X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=1312bfe06408a5d1a1c7842f4401c749fdd94d03;hb=069747eea98eaa9df53282368a481ac26fd2e0e6;hp=0a1d8c98d16812efc0a286f109f6b459fd7c85f5;hpb=ca39c711e9398043092cceeb3b4a7d8a0db1c0d8;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index 0a1d8c9..1312bfe 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,5 +1,5 @@ -/* $Id: extract.c,v 1.239 2006-11-21 14:32:38 adam Exp $ - Copyright (C) 1995-2006 +/* $Id: extract.c,v 1.258 2007-05-08 14:27:23 adam Exp $ + Copyright (C) 1995-2007 Index Data ApS This file is part of the Zebra server. @@ -40,6 +40,16 @@ static int log_level_extract = 0; 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) @@ -51,16 +61,26 @@ static void zebra_init_log_level(void) } } -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_flush_record_keys(ZebraHandle zh, zint sysno, + int cmd, zebra_rec_keys_t reckeys, + zint staticrank); +static void extract_flush_sort_keys(ZebraHandle zh, zint 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 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)) { @@ -139,9 +159,9 @@ static void searchRecordKey(ZebraHandle zh, #define FILE_MATCH_BLANK "\t " -static char *fileMatchStr (ZebraHandle zh, - zebra_rec_keys_t reckeys, - const char *fname, const char *spec) +static char *get_match_from_spec(ZebraHandle zh, + zebra_rec_keys_t reckeys, + const char *fname, const char *spec) { static char dstBuf[2048]; /* static here ??? */ char *dst = dstBuf; @@ -309,7 +329,19 @@ static void all_matches_add(struct recExtractCtrl *ctrl) "", 0); } -ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, +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) { ZEBRA_RES r = ZEBRA_OK; @@ -351,8 +383,11 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, } 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 */ @@ -404,27 +439,15 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, streamp = &stream; zebra_create_stream_fd(streamp, fd, 0); } - while(1) - { - 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); - if (r != ZEBRA_OK) - { - 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; @@ -440,14 +463,12 @@ ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, 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, - SYSNO *sysno, + zint *sysno, const char *match_criteria, - const char *fname, - int force_update, - int allow_update) + const char *fname) { struct ZebraRecStream stream; ZEBRA_RES res; @@ -481,48 +502,94 @@ ZEBRA_RES zebra_buffer_extract_record(ZebraHandle zh, return ZEBRA_FAIL; } - - 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); + 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, - SYSNO *sysno, + zint *sysno, const char *match_criteria, const char *fname, - int force_update, - int allow_update, RecType recType, - void *recTypeClientData) + void *recTypeClientData, + int *more) { - SYSNO sysno0 = 0; + zint sysno0 = 0; RecordAttr *recordAttr; struct recExtractCtrl extractCtrl; int r; const char *matchStr = 0; Record rec; - off_t start_offset = 0; + 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(); @@ -557,47 +624,75 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 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; } - if (!sysno) { + *more = 1; + 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, - zh->m_record_id); + matchStr = get_match_from_spec(zh, zh->reg->keys, pr_fname, + zh->m_record_id); if (!matchStr) { - yaz_log (YLOG_WARN, "Bad match criteria (recordID)"); + yaz_log (YLOG_LOG, "error %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); return ZEBRA_FAIL; } } @@ -607,12 +702,20 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, int db_ord = zebraExplain_get_database_ord(zh->reg->zei); char *rinfo = dict_lookup_ord(zh->reg->matchDict, db_ord, matchStr); + + + if (log_level_extract) + { + WRBUF w = wrbuf_hex_str(matchStr); + yaz_log(log_level_extract, "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)) { @@ -625,13 +728,20 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, 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); @@ -640,7 +750,11 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, *sysno = rec->sysno; recordAttr = rec_init_attr (zh->reg->zei, rec); - recordAttr->staticrank = extractCtrl.staticrank; + if (extractCtrl.staticrank < 0) + { + yaz_log(YLOG_WARN, "Negative staticrank for record. Set to 0"); + extractCtrl.staticrank = 0; + } if (matchStr) { @@ -649,10 +763,16 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, sizeof(*sysno), sysno); } - - extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); - extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, - recordAttr->staticrank); + 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, + extractCtrl.staticrank); +#endif + recordAttr->staticrank = extractCtrl.staticrank; zh->records_inserted++; } else @@ -660,7 +780,7 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, /* 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); @@ -673,6 +793,10 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, recordAttr = rec_init_attr (zh->reg->zei, rec); + /* decrease total size */ + zebraExplain_recordBytesIncrement (zh->reg->zei, + - recordAttr->recordSize); + zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys], rec->size[recInfo_delKeys], @@ -682,12 +806,18 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, rec->size[recInfo_sortKeys], 0); - extract_flushSortKeys (zh, *sysno, 0, sortKeys); - extract_flushRecordKeys (zh, *sysno, 0, delkeys, - recordAttr->staticrank); - if (delete_flag) + extract_flush_sort_keys(zh, *sysno, 0, sortKeys); +#if !FLUSH2 + extract_flush_record_keys(zh, *sysno, 0, delkeys, + recordAttr->staticrank); +#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, @@ -708,19 +838,28 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, } 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) ZINT_FORMAT); + yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT, recordType, + pr_fname, (zint) start_offset); + extract_flush_sort_keys(zh, *sysno, 1, zh->reg->sortKeys); + +#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; - extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); - extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, - recordAttr->staticrank); zh->records_updated++; } zebra_rec_keys_close(delkeys); @@ -756,18 +895,8 @@ ZEBRA_RES zebra_extract_record_stream(ZebraHandle zh, &rec->info[recInfo_sortKeys], &rec->size[recInfo_sortKeys]); - /* save file size of original record */ - zebraExplain_recordBytesIncrement (zh->reg->zei, - - recordAttr->recordSize); if (stream) { - off_t end_offset = stream->endf(stream, 0); - - if (!end_offset) - end_offset = stream->tellf(stream); - else - stream->seekf(stream, end_offset); - recordAttr->recordSize = end_offset - start_offset; zebraExplain_recordBytesIncrement(zh->reg->zei, recordAttr->recordSize); @@ -846,6 +975,8 @@ ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n) extractCtrl.flagShowRecords = 0; extractCtrl.match_criteria[0] = '\0'; extractCtrl.staticrank = 0; + extractCtrl.action = action_update; + extractCtrl.handle = handle; extractCtrl.first_record = 1; @@ -863,19 +994,32 @@ ZEBRA_RES zebra_extract_explain(void *handle, Record rec, data1_node *n) zebra_rec_keys_set_buf(delkeys, rec->info[recInfo_delKeys], rec->size[recInfo_delKeys], 0); - extract_flushRecordKeys (zh, rec->sysno, 0, 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], rec->size[recInfo_sortKeys], 0); - extract_flushSortKeys (zh, rec->sysno, 0, sortkeys); + extract_flush_sort_keys(zh, rec->sysno, 0, sortkeys); zebra_rec_keys_close(sortkeys); } - extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0); - extract_flushSortKeys (zh, rec->sysno, 1, zh->reg->sortKeys); - + 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]); zebra_rec_keys_get_buf(zh->reg->keys, &rec->info[recInfo_delKeys], @@ -1004,9 +1148,82 @@ void extract_rec_keys_adjust(ZebraHandle zh, int is_insert, } } -void extract_flushRecordKeys(ZebraHandle zh, SYSNO sysno, int cmd, - zebra_rec_keys_t reckeys, - zint staticrank) +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) { ZebraExplainInfo zei = zh->reg->zei; @@ -1021,12 +1238,17 @@ void extract_flushRecordKeys(ZebraHandle zh, SYSNO sysno, int cmd, if (!zh->reg->key_block) { - int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8")); + int mem = 1024*1024 * atoi( res_get_def( zh->res, "memmax", "8")); const char *key_tmp_dir = res_get_def (zh->res, "keyTmpDir", "."); - zh->reg->key_block = key_block_create(mem, key_tmp_dir); + int use_threads = atoi(res_get_def (zh->res, "threads", "1")); + zh->reg->key_block = key_block_create(mem, key_tmp_dir, use_threads); } 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; @@ -1041,7 +1263,6 @@ void extract_flushRecordKeys(ZebraHandle zh, SYSNO sysno, int cmd, } } - ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh, zebra_rec_keys_t reckeys, zebra_snippets *snippets) @@ -1054,8 +1275,7 @@ ZEBRA_RES zebra_rec_keys_to_snippets(ZebraHandle zh, struct it_key key; while (zebra_rec_keys_read(reckeys, &str, &slen, &key)) { - char dst_buf[IT_MAX_WORD]; - char *dst_term = dst_buf; + char *dst_term = 0; int ord; zint seqno; int index_type; @@ -1149,6 +1369,20 @@ static void extract_add_sort_string(RecWord *p, const char *str, int length) 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; @@ -1157,9 +1391,7 @@ static void extract_add_string(RecWord *p, const char *string, int length) 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); @@ -1173,6 +1405,14 @@ static void extract_add_string(RecWord *p, const char *string, int 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) @@ -1180,20 +1420,12 @@ static void extract_add_incomplete_field(RecWord *p) ZebraHandle zh = p->extractCtrl->handle; const char *b = p->term_buf; int remain = p->term_len; + int first = 1; const char **map = 0; if (remain > 0) map = zebra_maps_input(zh->reg->zebra_maps, p->index_type, &b, remain, 0); - if (map) - { - if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type)) - { - /* first in field marker */ - extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN); - p->seqno++; - } - } while (map) { char buf[IT_MAX_WORD+1]; @@ -1226,6 +1458,17 @@ static void extract_add_incomplete_field(RecWord *p) } if (!i) return; + + if (first) + { + first = 0; + if (zebra_maps_is_first_in_field(zh->reg->zebra_maps, p->index_type)) + { + /* first in field marker */ + extract_add_string(p, FIRST_IN_FIELD_STR, FIRST_IN_FIELD_LEN); + p->seqno++; + } + } extract_add_string (p, buf, i); p->seqno++; } @@ -1296,9 +1539,9 @@ static void extract_token_add(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); @@ -1346,27 +1589,33 @@ static void extract_schema_add(struct recExtractCtrl *p, Odr_oid *oid) zebraExplain_addSchema (zh->reg->zei, oid); } -void extract_flushSortKeys(ZebraHandle zh, SYSNO sysno, - int cmd, zebra_rec_keys_t reckeys) +void extract_flush_sort_keys(ZebraHandle zh, zint sysno, + int cmd, zebra_rec_keys_t reckeys) { +#if 0 + yaz_log(YLOG_LOG, "extract_flush_sort_keys cmd=%d sysno=" ZINT_FORMAT, + cmd, sysno); + extract_rec_keys_log(zh, cmd, reckeys, YLOG_LOG); +#endif + if (zebra_rec_keys_rewind(reckeys)) { - SortIdx sortIdx = zh->reg->sortIdx; + zebra_sort_index_t si = zh->reg->sort_index; size_t slen; const char *str; struct it_key key_in; - sortIdx_sysno (sortIdx, sysno); + zebra_sort_sysno(si, sysno); while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) { int ord = CAST_ZINT_TO_INT(key_in.mem[0]); - sortIdx_type(sortIdx, ord); + zebra_sort_type(si, ord); if (cmd == 1) - sortIdx_add(sortIdx, str, slen); + zebra_sort_add(si, str, slen); else - sortIdx_add(sortIdx, "", 1); + zebra_sort_delete(si); } } }