X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=66212ceeea825550494f8aef800902a01735fb26;hb=54ddb0c5a3a8157235271665cd874576e656eed4;hp=9d2742c928eb35ce7195ba94fe2bd4742a26bff1;hpb=3c37091eb3e508bff58b297eaa0feab5e92ff53b;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index 9d2742c..66212ce 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,5 +1,5 @@ -/* $Id: extract.c,v 1.204 2006-03-20 15:17:30 mike Exp $ - Copyright (C) 1995-2005 +/* $Id: extract.c,v 1.210 2006-05-10 12:31:08 adam Exp $ + Copyright (C) 1995-2006 Index Data ApS This file is part of the Zebra server. @@ -379,13 +379,13 @@ 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 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) { RecordAttr *recordAttr; int r; @@ -400,7 +400,7 @@ 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) @@ -447,7 +447,7 @@ 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 ... */ @@ -457,7 +457,7 @@ static int file_extract_record(ZebraHandle zh, yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type, fname, recordOffset); } - return 0; + return ZEBRA_FAIL; } else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER) { @@ -469,7 +469,7 @@ static int file_extract_record(ZebraHandle zh, PRINTF_OFF_T, zh->m_record_type, fname, recordOffset); } - return 0; + return ZEBRA_FAIL; } if (extractCtrl.match_criteria[0]) matchStr = extractCtrl.match_criteria; @@ -488,7 +488,7 @@ static int file_extract_record(ZebraHandle zh, if (!matchStr) { yaz_log(YLOG_WARN, "Bad match criteria"); - return 0; + return ZEBRA_FAIL; } } if (matchStr) @@ -508,12 +508,12 @@ 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; + return ZEBRA_OK; } if (! *sysno) @@ -524,15 +524,26 @@ static int file_extract_record(ZebraHandle zh, 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; + 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) + { + if (matchStr) + yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T + " " ZINT_FORMAT " %s" , + zh->m_record_type, + fname, recordOffset, *sysno, matchStr); + else + yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T + " " ZINT_FORMAT , + zh->m_record_type, + fname, recordOffset, *sysno); + } recordAttr = rec_init_attr (zh->reg->zei, rec); recordAttr->staticrank = extractCtrl.staticrank; @@ -542,6 +553,8 @@ 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 @@ -591,15 +604,26 @@ 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_LOG, "delete %s %s " PRINTF_OFF_T + " " ZINT_FORMAT, + zh->m_record_type, fname, 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); + { + if (matchStr) + yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T + " " ZINT_FORMAT " %s" , + zh->m_record_type, + fname, recordOffset, *sysno, matchStr); + else + yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T + " " ZINT_FORMAT , + zh->m_record_type, + fname, recordOffset, *sysno); + } zh->records_deleted++; if (matchStr) { @@ -610,14 +634,24 @@ 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); + { + if (matchStr) + yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T + " " ZINT_FORMAT " %s" , + zh->m_record_type, + fname, recordOffset, *sysno, matchStr); + else + yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T + " " ZINT_FORMAT , + zh->m_record_type, + fname, recordOffset, *sysno); + } recordAttr->staticrank = extractCtrl.staticrank; #if NATTR extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys); @@ -727,13 +761,14 @@ 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]; @@ -784,7 +819,7 @@ 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 +844,15 @@ 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 +863,15 @@ 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); @@ -921,7 +962,7 @@ 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; } @@ -1020,6 +1061,8 @@ 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 @@ -1308,6 +1351,58 @@ int explain_extract (void *handle, Record rec, data1_node *n) return 0; } +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 = 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; + + 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, int cmd, zebra_rec_keys_t reckeys, @@ -1315,6 +1410,8 @@ 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")); @@ -1361,7 +1458,7 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, 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; @@ -1553,7 +1650,8 @@ ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, ord = key.mem[0]; zebraExplain_lookup_ord(zh->reg->zei, ord, &index_type, - 0/* db */, 0/* set */, 0/* use */); + 0/* db */, 0/* set */, 0/* use */, + 0 /* string_index */); assert(index_type); zebra_term_untrans_iconv(zh, nmem, index_type, &dst_term, str); @@ -1582,7 +1680,7 @@ void print_rec_keys(ZebraHandle zh, zebra_rec_keys_t reckeys) assert(key.len <= 4 && key.len > 2); zebraExplain_lookup_ord(zh->reg->zei, - key.mem[0], &index_type, &db, 0, 0); + key.mem[0], &index_type, &db, 0, 0, 0); seqno = (int) key.mem[key.len-1]; @@ -2116,3 +2214,11 @@ void encode_key_flush (struct encode_info *i, FILE *outf) i->prevseq=0; } #endif +/* + * Local variables: + * c-basic-offset: 4 + * indent-tabs-mode: nil + * End: + * vim: shiftwidth=4 tabstop=8 expandtab + */ +