X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=aa695502246c84e03bf02c969aeae1dcaf557b18;hb=0ca8a9894b43c838980d650dd265f7a3da4ed219;hp=066262153574ee15179e9dbd6c725ea1441d5d57;hpb=1cd8e47d53d94462ab3a96f01ab200222004b901;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index 0662621..aa69550 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,10 +1,76 @@ /* - * Copyright (C) 1994-1995, Index Data I/S + * Copyright (C) 1994-1996, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: extract.c,v $ - * Revision 1.55 1996-05-09 07:28:55 quinn + * 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 + * Buffers used during file match got bigger. + * Compressed ISAM support everywhere. + * Bug fixes regarding masking characters in queries. + * Redesigned Regexp-2 queries. + * + * Revision 1.63 1996/10/29 14:09:39 adam + * Use of cisam system - enabled if setting isamc is 1. + * + * Revision 1.62 1996/10/11 10:57:01 adam + * New module recctrl. Used to manage records (extract/retrieval). + * Several files have been moved to the recctrl sub directory. + * + * Revision 1.61 1996/06/06 12:08:37 quinn + * Added showRecord function + * + * Revision 1.60 1996/06/04 10:18:12 adam + * Search/scan uses character mapping module. + * + * Revision 1.59 1996/05/14 15:47:07 adam + * Cleanup of various buffer size entities. + * + * Revision 1.58 1996/05/14 06:16:38 adam + * Compact use/set bytes used in search service. + * + * Revision 1.57 1996/05/13 14:23:04 adam + * Work on compaction of set/use bytes in dictionary. + * + * Revision 1.56 1996/05/09 09:54:42 adam + * Server supports maps from one logical attributes to a list of physical + * attributes. + * The extraction process doesn't make space consuming 'any' keys. + * + * Revision 1.55 1996/05/09 07:28:55 quinn * Work towards phrases and multiple registers * * Revision 1.54 1996/05/01 13:46:35 adam @@ -202,13 +268,11 @@ #include #include #include -#include -#include #include #include "index.h" -#include "recindex.h" +#include "zinfo.h" static Dict matchDict; @@ -225,6 +289,8 @@ static int records_updated = 0; static int records_deleted = 0; static int records_processed = 0; +static ZebTargetInfo *zti = NULL; + static void logRecord (int showFlag) { if (!showFlag) @@ -239,6 +305,8 @@ static void logRecord (int showFlag) void key_open (int mem) { + if (!mem) + mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024; if (mem < 50000) mem = 50000; key_buf = xmalloc (mem); @@ -255,12 +323,15 @@ void key_open (int mem) } assert (!records); records = rec_open (1); +#if 1 + zti = zebTargetInfo_open (records, 1); +#endif } struct encode_info { int sysno; int seqno; - char buf[512]; + char buf[768]; }; void encode_key_init (struct encode_info *i) @@ -317,20 +388,45 @@ void encode_key_write (char *k, struct encode_info *i, FILE *outf) } } +#define SORT_EXTRA 0 + +#if SORT_EXTRA +static int key_y_len; + +static int key_y_compare (const void *p1, const void *p2) +{ + int r; + + if ((r = key_compare (*(char**) p1 + key_y_len + 1, + *(char**) p2 + key_y_len + 1))) + return r; + return *(*(char**) p1 + key_y_len) - *(*(char**) p2 + key_y_len); +} + +static int key_x_compare (const void *p1, const void *p2) +{ + return strcmp (*(char**) p1, *(char**) p2); +} +#endif + void key_flush (void) { FILE *outf; char out_fname[200]; char *prevcp, *cp; struct encode_info encode_info; +#if SORT_EXTRA + int i; +#endif if (ptr_i <= 0) return; key_file_no++; 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); - sprintf (out_fname, TEMP_FNAME, key_file_no); + getFnameTmp (out_fname, key_file_no); if (!(outf = fopen (out_fname, "w"))) { @@ -354,6 +450,42 @@ void key_flush (void) else encode_key_write (cp + strlen(cp), &encode_info, outf); } +#else + qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_x_compare); + getFnameTmp (out_fname, key_file_no); + + if (!(outf = fopen (out_fname, "w"))) + { + logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname); + exit (1); + } + logf (LOG_LOG, "writing section %d", key_file_no); + i = ptr_i; + prevcp = key_buf[ptr_top-i]; + while (1) + if (!--i || strcmp (prevcp, key_buf[ptr_top-i])) + { + key_y_len = strlen(prevcp)+1; +#if 0 + logf (LOG_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, + sizeof(char*), key_y_compare); + cp = key_buf[ptr_top-ptr_i]; + --key_y_len; + encode_key_init (&encode_info); + encode_key_write (cp, &encode_info, outf); + while (--ptr_i > i) + { + cp = key_buf[ptr_top-ptr_i]; + encode_key_write (cp+key_y_len, &encode_info, outf); + } + if (!i) + break; + prevcp = key_buf[ptr_top-ptr_i]; + } +#endif if (fclose (outf)) { logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname); @@ -368,6 +500,9 @@ int key_close (void) { key_flush (); xfree (key_buf); +#if 1 + zebTargetInfo_close (zti, 1); +#endif rec_close (&records); dict_close (matchDict); @@ -388,6 +523,7 @@ struct recKeys { char *buf; char prevAttrSet; short prevAttrUse; + int prevSeqNo; } reckeys; static void addRecordKey (const RecWord *p) @@ -397,6 +533,7 @@ static void addRecordKey (const RecWord *p) short attrUse; size_t i; int lead = 0; + int diff = 0; if (reckeys.buf_used+1024 > reckeys.buf_max) { @@ -420,32 +557,46 @@ static void addRecordKey (const RecWord *p) lead |= 2; else reckeys.prevAttrUse = attrUse; +#if 1 + diff = 1 + p->seqno - reckeys.prevSeqNo; + if (diff >= 1 && diff <= 15) + lead |= (diff << 2); + else + diff = 0; +#endif + reckeys.prevSeqNo = p->seqno; + + *dst++ = lead; + if (!(lead & 1)) + { + memcpy (dst, &attrSet, sizeof(attrSet)); + dst += sizeof(attrSet); + } + if (!(lead & 2)) + { + memcpy (dst, &attrUse, sizeof(attrUse)); + dst += sizeof(attrUse); + } switch (p->which) { - case Word_String: case Word_Phrase: - *dst++ = lead; - - if (!(lead & 1)) - { - memcpy (dst, &attrSet, sizeof(attrSet)); - dst += sizeof(attrSet); - } - if (!(lead & 2)) - { - memcpy (dst, &attrUse, sizeof(attrUse)); - dst += sizeof(attrUse); - } - for (i = 0; p->u.string[i]; i++) - *dst++ = p->u.string[i]; - *dst++ = '\0'; + 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 < IT_MAX_WORD-3; i++) + *dst++ = p->u.string[i]; + *dst++ = '\0'; + if (!diff) + { memcpy (dst, &p->seqno, sizeof(p->seqno)); dst += sizeof(p->seqno); - - break; - default: - return; } reckeys.buf_used = dst - reckeys.buf; } @@ -455,12 +606,19 @@ static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys, { char attrSet = -1; short attrUse = -1; + int seqno = 0; int off = 0; + + if (zebTargetInfo_curDatabase (zti, databaseName)) + { + if (zebTargetInfo_newDatabase (zti, databaseName)) + abort (); + } while (off < reckeys->buf_used) { const char *src = reckeys->buf + off; struct it_key key; - int lead; + int lead, ch; lead = *src++; @@ -478,17 +636,26 @@ static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys, key_flush (); ++ptr_i; key_buf[ptr_top-ptr_i] = (char*)key_buf + key_buf_used; - key_buf_used += index_word_prefix ((char*)key_buf + key_buf_used, - attrSet, attrUse, databaseName); + + ch = zebTargetInfo_lookupSU (zti, attrSet, attrUse); + if (ch < 0) + ch = zebTargetInfo_addSU (zti, attrSet, attrUse); + assert (ch > 0); + ((char*) key_buf) [key_buf_used++] = ch; while (*src) - ((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++); + ((char*)key_buf) [key_buf_used++] = *src++; src++; ((char*)key_buf) [key_buf_used++] = '\0'; - ((char*) key_buf)[key_buf_used++] = cmd; - memcpy (&key.seqno, src, sizeof(key.seqno)); - src += sizeof(key.seqno); + 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*)key_buf + key_buf_used, &key, sizeof(key)); key_buf_used += sizeof(key); @@ -504,34 +671,42 @@ 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; @@ -550,21 +725,6 @@ static const char **searchRecordKey (struct recKeys *reckeys, return ws; } -static void addRecordKeyAny (const RecWord *p) -{ - if (p->attrSet != 1 || p->attrUse != 1016) - { - RecWord w; - - memcpy (&w, p, sizeof(w)); - w.attrSet = 1; - w.attrUse = 1016; - addRecordKey (&w); - } - if (p->attrSet != -1) - addRecordKey (p); -} - struct file_read_info { off_t file_max; off_t file_offset; @@ -596,6 +756,12 @@ 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; @@ -770,12 +936,13 @@ 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; }; @@ -786,9 +953,8 @@ 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); } @@ -805,41 +971,58 @@ 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 */ extractCtrl.fh = fi; extractCtrl.subType = subType; extractCtrl.init = wordInit; - extractCtrl.add = addRecordKeyAny; + extractCtrl.add = addRecordKey; reckeys.buf_used = 0; reckeys.prevAttrUse = -1; reckeys.prevAttrSet = -1; + reckeys.prevSeqNo = 0; recordOffset = fi->file_moffset; 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; + if (rGroup->flagShowRecords) + 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->flagShowRecords && + 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->flagShowRecords) + return 1; + logf (LOG_WARN, "No keys generated for file %s", fname); + logf (LOG_WARN, " The file is probably empty"); return 0; } } @@ -879,9 +1062,8 @@ static int recordExtract (SYSNO *sysno, const char *fname, logf (LOG_LOG, "Cannot delete new record"); 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; @@ -907,14 +1089,15 @@ static int recordExtract (SYSNO *sysno, const char *fname, 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++; @@ -928,14 +1111,15 @@ 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); @@ -979,16 +1163,18 @@ 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); @@ -1061,7 +1247,7 @@ 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; } @@ -1069,7 +1255,7 @@ 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; } @@ -1134,7 +1320,7 @@ 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; @@ -1147,7 +1333,6 @@ 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);