X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=9a609a991a57063b62e8d247b89f48055f92f47c;hb=a2c4cb0542fef348b998284ca9bf3b2348f5c5b7;hp=6718670c6830a73f97f37a88ba50c64b73d34924;hpb=289761234a123e7fb51df77ab9baf0a2cd101dbd;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index 6718670..9a609a9 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,5 +1,5 @@ -/* $Id: extract.c,v 1.203 2006-02-20 18:39:43 adam Exp $ - Copyright (C) 1995-2005 +/* $Id: extract.c,v 1.224 2006-06-23 11:21:38 adam Exp $ + Copyright (C) 1995-2006 Index Data ApS This file is part of the Zebra server. @@ -36,11 +36,36 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA #include #include -#if _FILE_OFFSET_BITS == 64 -#define PRINTF_OFF_T "%Ld" -#else -#define PRINTF_OFF_T "%ld" -#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 @@ -74,25 +99,24 @@ static void logRecord (ZebraHandle zh) ++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); } } +static void extract_add_index_string (RecWord *p, + zinfo_index_category_t cat, + const char *str, int length); + static void extract_set_store_data_prepare(struct recExtractCtrl *p); static void extract_init (struct recExtractCtrl *p, RecWord *w) { 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; @@ -101,23 +125,29 @@ static void extract_init (struct recExtractCtrl *p, RecWord *w) 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; + zinfo_index_category_t cat = zinfo_index_category_index; for (i = 0; ireg->zei, - attrSetS, attrUseS); + if (ch < 0) + ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '0', index_name); + if (ch < 0) + ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, 'p', index_name); + if (ch < 0) + ch = zebraExplain_lookup_attr_str(zh->reg->zei, cat, '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; @@ -130,7 +160,7 @@ static void searchRecordKey(ZebraHandle zh, if (key.mem[0] == ch) { - int woff; + zint woff; if (startSeq == -1) startSeq = seqno; @@ -227,9 +257,7 @@ static char *fileMatchStr (ZebraHandle zh, { 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++) @@ -242,7 +270,9 @@ static char *fileMatchStr (ZebraHandle zh, 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++) ; @@ -252,18 +282,8 @@ static char *fileMatchStr (ZebraHandle zh, 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 != ')') { @@ -379,14 +399,26 @@ static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl) ctrl->flagShowRecords = !zh->m_flag_rw; } -static int file_extract_record(ZebraHandle zh, - SYSNO *sysno, const char *fname, - int deleteFlag, - struct file_read_info *fi, - int force_update, - RecType recType, - void *recTypeClientData) +static void all_matches_add(struct recExtractCtrl *ctrl) { + RecWord word; + extract_init(ctrl, &word); + word.index_name = "_ALLRECORDS"; + word.index_type = 'w'; + word.seqno = 1; + extract_add_index_string (&word, zinfo_index_category_alwaysmatches, + "", 0); +} + +static ZEBRA_RES file_extract_record(ZebraHandle zh, + SYSNO *sysno, const char *fname, + int deleteFlag, + struct file_read_info *fi, + int force_update, + RecType recType, + void *recTypeClientData) +{ + const char *match_str_to_print = ""; RecordAttr *recordAttr; int r; const char *matchStr = 0; @@ -400,7 +432,7 @@ static int file_extract_record(ZebraHandle zh, { if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], zh->m_explain_database)) - return 0; + return ZEBRA_FAIL; } if (fi->fd != -1) @@ -408,11 +440,7 @@ static int file_extract_record(ZebraHandle zh, /* we are going to read from a file, so prepare the extraction */ zebra_rec_keys_reset(zh->reg->keys); -#if NATTR zebra_rec_keys_reset(zh->reg->sortKeys); -#else - zh->reg->sortKeys.buf_used = 0; -#endif recordOffset = fi->file_moffset; extractCtrl.handle = zh; extractCtrl.offset = fi->file_moffset; @@ -435,11 +463,11 @@ static int file_extract_record(ZebraHandle zh, init_extractCtrl(zh, &extractCtrl); if (!zh->m_flag_rw) - printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset); + printf ("File: %s " ZINT_FORMAT "\n", fname, (zint)recordOffset); if (zh->m_flag_rw) { char msg[512]; - sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset); + sprintf (msg, "%s:" ZINT_FORMAT , fname, (zint)recordOffset); yaz_log_init_prefix2 (msg); } @@ -447,17 +475,18 @@ static int file_extract_record(ZebraHandle zh, yaz_log_init_prefix2 (0); if (r == RECCTRL_EXTRACT_EOF) - return 0; + return ZEBRA_FAIL; else if (r == RECCTRL_EXTRACT_ERROR_GENERIC) { /* error occured during extraction ... */ if (zh->m_flag_rw && zh->records_processed < zh->m_file_verbose_limit) { - yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type, - fname, recordOffset); + yaz_log (YLOG_WARN, "fail %s %s " ZINT_FORMAT, + zh->m_record_type, + fname, (zint) recordOffset); } - return 0; + return ZEBRA_FAIL; } else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER) { @@ -466,16 +495,23 @@ static int file_extract_record(ZebraHandle zh, zh->records_processed < zh->m_file_verbose_limit) { yaz_log (YLOG_WARN, "no filter for %s %s " - PRINTF_OFF_T, zh->m_record_type, - fname, recordOffset); + ZINT_FORMAT, zh->m_record_type, + fname, (zint) recordOffset); } - return 0; + return ZEBRA_FAIL; } + all_matches_add(&extractCtrl); if (extractCtrl.match_criteria[0]) - matchStr = extractCtrl.match_criteria; + matchStr = extractCtrl.match_criteria; } - /* perform match if sysno not known and if match criteria is specified */ + /* if matchStr is set now - we assume it's printable . + For internal matchStr (see below) we don't print */ + if (matchStr) + match_str_to_print = matchStr; + + /* perform internal match if sysno not known and if match criteria is + specified already */ if (!sysno) { sysnotmp = 0; @@ -488,7 +524,12 @@ static int file_extract_record(ZebraHandle zh, if (!matchStr) { yaz_log(YLOG_WARN, "Bad match criteria"); - return 0; + + if (zebra_rec_keys_empty(zh->reg->keys)) + { + yaz_log(YLOG_WARN, "And no index keys"); + } + return ZEBRA_FAIL; } } if (matchStr) @@ -508,12 +549,12 @@ static int file_extract_record(ZebraHandle zh, /* the extraction process returned no information - the record is probably empty - unless flagShowRecords is in use */ if (!zh->m_flag_rw) - return 1; + return ZEBRA_OK; if (zh->records_processed < zh->m_file_verbose_limit) - yaz_log (YLOG_WARN, "empty %s %s " PRINTF_OFF_T, zh->m_record_type, - fname, recordOffset); - return 1; + yaz_log(YLOG_WARN, "empty %s %s " ZINT_FORMAT, zh->m_record_type, + fname, (zint)recordOffset); + return ZEBRA_OK; } if (! *sysno) @@ -521,18 +562,23 @@ static int file_extract_record(ZebraHandle zh, /* new record */ if (deleteFlag) { - yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type, - fname, recordOffset); - yaz_log (YLOG_WARN, "cannot delete record above (seems new)"); - return 1; + yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT, zh->m_record_type, + fname, (zint)recordOffset); + yaz_log(YLOG_WARN, "cannot delete record above (seems new)"); + return ZEBRA_OK; } - if (zh->records_processed < zh->m_file_verbose_limit) - yaz_log (YLOG_LOG, "add %s %s " PRINTF_OFF_T, zh->m_record_type, - fname, recordOffset); - rec = rec_new (zh->reg->records); + rec = rec_new (zh->reg->records); + *sysno = rec->sysno; - + + if (zh->records_processed < zh->m_file_verbose_limit) + { + yaz_log(YLOG_LOG, "add %s %s " ZINT_FORMAT + " " ZINT_FORMAT " %s" , + zh->m_record_type, + fname, (zint) recordOffset, *sysno, match_str_to_print); + } recordAttr = rec_init_attr (zh->reg->zei, rec); recordAttr->staticrank = extractCtrl.staticrank; @@ -542,11 +588,9 @@ static int file_extract_record(ZebraHandle zh, dict_insert_ord(zh->reg->matchDict, db_ord, matchStr, sizeof(*sysno), sysno); } -#if NATTR + + extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); -#else - extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys); -#endif extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, recordAttr->staticrank); zh->records_inserted++; @@ -556,11 +600,7 @@ static int file_extract_record(ZebraHandle zh, /* record already exists */ zebra_rec_keys_t delkeys = zebra_rec_keys_open(); -#if NATTR zebra_rec_keys_t sortKeys = zebra_rec_keys_open(); -#else - struct sortKeys sortKeys; -#endif rec = rec_get (zh->reg->records, *sysno); assert (rec); @@ -572,18 +612,11 @@ static int file_extract_record(ZebraHandle zh, rec->size[recInfo_delKeys], 0); -#if NATTR zebra_rec_keys_set_buf(sortKeys, rec->info[recInfo_sortKeys], rec->size[recInfo_sortKeys], 0); extract_flushSortKeys (zh, *sysno, 0, sortKeys); -#else - sortKeys.buf_used = rec->size[recInfo_sortKeys]; - sortKeys.buf = rec->info[recInfo_sortKeys]; - extract_flushSortKeys (zh, *sysno, 0, &sortKeys); -#endif - extract_flushRecordKeys (zh, *sysno, 0, delkeys, recordAttr->staticrank); /* old values */ if (deleteFlag) @@ -591,15 +624,20 @@ static int file_extract_record(ZebraHandle zh, /* record going to be deleted */ if (zebra_rec_keys_empty(delkeys)) { - yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, - zh->m_record_type, fname, recordOffset); - yaz_log (YLOG_WARN, "cannot delete file above, storeKeys false (1)"); + yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT + " " ZINT_FORMAT, + zh->m_record_type, fname, (zint)recordOffset, *sysno); + yaz_log(YLOG_WARN, "cannot delete file above, storeKeys false (1)"); } else { if (zh->records_processed < zh->m_file_verbose_limit) - yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, - zh->m_record_type, fname, recordOffset); + { + yaz_log(YLOG_LOG, "delete %s %s " ZINT_FORMAT + " " ZINT_FORMAT " %s" , + zh->m_record_type, fname, (zint) recordOffset, + *sysno, match_str_to_print); + } zh->records_deleted++; if (matchStr) { @@ -610,28 +648,26 @@ static int file_extract_record(ZebraHandle zh, } rec_rm (&rec); logRecord (zh); - return 1; + return ZEBRA_OK; } else { /* flush new keys for sort&search etc */ if (zh->records_processed < zh->m_file_verbose_limit) - yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T, - zh->m_record_type, fname, recordOffset); + { + yaz_log(YLOG_LOG, "update %s %s " ZINT_FORMAT + " " ZINT_FORMAT " %s" , + zh->m_record_type, fname, (zint) recordOffset, + *sysno, match_str_to_print); + } recordAttr->staticrank = extractCtrl.staticrank; -#if NATTR extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); -#else - extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys); -#endif extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, recordAttr->staticrank); zh->records_updated++; } zebra_rec_keys_close(delkeys); -#if NATTR zebra_rec_keys_close(sortKeys); -#endif } /* update file type */ xfree (rec->info[recInfo_fileType]); @@ -660,16 +696,9 @@ static int file_extract_record(ZebraHandle zh, /* update sort keys */ xfree (rec->info[recInfo_sortKeys]); -#if NATTR zebra_rec_keys_get_buf(zh->reg->sortKeys, &rec->info[recInfo_sortKeys], &rec->size[recInfo_sortKeys]); -#else - rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used; - rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf; - zh->reg->sortKeys.buf = NULL; - zh->reg->sortKeys.buf_max = 0; -#endif /* save file size of original record */ zebraExplain_recordBytesIncrement (zh->reg->zei, @@ -699,8 +728,8 @@ static int file_extract_record(ZebraHandle zh, xmalloc (recordAttr->recordSize); if (lseek (fi->fd, recordOffset, SEEK_SET) < 0) { - yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s", - recordOffset, fname); + yaz_log(YLOG_ERRNO|YLOG_FATAL, "seek to " ZINT_FORMAT " in %s", + (zint)recordOffset, fname); exit (1); } if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize) @@ -727,13 +756,14 @@ static int file_extract_record(ZebraHandle zh, /* commit this record */ rec_put (zh->reg->records, &rec); logRecord (zh); - return 1; + return ZEBRA_OK; } -int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, - int deleteFlag) +ZEBRA_RES zebra_extract_file(ZebraHandle zh, SYSNO *sysno, const char *fname, + int deleteFlag) { - int r, i, fd; + ZEBRA_RES r = ZEBRA_OK; + int i, fd; char gprefix[128]; char ext[128]; char ext_res[128]; @@ -742,12 +772,14 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, 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'; @@ -784,7 +816,7 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, &recTypeClientData))) { yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type); - return 0; + return ZEBRA_FAIL; } switch(recType->version) @@ -809,16 +841,15 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, else strcpy (full_rep, fname); - if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1) { yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep); zh->m_record_type = original_record_type; - return 0; + return ZEBRA_FAIL; } } fi = file_read_start (fd); - do + while(1) { fi->file_moffset = fi->file_offset; fi->file_more = 0; /* file_end not called (yet) */ @@ -829,8 +860,15 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, fi->file_offset = fi->file_moffset; lseek(fi->fd, fi->file_moffset, SEEK_SET); } + if (r != ZEBRA_OK) + { + break; + } + if (sysno) + { + break; + } } - while (r && !sysno); file_read_stop (fi); if (fd != -1) close (fd); @@ -844,16 +882,16 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, 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; @@ -868,6 +906,8 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, 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 = ""; /* make it printable if file is omitted */ @@ -887,12 +927,8 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, extractCtrl.fh = &fc; zebra_rec_keys_reset(zh->reg->keys); - -#if NATTR zebra_rec_keys_reset(zh->reg->sortKeys); -#else - zh->reg->sortKeys.buf_used = 0; -#endif + if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0])) { if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], @@ -902,7 +938,7 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, 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); } @@ -913,7 +949,8 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, 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; @@ -921,7 +958,7 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, if (!recType) { - yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type); + yaz_log (YLOG_WARN, "No such record type: %s", recordType); return ZEBRA_FAIL; } @@ -954,6 +991,8 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, return ZEBRA_FAIL; } + all_matches_add(&extractCtrl); + if (extractCtrl.match_criteria[0]) match_criteria = extractCtrl.match_criteria; @@ -1020,15 +1059,9 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, dict_insert_ord(zh->reg->matchDict, db_ord, matchStr, sizeof(*sysno), sysno); } -#if NATTR + + extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); -#else - extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys); -#endif - -#if 0 - print_rec_keys(zh, zh->reg->keys); -#endif extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, recordAttr->staticrank); zh->records_inserted++; @@ -1037,12 +1070,7 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, { /* record already exists */ zebra_rec_keys_t delkeys = zebra_rec_keys_open(); -#if NATTR zebra_rec_keys_t sortKeys = zebra_rec_keys_open(); -#else - struct sortKeys sortKeys; -#endif - if (!allow_update) { yaz_log (YLOG_LOG, "skipped %s %s %ld", @@ -1060,21 +1088,12 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, rec->info[recInfo_delKeys], rec->size[recInfo_delKeys], 0); -#if NATTR zebra_rec_keys_set_buf(sortKeys, rec->info[recInfo_sortKeys], rec->size[recInfo_sortKeys], 0); -#else - sortKeys.buf_used = rec->size[recInfo_sortKeys]; - sortKeys.buf = rec->info[recInfo_sortKeys]; -#endif -#if NATTR extract_flushSortKeys (zh, *sysno, 0, sortKeys); -#else - extract_flushSortKeys (zh, *sysno, 0, &sortKeys); -#endif extract_flushRecordKeys (zh, *sysno, 0, delkeys, recordAttr->staticrank); if (delete_flag) @@ -1082,16 +1101,16 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, /* 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) { @@ -1107,22 +1126,16 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, 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; -#if NATTR extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); -#else - extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys); -#endif extract_flushRecordKeys (zh, *sysno, 1, zh->reg->keys, recordAttr->staticrank); zh->records_updated++; } zebra_rec_keys_close(delkeys); -#if NATTR zebra_rec_keys_close(sortKeys); -#endif } /* update file type */ xfree (rec->info[recInfo_fileType]); @@ -1150,16 +1163,9 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, /* update sort keys */ xfree (rec->info[recInfo_sortKeys]); -#if NATTR zebra_rec_keys_get_buf(zh->reg->sortKeys, &rec->info[recInfo_sortKeys], &rec->size[recInfo_sortKeys]); -#else - rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used; - rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf; - zh->reg->sortKeys.buf = NULL; - zh->reg->sortKeys.buf_max = 0; -#endif /* save file size of original record */ zebraExplain_recordBytesIncrement (zh->reg->zei, @@ -1214,7 +1220,7 @@ ZEBRA_RES buffer_extract_record(ZebraHandle zh, 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; @@ -1229,12 +1235,8 @@ int explain_extract (void *handle, Record rec, data1_node *n) } zebra_rec_keys_reset(zh->reg->keys); - -#if NATTR zebra_rec_keys_reset(zh->reg->sortKeys); -#else - zh->reg->sortKeys.buf_used = 0; -#endif + extractCtrl.init = extract_init; extractCtrl.tokenAdd = extract_token_add; extractCtrl.schemaAdd = extract_schema_add; @@ -1257,36 +1259,23 @@ int explain_extract (void *handle, Record rec, data1_node *n) { zebra_rec_keys_t delkeys = zebra_rec_keys_open(); -#if NATTR zebra_rec_keys_t sortkeys = zebra_rec_keys_open(); -#else - struct sortKeys sortkeys; -#endif 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); -#if NATTR + zebra_rec_keys_set_buf(sortkeys, rec->info[recInfo_sortKeys], rec->size[recInfo_sortKeys], 0); extract_flushSortKeys (zh, rec->sysno, 0, sortkeys); zebra_rec_keys_close(sortkeys); -#else - sortkeys.buf_used = rec->size[recInfo_sortKeys]; - sortkeys.buf = rec->info[recInfo_sortKeys]; - extract_flushSortKeys (zh, rec->sysno, 0, &sortkeys); -#endif } extract_flushRecordKeys (zh, rec->sysno, 1, zh->reg->keys, 0); -#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]); zebra_rec_keys_get_buf(zh->reg->keys, @@ -1294,18 +1283,62 @@ int explain_extract (void *handle, Record rec, data1_node *n) &rec->size[recInfo_delKeys]); xfree (rec->info[recInfo_sortKeys]); -#if NATTR zebra_rec_keys_get_buf(zh->reg->sortKeys, &rec->info[recInfo_sortKeys], &rec->size[recInfo_sortKeys]); -#else - rec->size[recInfo_sortKeys] = zh->reg->sortKeys.buf_used; - rec->info[recInfo_sortKeys] = zh->reg->sortKeys.buf; - zh->reg->sortKeys.buf = NULL; - zh->reg->sortKeys.buf_max = 0; -#endif + return ZEBRA_OK; +} + +void extract_rec_keys_adjust(ZebraHandle zh, int is_insert, + zebra_rec_keys_t reckeys) +{ + ZebraExplainInfo zei = zh->reg->zei; + struct ord_stat { + int no; + int ord; + struct ord_stat *next; + }; + + if (zebra_rec_keys_rewind(reckeys)) + { + struct ord_stat *ord_list = 0; + struct ord_stat *p; + size_t slen; + const char *str; + struct it_key key_in; + while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) + { + int ord = CAST_ZINT_TO_INT(key_in.mem[0]); + + for (p = ord_list; p ; p = p->next) + if (p->ord == ord) + { + p->no++; + break; + } + if (!p) + { + p = xmalloc(sizeof(*p)); + p->no = 1; + p->ord = ord; + p->next = ord_list; + ord_list = p; + } + } + + p = ord_list; + while (p) + { + struct ord_stat *p1 = p; - return 0; + if (is_insert) + zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1); + else + zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1); + p = p->next; + xfree(p1); + } + } } void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, @@ -1315,6 +1348,8 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, { ZebraExplainInfo zei = zh->reg->zei; + extract_rec_keys_adjust(zh, cmd, reckeys); + if (!zh->reg->key_buf) { int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8")); @@ -1357,11 +1392,11 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, (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); - + /* copy the 0-terminated stuff from str to output */ memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen); zh->reg->key_buf_used += slen; @@ -1374,8 +1409,8 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, { if (staticrank < 0) { - yaz_log(YLOG_WARN, "staticrank = %d. Setting to 0", - staticrank); + yaz_log(YLOG_WARN, "staticrank = %ld. Setting to 0", + (long) staticrank); staticrank = 0; } *keyp++ = staticrank; @@ -1414,19 +1449,14 @@ void extract_flushWriteKeys (ZebraHandle zh, int final) #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 ", - zh->reg->key_buf); - yaz_log (YLOG_DEBUG, " ptr=%d ",zh->reg->ptr_i); - yaz_log (YLOG_DEBUG, " 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, @@ -1546,14 +1576,16 @@ ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, { 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/* db */, 0 /* string_index */); assert(index_type); zebra_term_untrans_iconv(zh, nmem, index_type, &dst_term, str); @@ -1576,25 +1608,26 @@ void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys) 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); + 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, zinfo_index_category_t cat, + const char *str, int length) { struct it_key key; @@ -1602,24 +1635,13 @@ void extract_add_index_string (RecWord *p, const char *str, int length) 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, cat, p->index_type, p->index_name); + if (ch < 0) + ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name); + key.len = 4; key.mem[0] = ch; key.mem[1] = p->record_id; @@ -1646,78 +1668,66 @@ void extract_add_index_string (RecWord *p, const char *str, int length) zebra_rec_keys_write(zh->reg->keys, str, length, &key); } -#if NATTR -static void extract_add_sort_string (RecWord *p, const char *str, int length) +static void extract_add_sort_string(RecWord *p, const char *str, int length) { struct it_key key; ZebraHandle zh = p->extractCtrl->handle; ZebraExplainInfo zei = zh->reg->zei; int ch; + zinfo_index_category_t cat = zinfo_index_category_sort; + - 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 - { - return; - } + if (!p->index_name) + return; + + ch = zebraExplain_lookup_attr_str(zei, cat, p->index_type, p->index_name); + if (ch < 0) + ch = zebraExplain_add_attr_str(zei, cat, p->index_type, p->index_name); key.len = 4; key.mem[0] = ch; key.mem[1] = p->record_id; key.mem[2] = p->section_id; key.mem[3] = p->seqno; - zebra_rec_keys_write(zh->reg->sortKeys, str, length, &key); -} -#else -static void extract_add_sort_string (RecWord *p, const char *str, int length) -{ - ZebraHandle zh = p->extractCtrl->handle; - struct sortKeys *sk = &zh->reg->sortKeys; - int off = 0; - - while (off < sk->buf_used) - { - int set, use, slen; - - off += key_SU_decode(&set, (unsigned char *) sk->buf + off); - off += key_SU_decode(&use, (unsigned char *) sk->buf + off); - off += key_SU_decode(&slen, (unsigned char *) sk->buf + off); - off += slen; - if (p->attrSet == set && p->attrUse == use) - return; - } - assert (off == sk->buf_used); - - if (sk->buf_used + IT_MAX_WORD > sk->buf_max) +#if 0 + if (1) { - char *b; - - b = (char *) xmalloc (sk->buf_max += 128000); - if (sk->buf_used > 0) - memcpy (b, sk->buf, sk->buf_used); - xfree (sk->buf); - sk->buf = b; + char strz[80]; + int i; + + strz[0] = 0; + for (i = 0; iattrSet, p->attrUse, p->record_id, p->section_id, p->seqno, + strz); } - off += key_SU_encode(p->attrSet, sk->buf + off); - off += key_SU_encode(p->attrUse, sk->buf + off); - off += key_SU_encode(length, sk->buf + off); - memcpy (sk->buf + off, str, length); - sk->buf_used = off + length; -} #endif + 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)) extract_add_sort_string (p, string, length); else - extract_add_index_string (p, string, length); + { + extract_add_index_string(p, zinfo_index_category_index, + string, length); + if (zebra_maps_is_alwaysmatches(p->zebra_maps, p->index_type)) + { + RecWord word; + memcpy(&word, p, sizeof(word)); + + word.seqno = 1; + extract_add_index_string( + &word, zinfo_index_category_alwaysmatches, "", 0); + } + } } static void extract_add_incomplete_field (RecWord *p) @@ -1726,8 +1736,6 @@ static void extract_add_incomplete_field (RecWord *p) 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); @@ -1775,9 +1783,6 @@ static void extract_add_complete_field (RecWord *p) 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); @@ -1812,7 +1817,6 @@ static void extract_add_complete_field (RecWord *p) { 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++); } @@ -1831,15 +1835,14 @@ static void extract_add_complete_field (RecWord *p) 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))) { @@ -1877,95 +1880,44 @@ static void extract_set_store_data_prepare(struct recExtractCtrl *p) 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); } -#if NATTR -#error not done yet with zebra_rec_keys_t void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno, - int cmd, struct recKeys *reckeys) + int cmd, zebra_rec_keys_t reckeys) { - SortIdx sortIdx = zh->reg->sortIdx; - void *decode_handle = iscz1_start(); - int off = 0; - int ch = 0; - - while (off < reckeys->buf_used) + if (zebra_rec_keys_rewind(reckeys)) { - const char *src = reckeys->buf + off; - struct it_key key; - char *dst = (char*) &key; - - iscz1_decode(decode_handle, &dst, &src); - assert(key.len == 4); - - ch = (int) key.mem[0]; /* ordinal for field/use/attribute */ - - sortIdx_type(sortIdx, ch); - if (cmd == 1) - sortIdx_add(sortIdx, src, strlen(src)); - else - sortIdx_add(sortIdx, "", 1); - - src += strlen(src); - src++; - - off = src - reckeys->buf; - } - assert (off == reckeys->buf_used); - iscz1_stop(decode_handle); -} -#else -void extract_flushSortKeys (ZebraHandle zh, SYSNO sysno, - int cmd, struct sortKeys *sk) -{ - SortIdx sortIdx = zh->reg->sortIdx; - int off = 0; + SortIdx sortIdx = zh->reg->sortIdx; + size_t slen; + const char *str; + struct it_key key_in; - sortIdx_sysno (sortIdx, sysno); + sortIdx_sysno (sortIdx, sysno); - while (off < sk->buf_used) - { - int set, use, slen; - - off += key_SU_decode(&set, (unsigned char *) sk->buf + off); - off += key_SU_decode(&use, (unsigned char *) sk->buf + off); - off += key_SU_decode(&slen, (unsigned char *) sk->buf + off); - - sortIdx_type(sortIdx, use); - if (cmd == 1) - sortIdx_add(sortIdx, sk->buf + off, slen); - else - sortIdx_add(sortIdx, "", 1); - off += slen; + while (zebra_rec_keys_read(reckeys, &str, &slen, &key_in)) + { + int ord = CAST_ZINT_TO_INT(key_in.mem[0]); + + sortIdx_type(sortIdx, ord); + if (cmd == 1) + sortIdx_add(sortIdx, str, slen); + else + sortIdx_add(sortIdx, "", 1); + } } } -#endif -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; @@ -2009,110 +1961,17 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf) #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) +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab */ -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