X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fextract.c;h=de6aa9e776431ecbab32bc66729547e9b997c4f8;hp=b4bd6639cb3b91a480c3127de27bece580694fdb;hb=47eeb5384a8fae8bcac8afb8a84782ca094683f9;hpb=94731f54d976f745cc604bd7e00e7de3fd8151fa diff --git a/index/extract.c b/index/extract.c index b4bd663..de6aa9e 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,6 +1,6 @@ -/* $Id: extract.c,v 1.135 2003-02-25 21:51:05 adam Exp $ - Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002 - Index Data Aps +/* $Id: extract.c,v 1.186 2005-06-14 20:28:54 adam Exp $ + Copyright (C) 1995-2005 + Index Data ApS This file is part of the Zebra server. @@ -20,12 +20,13 @@ Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */ - #include #include +#include #ifdef WIN32 #include -#else +#endif +#if HAVE_UNISTD_H #include #endif #include @@ -72,20 +73,26 @@ static void logRecord (ZebraHandle zh) ++zh->records_processed; if (!(zh->records_processed % 1000)) { - logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d", + 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_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; w->attrSet = VAL_BIB1; w->attrUse = 1016; + w->attrStr = 0; w->reg_type = 'w'; w->extractCtrl = p; + w->record_id = 0; + w->section_id = 0; } static const char **searchRecordKey (ZebraHandle zh, @@ -93,81 +100,45 @@ static const char **searchRecordKey (ZebraHandle zh, int attrSetS, int attrUseS) { static const char *ws[32]; + void *decode_handle = iscz1_start(); int off = 0; int startSeq = -1; - int i; int seqno = 0; -#if SU_SCHEME - int chS, ch; -#else - short attrUse; - char attrSet; -#endif + int i; for (i = 0; i<32; i++) ws[i] = NULL; -#if SU_SCHEME - chS = zebraExplain_lookupSU (zh->reg->zei, attrSetS, attrUseS); - if (chS < 0) - return ws; -#endif while (off < reckeys->buf_used) { - const char *src = reckeys->buf + off; - const char *wstart; - int lead; - - lead = *src++; -#if SU_SCHEME - if ((lead & 3)<3) - { - memcpy (&ch, src, sizeof(ch)); - src += sizeof(ch); - } -#else - if (!(lead & 1)) - { - memcpy (&attrSet, src, sizeof(attrSet)); - src += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (&attrUse, src, sizeof(attrUse)); - src += sizeof(attrUse); - } -#endif - wstart = src; - while (*src++) - ; - if (lead & 60) - seqno += ((lead>>2) & 15)-1; - else - { - memcpy (&seqno, src, sizeof(seqno)); - src += sizeof(seqno); - } - if ( -#if SU_SCHEME - ch == chS -#else - attrUseS == attrUse && attrSetS == attrSet -#endif - ) + struct it_key key; + char *dst = (char*) &key; + int attrSet, attrUse; + + iscz1_decode(decode_handle, &dst, &src); + assert(key.len <= 4 && key.len > 2); + + attrSet = (int) key.mem[0] >> 16; + attrUse = (int) key.mem[0] & 65535; + seqno = (int) key.mem[key.len-1]; + + if (attrUseS == attrUse && attrSetS == attrSet) { int woff; - if (startSeq == -1) startSeq = seqno; woff = seqno - startSeq; if (woff >= 0 && woff < 31) - ws[woff] = wstart; + ws[woff] = src; } + while (*src++) + ; off = src - reckeys->buf; } + iscz1_stop(decode_handle); assert (off == reckeys->buf_used); return ws; } @@ -178,8 +149,6 @@ struct file_read_info { off_t file_moffset; /* offset of rec/rec boundary */ int file_more; int fd; - char *sdrbuf; - int sdrmax; }; static struct file_read_info *file_read_start (int fd) @@ -190,8 +159,8 @@ static struct file_read_info *file_read_start (int fd) fi->fd = fd; fi->file_max = 0; fi->file_moffset = 0; - fi->sdrbuf = 0; - fi->sdrmax = 0; + fi->file_offset = 0; + fi->file_more = 0; return fi; } @@ -204,8 +173,6 @@ static off_t file_seek (void *handle, off_t offset) { struct file_read_info *p = (struct file_read_info *) handle; p->file_offset = offset; - if (p->sdrbuf) - return offset; return lseek (p->fd, offset, SEEK_SET); } @@ -220,16 +187,7 @@ static int file_read (void *handle, char *buf, size_t count) struct file_read_info *p = (struct file_read_info *) handle; int fd = p->fd; int r; - if (p->sdrbuf) - { - r = count; - if (r > p->sdrmax - p->file_offset) - r = p->sdrmax - p->file_offset; - if (r) - memcpy (buf, p->sdrbuf + p->file_offset, r); - } - else - r = read (fd, buf, count); + r = read (fd, buf, count); if (r > 0) { p->file_offset += r; @@ -239,27 +197,19 @@ static int file_read (void *handle, char *buf, size_t count) return r; } -static void file_begin (void *handle) -{ - struct file_read_info *p = (struct file_read_info *) handle; - - p->file_offset = p->file_moffset; - if (!p->sdrbuf && p->file_moffset) - lseek (p->fd, p->file_moffset, SEEK_SET); - p->file_more = 0; -} - static void file_end (void *handle, off_t offset) { struct file_read_info *p = (struct file_read_info *) handle; - assert (p->file_more == 0); - p->file_more = 1; - p->file_moffset = offset; + if (offset != p->file_moffset) + { + p->file_moffset = offset; + p->file_more = 1; + } } static char *fileMatchStr (ZebraHandle zh, - struct recKeys *reckeys, struct recordGroup *rGroup, + struct recKeys *reckeys, const char *fname, const char *spec) { static char dstBuf[2048]; /* static here ??? */ @@ -317,8 +267,8 @@ static char *fileMatchStr (ZebraHandle zh, } else { - logf (LOG_WARN, "Missing ) in match criteria %s in group %s", - spec, rGroup->groupName ? rGroup->groupName : "none"); + yaz_log (YLOG_WARN, "Missing ) in match criteria %s in group %s", + spec, zh->m_group ? zh->m_group : "none"); return NULL; } s++; @@ -336,7 +286,7 @@ static char *fileMatchStr (ZebraHandle zh, } if (first) { - logf (LOG_WARN, "Record didn't contain match" + yaz_log (YLOG_WARN, "Record didn't contain match" " fields in (%s,%s)", attset_str, attname_str); return NULL; } @@ -358,14 +308,14 @@ static char *fileMatchStr (ZebraHandle zh, s = s1; if (!strcmp (special, "group")) - spec_src = rGroup->groupName; + spec_src = zh->m_group; else if (!strcmp (special, "database")) - spec_src = rGroup->databaseName; + spec_src = zh->basenames[0]; else if (!strcmp (special, "filename")) { spec_src = fname; } else if (!strcmp (special, "type")) - spec_src = rGroup->recordType; + spec_src = zh->m_record_type; else spec_src = NULL; if (spec_src) @@ -393,16 +343,16 @@ static char *fileMatchStr (ZebraHandle zh, } else { - logf (LOG_WARN, "Syntax error in match criteria %s in group %s", - spec, rGroup->groupName ? rGroup->groupName : "none"); + yaz_log (YLOG_WARN, "Syntax error in match criteria %s in group %s", + spec, zh->m_group ? zh->m_group : "none"); return NULL; } *dst++ = 1; } if (dst == dstBuf) { - logf (LOG_WARN, "No match criteria for record %s in group %s", - fname, rGroup->groupName ? rGroup->groupName : "none"); + yaz_log (YLOG_WARN, "No match criteria for record %s in group %s", + fname, zh->m_group ? zh->m_group : "none"); return NULL; } *dst = '\0'; @@ -414,68 +364,88 @@ struct recordLogInfo { int recordOffset; struct recordGroup *rGroup; }; - -static int recordExtract (ZebraHandle zh, - SYSNO *sysno, const char *fname, - struct recordGroup *rGroup, int deleteFlag, - struct file_read_info *fi, - RecType recType, char *subType, void *clientData) + +void create_rec_keys_codec(struct recKeys *keys) +{ + keys->buf_used = 0; + iscz1_reset(keys->codec_handle); +} + +static void init_extractCtrl(ZebraHandle zh, struct recExtractCtrl *ctrl) +{ + int i; + for (i = 0; i<256; i++) + { + if (zebra_maps_is_positioned(zh->reg->zebra_maps, i)) + ctrl->seqno[i] = 1; + else + ctrl->seqno[i] = 0; + } + ctrl->zebra_maps = zh->reg->zebra_maps; + 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) { RecordAttr *recordAttr; int r; - char *matchStr; + const char *matchStr = 0; SYSNO sysnotmp; Record rec; off_t recordOffset = 0; + + /* announce database */ + if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0])) + { + if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], + zh->m_explain_database)) + return 0; + } if (fi->fd != -1) { struct recExtractCtrl extractCtrl; /* we are going to read from a file, so prepare the extraction */ - int i; + create_rec_keys_codec(&zh->reg->keys); - zh->reg->keys.buf_used = 0; - zh->reg->keys.prevAttrUse = -1; - zh->reg->keys.prevAttrSet = -1; - zh->reg->keys.prevSeqNo = 0; zh->reg->sortKeys.buf_used = 0; - zh->reg->sortKeys.buf_max = 0; - zh->reg->sortKeys.buf = 0; recordOffset = fi->file_moffset; + extractCtrl.handle = zh; extractCtrl.offset = fi->file_moffset; extractCtrl.readf = file_read; extractCtrl.seekf = file_seek; extractCtrl.tellf = file_tell; extractCtrl.endf = file_end; extractCtrl.fh = fi; - extractCtrl.subType = subType; extractCtrl.init = extract_init; extractCtrl.tokenAdd = extract_token_add; extractCtrl.schemaAdd = extract_schema_add; extractCtrl.dh = zh->reg->dh; - extractCtrl.handle = zh; - for (i = 0; i<256; i++) - { - if (zebra_maps_is_positioned(zh->reg->zebra_maps, i)) - extractCtrl.seqno[i] = 1; - else - extractCtrl.seqno[i] = 0; - } - extractCtrl.zebra_maps = zh->reg->zebra_maps; - extractCtrl.flagShowRecords = !rGroup->flagRw; + extractCtrl.match_criteria[0] = '\0'; + extractCtrl.first_record = fi->file_offset ? 0 : 1; + + extract_set_store_data_prepare(&extractCtrl); - if (!rGroup->flagRw) + init_extractCtrl(zh, &extractCtrl); + + if (!zh->m_flag_rw) printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset); - if (rGroup->flagRw) + if (zh->m_flag_rw) { char msg[512]; sprintf (msg, "%s:" PRINTF_OFF_T , fname, recordOffset); yaz_log_init_prefix2 (msg); } - r = (*recType->extract)(clientData, &extractCtrl); + r = (*recType->extract)(recTypeClientData, &extractCtrl); yaz_log_init_prefix2 (0); if (r == RECCTRL_EXTRACT_EOF) @@ -483,10 +453,10 @@ static int recordExtract (ZebraHandle zh, else if (r == RECCTRL_EXTRACT_ERROR_GENERIC) { /* error occured during extraction ... */ - if (rGroup->flagRw && - zh->records_processed < rGroup->fileVerboseLimit) + if (zh->m_flag_rw && + zh->records_processed < zh->m_file_verbose_limit) { - logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, rGroup->recordType, + yaz_log (YLOG_WARN, "fail %s %s " PRINTF_OFF_T, zh->m_record_type, fname, recordOffset); } return 0; @@ -494,53 +464,57 @@ static int recordExtract (ZebraHandle zh, else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER) { /* error occured during extraction ... */ - if (rGroup->flagRw && - zh->records_processed < rGroup->fileVerboseLimit) + if (zh->m_flag_rw && + zh->records_processed < zh->m_file_verbose_limit) { - logf (LOG_WARN, "no filter for %s %s " - PRINTF_OFF_T, rGroup->recordType, + yaz_log (YLOG_WARN, "no filter for %s %s " + PRINTF_OFF_T, zh->m_record_type, fname, recordOffset); } return 0; } - if (zh->reg->keys.buf_used == 0) - { - /* the extraction process returned no information - the record - is probably empty - unless flagShowRecords is in use */ - if (!rGroup->flagRw) - return 1; - - logf (LOG_WARN, "empty %s %s " PRINTF_OFF_T, rGroup->recordType, - fname, recordOffset); - return 1; - } + if (extractCtrl.match_criteria[0]) + matchStr = extractCtrl.match_criteria; } /* perform match if sysno not known and if match criteria is specified */ - - matchStr = NULL; if (!sysno) { sysnotmp = 0; sysno = &sysnotmp; - if (rGroup->recordId && *rGroup->recordId) + + if (matchStr == 0 && zh->m_record_id && *zh->m_record_id) { - char *rinfo; - matchStr = fileMatchStr (zh, &zh->reg->keys, rGroup, fname, - rGroup->recordId); - if (matchStr) - { - rinfo = dict_lookup (zh->reg->matchDict, matchStr); - if (rinfo) - memcpy (sysno, rinfo+1, sizeof(*sysno)); - } - else - { - logf (LOG_WARN, "Bad match criteria"); - return 0; - } - } + matchStr = fileMatchStr (zh, &zh->reg->keys, fname, + zh->m_record_id); + if (!matchStr) + { + yaz_log(YLOG_WARN, "Bad match criteria"); + return 0; + } + } + if (matchStr) + { + char *rinfo = dict_lookup (zh->reg->matchDict, matchStr); + if (rinfo) + { + assert(*rinfo == sizeof(*sysno)); + memcpy (sysno, rinfo+1, sizeof(*sysno)); + } + } + } + if (! *sysno && zh->reg->keys.buf_used == 0) + { + /* the extraction process returned no information - the record + is probably empty - unless flagShowRecords is in use */ + if (!zh->m_flag_rw) + return 1; + + 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; } if (! *sysno) @@ -548,13 +522,13 @@ static int recordExtract (ZebraHandle zh, /* new record */ if (deleteFlag) { - logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, rGroup->recordType, + yaz_log (YLOG_LOG, "delete %s %s " PRINTF_OFF_T, zh->m_record_type, fname, recordOffset); - logf (LOG_WARN, "cannot delete record above (seems new)"); + yaz_log (YLOG_WARN, "cannot delete record above (seems new)"); return 1; } - if (zh->records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType, + 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); @@ -582,12 +556,13 @@ static int recordExtract (ZebraHandle zh, recordAttr = rec_init_attr (zh->reg->zei, rec); - if (recordAttr->runNumber == + if (!force_update && recordAttr->runNumber == zebraExplain_runNumberIncrement (zh->reg->zei, 0)) { - yaz_log (LOG_LOG, "run number = %d", recordAttr->runNumber); - yaz_log (LOG_LOG, "skipped %s %s " PRINTF_OFF_T, - rGroup->recordType, fname, recordOffset); + yaz_log (YLOG_LOG, "run number = " ZINT_FORMAT, + recordAttr->runNumber); + yaz_log (YLOG_LOG, "skipped %s %s " PRINTF_OFF_T, + zh->m_record_type, fname, recordOffset); extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys); rec_rm (&rec); logRecord (zh); @@ -606,15 +581,15 @@ static int recordExtract (ZebraHandle zh, /* record going to be deleted */ if (!delkeys.buf_used) { - logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, - rGroup->recordType, fname, recordOffset); - logf (LOG_WARN, "cannot delete file above, storeKeys false"); + 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"); } else { - if (zh->records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, - rGroup->recordType, fname, recordOffset); + 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); zh->records_deleted++; if (matchStr) dict_delete (zh->reg->matchDict, matchStr); @@ -629,15 +604,15 @@ static int recordExtract (ZebraHandle zh, /* record going to be updated */ if (!delkeys.buf_used) { - logf (LOG_LOG, "update %s %s " PRINTF_OFF_T, - rGroup->recordType, fname, recordOffset); - logf (LOG_WARN, "cannot update file above, storeKeys false"); + yaz_log (YLOG_LOG, "update %s %s " PRINTF_OFF_T, + zh->m_record_type, fname, recordOffset); + yaz_log (YLOG_WARN, "cannot update file above, storeKeys false"); } else { - if (zh->records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "update %s %s " PRINTF_OFF_T, - rGroup->recordType, fname, recordOffset); + 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); extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys); extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys); zh->records_updated++; @@ -647,7 +622,7 @@ static int recordExtract (ZebraHandle zh, /* update file type */ xfree (rec->info[recInfo_fileType]); rec->info[recInfo_fileType] = - rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]); + rec_strdup (zh->m_record_type, &rec->size[recInfo_fileType]); /* update filename */ xfree (rec->info[recInfo_filename]); @@ -656,7 +631,7 @@ static int recordExtract (ZebraHandle zh, /* update delete keys */ xfree (rec->info[recInfo_delKeys]); - if (zh->reg->keys.buf_used > 0 && rGroup->flagStoreKeys == 1) + if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1) { rec->size[recInfo_delKeys] = zh->reg->keys.buf_used; rec->info[recInfo_delKeys] = zh->reg->keys.buf; @@ -692,21 +667,27 @@ static int recordExtract (ZebraHandle zh, /* update store data */ xfree (rec->info[recInfo_storeData]); - if (rGroup->flagStoreData == 1) + if (zh->store_data_buf) + { + rec->size[recInfo_storeData] = zh->store_data_size; + rec->info[recInfo_storeData] = zh->store_data_buf; + zh->store_data_buf = 0; + } + else if (zh->m_store_data) { rec->size[recInfo_storeData] = recordAttr->recordSize; rec->info[recInfo_storeData] = (char *) xmalloc (recordAttr->recordSize); if (lseek (fi->fd, recordOffset, SEEK_SET) < 0) { - logf (LOG_ERRNO|LOG_FATAL, "seek to " PRINTF_OFF_T " in %s", + yaz_log (YLOG_ERRNO|YLOG_FATAL, "seek to " PRINTF_OFF_T " in %s", recordOffset, fname); exit (1); } if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize) < recordAttr->recordSize) { - logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s", + yaz_log (YLOG_ERRNO|YLOG_FATAL, "read %d bytes of %s", recordAttr->recordSize, fname); exit (1); } @@ -719,7 +700,7 @@ static int recordExtract (ZebraHandle zh, /* update database name */ xfree (rec->info[recInfo_databaseName]); rec->info[recInfo_databaseName] = - rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]); + rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]); /* update offset */ recordAttr->recordOffset = recordOffset; @@ -731,27 +712,23 @@ static int recordExtract (ZebraHandle zh, } int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, - const struct recordGroup *rGroupP, int deleteFlag) + int deleteFlag) { int r, i, fd; char gprefix[128]; char ext[128]; char ext_res[128]; - char subType[128]; - RecType recType; - struct recordGroup rGroupM; - struct recordGroup *rGroup = &rGroupM; struct file_read_info *fi; - void *clientData; + const char *original_record_type = 0; + RecType recType; + void *recTypeClientData; - memcpy (rGroup, rGroupP, sizeof(*rGroupP)); - - if (!rGroup->groupName || !*rGroup->groupName) + if (!zh->m_group || !*zh->m_group) *gprefix = '\0'; else - sprintf (gprefix, "%s.", rGroup->groupName); - - logf (LOG_DEBUG, "fileExtract %s", fname); + sprintf (gprefix, "%s.", zh->m_group); + + yaz_log (YLOG_DEBUG, "fileExtract %s", fname); /* determine file extension */ *ext = '\0'; @@ -764,99 +741,40 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, break; } /* determine file type - depending on extension */ - if (!rGroup->recordType) + original_record_type = zh->m_record_type; + if (!zh->m_record_type) { sprintf (ext_res, "%srecordType.%s", gprefix, ext); - if (!(rGroup->recordType = res_get (zh->res, ext_res))) - { - sprintf (ext_res, "%srecordType", gprefix); - rGroup->recordType = res_get (zh->res, ext_res); - } + zh->m_record_type = res_get (zh->res, ext_res); } - if (!rGroup->recordType) + if (!zh->m_record_type) { - if (zh->records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "? %s", fname); + if (zh->records_processed < zh->m_file_verbose_limit) + yaz_log (YLOG_LOG, "? %s", fname); return 0; } - if (!*rGroup->recordType) - return 0; - if (!(recType = - recType_byName (zh->reg->recTypes, rGroup->recordType, subType, - &clientData))) - { - logf (LOG_WARN, "No such record type: %s", rGroup->recordType); - return 0; - } - /* determine match criteria */ - if (!rGroup->recordId) + if (!zh->m_record_id) { sprintf (ext_res, "%srecordId.%s", gprefix, ext); - rGroup->recordId = res_get (zh->res, ext_res); + zh->m_record_id = res_get (zh->res, ext_res); } - /* determine database name */ - if (!rGroup->databaseName) - { - sprintf (ext_res, "%sdatabase.%s", gprefix, ext); - if (!(rGroup->databaseName = res_get (zh->res, ext_res))) - { - sprintf (ext_res, "%sdatabase", gprefix); - rGroup->databaseName = res_get (zh->res, ext_res); - } - } - if (!rGroup->databaseName) - rGroup->databaseName = "Default"; - - /* determine if explain database */ - - sprintf (ext_res, "%sexplainDatabase", gprefix); - rGroup->explainDatabase = - atoi (res_get_def (zh->res, ext_res, "0")); - - /* announce database */ - if (zebraExplain_curDatabase (zh->reg->zei, rGroup->databaseName)) - { - if (zebraExplain_newDatabase (zh->reg->zei, rGroup->databaseName, - rGroup->explainDatabase)) - return 0; - } - - if (rGroup->flagStoreData == -1) + if (!(recType = + recType_byName (zh->reg->recTypes, zh->res, zh->m_record_type, + &recTypeClientData))) { - const char *sval; - sprintf (ext_res, "%sstoreData.%s", gprefix, ext); - if (!(sval = res_get (zh->res, ext_res))) - { - sprintf (ext_res, "%sstoreData", gprefix); - sval = res_get (zh->res, ext_res); - } - if (sval) - rGroup->flagStoreData = atoi (sval); + yaz_log(YLOG_WARN, "No such record type: %s", zh->m_record_type); + return 0; } - if (rGroup->flagStoreData == -1) - rGroup->flagStoreData = 0; - if (rGroup->flagStoreKeys == -1) + switch(recType->version) { - const char *sval; - - sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext); - sval = res_get (zh->res, ext_res); - if (!sval) - { - sprintf (ext_res, "%sstoreKeys", gprefix); - sval = res_get (zh->res, ext_res); - } - if (!sval) - sval = res_get (zh->res, "storeKeys"); - if (sval) - rGroup->flagStoreKeys = atoi (sval); + case 0: + break; + default: + yaz_log(YLOG_WARN, "Bad filter version: %s", zh->m_record_type); } - if (rGroup->flagStoreKeys == -1) - rGroup->flagStoreKeys = 0; - if (sysno && deleteFlag) fd = -1; else @@ -875,77 +793,63 @@ int fileExtract (ZebraHandle zh, SYSNO *sysno, const char *fname, if ((fd = open (full_rep, O_BINARY|O_RDONLY)) == -1) { - logf (LOG_WARN|LOG_ERRNO, "open %s", full_rep); + yaz_log (YLOG_WARN|YLOG_ERRNO, "open %s", full_rep); + zh->m_record_type = original_record_type; return 0; } } fi = file_read_start (fd); do { - file_begin (fi); - r = recordExtract (zh, sysno, fname, rGroup, deleteFlag, fi, - recType, subType, clientData); - } while (r && !sysno && fi->file_more); + fi->file_moffset = fi->file_offset; + fi->file_more = 0; /* file_end not called (yet) */ + r = file_extract_record (zh, sysno, fname, deleteFlag, fi, 1, + recType, recTypeClientData); + if (fi->file_more) + { /* file_end has been called so reset offset .. */ + fi->file_offset = fi->file_moffset; + lseek(fi->fd, fi->file_moffset, SEEK_SET); + } + } + while (r && !sysno); file_read_stop (fi); if (fd != -1) close (fd); + zh->m_record_type = original_record_type; return r; } -int extract_rec_in_mem (ZebraHandle zh, const char *recordType, - const char *buf, size_t buf_size, - const char *databaseName, int delete_flag, - int test_mode, int *sysno, - int store_keys, int store_data, - const char *match_criteria) -{ - struct recordGroup rGroup; - rGroup.groupName = NULL; - rGroup.databaseName = (char *)databaseName; - rGroup.path = NULL; - rGroup.recordId = NULL; - rGroup.recordType = (char *)recordType; - rGroup.flagStoreData = store_data; - rGroup.flagStoreKeys = store_keys; - rGroup.flagRw = 1; - rGroup.databaseNamePath = 0; - rGroup.explainDatabase = 0; - rGroup.fileVerboseLimit = 100000; - rGroup.followLinks = -1; - return (bufferExtractRecord (zh, - buf, buf_size, - &rGroup, - delete_flag, - test_mode, - sysno, - match_criteria, - "")); -} + /* If sysno is provided, then it's used to identify the reocord. If not, and match_criteria is provided, then sysno is guessed If not, and a record is provided, then sysno is got from there */ -int bufferExtractRecord (ZebraHandle zh, - const char *buf, size_t buf_size, - struct recordGroup *rGroup, - int delete_flag, - int test_mode, - int *sysno, - const char *match_criteria, - const char *fname) - +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) { RecordAttr *recordAttr; struct recExtractCtrl extractCtrl; - int i, r; - char *matchStr = 0; - RecType recType; - char subType[1024]; + int r; + const char *matchStr = 0; + RecType recType = NULL; void *clientData; Record rec; long recordOffset = 0; struct zebra_fetch_control fc; + const char *pr_fname = fname; /* filename to print .. */ + int show_progress = zh->records_processed < zh->m_file_verbose_limit ? 1:0; + + if (!pr_fname) + pr_fname = ""; /* make it printable if file is omitted */ fc.fd = -1; fc.record_int_buf = buf; @@ -959,99 +863,108 @@ int bufferExtractRecord (ZebraHandle zh, extractCtrl.seekf = zebra_record_int_seek; extractCtrl.tellf = zebra_record_int_tell; extractCtrl.endf = zebra_record_int_end; + extractCtrl.first_record = 1; extractCtrl.fh = &fc; - /* announce database */ - if (zebraExplain_curDatabase (zh->reg->zei, rGroup->databaseName)) + create_rec_keys_codec(&zh->reg->keys); + + zh->reg->sortKeys.buf_used = 0; + + if (zebraExplain_curDatabase (zh->reg->zei, zh->basenames[0])) { - if (zebraExplain_newDatabase (zh->reg->zei, rGroup->databaseName, 0)) - return 0; + if (zebraExplain_newDatabase (zh->reg->zei, zh->basenames[0], + zh->m_explain_database)) + return ZEBRA_FAIL; } - - if (!(rGroup->recordType)) { - logf (LOG_WARN, "No such record type defined"); - return 0; + + if (recordType && *recordType) + { + yaz_log (YLOG_DEBUG, "Record type explicitly specified: %s", recordType); + recType = recType_byName (zh->reg->recTypes, zh->res, recordType, + &clientData); + } + else + { + if (!(zh->m_record_type)) + { + 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); + recType = recType_byName (zh->reg->recTypes, zh->res, + zh->m_record_type, &clientData); + recordType = zh->m_record_type; } - - if (!(recType = - recType_byName (zh->reg->recTypes, rGroup->recordType, subType, - &clientData))) + + if (!recType) { - logf (LOG_WARN, "No such record type: %s", rGroup->recordType); - return 0; + yaz_log (YLOG_WARN, "No such record type: %s", zh->m_record_type); + return ZEBRA_FAIL; } - zh->reg->keys.buf_used = 0; - zh->reg->keys.prevAttrUse = -1; - zh->reg->keys.prevAttrSet = -1; - zh->reg->keys.prevSeqNo = 0; - zh->reg->sortKeys.buf_used = 0; - zh->reg->sortKeys.buf_max = 0; - zh->reg->sortKeys.buf = 0; - - extractCtrl.subType = subType; + extractCtrl.init = extract_init; extractCtrl.tokenAdd = extract_token_add; extractCtrl.schemaAdd = extract_schema_add; extractCtrl.dh = zh->reg->dh; extractCtrl.handle = zh; - extractCtrl.zebra_maps = zh->reg->zebra_maps; - extractCtrl.flagShowRecords = 0; - for (i = 0; i<256; i++) - { - if (zebra_maps_is_positioned(zh->reg->zebra_maps, i)) - extractCtrl.seqno[i] = 1; - else - extractCtrl.seqno[i] = 0; - } + extractCtrl.match_criteria[0] = '\0'; + + init_extractCtrl(zh, &extractCtrl); + + extract_set_store_data_prepare(&extractCtrl); r = (*recType->extract)(clientData, &extractCtrl); if (r == RECCTRL_EXTRACT_EOF) - return 0; + return ZEBRA_FAIL; else if (r == RECCTRL_EXTRACT_ERROR_GENERIC) { /* error occured during extraction ... */ - yaz_log (LOG_WARN, "extract error: generic"); - return 0; + yaz_log (YLOG_WARN, "extract error: generic"); + return ZEBRA_FAIL; } else if (r == RECCTRL_EXTRACT_ERROR_NO_SUCH_FILTER) { /* error occured during extraction ... */ - yaz_log (LOG_WARN, "extract error: no such filter"); - return 0; + yaz_log (YLOG_WARN, "extract error: no such filter"); + return ZEBRA_FAIL; + } + /* match criteria */ + matchStr = NULL; + + if (extractCtrl.match_criteria[0]) + match_criteria = extractCtrl.match_criteria; + + if (! *sysno) { + char *rinfo; + 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); + if (!matchStr) + { + yaz_log (YLOG_WARN, "Bad match criteria (recordID)"); + return ZEBRA_FAIL; + } + } + } + if (matchStr) { + rinfo = dict_lookup (zh->reg->matchDict, matchStr); + if (rinfo) + { + assert(*rinfo == sizeof(*sysno)); + memcpy (sysno, rinfo+1, sizeof(*sysno)); + } + } } if (zh->reg->keys.buf_used == 0) { /* the extraction process returned no information - the record is probably empty - unless flagShowRecords is in use */ if (test_mode) - return 1; - logf (LOG_WARN, "No keys generated for record"); - logf (LOG_WARN, " The file is probably empty"); - return 1; - } - /* match criteria */ - matchStr = NULL; - - if (! *sysno) { - char *rinfo; - if (strlen(match_criteria) > 0) { - matchStr = (char *)match_criteria; - } else { - if (rGroup->recordId && *rGroup->recordId) { - matchStr = fileMatchStr (zh, &zh->reg->keys, rGroup, fname, - rGroup->recordId); - } - } - if (matchStr) { - rinfo = dict_lookup (zh->reg->matchDict, matchStr); - if (rinfo) - memcpy (sysno, rinfo+1, sizeof(*sysno)); - } else { - logf (LOG_WARN, "Bad match criteria (recordID)"); - return 0; - } - + return ZEBRA_OK; } if (! *sysno) @@ -1059,13 +972,15 @@ int bufferExtractRecord (ZebraHandle zh, /* new record */ if (delete_flag) { - logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); - logf (LOG_WARN, "cannot delete record above (seems new)"); - return 1; + if (show_progress) + yaz_log (YLOG_LOG, "delete %s %s %ld", recordType, + pr_fname, (long) recordOffset); + yaz_log (YLOG_WARN, "cannot delete record above (seems new)"); + return ZEBRA_FAIL; } - logf (LOG_LOG, "add %s %s %ld", rGroup->recordType, fname, - (long) recordOffset); + if (show_progress) + yaz_log (YLOG_LOG, "add %s %s %ld", recordType, pr_fname, + (long) recordOffset); rec = rec_new (zh->reg->records); *sysno = rec->sysno; @@ -1081,28 +996,41 @@ int bufferExtractRecord (ZebraHandle zh, extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys); zh->records_inserted++; - } + } else { /* record already exists */ struct recKeys delkeys; struct sortKeys sortKeys; + if (!allow_update) + { + if (show_progress) + yaz_log (YLOG_LOG, "skipped %s %s %ld", + recordType, pr_fname, (long) recordOffset); + logRecord(zh); + return ZEBRA_FAIL; + } + rec = rec_get (zh->reg->records, *sysno); assert (rec); recordAttr = rec_init_attr (zh->reg->zei, rec); - - if (recordAttr->runNumber == - zebraExplain_runNumberIncrement (zh->reg->zei, 0)) - { - logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); - extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys); - rec_rm (&rec); - logRecord(zh); - return 1; + + if (!force_update) { + if (recordAttr->runNumber == + zebraExplain_runNumberIncrement (zh->reg->zei, 0)) + { + if (show_progress) + yaz_log (YLOG_LOG, "skipped %s %s %ld", recordType, + pr_fname, (long) recordOffset); + extract_flushSortKeys (zh, *sysno, -1, &zh->reg->sortKeys); + rec_rm (&rec); + logRecord(zh); + return ZEBRA_FAIL; + } } + delkeys.buf_used = rec->size[recInfo_delKeys]; delkeys.buf = rec->info[recInfo_delKeys]; @@ -1116,14 +1044,19 @@ int bufferExtractRecord (ZebraHandle zh, /* record going to be deleted */ if (!delkeys.buf_used) { - logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); - logf (LOG_WARN, "cannot delete file above, storeKeys false"); - } + if (show_progress) + { + yaz_log (YLOG_LOG, "delete %s %s %ld", recordType, + pr_fname, (long) recordOffset); + yaz_log (YLOG_WARN, "cannot delete file above, " + "storeKeys false"); + } + } else { - logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + if (show_progress) + yaz_log (YLOG_LOG, "delete %s %s %ld", recordType, + pr_fname, (long) recordOffset); zh->records_deleted++; if (matchStr) dict_delete (zh->reg->matchDict, matchStr); @@ -1131,21 +1064,25 @@ int bufferExtractRecord (ZebraHandle zh, } rec_rm (&rec); logRecord(zh); - return 1; + return ZEBRA_OK; } else { /* record going to be updated */ if (!delkeys.buf_used) { - logf (LOG_LOG, "update %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); - logf (LOG_WARN, "cannot update file above, storeKeys false"); - } + if (show_progress) + { + yaz_log (YLOG_LOG, "update %s %s %ld", recordType, + pr_fname, (long) recordOffset); + yaz_log (YLOG_WARN, "cannot update file above, storeKeys false"); + } + } else { - logf (LOG_LOG, "update %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + if (show_progress) + yaz_log (YLOG_LOG, "update %s %s %ld", recordType, + pr_fname, (long) recordOffset); extract_flushSortKeys (zh, *sysno, 1, &zh->reg->sortKeys); extract_flushRecordKeys (zh, *sysno, 1, &zh->reg->keys); zh->records_updated++; @@ -1155,7 +1092,7 @@ int bufferExtractRecord (ZebraHandle zh, /* update file type */ xfree (rec->info[recInfo_fileType]); rec->info[recInfo_fileType] = - rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]); + rec_strdup (recordType, &rec->size[recInfo_fileType]); /* update filename */ xfree (rec->info[recInfo_filename]); @@ -1164,7 +1101,7 @@ int bufferExtractRecord (ZebraHandle zh, /* update delete keys */ xfree (rec->info[recInfo_delKeys]); - if (zh->reg->keys.buf_used > 0 && rGroup->flagStoreKeys == 1) + if (zh->reg->keys.buf_used > 0 && zh->m_store_keys == 1) { rec->size[recInfo_delKeys] = zh->reg->keys.buf_used; rec->info[recInfo_delKeys] = zh->reg->keys.buf; @@ -1204,28 +1141,20 @@ int bufferExtractRecord (ZebraHandle zh, /* update store data */ xfree (rec->info[recInfo_storeData]); - if (rGroup->flagStoreData == 1) + + /* update store data */ + if (zh->store_data_buf) + { + rec->size[recInfo_storeData] = zh->store_data_size; + rec->info[recInfo_storeData] = zh->store_data_buf; + zh->store_data_buf = 0; + } + else if (zh->m_store_data) { rec->size[recInfo_storeData] = recordAttr->recordSize; rec->info[recInfo_storeData] = (char *) xmalloc (recordAttr->recordSize); -#if 1 memcpy (rec->info[recInfo_storeData], buf, recordAttr->recordSize); -#else - if (lseek (fi->fd, recordOffset, SEEK_SET) < 0) - { - logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s", - (long) recordOffset, fname); - exit (1); - } - if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize) - < recordAttr->recordSize) - { - logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s", - recordAttr->recordSize, fname); - exit (1); - } -#endif } else { @@ -1235,7 +1164,7 @@ int bufferExtractRecord (ZebraHandle zh, /* update database name */ xfree (rec->info[recInfo_databaseName]); rec->info[recInfo_databaseName] = - rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]); + rec_strdup (zh->basenames[0], &rec->size[recInfo_databaseName]); /* update offset */ recordAttr->recordOffset = recordOffset; @@ -1243,16 +1172,13 @@ int bufferExtractRecord (ZebraHandle zh, /* commit this record */ rec_put (zh->reg->records, &rec); logRecord(zh); - return 0; + return ZEBRA_OK; } - - int explain_extract (void *handle, Record rec, data1_node *n) { ZebraHandle zh = (ZebraHandle) handle; struct recExtractCtrl extractCtrl; - int i; if (zebraExplain_curDatabase (zh->reg->zei, rec->info[recInfo_databaseName])) @@ -1263,23 +1189,26 @@ int explain_extract (void *handle, Record rec, data1_node *n) abort (); } - zh->reg->keys.buf_used = 0; - zh->reg->keys.prevAttrUse = -1; - zh->reg->keys.prevAttrSet = -1; - zh->reg->keys.prevSeqNo = 0; + create_rec_keys_codec(&zh->reg->keys); + zh->reg->sortKeys.buf_used = 0; extractCtrl.init = extract_init; extractCtrl.tokenAdd = extract_token_add; extractCtrl.schemaAdd = extract_schema_add; extractCtrl.dh = zh->reg->dh; - for (i = 0; i<256; i++) - extractCtrl.seqno[i] = 0; - extractCtrl.zebra_maps = zh->reg->zebra_maps; + + init_extractCtrl(zh, &extractCtrl); + extractCtrl.flagShowRecords = 0; + extractCtrl.match_criteria[0] = '\0'; extractCtrl.handle = handle; + extractCtrl.first_record = 1; - grs_extract_tree(&extractCtrl, n); + extract_set_store_data_prepare(&extractCtrl); + + if (n) + grs_extract_tree(&extractCtrl, n); if (rec->size[recInfo_delKeys]) { @@ -1316,12 +1245,7 @@ int explain_extract (void *handle, Record rec, data1_node *n) void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, int cmd, struct recKeys *reckeys) { -#if SU_SCHEME -#else - unsigned char attrSet = (unsigned char) -1; - unsigned short attrUse = (unsigned short) -1; -#endif - int seqno = 0; + void *decode_handle = iscz1_start(); int off = 0; int ch = 0; ZebraExplainInfo zei = zh->reg->zei; @@ -1331,7 +1255,7 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8")); if (mem <= 0) { - logf(LOG_WARN, "Invalid memory setting, using default 8 MB"); + yaz_log(YLOG_WARN, "Invalid memory setting, using default 8 MB"); mem= 1024*1024*8; } /* FIXME: That "8" should be in a default settings include */ @@ -1343,97 +1267,120 @@ void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno, zh->reg->key_file_no = 0; } zebraExplain_recordCountIncrement (zei, cmd ? 1 : -1); + while (off < reckeys->buf_used) { const char *src = reckeys->buf + off; struct it_key key; - int lead; - - lead = *src++; + char *dst = (char*) &key; + + iscz1_decode(decode_handle, &dst, &src); + assert(key.len == 4); -#if SU_SCHEME - if ((lead & 3) < 3) - { - memcpy (&ch, src, sizeof(ch)); - src += sizeof(ch); - } -#else - if (!(lead & 1)) - { - memcpy (&attrSet, src, sizeof(attrSet)); - src += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (&attrUse, src, sizeof(attrUse)); - src += sizeof(attrUse); - } -#endif if (zh->reg->key_buf_used + 1024 > (zh->reg->ptr_top -zh->reg->ptr_i)*sizeof(char*)) - extract_flushWriteKeys (zh); + extract_flushWriteKeys (zh,0); ++(zh->reg->ptr_i); + assert(zh->reg->ptr_i > 0); (zh->reg->key_buf)[zh->reg->ptr_top - zh->reg->ptr_i] = (char*)zh->reg->key_buf + zh->reg->key_buf_used; -#if SU_SCHEME -#else - ch = zebraExplain_lookupSU (zei, attrSet, attrUse); - if (ch < 0) - ch = zebraExplain_addSU (zei, attrSet, attrUse); -#endif - assert (ch > 0); + + ch = (int) key.mem[0]; /* ordinal for field/use/attribute */ + zh->reg->key_buf_used += key_SU_encode (ch,((char*)zh->reg->key_buf) + zh->reg->key_buf_used); - while (*src) ((char*)zh->reg->key_buf) [(zh->reg->key_buf_used)++] = *src++; src++; ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = '\0'; ((char*)(zh->reg->key_buf))[(zh->reg->key_buf_used)++] = cmd; - if (lead & 60) - seqno += ((lead>>2) & 15)-1; - else - { - memcpy (&seqno, src, sizeof(seqno)); - src += sizeof(seqno); - } - key.seqno = seqno; - key.sysno = sysno; - memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used, &key, sizeof(key)); + key.len = 3; + if (key.mem[1]) /* filter specified record ID */ + key.mem[0] = key.mem[1]; + else + key.mem[0] = sysno; + key.mem[1] = key.mem[2]; /* section_id */ + key.mem[2] = key.mem[3]; /* sequence .. */ + + memcpy ((char*)zh->reg->key_buf + zh->reg->key_buf_used, + &key, sizeof(key)); (zh->reg->key_buf_used) += sizeof(key); off = src - reckeys->buf; } assert (off == reckeys->buf_used); + iscz1_stop(decode_handle); } -void extract_flushWriteKeys (ZebraHandle zh) +void extract_flushWriteKeys (ZebraHandle zh, int final) + /* optimizing: if final=1, and no files written yet */ + /* push the keys directly to merge, sidestepping the */ + /* temp file altogether. Speeds small updates */ { FILE *outf; char out_fname[200]; char *prevcp, *cp; struct encode_info encode_info; int ptr_i = zh->reg->ptr_i; + int temp_policy; #if SORT_EXTRA int i; #endif if (!zh->reg->key_buf || ptr_i <= 0) + { + yaz_log (YLOG_DEBUG, " 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)++; - logf (LOG_LOG, "sorting section %d", (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", + 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, sizeof(char*), key_qsort_compare); + + /* zebra.cfg: tempfiles: + Y: always use temp files (old way) + A: use temp files, if more than one (auto) + = if this is both the last and the first + N: never bother with temp files (new) */ + + temp_policy=toupper(res_get_def(zh->res,"tempfiles","auto")[0]); + if (temp_policy != 'Y' && temp_policy != 'N' && temp_policy != 'A') { + yaz_log (YLOG_WARN, "Illegal tempfiles setting '%c'. using 'Auto' ", + temp_policy); + temp_policy='A'; + } + + if ( ( temp_policy =='N' ) || /* always from memory */ + ( ( temp_policy =='A' ) && /* automatic */ + (zh->reg->key_file_no == 1) && /* this is first time */ + (final) ) ) /* and last (=only) time */ + { /* go directly from memory */ + zh->reg->key_file_no =0; /* signal not to read files */ + zebra_index_merge(zh); + zh->reg->ptr_i = 0; + zh->reg->key_buf_used = 0; + return; + } + + /* Not doing directly from memory, write into a temp file */ extract_get_fname_tmp (zh, out_fname, zh->reg->key_file_no); if (!(outf = fopen (out_fname, "wb"))) { - logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname); exit (1); } - logf (LOG_LOG, "writing section %d", zh->reg->key_file_no); + yaz_log (YLOG_LOG, "writing section %d", zh->reg->key_file_no); prevcp = cp = (zh->reg->key_buf)[zh->reg->ptr_top - ptr_i]; encode_key_init (&encode_info); @@ -1459,10 +1406,10 @@ void extract_flushWriteKeys (ZebraHandle zh) if (!(outf = fopen (out_fname, "wb"))) { - logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fopen %s", out_fname); exit (1); } - logf (LOG_LOG, "writing section %d", key_file_no); + yaz_log (YLOG_LOG, "writing section %d", key_file_no); i = ptr_i; prevcp = key_buf[ptr_top-i]; while (1) @@ -1470,7 +1417,7 @@ void extract_flushWriteKeys (ZebraHandle zh) { key_y_len = strlen(prevcp)+1; #if 0 - logf (LOG_LOG, "key_y_len: %2d %02x %02x %s", + yaz_log (YLOG_LOG, "key_y_len: %2d %02x %02x %s", key_y_len, prevcp[0], prevcp[1], 2+prevcp); #endif qsort (key_buf + ptr_top-ptr_i, ptr_i - i, @@ -1492,32 +1439,25 @@ void extract_flushWriteKeys (ZebraHandle zh) #endif if (fclose (outf)) { - logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname); + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fclose %s", out_fname); exit (1); } - logf (LOG_LOG, "finished section %d", zh->reg->key_file_no); + yaz_log (YLOG_LOG, "finished section %d", zh->reg->key_file_no); zh->reg->ptr_i = 0; zh->reg->key_buf_used = 0; } -void extract_add_index_string (RecWord *p, const char *string, - int length) +void extract_add_it_key (ZebraHandle zh, + int reg_type, + const char *str, int slen, struct it_key *key) { char *dst; - unsigned char attrSet; - unsigned short attrUse; - int lead = 0; - int diff = 0; - int *pseqno = &p->seqno; - ZebraHandle zh = p->extractCtrl->handle; - ZebraExplainInfo zei = zh->reg->zei; struct recKeys *keys = &zh->reg->keys; + const char *src = (char*) key; if (keys->buf_used+1024 > keys->buf_max) { - char *b; - - b = (char *) xmalloc (keys->buf_max += 128000); + char *b = (char *) xmalloc (keys->buf_max += 128000); if (keys->buf_used > 0) memcpy (b, keys->buf, keys->buf_used); xfree (keys->buf); @@ -1525,67 +1465,128 @@ void extract_add_index_string (RecWord *p, const char *string, } dst = keys->buf + keys->buf_used; - attrSet = p->attrSet; - if (keys->buf_used > 0 && keys->prevAttrSet == attrSet) - lead |= 1; - else - keys->prevAttrSet = attrSet; - attrUse = p->attrUse; - if (keys->buf_used > 0 && keys->prevAttrUse == attrUse) - lead |= 2; - else - keys->prevAttrUse = attrUse; -#if 1 - diff = 1 + *pseqno - keys->prevSeqNo; - if (diff >= 1 && diff <= 15) - lead |= (diff << 2); - else - diff = 0; -#endif - keys->prevSeqNo = *pseqno; - - *dst++ = lead; + iscz1_encode(keys->codec_handle, &dst, &src); -#if SU_SCHEME - if ((lead & 3) < 3) + *dst++ = reg_type; + memcpy (dst, str, slen); + dst += slen; + *dst++ = '\0'; + keys->buf_used = dst - keys->buf; +} + +ZEBRA_RES zebra_snippets_rec_keys(ZebraHandle zh, struct recKeys *reckeys, + zebra_snippets *snippets) +{ + void *decode_handle = iscz1_start(); + int off = 0; + int seqno = 0; + NMEM nmem = nmem_create(); + + yaz_log(YLOG_LOG, "zebra_rec_keys_snippets buf=%p sz=%d", reckeys->buf, + reckeys->buf_used); + assert(reckeys->buf); + while (off < reckeys->buf_used) { - int ch = zebraExplain_lookupSU (zei, attrSet, attrUse); - if (ch < 0) - { - ch = zebraExplain_addSU (zei, attrSet, attrUse); - yaz_log (LOG_DEBUG, "addSU set=%d use=%d SU=%d", - attrSet, attrUse, ch); - } - assert (ch > 0); - memcpy (dst, &ch, sizeof(ch)); - dst += sizeof(ch); + const char *src = reckeys->buf + off; + struct it_key key; + char *dst = (char*) &key; + char dst_buf[IT_MAX_WORD]; + char *dst_term = dst_buf; + + iscz1_decode(decode_handle, &dst, &src); + assert(key.len <= 4 && key.len > 2); + + seqno = (int) key.mem[key.len-1]; + + zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1); + zebra_snippets_append(snippets, seqno, src[0], key.mem[0], dst_term); + while (*src++) + ; + off = src - reckeys->buf; + nmem_reset(nmem); } -#else - if (!(lead & 1)) + nmem_destroy(nmem); + iscz1_stop(decode_handle); + return ZEBRA_OK; +} + +void print_rec_keys(ZebraHandle zh, struct recKeys *reckeys) +{ + void *decode_handle = iscz1_start(); + int off = 0; + int seqno = 0; + NMEM nmem = nmem_create(); + + yaz_log(YLOG_LOG, "print_rec_keys buf=%p sz=%d", reckeys->buf, + reckeys->buf_used); + assert(reckeys->buf); + while (off < reckeys->buf_used) { - memcpy (dst, &attrSet, sizeof(attrSet)); - dst += sizeof(attrSet); + const char *src = reckeys->buf + off; + struct it_key key; + char *dst = (char*) &key; + int attrSet, attrUse; + char dst_buf[IT_MAX_WORD]; + char *dst_term = dst_buf; + + iscz1_decode(decode_handle, &dst, &src); + assert(key.len <= 4 && key.len > 2); + + attrSet = (int) key.mem[0] >> 16; + attrUse = (int) key.mem[0] & 65535; + seqno = (int) key.mem[key.len-1]; + + zebra_term_untrans_iconv(zh, nmem, src[0], &dst_term, src+1); + + yaz_log(YLOG_LOG, "ord=" ZINT_FORMAT " seqno=%d term=%s", + key.mem[0], seqno, dst_term); + while (*src++) + ; + off = src - reckeys->buf; + nmem_reset(nmem); } - if (!(lead & 2)) + nmem_destroy(nmem); + iscz1_stop(decode_handle); +} + +void extract_add_index_string (RecWord *p, const char *str, int length) +{ + struct it_key key; + + ZebraHandle zh = p->extractCtrl->handle; + ZebraExplainInfo zei = zh->reg->zei; + int ch; + + if (p->attrStr) { - memcpy (dst, &attrUse, sizeof(attrUse)); - dst += sizeof(attrUse); + ch = zebraExplain_lookup_attr_str(zei, p->attrStr); + if (ch < 0) + ch = zebraExplain_add_attr_str(zei, p->attrStr); } -#endif - *dst++ = p->reg_type; - memcpy (dst, string, length); - dst += length; - *dst++ = '\0'; - - if (!diff) + else { - memcpy (dst, pseqno, sizeof(*pseqno)); - dst += sizeof(*pseqno); + ch = zebraExplain_lookup_attr_su(zei, p->attrSet, p->attrUse); + if (ch < 0) + ch = zebraExplain_add_attr_su(zei, p->attrSet, p->attrUse); } - keys->buf_used = dst - keys->buf; + key.len = 4; + key.mem[0] = ch; + key.mem[1] = p->record_id; + key.mem[2] = p->section_id; + key.mem[3] = p->seqno; + +#if 0 + /* just for debugging .. */ + yaz_log(YLOG_LOG, "add: set=%d use=%d " + "record_id=%lld section_id=%lld seqno=%lld", + p->attrSet, p->attrUse, p->record_id, p->section_id, p->seqno); +#endif + + extract_add_it_key(p->extractCtrl->handle, p->reg_type, str, + length, &key); } -static void extract_add_sort_string (RecWord *p, const char *string, +static void extract_add_sort_string (RecWord *p, const char *str, int length) { ZebraHandle zh = p->extractCtrl->handle; @@ -1618,7 +1619,7 @@ static void extract_add_sort_string (RecWord *p, const char *string, 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, string, length); + memcpy (sk->buf + off, str, length); sk->buf_used = off + length; } @@ -1633,12 +1634,14 @@ void extract_add_string (RecWord *p, const char *string, int length) static void extract_add_incomplete_field (RecWord *p) { - const char *b = p->string; - int remain = p->length; + const char *b = p->term_buf; + 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->reg_type, &b, remain); + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0); while (map) { @@ -1648,9 +1651,9 @@ static void extract_add_incomplete_field (RecWord *p) /* Skip spaces */ while (map && *map && **map == *CHR_SPACE) { - remain = p->length - (b - p->string); + remain = p->term_len - (b - p->term_buf); if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0); else map = 0; } @@ -1663,9 +1666,9 @@ static void extract_add_incomplete_field (RecWord *p) while (i < IT_MAX_WORD && *cp) buf[i++] = *(cp++); - remain = p->length - (b - p->string); + remain = p->term_len - (b - p->term_buf); if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, 0); else map = 0; } @@ -1678,21 +1681,28 @@ static void extract_add_incomplete_field (RecWord *p) static void extract_add_complete_field (RecWord *p) { - const char *b = p->string; + const char *b = p->term_buf; char buf[IT_MAX_WORD+1]; const char **map = 0; - int i = 0, remain = p->length; + 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->reg_type, &b, remain); + map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain, 1); while (remain > 0 && i < IT_MAX_WORD) { while (map && *map && **map == *CHR_SPACE) { - remain = p->length - (b - p->string); + remain = p->term_len - (b - p->term_buf); + if (remain > 0) - map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + { + int first = i ? 0 : 1; /* first position */ + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain, first); + } else map = 0; } @@ -1705,14 +1715,24 @@ static void extract_add_complete_field (RecWord *p) { const char *cp = *map; - if (i >= IT_MAX_WORD) - break; - while (i < IT_MAX_WORD && *cp) - buf[i++] = *(cp++); - remain = p->length - (b - p->string); + if (**map == *CHR_CUT) + { + i = 0; + } + else + { + 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++); + } + remain = p->term_len - (b - p->term_buf); if (remain > 0) + { map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, - remain); + remain, 0); + } else map = 0; } @@ -1725,17 +1745,17 @@ static void extract_add_complete_field (RecWord *p) void extract_token_add (RecWord *p) { WRBUF wrbuf; - #if 0 - yaz_log (LOG_LOG, "reg_type=%c attrSet=%d attrUse=%d seqno=%d s=%.*s", + 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 ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0, - p->string, p->length))) + p->term_buf, p->term_len))) { - p->string = wrbuf_buf(wrbuf); - p->length = wrbuf_len(wrbuf); + p->term_buf = wrbuf_buf(wrbuf); + p->term_len = wrbuf_len(wrbuf); } if (zebra_maps_is_complete (p->zebra_maps, p->reg_type)) extract_add_complete_field (p); @@ -1743,9 +1763,34 @@ void extract_token_add (RecWord *p) extract_add_incomplete_field(p); } +static void extract_set_store_data_cb(struct recExtractCtrl *p, + void *buf, size_t sz) +{ + ZebraHandle zh = (ZebraHandle) p->handle; + + xfree(zh->store_data_buf); + zh->store_data_buf = 0; + zh->store_data_size = 0; + if (buf && sz) + { + zh->store_data_buf = xmalloc(sz); + zh->store_data_size = sz; + memcpy(zh->store_data_buf, buf, sz); + } +} + +static void extract_set_store_data_prepare(struct recExtractCtrl *p) +{ + ZebraHandle zh = (ZebraHandle) p->handle; + xfree(zh->store_data_buf); + zh->store_data_buf = 0; + zh->store_data_size = 0; + p->setStoreData = extract_set_store_data_cb; +} + void extract_schema_add (struct recExtractCtrl *p, Odr_oid *oid) { - ZebraHandle zh = (ZebraHandle) (p->handle); + ZebraHandle zh = (ZebraHandle) p->handle; zebraExplain_addSchema (zh->reg->zei, oid); } @@ -1783,32 +1828,10 @@ void encode_key_init (struct encode_info *i) i->prevseq=0; i->prevcmd=-1; i->keylen=0; + i->encode_handle = iscz1_start(); } -char *encode_key_int (int d, char *bp) -{ - if (d <= 63) - *bp++ = d; - else if (d <= 16383) - { - *bp++ = 64 + (d>>8); - *bp++ = d & 255; - } - else if (d <= 4194303) - { - *bp++ = 128 + (d>>16); - *bp++ = (d>>8) & 255; - *bp++ = d & 255; - } - else - { - *bp++ = 192 + (d>>24); - *bp++ = (d>>16) & 255; - *bp++ = (d>>8) & 255; - *bp++ = d & 255; - } - return bp; -} +#define OLDENCODE 1 #ifdef OLDENCODE /* this is the old encode_key_write @@ -1818,31 +1841,28 @@ char *encode_key_int (int d, char *bp) void encode_key_write (char *k, struct encode_info *i, FILE *outf) { struct it_key key; - char *bp = i->buf; + char *bp = i->buf, *bp0; + const char *src = (char *) &key; + /* copy term to output buf */ while ((*bp++ = *k++)) ; - memcpy (&key, k+1, sizeof(struct it_key)); - bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp); - if (i->sysno != key.sysno) - { - i->sysno = key.sysno; - i->seqno = 0; - } - else if (!i->seqno && !key.seqno && i->cmd == *k) - return; - bp = encode_key_int (key.seqno - i->seqno, bp); - i->seqno = key.seqno; - i->cmd = *k; + /* and copy & align key so we can mangle */ + memcpy (&key, k+1, sizeof(struct it_key)); /* *k is insert/delete */ + + bp0 = bp++; + iscz1_encode(i->encode_handle, &bp, &src); + *bp0 = (*k * 128) + bp - bp0 - 1; /* length and insert/delete combined */ if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) { - logf (LOG_FATAL|LOG_ERRNO, "fwrite"); + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite"); exit (1); } } void encode_key_flush (struct encode_info *i, FILE *outf) { /* dummy routine */ + iscz1_stop(i->encode_handle); } #else @@ -1904,7 +1924,7 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf) i->cmd = i->prevcmd; if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) { - logf (LOG_FATAL|LOG_ERRNO, "fwrite"); + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite"); exit (1); } i->keylen=0; /* ok, it's written, forget it */ @@ -1938,7 +1958,7 @@ void encode_key_flush (struct encode_info *i, FILE *outf) i->cmd = i->prevcmd; if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) { - logf (LOG_FATAL|LOG_ERRNO, "fwrite"); + yaz_log (YLOG_FATAL|YLOG_ERRNO, "fwrite"); exit (1); } i->keylen=0; /* ok, it's written, forget it */