X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=269c21f417a35baaf11c9f6886eb43ff8d74fca3;hb=297ba5c5c265a5f869f43a0a211bf9f48f700add;hp=ebd04ac0c4bc7e590b79b2f0aa183473916fdfd5;hpb=3c5f6226f97612c0d6ac40591f600587c5ffa858;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index ebd04ac..269c21f 100644 --- a/index/extract.c +++ b/index/extract.c @@ -4,7 +4,48 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: extract.c,v $ - * Revision 1.90 1999-02-02 14:50:52 adam + * 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 @@ -343,6 +384,12 @@ #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 @@ -388,24 +435,32 @@ 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); @@ -426,6 +481,7 @@ int key_open (struct recordGroup *rGroup, int mem) struct encode_info { int sysno; int seqno; + int cmd; char buf[768]; }; @@ -433,6 +489,7 @@ 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) @@ -474,8 +531,11 @@ 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"); @@ -521,7 +581,7 @@ 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"))) { @@ -596,7 +656,7 @@ 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); @@ -614,6 +674,7 @@ 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 { @@ -646,7 +707,7 @@ 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); @@ -696,7 +757,8 @@ 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) @@ -707,11 +769,11 @@ 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); @@ -770,6 +832,7 @@ 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) @@ -820,6 +883,13 @@ 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 @@ -975,7 +1045,8 @@ struct file_read_info { 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; @@ -992,7 +1063,7 @@ 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; @@ -1001,13 +1072,13 @@ static off_t file_seek (void *handle, off_t offset) 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; if (p->sdrbuf) @@ -1031,7 +1102,7 @@ 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->sdrbuf && p->file_moffset) @@ -1041,7 +1112,7 @@ static void file_begin (void *handle) 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; @@ -1205,7 +1276,7 @@ struct recordLogInfo { static void recordLogPreamble (int level, const char *msg, void *info) { - struct recordLogInfo *p = info; + struct recordLogInfo *p = (struct recordLogInfo *) info; FILE *outf = log_file (); if (level & LOG_LOG) @@ -1223,7 +1294,7 @@ 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; @@ -1254,34 +1325,41 @@ 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; } @@ -1291,8 +1369,9 @@ 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; } } @@ -1329,14 +1408,14 @@ 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; @@ -1364,8 +1443,8 @@ 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; @@ -1379,15 +1458,15 @@ 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); @@ -1402,15 +1481,15 @@ 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++; } @@ -1463,11 +1542,12 @@ 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) @@ -1509,6 +1589,7 @@ 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)); @@ -1548,7 +1629,8 @@ int fileExtract (SYSNO *sysno, const char *fname, if (!*rGroup->recordType) return 0; if (!(recType = - recType_byName (rGroup->recTypes, rGroup->recordType, subType))) + recType_byName (rGroup->recTypes, rGroup->recordType, subType, + &clientData))) { logf (LOG_WARN, "No such record type: %s", rGroup->recordType); return 0; @@ -1585,7 +1667,7 @@ int fileExtract (SYSNO *sysno, const char *fname, { if (zebraExplain_newDatabase (zti, rGroup->databaseName, rGroup->explainDatabase)) - abort (); + return 0; } if (rGroup->flagStoreData == -1) @@ -1619,57 +1701,6 @@ int fileExtract (SYSNO *sysno, const char *fname, if (rGroup->flagStoreKeys == -1) rGroup->flagStoreKeys = 0; -#if ZEBRASDR - if (1) - { - ZebraSdrHandle h; - char xname[128], *xp; - - strncpy (xname, fname, 127); - if ((xp = strchr (xname, '.'))) - *xp = '\0'; - - h = zebraSdr_open (xname); - if (!h) - { - logf (LOG_WARN, "sdr open %s", xname); - return 0; - } - for (;;) - { - unsigned char *buf; - char sdr_name[128]; - int r, segmentno; - - segmentno = zebraSdr_segment (h, 0); - sprintf (sdr_name, "%%%s.%d", xname, segmentno); - logf (LOG_LOG, "SDR: %s", sdr_name); - -#if 1 - if (segmentno > 20) - break; -#endif - r = zebraSdr_read (h, &buf); - - if (!r) - break; - - fi = file_read_start (0); - fi->sdrbuf = buf; - fi->sdrmax = r; - do - { - file_begin (fi); - r = recordExtract (sysno, sdr_name, rGroup, deleteFlag, fi, - recType, subType); - } while (r && !sysno && fi->file_more); - file_read_stop (fi); - free (buf); - } - zebraSdr_close (h); - return 1; - } -#endif if (sysno && deleteFlag) fd = -1; else @@ -1685,7 +1716,7 @@ 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) @@ -1711,8 +1742,8 @@ 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;