X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=3a9c4b11ff9fc3218278298771e57e79b194392b;hb=28f3461a79a4569d736f4ffc66c6ae207e38c2ab;hp=f3bceaee76cdd1a2970137d12ba96987bd725a06;hpb=85a2a0b28cb516d28ac70b7824f2b7d4b07e56ae;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index f3bceae..3a9c4b1 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,10 +1,65 @@ /* - * Copyright (C) 1994-1998, Index Data I/S + * Copyright (C) 1994-1999, Index Data * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: extract.c,v $ - * Revision 1.82 1998-05-20 10:12:15 adam + * 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. + * + * Revision 1.86 1998/10/13 20:33:53 adam + * Fixed one log message and change use ordinal to be an unsigned char. + * + * Revision 1.85 1998/09/22 10:03:41 adam + * Changed result sets to be persistent in the sense that they can + * be re-searched if needed. + * Fixed memory leak in rsm_or. + * + * Revision 1.84 1998/06/11 15:42:22 adam + * Changed the way use attributes are specified in the recordId + * specification. + * + * Revision 1.83 1998/06/08 14:43:10 adam + * Added suport for EXPLAIN Proxy servers - added settings databasePath + * and explainDatabase to facilitate this. Increased maximum number + * of databases and attributes in one register. + * + * Revision 1.82 1998/05/20 10:12:15 adam * Implemented automatic EXPLAIN database maintenance. * Modified Zebra to work with ASN.1 compiled version of YAZ. * @@ -299,7 +354,7 @@ */ #include #include -#ifdef WINDOWS +#ifdef WIN32 #include #else #include @@ -313,6 +368,14 @@ #include "zinfo.h" +#ifndef ZEBRASDR +#define ZEBRASDR 0 +#endif + +#if ZEBRASDR +#include "zebrasdr.h" +#endif + static Dict matchDict; static Records records = NULL; @@ -350,24 +413,32 @@ 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; 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); @@ -388,6 +459,7 @@ int key_open (struct recordGroup *rGroup, int mem) struct encode_info { int sysno; int seqno; + int cmd; char buf[768]; }; @@ -395,6 +467,7 @@ 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) @@ -436,8 +509,11 @@ 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"); @@ -576,6 +652,7 @@ 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 { @@ -598,8 +675,8 @@ static struct recKeys { static void addIndexString (RecWord *p, const char *string, int length) { char *dst; - char attrSet; - short attrUse; + unsigned char attrSet; + unsigned short attrUse; int lead = 0; int diff = 0; int *pseqno = &p->seqnos[p->reg_type]; @@ -608,7 +685,7 @@ 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); @@ -658,7 +735,8 @@ 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) @@ -669,13 +747,14 @@ 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 (p->length); - sk->length = p->length; - memcpy (sk->string, p->string, p->length); + sk->string = (char *) xmalloc (length); + sk->length = length; + memcpy (sk->string, string, length); + sk->attrSet = p->attrSet; sk->attrUse = p->attrUse; } @@ -781,6 +860,13 @@ 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 @@ -806,8 +892,8 @@ static void flushSortKeys (SYSNO sysno, int cmd) static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys) { - char attrSet = -1; - short attrUse = -1; + unsigned char attrSet = (unsigned char) -1; + unsigned short attrUse = (unsigned short) -1; int seqno = 0; int off = 0; @@ -839,7 +925,8 @@ static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys) if (ch < 0) ch = zebraExplain_addSU (zti, attrSet, attrUse); assert (ch > 0); - ((char*) key_buf) [key_buf_used++] = ch; + key_buf_used += key_SU_code (ch, ((char*)key_buf) + key_buf_used); + while (*src) ((char*)key_buf) [key_buf_used++] = *src++; src++; @@ -924,20 +1011,25 @@ 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; } @@ -948,23 +1040,34 @@ 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; @@ -976,43 +1079,30 @@ 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; p->file_moffset = offset; } -static int atois (const char **s) -{ - int val = 0, c; - while ( (c=**s) >= '0' && c <= '9') - { - val = val*10 + c - '0'; - ++(*s); - } - return val; -} - static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, - const char *fname, - const char *spec) + const char *fname, const char *spec) { static char dstBuf[2048]; char *dst = dstBuf; const char *s = spec; static const char **w; - int i; while (1) { @@ -1022,21 +1112,39 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, break; if (*s == '(') { + char attset_str[64], attname_str[64]; + data1_attset *attset; + int i; char matchFlag[32]; - int attrSet, attrUse; + int attSet = 1, attUse = 1; int first = 1; s++; - attrSet = atois (&s); - if (*s != ',') - { - logf (LOG_WARN, "Missing , in match criteria %s in group %s", - spec, rGroup->groupName ? rGroup->groupName : "none"); - return NULL; - } - s++; - attrUse = atois (&s); - w = searchRecordKey (reckeys, attrSet, attrUse); + for (i = 0; *s && *s != ',' && *s != ')'; s++) + if (i < 63) + attset_str[i++] = *s; + attset_str[i] = '\0'; + + if (*s == ',') + { + s++; + for (i = 0; *s && *s != ')'; s++) + if (i < 63) + attname_str[i++] = *s; + attname_str[i] = '\0'; + } + + if ((attset = data1_get_attset (rGroup->dh, attset_str))) + { + data1_att *att; + attSet = attset->reference; + att = data1_getattbyname(rGroup->dh, attset, attname_str); + if (att) + attUse = att->value; + else + attUse = atoi (attname_str); + } + w = searchRecordKey (reckeys, attSet, attUse); assert (w); if (*s == ')') @@ -1066,7 +1174,7 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, if (first) { logf (LOG_WARN, "Record didn't contain match" - " fields in (%d,%d)", attrSet, attrUse); + " fields in (%s,%s)", attset_str, attname_str); return NULL; } } @@ -1145,7 +1253,7 @@ 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) @@ -1163,7 +1271,7 @@ 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; @@ -1194,11 +1302,16 @@ 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; @@ -1210,18 +1323,20 @@ static int recordExtract (SYSNO *sysno, const char *fname, 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 %ld", rGroup->recordType, + fname, (long) recordOffset); } return 0; } @@ -1233,7 +1348,7 @@ static int recordExtract (SYSNO *sysno, const char *fname, return 1; logf (LOG_WARN, "No keys generated for file %s", fname); logf (LOG_WARN, " The file is probably empty"); - return 0; + return 1; } } @@ -1403,11 +1518,12 @@ 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", fname, - (long) recordOffset); + 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) @@ -1449,6 +1565,7 @@ 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)); @@ -1460,12 +1577,10 @@ int fileExtract (SYSNO *sysno, const char *fname, logf (LOG_DEBUG, "fileExtract %s", fname); /* determine file extension */ + *ext = '\0'; for (i = strlen(fname); --i >= 0; ) if (fname[i] == '/') - { - strcpy (ext, ""); break; - } else if (fname[i] == '.') { strcpy (ext, fname+i+1); @@ -1478,21 +1593,20 @@ int fileExtract (SYSNO *sysno, const char *fname, if (!(rGroup->recordType = res_get (common_resource, ext_res))) { sprintf (ext_res, "%srecordType", gprefix); - if (!(rGroup->recordType = res_get (common_resource, ext_res))) - { - if (records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "? %s", fname); - return 0; - } + rGroup->recordType = res_get (common_resource, ext_res); } } if (!rGroup->recordType) { if (records_processed < rGroup->fileVerboseLimit) - logf (LOG_LOG, "? record %s", fname); + logf (LOG_LOG, "? %s", fname); return 0; } - if (!(recType = recType_byName (rGroup->recordType, subType))) + if (!*rGroup->recordType) + return 0; + if (!(recType = + recType_byName (rGroup->recTypes, rGroup->recordType, subType, + &clientData))) { logf (LOG_WARN, "No such record type: %s", rGroup->recordType); return 0; @@ -1518,10 +1632,18 @@ int fileExtract (SYSNO *sysno, const char *fname, if (!rGroup->databaseName) rGroup->databaseName = "Default"; + /* determine if explain database */ + + sprintf (ext_res, "%sexplainDatabase", gprefix); + rGroup->explainDatabase = + atoi (res_get_def (common_resource, ext_res, "0")); + + /* announce database */ if (zebraExplain_curDatabase (zti, rGroup->databaseName)) { - if (zebraExplain_newDatabase (zti, rGroup->databaseName)) - abort (); + if (zebraExplain_newDatabase (zti, rGroup->databaseName, + rGroup->explainDatabase)) + return 0; } if (rGroup->flagStoreData == -1) @@ -1555,6 +1677,59 @@ int fileExtract (SYSNO *sysno, const char *fname, if (rGroup->flagStoreKeys == -1) rGroup->flagStoreKeys = 0; +#if ZEBRASDR + if (rGroup->useSDR) + { + ZebraSdrHandle h; + char xname[128], *xp; + + strncpy (xname, fname, 127); + if (!(xp = strchr (xname, '.'))) + return 0; + *xp = '\0'; + if (strcmp (xp+1, "sdr.bits")) + return 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); + +#if 0 + 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 @@ -1570,7 +1745,7 @@ 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) @@ -1586,7 +1761,7 @@ static int explain_extract (void *handle, Record rec, data1_node *n) if (zebraExplain_curDatabase (zti, rec->info[recInfo_databaseName])) { - if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName])) + if (zebraExplain_newDatabase (zti, rec->info[recInfo_databaseName], 0)) abort (); } @@ -1596,8 +1771,8 @@ 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;