X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=9b85afacf93b34bda8cf50e947412244fd9cddd9;hb=9b4f970d2816490fa4cc850fabc51f494ee5612b;hp=711e3606a9289cb4ae15307b6535a9ba0c6fa4e3;hpb=023b1af5942f75cf6d2b70ffd42df4a3c5d16dae;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index 711e360..9b85afa 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,10 +1,70 @@ /* - * Copyright (C) 1994-1998, Index Data + * Copyright (C) 1994-2000, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: extract.c,v $ - * Revision 1.87 1998-10-15 13:10:33 adam + * Revision 1.106 2000-12-05 12:22:53 adam + * Termlist source implemented (so that we can index values of XML/SGML + * attributes). + * + * Revision 1.105 2000/12/05 10:01:44 adam + * Fixed bug regarding user-defined attribute sets. + * + * Revision 1.104 2000/09/05 14:04:05 adam + * Updates for prefix 'yaz_' for YAZ log functions. + * + * Revision 1.103 2000/05/18 12:01:36 adam + * System call times(2) used again. More 64-bit fixes. + * + * Revision 1.102 2000/05/15 15:32:33 adam + * Added 64 bit file input. + * + * Revision 1.101 2000/05/15 13:02:39 adam + * Minor change. + * + * Revision 1.100 2000/03/20 19:08:36 adam + * Added remote record import using Z39.50 extended services and Segment + * Requests. + * + * Revision 1.99 2000/02/24 10:57:02 adam + * Sequence number incremented after each incomplete-field. + * + * Revision 1.98 1999/09/07 07:19:21 adam + * Work on character mapping. Implemented replace rules. + * + * Revision 1.97 1999/07/06 12:28:04 adam + * Updated record index structure. Format includes version ID. Compression + * algorithm ID is stored for each record block. + * + * Revision 1.96 1999/05/26 07:49:13 adam + * C++ compilation. + * + * Revision 1.95 1999/05/21 12:00:17 adam + * Better diagnostics for extraction process. + * + * Revision 1.94 1999/05/20 12:57:18 adam + * Implemented TCL filter. Updated recctrl system. + * + * Revision 1.93 1999/05/15 14:36:38 adam + * Updated dictionary. Implemented "compression" of dictionary. + * + * Revision 1.92 1999/03/09 16:27:49 adam + * More work on SDRKit integration. + * + * Revision 1.91 1999/02/12 13:29:22 adam + * Implemented position-flag for registers. + * + * Revision 1.90 1999/02/02 14:50:52 adam + * Updated WIN32 code specific sections. Changed header. + * + * Revision 1.89 1998/10/28 10:54:38 adam + * SDRKit integration. + * + * Revision 1.88 1998/10/16 08:14:29 adam + * Updated record control system. + * + * Revision 1.87 1998/10/15 13:10:33 adam * Fixed bug in Zebra that caused it to stop indexing when empty * record was read. * @@ -320,7 +380,7 @@ */ #include #include -#ifdef WINDOWS +#ifdef WIN32 #include #else #include @@ -334,6 +394,20 @@ #include "zinfo.h" +#if _FILE_OFFSET_BITS == 64 +#define PRINTF_OFF_T "%Ld" +#else +#define PRINTF_OFF_T "%ld" +#endif + +#ifndef ZEBRASDR +#define ZEBRASDR 0 +#endif + +#if ZEBRASDR +#include "zebrasdr.h" +#endif + static Dict matchDict; static Records records = NULL; @@ -352,6 +426,7 @@ static int records_processed = 0; static ZebraExplainInfo zti = NULL; + static void logRecord (int showFlag) { if (!showFlag) @@ -371,24 +446,32 @@ int key_open (struct recordGroup *rGroup, int mem) BFiles bfs = rGroup->bfs; int rw = rGroup->flagRw; data1_handle dh = rGroup->dh; + char *recordCompression; + int record_compression = REC_COMPRESS_NONE; if (!mem) - mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024; + mem = atoi(res_get_def (common_resource, "memMax", "16"))*1024*1024; if (mem < 50000) mem = 50000; - key_buf = xmalloc (mem); + key_buf = (char **) xmalloc (mem); ptr_top = mem/sizeof(char*); ptr_i = 0; key_buf_used = 0; key_file_no = 0; - if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw))) + if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw, 0))) { logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT); return -1; } assert (!records); - records = rec_open (bfs, rw); + recordCompression = res_get_def (common_resource, + "recordCompression", "none"); + if (!strcmp (recordCompression, "none")) + record_compression = REC_COMPRESS_NONE; + if (!strcmp (recordCompression, "bzip2")) + record_compression = REC_COMPRESS_BZIP2; + records = rec_open (bfs, rw, record_compression); if (!records) { dict_close (matchDict); @@ -409,6 +492,7 @@ int key_open (struct recordGroup *rGroup, int mem) struct encode_info { int sysno; int seqno; + int cmd; char buf[768]; }; @@ -416,6 +500,7 @@ void encode_key_init (struct encode_info *i) { i->sysno = 0; i->seqno = 0; + i->cmd = -1; } char *encode_key_int (int d, char *bp) @@ -457,8 +542,11 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf) 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; if (fwrite (i->buf, bp - i->buf, 1, outf) != 1) { logf (LOG_FATAL|LOG_ERRNO, "fwrite"); @@ -504,7 +592,7 @@ void key_flush (void) logf (LOG_LOG, "sorting section %d", key_file_no); #if !SORT_EXTRA qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare); - getFnameTmp (out_fname, key_file_no); + getFnameTmp (common_resource, out_fname, key_file_no); if (!(outf = fopen (out_fname, "wb"))) { @@ -579,7 +667,7 @@ int key_close (struct recordGroup *rGroup) int rw = rGroup->flagRw; if (rw) zebraExplain_runNumberIncrement (zti, 1); - zebraExplain_close (zti, rw, 0); + zebraExplain_close (zti, rw); key_flush (); xfree (key_buf); rec_close (&records); @@ -597,6 +685,7 @@ static void wordInit (struct recExtractCtrl *p, RecWord *w) w->attrSet = VAL_BIB1; w->attrUse = 1016; w->reg_type = 'w'; + w->extractCtrl = p; } static struct sortKey { @@ -629,7 +718,7 @@ static void addIndexString (RecWord *p, const char *string, int length) { char *b; - b = xmalloc (reckeys.buf_max += 128000); + b = (char *) xmalloc (reckeys.buf_max += 128000); if (reckeys.buf_used > 0) memcpy (b, reckeys.buf, reckeys.buf_used); xfree (reckeys.buf); @@ -658,6 +747,19 @@ static void addIndexString (RecWord *p, const char *string, int length) *dst++ = lead; +#if SU_SCHEME + if ((lead & 3) < 3) + { + int ch = zebraExplain_lookupSU (zti, attrSet, attrUse); + if (ch < 0) + { + ch = zebraExplain_addSU (zti, attrSet, attrUse); + } + assert (ch > 0); + memcpy (dst, &ch, sizeof(ch)); + dst += sizeof(ch); + } +#else if (!(lead & 1)) { memcpy (dst, &attrSet, sizeof(attrSet)); @@ -668,6 +770,7 @@ static void addIndexString (RecWord *p, const char *string, int length) memcpy (dst, &attrUse, sizeof(attrUse)); dst += sizeof(attrUse); } +#endif *dst++ = p->reg_type; memcpy (dst, string, length); dst += length; @@ -679,7 +782,8 @@ static void addIndexString (RecWord *p, const char *string, int length) dst += sizeof(*pseqno); } reckeys.buf_used = dst - reckeys.buf; - (*pseqno)++; + if (*pseqno) + (*pseqno)++; } static void addSortString (RecWord *p, const char *string, int length) @@ -690,11 +794,11 @@ static void addSortString (RecWord *p, const char *string, int length) if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse) return; - sk = xmalloc (sizeof(*sk)); + sk = (struct sortKey *) xmalloc (sizeof(*sk)); sk->next = sortKeys; sortKeys = sk; - sk->string = xmalloc (length); + sk->string = (char *) xmalloc (length); sk->length = length; memcpy (sk->string, string, length); @@ -753,6 +857,7 @@ static void addIncompleteField (RecWord *p) return; addString (p, buf, i); } + (p->seqnos[p->reg_type])++; /* to separate this from next one */ } static void addCompleteField (RecWord *p) @@ -803,6 +908,13 @@ static void addCompleteField (RecWord *p) static void addRecordKey (RecWord *p) { + WRBUF wrbuf; + if ((wrbuf = zebra_replace(p->zebra_maps, p->reg_type, 0, + p->string, p->length))) + { + p->string = wrbuf_buf(wrbuf); + p->length = wrbuf_len(wrbuf); + } if (zebra_maps_is_complete (p->zebra_maps, p->reg_type)) addCompleteField (p); else @@ -828,20 +940,31 @@ static void flushSortKeys (SYSNO sysno, int cmd) static void flushRecordKeys (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; int off = 0; + int ch = 0; zebraExplain_recordCountIncrement (zti, cmd ? 1 : -1); while (off < reckeys->buf_used) { const char *src = reckeys->buf + off; struct it_key key; - int lead, ch; + 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)); @@ -852,14 +975,23 @@ static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys) memcpy (&attrUse, src, sizeof(attrUse)); src += sizeof(attrUse); } +#endif if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*)) key_flush (); ++ptr_i; + key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used; +#if SU_SCHEME +#else ch = zebraExplain_lookupSU (zti, attrSet, attrUse); if (ch < 0) + { ch = zebraExplain_addSU (zti, attrSet, attrUse); + yaz_log (LOG_LOG, "addSU cmd=%d set=%d use=%d SU=%d", + cmd, attrSet, attrUse, ch); + } +#endif assert (ch > 0); key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used); @@ -892,13 +1024,22 @@ static const char **searchRecordKey (struct recKeys *reckeys, int off = 0; int startSeq = -1; int i; + int seqno = 0; +#if SU_SCHEME + int chS, ch; +#else short attrUse; char attrSet; - int seqno = 0; +#endif for (i = 0; i<32; i++) ws[i] = NULL; - + +#if SU_SCHEME + chS = zebraExplain_lookupSU (zti, attrSetS, attrUseS); + if (chS < 0) + return ws; +#endif while (off < reckeys->buf_used) { @@ -907,7 +1048,13 @@ static const char **searchRecordKey (struct recKeys *reckeys, 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)); @@ -918,6 +1065,7 @@ static const char **searchRecordKey (struct recKeys *reckeys, memcpy (&attrUse, src, sizeof(attrUse)); src += sizeof(attrUse); } +#endif wstart = src; while (*src++) ; @@ -928,7 +1076,13 @@ static const char **searchRecordKey (struct recKeys *reckeys, memcpy (&seqno, src, sizeof(seqno)); src += sizeof(seqno); } - if (attrUseS == attrUse && attrSetS == attrSet) + if ( +#if SU_SCHEME + ch == chS +#else + attrUseS == attrUse && attrSetS == attrSet +#endif + ) { int woff; @@ -947,20 +1101,25 @@ static const char **searchRecordKey (struct recKeys *reckeys, } struct file_read_info { - off_t file_max; - off_t file_offset; - off_t file_moffset; + off_t file_max; /* maximum offset so far */ + off_t file_offset; /* current offset */ + 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) { - struct file_read_info *fi = xmalloc (sizeof(*fi)); + struct file_read_info *fi = (struct file_read_info *) + xmalloc (sizeof(*fi)); fi->fd = fd; fi->file_max = 0; fi->file_moffset = 0; + fi->sdrbuf = 0; + fi->sdrmax = 0; return fi; } @@ -971,23 +1130,34 @@ static void file_read_stop (struct file_read_info *fi) static off_t file_seek (void *handle, off_t offset) { - struct file_read_info *p = handle; + 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); } static off_t file_tell (void *handle) { - struct file_read_info *p = handle; + struct file_read_info *p = (struct file_read_info *) handle; return p->file_offset; } static int file_read (void *handle, char *buf, size_t count) { - struct file_read_info *p = handle; + struct file_read_info *p = (struct file_read_info *) handle; int fd = p->fd; int r; - r = read (fd, buf, count); + 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); if (r > 0) { p->file_offset += r; @@ -999,17 +1169,17 @@ static int file_read (void *handle, char *buf, size_t count) static void file_begin (void *handle) { - struct file_read_info *p = handle; + struct file_read_info *p = (struct file_read_info *) handle; p->file_offset = p->file_moffset; - if (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 = handle; + struct file_read_info *p = (struct file_read_info *) handle; assert (p->file_more == 0); p->file_more = 1; @@ -1173,8 +1343,8 @@ struct recordLogInfo { static void recordLogPreamble (int level, const char *msg, void *info) { - struct recordLogInfo *p = info; - FILE *outf = log_file (); + struct recordLogInfo *p = (struct recordLogInfo *) info; + FILE *outf = yaz_log_file (); if (level & LOG_LOG) return ; @@ -1191,7 +1361,7 @@ void addSchema (struct recExtractCtrl *p, Odr_oid *oid) static int recordExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup, int deleteFlag, struct file_read_info *fi, - RecType recType, char *subType) + RecType recType, char *subType, void *clientData) { RecordAttr *recordAttr; int r; @@ -1222,34 +1392,41 @@ static int recordExtract (SYSNO *sysno, const char *fname, extractCtrl.fh = fi; extractCtrl.subType = subType; extractCtrl.init = wordInit; - extractCtrl.addWord = addRecordKey; - extractCtrl.addSchema = addSchema; + extractCtrl.tokenAdd = addRecordKey; + extractCtrl.schemaAdd = addSchema; extractCtrl.dh = rGroup->dh; for (i = 0; i<256; i++) - extractCtrl.seqno[i] = 0; + { + if (zebra_maps_is_positioned(rGroup->zebra_maps, i)) + extractCtrl.seqno[i] = 1; + else + extractCtrl.seqno[i] = 0; + } extractCtrl.zebra_maps = rGroup->zebra_maps; extractCtrl.flagShowRecords = !rGroup->flagRw; if (!rGroup->flagRw) - printf ("File: %s %ld\n", fname, (long) recordOffset); + printf ("File: %s " PRINTF_OFF_T "\n", fname, recordOffset); logInfo.fname = fname; logInfo.recordOffset = recordOffset; logInfo.rGroup = rGroup; log_event_start (recordLogPreamble, &logInfo); - r = (*recType->extract)(&extractCtrl); + r = (*recType->extract)(clientData, &extractCtrl); log_event_start (NULL, NULL); - if (r) - { + if (r == RECCTRL_EXTRACT_EOF) + return 0; + else if (r == RECCTRL_EXTRACT_ERROR) + { /* error occured during extraction ... */ if (rGroup->flagRw && records_processed < rGroup->fileVerboseLimit) { - logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType, - fname, (long) recordOffset, r); + logf (LOG_WARN, "fail %s %s " PRINTF_OFF_T, rGroup->recordType, + fname, recordOffset); } return 0; } @@ -1259,8 +1436,9 @@ static int recordExtract (SYSNO *sysno, const char *fname, is probably empty - unless flagShowRecords is in use */ if (!rGroup->flagRw) return 1; - logf (LOG_WARN, "No keys generated for file %s", fname); - logf (LOG_WARN, " The file is probably empty"); + + logf (LOG_WARN, "empty %s %s " PRINTF_OFF_T, rGroup->recordType, + fname, recordOffset); return 1; } } @@ -1297,14 +1475,14 @@ static int recordExtract (SYSNO *sysno, const char *fname, /* new record */ if (deleteFlag) { - logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, rGroup->recordType, + fname, recordOffset); logf (LOG_WARN, "cannot delete record above (seems new)"); return 1; } if (records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "add %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + logf (LOG_LOG, "add %s %s " PRINTF_OFF_T, rGroup->recordType, + fname, recordOffset); rec = rec_new (records); *sysno = rec->sysno; @@ -1332,8 +1510,8 @@ static int recordExtract (SYSNO *sysno, const char *fname, if (recordAttr->runNumber == zebraExplain_runNumberIncrement (zti, 0)) { - logf (LOG_LOG, "skipped %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + logf (LOG_LOG, "skipped %s %s " PRINTF_OFF_T, rGroup->recordType, + fname, recordOffset); rec_rm (&rec); logRecord (0); return 1; @@ -1347,15 +1525,15 @@ static int recordExtract (SYSNO *sysno, const char *fname, /* record going to be deleted */ if (!delkeys.buf_used) { - logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, + rGroup->recordType, fname, recordOffset); logf (LOG_WARN, "cannot delete file above, storeKeys false"); } else { if (records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + logf (LOG_LOG, "delete %s %s " PRINTF_OFF_T, + rGroup->recordType, fname, recordOffset); records_deleted++; if (matchStr) dict_delete (matchDict, matchStr); @@ -1370,15 +1548,15 @@ static int recordExtract (SYSNO *sysno, const char *fname, /* record going to be updated */ if (!delkeys.buf_used) { - logf (LOG_LOG, "update %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + logf (LOG_LOG, "update %s %s " PRINTF_OFF_T, + rGroup->recordType, fname, recordOffset); logf (LOG_WARN, "cannot update file above, storeKeys false"); } else { if (records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "update %s %s %ld", rGroup->recordType, - fname, (long) recordOffset); + logf (LOG_LOG, "update %s %s " PRINTF_OFF_T, + rGroup->recordType, fname, recordOffset); flushRecordKeys (*sysno, 1, &reckeys); records_updated++; } @@ -1431,11 +1609,12 @@ static int recordExtract (SYSNO *sysno, const char *fname, if (rGroup->flagStoreData == 1) { rec->size[recInfo_storeData] = recordAttr->recordSize; - rec->info[recInfo_storeData] = xmalloc (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 %ld in %s", - (long) recordOffset, fname); + logf (LOG_ERRNO|LOG_FATAL, "seek to " PRINTF_OFF_T " in %s", + recordOffset, fname); exit (1); } if (read (fi->fd, rec->info[recInfo_storeData], recordAttr->recordSize) @@ -1477,6 +1656,7 @@ int fileExtract (SYSNO *sysno, const char *fname, struct recordGroup rGroupM; struct recordGroup *rGroup = &rGroupM; struct file_read_info *fi; + void *clientData; memcpy (rGroup, rGroupP, sizeof(*rGroupP)); @@ -1515,7 +1695,9 @@ int fileExtract (SYSNO *sysno, const char *fname, } if (!*rGroup->recordType) return 0; - if (!(recType = recType_byName (rGroup->recordType, subType))) + if (!(recType = + recType_byName (rGroup->recTypes, rGroup->recordType, subType, + &clientData))) { logf (LOG_WARN, "No such record type: %s", rGroup->recordType); return 0; @@ -1552,7 +1734,7 @@ int fileExtract (SYSNO *sysno, const char *fname, { if (zebraExplain_newDatabase (zti, rGroup->databaseName, rGroup->explainDatabase)) - abort (); + return 0; } if (rGroup->flagStoreData == -1) @@ -1601,7 +1783,7 @@ int fileExtract (SYSNO *sysno, const char *fname, { file_begin (fi); r = recordExtract (sysno, fname, rGroup, deleteFlag, fi, - recType, subType); + recType, subType, clientData); } while (r && !sysno && fi->file_more); file_read_stop (fi); if (fd != -1) @@ -1627,8 +1809,8 @@ static int explain_extract (void *handle, Record rec, data1_node *n) reckeys.prevSeqNo = 0; extractCtrl.init = wordInit; - extractCtrl.addWord = addRecordKey; - extractCtrl.addSchema = addSchema; + extractCtrl.tokenAdd = addRecordKey; + extractCtrl.schemaAdd = addSchema; extractCtrl.dh = rGroup->dh; for (i = 0; i<256; i++) extractCtrl.seqno[i] = 0;