = ZEBRA_OK;
+ while (1)
+ {
+ int more = 0;
+ res = zebra_extract_record_stream(zh, stream,
+ action,
+ test_mode,
+ recordType,
+ sysno,
+ match_criteria,
+ fname,
+ recType, recTypeClientData, &more);
+ if (!more)
+ {
+ res = ZEBRA_OK;
+ break;
+ }
+ if (res != ZEBRA_OK)
+ break;
+ if (sysno)
+ break;
+ }
+ return res;
+}
+
+
+static WRBUF wrbuf_hex_str(const char *cstr)
+{
+ size_t i;
+ WRBUF w = wrbuf_alloc();
+ for (i = 0; cstr[i]; i++)
+ {
+ if (cstr[i] < ' ' || cstr[i] > 126)
+ wrbuf_printf(w, "\\%02X", cstr[i] & 0xff);
+ else
+ wrbuf_putc(w, cstr[i]);
+ }
+ return w;
+}
ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh,
struct ZebraRecStream *stream,
- int delete_flag,
+ enum zebra_recctrl_action_t action,
int test_mode,
const char *recordType,
zint *sysno,
const char *match_criteria,
const char *fname,
- int force_update,
- int allow_update,
RecType recType,
void *recTypeClientData,
int *more)
Record rec;
off_t start_offset = 0, end_offset = 0;
const char *pr_fname = fname; /* filename to print .. */
- int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0;
+ int show_progress = zh->records_processed + zh->records_skipped
+ < zh->m_file_verbose_limit ? 1:0;
zebra_init_log_level();
extractCtrl.handle = zh;
extractCtrl.match_criteria[0] = '\0';
extractCtrl.staticrank = 0;
+ extractCtrl.action = action;
init_extractCtrl(zh, &extractCtrl);
-
+
extract_set_store_data_prepare(&extractCtrl);
r = (*recType->extract)(recTypeClientData, &extractCtrl);
+
+ if (action == action_update)
+ {
+ action = extractCtrl.action;
+ }
- if (r == RECCTRL_EXTRACT_EOF)
- return ZEBRA_FAIL;
- else if (r == RECCTRL_EXTRACT_ERROR_GENERIC)
+ switch (r)
{
+ case RECCTRL_EXTRACT_EOF:
+ return ZEBRA_FAIL;
+ case RECCTRL_EXTRACT_ERROR_GENERIC:
/* error occured during extraction ... */
yaz_log (YLOG_WARN, "extract error: generic");
return ZEBRA_FAIL;
- }
- else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER)
- {
+ case RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER:
/* error occured during extraction ... */
yaz_log (YLOG_WARN, "extract error: no such filter");
return ZEBRA_FAIL;
+ case RECCTRL_EXTRACT_SKIP:
+ if (show_progress)
+ yaz_log (YLOG_LOG, "skip %s %s " ZINT_FORMAT,
+ recordType, pr_fname, (zint) start_offset);
+ *more = 1;
+
+ end_offset = stream->endf(stream, 0);
+ if (end_offset)
+ stream->seekf(stream, end_offset);
+
+ return ZEBRA_OK;
+ case RECCTRL_EXTRACT_OK:
+ break;
+ default:
+ yaz_log (YLOG_WARN, "extract error: unknown error: %d", r);
+ return ZEBRA_FAIL;
}
-
+ end_offset = stream->endf(stream, 0);
+ if (end_offset)
+ stream->seekf(stream, end_offset);
+ else
+ end_offset = stream->tellf(stream);
+
all_matches_add(&extractCtrl);
if (extractCtrl.match_criteria[0])
match_criteria = extractCtrl.match_criteria;
-
-
- end_offset = stream->endf(stream, 0);
-
- if (!end_offset)
- end_offset = stream->tellf(stream);
- else
- stream->seekf(stream, end_offset);
-
}
-
*more = 1;
if (!sysno)
{
int db_ord = zebraExplain_get_database_ord(zh->reg->zei);
char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord,
matchStr);
+
+
+ if (1)
+ {
+ WRBUF w = wrbuf_hex_str(matchStr);
+ yaz_log(YLOG_LOG, "matchStr: %s", wrbuf_cstr(w));
+ wrbuf_destroy(w);
+ }
if (rinfo)
{
assert(*rinfo == sizeof(*sysno));
memcpy (sysno, rinfo+1, sizeof(*sysno));
}
- }
+ }
}
if (zebra_rec_keys_empty(zh->reg->keys))
{
if (! *sysno)
{
/* new record */
- if (delete_flag)
+ if (action == action_delete)
{
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_replace)
+ {
+ yaz_log (YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
+ pr_fname, (zint) start_offset);
+ yaz_log (YLOG_WARN, "cannot update record above (seems new)");
+ return ZEBRA_FAIL;
+ }
if (show_progress)
yaz_log (YLOG_LOG, "add %s %s " ZINT_FORMAT, recordType, pr_fname,
(zint) start_offset);
yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0");
extractCtrl.staticrank = 0;
}
- recordAttr->staticrank = extractCtrl.staticrank;
if (matchStr)
{
}
extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
+#if FLUSH2
+ extract_flush_record_keys2(zh, *sysno,
+ zh->reg->keys, extractCtrl.staticrank,
+ 0, recordAttr->staticrank);
+#else
extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
- recordAttr->staticrank);
+ extractCtrl.staticrank);
+#endif
+ recordAttr->staticrank = extractCtrl.staticrank;
zh->records_inserted++;
}
else
/* record already exists */
zebra_rec_keys_t delkeys = zebra_rec_keys_open();
zebra_rec_keys_t sortKeys = zebra_rec_keys_open();
- if (!allow_update)
+ if (action == action_insert)
{
yaz_log (YLOG_LOG, "skipped %s %s " ZINT_FORMAT,
recordType, pr_fname, (zint) start_offset);
0);
extract_flush_sort_keys(zh, *sysno, 0, sortKeys);
+#if !FLUSH2
extract_flush_record_keys(zh, *sysno, 0, delkeys,
recordAttr->staticrank);
- if (delete_flag)
+#endif
+ if (action == action_delete)
{
/* record going to be deleted */
+#if FLUSH2
+ extract_flush_record_keys2(zh, *sysno, 0, recordAttr->staticrank,
+ delkeys, recordAttr->staticrank);
+#endif
if (zebra_rec_keys_empty(delkeys))
{
yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, recordType,
}
rec_del (zh->reg->records, &rec);
}
+ zebra_rec_keys_close(delkeys);
+ zebra_rec_keys_close(sortKeys);
rec_free(&rec);
logRecord(zh);
return ZEBRA_OK;
}
else
- {
+ { /* update or special_update */
if (show_progress)
yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType,
pr_fname, (zint) start_offset);
- recordAttr->staticrank = extractCtrl.staticrank;
extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys);
- extract_flush_record_keys(zh, *sysno, 1, zh->reg->keys,
- recordAttr->staticrank);
+
+#if FLUSH2
+ extract_flush_record_keys2(zh, *sysno,
+ zh->reg->keys, extractCtrl.staticrank,
+ delkeys, recordAttr->staticrank);
+#else
+ extract_flush_record_keys(zh, *sysno, 1,
+ zh->reg->keys, extractCtrl.staticrank);
+#endif
+ recordAttr->staticrank = extractCtrl.staticrank;
zh->records_updated++;
}
zebra_rec_keys_close(delkeys);
extractCtrl.flagShowRecords = 0;
extractCtrl.match_criteria[0] = '\0';
extractCtrl.staticrank = 0;
+ extractCtrl.action = action_update;
+
extractCtrl.handle = handle;
extractCtrl.first_record = 1;
zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys],
rec->size[recInfo_delKeys],
0);
+#if FLUSH2
+ extract_flush_record_keys2(zh, rec->sysno,
+ zh->reg->keys, 0, delkeys, 0);
+#else
extract_flush_record_keys(zh, rec->sysno, 0, delkeys, 0);
+ extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
+#endif
zebra_rec_keys_close(delkeys);
zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys],
extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys);
zebra_rec_keys_close(sortkeys);
}
- extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
+ else
+ {
+#if FLUSH2
+ extract_flush_record_keys2(zh, rec->sysno, zh->reg->keys, 0, 0, 0);
+#else
+ extract_flush_record_keys(zh, rec->sysno, 1, zh->reg->keys, 0);
+#endif
+ }
extract_flush_sort_keys(zh, rec->sysno, 1, zh->reg->sortKeys);
xfree (rec->info[recInfo_delKeys]);
}
}
+void extract_flush_record_keys2(ZebraHandle zh, zint sysno,
+ zebra_rec_keys_t ins_keys, zint ins_rank,
+ zebra_rec_keys_t del_keys, zint del_rank)
+{
+ ZebraExplainInfo zei = zh->reg->zei;
+ int normal = 0;
+ int optimized = 0;
+
+ if (!zh->reg->key_block)
+ {
+ int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8"));
+ const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", ".");
+ int use_threads = atoi(res_get_def (zh->res, "threads", "1"));
+ zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads);
+ }
+
+ if (ins_keys)
+ {
+ extract_rec_keys_adjust(zh, 1, ins_keys);
+ if (!del_keys)
+ zebraExplain_recordCountIncrement (zei, 1);
+ zebra_rec_keys_rewind(ins_keys);
+ }
+ if (del_keys)
+ {
+ extract_rec_keys_adjust(zh, 0, del_keys);
+ if (!ins_keys)
+ zebraExplain_recordCountIncrement (zei, -1);
+ zebra_rec_keys_rewind(del_keys);
+ }
+
+ while (1)
+ {
+ size_t del_slen;
+ const char *del_str;
+ struct it_key del_key_in;
+ int del = 0;
+
+ size_t ins_slen;
+ const char *ins_str;
+ struct it_key ins_key_in;
+ int ins = 0;
+
+ if (del_keys)
+ del = zebra_rec_keys_read(del_keys, &del_str, &del_slen,
+ &del_key_in);
+ if (ins_keys)
+ ins = zebra_rec_keys_read(ins_keys, &ins_str, &ins_slen,
+ &ins_key_in);
+
+ if (del && ins && ins_rank == del_rank
+ && !key_compare(&del_key_in, &ins_key_in)
+ && ins_slen == del_slen && !memcmp(del_str, ins_str, del_slen))
+ {
+ optimized++;
+ continue;
+ }
+ if (!del && !ins)
+ break;
+
+ normal++;
+ if (del)
+ key_block_write(zh->reg->key_block, sysno,
+ &del_key_in, 0, del_str, del_slen,
+ del_rank, zh->m_staticrank);
+ if (ins)
+ key_block_write(zh->reg->key_block, sysno,
+ &ins_key_in, 1, ins_str, ins_slen,
+ ins_rank, zh->m_staticrank);
+ }
+ yaz_log(log_level_extract, "normal=%d optimized=%d", normal, optimized);
+}
+
void extract_flush_record_keys(ZebraHandle zh, zint sysno, int cmd,
zebra_rec_keys_t reckeys,
zint staticrank)
}
zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1);
+#if 0
+ yaz_log(YLOG_LOG, "sysno=" ZINT_FORMAT " cmd=%d", sysno, cmd);
+ print_rec_keys(zh, reckeys);
+#endif
if (zebra_rec_keys_rewind(reckeys))
{
size_t slen;
}
}
-
ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh,
zebra_rec_keys_t reckeys,
zebra_snippets *snippets)
zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key);
}
+static void extract_add_staticrank_string(RecWord *p,
+ const char *str, int length)
+{
+ char valz[40];
+ struct recExtractCtrl *ctrl = p->extractCtrl;
+
+ if (length > sizeof(valz)-1)
+ length = sizeof(valz)-1;
+
+ memcpy(valz, str, length);
+ valz[length] = '\0';
+ ctrl->staticrank = atozint(valz);
+}
+
static void extract_add_string(RecWord *p, const char *string, int length)
{
ZebraHandle zh = p->extractCtrl->handle;
if (!p->index_name)
return;
- if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
- extract_add_sort_string(p, string, length);
- else
+ if (zebra_maps_is_index(zh->reg->zebra_maps, p->index_type))
{
extract_add_index_string(p, zinfo_index_category_index,
string, length);
&word, zinfo_index_category_alwaysmatches, "", 0);
}
}
+ else if (zebra_maps_is_sort(zh->reg->zebra_maps, p->index_type))
+ {
+ extract_add_sort_string(p, string, length);
+ }
+ else if (zebra_maps_is_staticrank(zh->reg->zebra_maps, p->index_type))
+ {
+ extract_add_staticrank_string(p, string, length);
+ }
}
static void extract_add_incomplete_field(RecWord *p)
ZebraHandle zh = p->extractCtrl->handle;
WRBUF wrbuf;
- if (log_level_extract)
+ if (log_level_details)
{
- yaz_log(log_level_extract, "extract_token_add "
+ yaz_log(log_level_details, "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);
projects / idzebra-moved-to-github.git / blobdiff