X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fextract.c;h=829bb68df75758c4a69bd3398741ba2f183fb720;hp=2eb7fd54365fea57814cb07b2747eb8e46e9bfaf;hb=5c693d36af8be6f6642257160b3c6441d2e2d762;hpb=8a10428f21d9a1c308a85eae620d2de8c7765577 diff --git a/index/extract.c b/index/extract.c index 2eb7fd5..829bb68 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,10 +1,59 @@ /* - * Copyright (C) 1994-1996, Index Data I/S + * Copyright (C) 1994-1998, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: extract.c,v $ - * Revision 1.65 1996-11-14 08:57:56 adam + * Revision 1.78 1998-02-10 12:03:05 adam + * Implemented Sort. + * + * Revision 1.77 1998/01/12 15:04:08 adam + * The test option (-s) only uses read-lock (and not write lock). + * + * Revision 1.76 1997/10/27 14:33:04 adam + * Moved towards generic character mapping depending on "structure" + * field in abstract syntax file. Fixed a few memory leaks. Fixed + * bug with negative integers when doing searches with relational + * operators. + * + * Revision 1.75 1997/09/17 12:19:12 adam + * Zebra version corresponds to YAZ version 1.4. + * Changed Zebra server so that it doesn't depend on global common_resource. + * + * Revision 1.74 1997/09/09 13:38:06 adam + * Partial port to WIN95/NT. + * + * Revision 1.73 1997/09/04 13:57:20 adam + * New file extract/retrieve method tellf (added). + * Added O_BINARY for open calls. + * + * Revision 1.72 1997/07/15 16:32:29 adam + * Bug fix: Match handler didn't terminate the resulting string! + * + * Revision 1.71 1997/07/15 16:28:41 adam + * Bug fix: storeData didn't work with files with multiple records. + * Bug fix: fixed memory management with records; not really well + * thought through. + * + * Revision 1.70 1997/07/01 13:00:42 adam + * Bug fix in routine searchRecordKey: uninitialized variables. + * + * Revision 1.69 1997/04/29 09:26:03 adam + * Bug fix: generic recordId handling didn't work for compressed internal + * keys. + * + * Revision 1.68 1997/02/12 20:39:45 adam + * Implemented options -f that limits the log to the first + * records. + * Changed some log messages also. + * + * Revision 1.67 1996/11/15 15:02:14 adam + * Minor changes regarding logging. + * + * Revision 1.66 1996/11/14 09:52:21 adam + * Strings in record keys bound by IT_MAX_WORD. + * + * Revision 1.65 1996/11/14 08:57:56 adam * Reduction of storeKeys area. * * Revision 1.64 1996/11/08 11:10:16 adam @@ -236,10 +285,16 @@ */ #include #include +#ifdef WINDOWS +#include +#else #include +#endif #include #include +#include +#include #include "index.h" #include "zinfo.h" @@ -247,6 +302,7 @@ static Dict matchDict; static Records records = NULL; +static SortIdx sortIdx = NULL; static char **key_buf; static size_t ptr_top; @@ -273,7 +329,7 @@ static void logRecord (int showFlag) } } -void key_open (int mem) +int key_open (BFiles bfs, int mem, int rw) { if (!mem) mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024; @@ -286,16 +342,27 @@ void key_open (int mem) key_buf_used = 0; key_file_no = 0; - if (!(matchDict = dict_open (GMATCH_DICT, 50, 1))) + if (!(matchDict = dict_open (bfs, GMATCH_DICT, 50, rw))) { logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT); - exit (1); + return -1; } assert (!records); - records = rec_open (1); -#if 1 - zti = zebTargetInfo_open (records, 1); -#endif + records = rec_open (bfs, rw); + if (!records) + { + dict_close (matchDict); + return -1; + } + zti = zebTargetInfo_open (records, rw); + if (!zti) + { + rec_close (&records); + dict_close (matchDict); + return -1; + } + sortIdx = sortIdx_open (bfs, 1); + return 0; } struct encode_info { @@ -400,7 +467,7 @@ void key_flush (void) if (!(outf = fopen (out_fname, "w"))) { - logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname); + logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); exit (1); } logf (LOG_LOG, "writing section %d", key_file_no); @@ -426,7 +493,7 @@ void key_flush (void) if (!(outf = fopen (out_fname, "w"))) { - logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname); + logf (LOG_FATAL|LOG_ERRNO, "fopen %s", out_fname); exit (1); } logf (LOG_LOG, "writing section %d", key_file_no); @@ -466,7 +533,7 @@ void key_flush (void) key_buf_used = 0; } -int key_close (void) +int key_close () { key_flush (); xfree (key_buf); @@ -475,19 +542,29 @@ int key_close (void) #endif rec_close (&records); dict_close (matchDict); + sortIdx_close (sortIdx); logRecord (1); return key_file_no; } -static void wordInit (RecWord *p) +static void wordInit (struct recExtractCtrl *p, RecWord *w) { - p->attrSet = 1; - p->attrUse = 1016; - p->which = Word_String; + w->zebra_maps = p->zebra_maps; + w->attrSet = 1; + w->attrUse = 1016; + w->reg_type = 'w'; } -struct recKeys { +static struct sortKey { + char *string; + int length; + int attrSet; + int attrUse; + struct sortKey *next; +} *sortKeys = NULL; + +static struct recKeys { int buf_used; int buf_max; char *buf; @@ -496,12 +573,11 @@ struct recKeys { int prevSeqNo; } reckeys; -static void addRecordKey (const RecWord *p) +static void addIndexString (RecWord *p, const char *string, int length) { char *dst; char attrSet; short attrUse; - size_t i; int lead = 0; int diff = 0; @@ -548,19 +624,9 @@ static void addRecordKey (const RecWord *p) memcpy (dst, &attrUse, sizeof(attrUse)); dst += sizeof(attrUse); } - switch (p->which) - { - case Word_String: - *dst++ = 'w'; - break; - case Word_Phrase: - *dst++ = 'p'; - break; - case Word_Numeric: - *dst++ = 'n'; - } - for (i = 0; p->u.string[i]; i++) - *dst++ = p->u.string[i]; + *dst++ = p->reg_type; + memcpy (dst, string, length); + dst += length; *dst++ = '\0'; if (!diff) @@ -569,6 +635,149 @@ static void addRecordKey (const RecWord *p) dst += sizeof(p->seqno); } reckeys.buf_used = dst - reckeys.buf; + (p->seqno)++; +} + +static void addSortString (RecWord *p, const char *string, int length) +{ + struct sortKey *sk; + + for (sk = sortKeys; sk; sk = sk->next) + if (sk->attrSet == p->attrSet && sk->attrUse == p->attrUse) + return; + + sk = 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->attrSet = p->attrSet; + sk->attrUse = p->attrUse; +} + +static void addString (RecWord *p, const char *string, int length) +{ + if (zebra_maps_is_sort (p->zebra_maps, p->reg_type)) + addSortString (p, string, length); + else + addIndexString (p, string, length); +} + +static void addIncompleteField (RecWord *p) +{ + const char *b = p->string; + int remain = p->length; + const char **map = 0; + + if (remain > 0) + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + + while (map) + { + char buf[IT_MAX_WORD+1]; + int i, remain; + + /* Skip spaces */ + while (map && *map && **map == *CHR_SPACE) + { + remain = p->length - (b - p->string); + if (remain > 0) + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + else + map = 0; + } + if (!map) + break; + i = 0; + while (map && *map && **map != *CHR_SPACE) + { + const char *cp = *map; + + while (i < IT_MAX_WORD && *cp) + buf[i++] = *(cp++); + remain = p->length - (b - p->string); + if (remain > 0) + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + else + map = 0; + } + if (!i) + return; + addString (p, buf, i); + } +} + +static void addCompleteField (RecWord *p) +{ + const char *b = p->string; + char buf[IT_MAX_WORD+1]; + const char **map = 0; + int i = 0, remain = p->length; + + if (remain > 0) + map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, remain); + + while (remain > 0 && i < IT_MAX_WORD) + { + while (map && *map && **map == *CHR_SPACE) + { + remain = p->length - (b - p->string); + if (remain > 0) + map = zebra_maps_input(p->zebra_maps, p->reg_type, &b, remain); + else + map = 0; + } + if (!map) + break; + + if (i && i < IT_MAX_WORD) + buf[i++] = *CHR_SPACE; + while (map && *map && **map != *CHR_SPACE) + { + 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 (remain > 0) + map = zebra_maps_input (p->zebra_maps, p->reg_type, &b, + remain); + else + map = 0; + } + } + if (!i) + return; + addString (p, buf, i); +} + +static void addRecordKey (RecWord *p) +{ + if (zebra_maps_is_complete (p->zebra_maps, p->reg_type)) + addCompleteField (p); + else + addIncompleteField(p); +} + +static void flushSortKeys (SYSNO sysno, int cmd) +{ + struct sortKey *sk = sortKeys; + + sortIdx_sysno (sortIdx, sysno); + while (sk) + { + struct sortKey *sk_next = sk->next; + sortIdx_type (sortIdx, sk->attrUse); + sortIdx_add (sortIdx, sk->string, sk->length); + xfree (sk->string); + xfree (sk); + sk = sk_next; + } + sortKeys = NULL; } static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys, @@ -641,34 +850,42 @@ static const char **searchRecordKey (struct recKeys *reckeys, int off = 0; int startSeq = -1; int i; + short attrUse; + char attrSet; + int seqno = 0; for (i = 0; i<32; i++) ws[i] = NULL; while (off < reckeys->buf_used) { - const char *src = reckeys->buf + off; - char attrSet; - short attrUse; - int seqno; - const char *wstart; - - memcpy (&attrSet, src, sizeof(attrSet)); - src += sizeof(attrSet); - memcpy (&attrUse, src, sizeof(attrUse)); - src += sizeof(attrUse); + const char *src = reckeys->buf + off; + const char *wstart; + int lead; + + lead = *src++; + if (!(lead & 1)) + { + memcpy (&attrSet, src, sizeof(attrSet)); + src += sizeof(attrSet); + } + if (!(lead & 2)) + { + memcpy (&attrUse, src, sizeof(attrUse)); + src += sizeof(attrUse); + } wstart = src; while (*src++) ; - - memcpy (&seqno, src, sizeof(seqno)); - src += sizeof(seqno); - -#if 0 - logf (LOG_LOG, "(%d,%d) %d %s", attrSet, attrUse, seqno, wstart); -#endif + if (lead & 60) + seqno += ((lead>>2) & 15)-1; + else + { + memcpy (&seqno, src, sizeof(seqno)); + src += sizeof(seqno); + } if (attrUseS == attrUse && attrSetS == attrSet) { int woff; @@ -718,6 +935,12 @@ static off_t file_seek (void *handle, off_t offset) return lseek (p->fd, offset, SEEK_SET); } +static off_t file_tell (void *handle) +{ + struct file_read_info *p = handle; + return p->file_offset; +} + static int file_read (void *handle, char *buf, size_t count) { struct file_read_info *p = handle; @@ -892,12 +1115,13 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, fname, rGroup->groupName ? rGroup->groupName : "none"); return NULL; } + *dst = '\0'; return dstBuf; } struct recordLogInfo { const char *fname; - char *op; + int recordOffset; struct recordGroup *rGroup; }; @@ -908,9 +1132,8 @@ static void recordLogPreamble (int level, const char *msg, void *info) if (level & LOG_LOG) return ; - if (p->op) - fprintf (outf, "%s of ", p->op); - fprintf (outf, "%s type %s\n", p->rGroup->recordType, p->fname); + fprintf (outf, "File %s, offset %d, type %s\n", + p->rGroup->recordType, p->recordOffset, p->fname); log_event_start (NULL, NULL); } @@ -927,11 +1150,6 @@ static int recordExtract (SYSNO *sysno, const char *fname, Record rec; struct recordLogInfo logInfo; - logInfo.fname = fname; - logInfo.op = NULL; - logInfo.rGroup = rGroup; - log_event_start (recordLogPreamble, &logInfo); - if (fi->fd != -1) { /* we are going to read from a file, so prepare the extraction */ @@ -939,6 +1157,7 @@ static int recordExtract (SYSNO *sysno, const char *fname, extractCtrl.subType = subType; extractCtrl.init = wordInit; extractCtrl.add = addRecordKey; + extractCtrl.dh = rGroup->dh; reckeys.buf_used = 0; reckeys.prevAttrUse = -1; @@ -949,22 +1168,41 @@ static int recordExtract (SYSNO *sysno, const char *fname, extractCtrl.offset = recordOffset; extractCtrl.readf = file_read; extractCtrl.seekf = file_seek; + extractCtrl.tellf = file_tell; extractCtrl.endf = file_end; - extractCtrl.map_chrs_input = map_chrs_input; - extractCtrl.flagShowRecords = rGroup->flagShowRecords; + extractCtrl.zebra_maps = rGroup->zebra_maps; + extractCtrl.flagShowRecords = !rGroup->flagRw; + if (!rGroup->flagRw) + printf ("File: %s %ld\n", fname, (long) recordOffset); + + logInfo.fname = fname; + logInfo.recordOffset = recordOffset; + logInfo.rGroup = rGroup; + log_event_start (recordLogPreamble, &logInfo); + r = (*recType->extract)(&extractCtrl); + log_event_start (NULL, NULL); + if (r) { /* error occured during extraction ... */ - logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r); + if (rGroup->flagRw && + records_processed < rGroup->fileVerboseLimit) + { + logf (LOG_WARN, "fail %s %s %ld code = %d", rGroup->recordType, + fname, (long) recordOffset, r); + } return 0; } if (reckeys.buf_used == 0) { /* the extraction process returned no information - the record - is probably empty */ - logf (LOG_WARN, "Empty file %s", 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"); return 0; } } @@ -1001,12 +1239,13 @@ static int recordExtract (SYSNO *sysno, const char *fname, /* new record */ if (deleteFlag) { - logf (LOG_LOG, "Cannot delete new record"); + logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, + fname, (long) recordOffset); + logf (LOG_WARN, "cannot delete record above (seems new)"); return 1; } - logInfo.op = "add"; - if (rGroup->fileVerboseFlag) - logf (LOG_LOG, "add %s %s+%ld", rGroup->recordType, + if (records_processed < rGroup->fileVerboseLimit) + logf (LOG_LOG, "add %s %s %ld", rGroup->recordType, fname, (long) recordOffset); rec = rec_new (records); *sysno = rec->sysno; @@ -1016,6 +1255,7 @@ static int recordExtract (SYSNO *sysno, const char *fname, dict_insert (matchDict, matchStr, sizeof(*sysno), sysno); } flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); + flushSortKeys (*sysno, 1); records_inserted++; } @@ -1028,18 +1268,20 @@ static int recordExtract (SYSNO *sysno, const char *fname, assert (rec); delkeys.buf_used = rec->size[recInfo_delKeys]; delkeys.buf = rec->info[recInfo_delKeys]; + flushSortKeys (*sysno, 0); flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]); if (deleteFlag) { /* record going to be deleted */ - logInfo.op = "delete"; if (!delkeys.buf_used) { - logf (LOG_WARN, "cannot delete; storeKeys false"); + logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, + fname, (long) recordOffset); + logf (LOG_WARN, "cannot delete file above, storeKeys false"); } else { - if (rGroup->fileVerboseFlag) + if (records_processed < rGroup->fileVerboseLimit) logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, fname, (long) recordOffset); records_deleted++; @@ -1053,14 +1295,15 @@ static int recordExtract (SYSNO *sysno, const char *fname, else { /* record going to be updated */ - logInfo.op = "update"; if (!delkeys.buf_used) { - logf (LOG_WARN, "cannot update; storeKeys false"); + logf (LOG_LOG, "update %s %s %ld", rGroup->recordType, + fname, (long) recordOffset); + logf (LOG_WARN, "cannot update file above, storeKeys false"); } else { - if (rGroup->fileVerboseFlag) + if (records_processed < rGroup->fileVerboseLimit) logf (LOG_LOG, "update %s %s %ld", rGroup->recordType, fname, (long) recordOffset); flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); @@ -1104,16 +1347,18 @@ static int recordExtract (SYSNO *sysno, const char *fname, xfree (rec->info[recInfo_storeData]); if (rGroup->flagStoreData == 1) { - rec->size[recInfo_storeData] = fi->file_max; - rec->info[recInfo_storeData] = xmalloc (fi->file_max); + int size = fi->file_moffset - recordOffset; + if (!size) + size = fi->file_max - recordOffset; + rec->size[recInfo_storeData] = size; + rec->info[recInfo_storeData] = xmalloc (size); if (lseek (fi->fd, recordOffset, SEEK_SET) < 0) { logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s", fname, (long) recordOffset); exit (1); } - if (read (fi->fd, rec->info[recInfo_storeData], fi->file_max) - < fi->file_max) + if (read (fi->fd, rec->info[recInfo_storeData], size) < size) { logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s", fi->file_max, fname); @@ -1186,7 +1431,7 @@ int fileExtract (SYSNO *sysno, const char *fname, sprintf (ext_res, "%srecordType", gprefix); if (!(rGroup->recordType = res_get (common_resource, ext_res))) { - if (rGroup->fileVerboseFlag) + if (records_processed < rGroup->fileVerboseLimit) logf (LOG_LOG, "? %s", fname); return 0; } @@ -1194,7 +1439,7 @@ int fileExtract (SYSNO *sysno, const char *fname, } if (!rGroup->recordType) { - if (rGroup->fileVerboseFlag) + if (records_processed < rGroup->fileVerboseLimit) logf (LOG_LOG, "? record %s", fname); return 0; } @@ -1259,7 +1504,7 @@ int fileExtract (SYSNO *sysno, const char *fname, fd = -1; else { - if ((fd = open (fname, O_RDONLY)) == -1) + if ((fd = open (fname, O_BINARY|O_RDONLY)) == -1) { logf (LOG_WARN|LOG_ERRNO, "open %s", fname); return 0; @@ -1272,7 +1517,6 @@ int fileExtract (SYSNO *sysno, const char *fname, r = recordExtract (sysno, fname, rGroup, deleteFlag, fi, recType, subType); } while (r && !sysno && fi->file_more); - log_event_start (NULL, NULL); file_read_stop (fi); if (fd != -1) close (fd);