X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=blobdiff_plain;f=index%2Fextract.c;h=829bb68df75758c4a69bd3398741ba2f183fb720;hp=662b52800da2c170ed8aec782f0036311cdc93df;hb=5c693d36af8be6f6642257160b3c6441d2e2d762;hpb=9bf9498168489df4a3df3e01ce95f46643b7ea9e diff --git a/index/extract.c b/index/extract.c index 662b528..829bb68 100644 --- a/index/extract.c +++ b/index/extract.c @@ -1,10 +1,109 @@ /* - * Copyright (C) 1994-1995, Index Data I/S + * Copyright (C) 1994-1998, Index Data I/S * All rights reserved. * Sebastian Hammer, Adam Dickmeiss * * $Log: extract.c,v $ - * Revision 1.51 1996-03-19 11:08:42 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 + * 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 + * First work on multiple records in one file. + * New option, -offset, to the "unread" command in the filter module. + * + * Revision 1.53 1996/04/26 12:09:43 adam + * Added a few comments. + * + * Revision 1.52 1996/04/25 13:27:57 adam + * Function recordExtract modified so that files with no keys (possibly empty) + * are ignored. + * + * Revision 1.51 1996/03/19 11:08:42 adam * Bug fix: Log preamble wasn't always turned off after recordExtract. * * Revision 1.50 1996/02/12 18:45:36 adam @@ -186,19 +285,24 @@ */ #include #include +#ifdef WINDOWS +#include +#else #include +#endif #include -#include -#include #include +#include +#include #include "index.h" -#include "recindex.h" +#include "zinfo.h" static Dict matchDict; static Records records = NULL; +static SortIdx sortIdx = NULL; static char **key_buf; static size_t ptr_top; @@ -209,9 +313,26 @@ static int key_file_no; static int records_inserted = 0; static int records_updated = 0; static int records_deleted = 0; +static int records_processed = 0; -void key_open (int mem) +static ZebTargetInfo *zti = NULL; + +static void logRecord (int showFlag) +{ + if (!showFlag) + ++records_processed; + if (showFlag || !(records_processed % 1000)) + { + logf (LOG_LOG, "Records: %7d i/u/d %d/%d/%d", + records_processed, records_inserted, records_updated, + records_deleted); + } +} + +int key_open (BFiles bfs, int mem, int rw) { + if (!mem) + mem = atoi(res_get_def (common_resource, "memMax", "4"))*1024*1024; if (mem < 50000) mem = 50000; key_buf = xmalloc (mem); @@ -221,19 +342,33 @@ 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); + 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 { int sysno; int seqno; - char buf[512]; + char buf[768]; }; void encode_key_init (struct encode_info *i) @@ -290,24 +425,49 @@ 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"))) { - 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); @@ -327,6 +487,42 @@ 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 %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); @@ -337,41 +533,53 @@ void key_flush (void) key_buf_used = 0; } -int key_close (void) +int key_close () { key_flush (); xfree (key_buf); +#if 1 + zebTargetInfo_close (zti, 1); +#endif rec_close (&records); dict_close (matchDict); + sortIdx_close (sortIdx); - logf (LOG_LOG, "Records inserted %6d", records_inserted); - logf (LOG_LOG, "Records updated %6d", records_updated); - logf (LOG_LOG, "Records deleted %6d", records_deleted); + 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; char prevAttrSet; short prevAttrUse; + 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; if (reckeys.buf_used+1024 > reckeys.buf_max) { @@ -395,34 +603,181 @@ 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; - switch (p->which) - { - case Word_String: - *dst++ = lead; + *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'; + if (!(lead & 1)) + { + memcpy (dst, &attrSet, sizeof(attrSet)); + dst += sizeof(attrSet); + } + if (!(lead & 2)) + { + memcpy (dst, &attrUse, sizeof(attrUse)); + dst += sizeof(attrUse); + } + *dst++ = p->reg_type; + memcpy (dst, string, length); + dst += length; + *dst++ = '\0'; + if (!diff) + { memcpy (dst, &p->seqno, sizeof(p->seqno)); dst += sizeof(p->seqno); - - break; - default: - return; } 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, @@ -430,12 +785,19 @@ 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++; @@ -453,17 +815,26 @@ 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); @@ -479,34 +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; @@ -525,30 +904,12 @@ 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); -} - -#define FILE_READ_BUFSIZE 4096 struct file_read_info { - int file_noread; + off_t file_max; + off_t file_offset; + off_t file_moffset; + int file_more; int fd; -#if FILE_READ_BUFSIZE - char *file_buf; - int file_offset; - int file_bufsize; -#endif }; static struct file_read_info *file_read_start (int fd) @@ -556,82 +917,62 @@ static struct file_read_info *file_read_start (int fd) struct file_read_info *fi = xmalloc (sizeof(*fi)); fi->fd = fd; - fi->file_noread = 0; -#if FILE_READ_BUFSIZE - fi->file_offset = 0; - fi->file_buf = xmalloc (FILE_READ_BUFSIZE); - fi->file_bufsize = read (fd, fi->file_buf, FILE_READ_BUFSIZE); -#endif + fi->file_max = 0; + fi->file_moffset = 0; return fi; } static void file_read_stop (struct file_read_info *fi) { assert (fi); -#if FILE_READ_BUFSIZE - xfree (fi->file_buf); - fi->file_buf = NULL; -#endif xfree (fi); } +static off_t file_seek (void *handle, off_t offset) +{ + struct file_read_info *p = handle; + p->file_offset = 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; int fd = p->fd; -#if FILE_READ_BUFSIZE - int l = p->file_bufsize - p->file_offset; - - if (count > l) - { - int r; - if (l > 0) - memcpy (buf, p->file_buf + p->file_offset, l); - count = count-l; - if (count > FILE_READ_BUFSIZE) - { - if ((r = read (fd, buf + l, count)) == -1) - { - logf (LOG_FATAL|LOG_ERRNO, "read"); - exit (1); - } - p->file_bufsize = 0; - p->file_offset = 0; - p->file_noread += l+r; - return l+r; - } - p->file_bufsize = r = read (fd, p->file_buf, FILE_READ_BUFSIZE); - if (r == -1) - { - logf (LOG_FATAL|LOG_ERRNO, "read"); - exit (1); - } - else if (r <= count) - { - p->file_offset = r; - memcpy (buf + l, p->file_buf, r); - p->file_noread += l+r; - return l+r; - } - else - { - p->file_offset = count; - memcpy (buf + l, p->file_buf, count - l); - p->file_noread += count; - return count; - } - } - memcpy (buf, p->file_buf + p->file_offset, count); - p->file_offset += count; - p->file_noread += count; - return count; -#else int r; r = read (fd, buf, count); if (r > 0) - p->file_noread += r; + { + p->file_offset += r; + if (p->file_offset > p->file_max) + p->file_max = p->file_offset; + } return r; -#endif +} + +static void file_begin (void *handle) +{ + struct file_read_info *p = handle; + + p->file_offset = p->file_moffset; + if (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; + + assert (p->file_more == 0); + p->file_more = 1; + p->file_moffset = offset; } static int atois (const char **s) @@ -774,12 +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; }; @@ -790,9 +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); } @@ -805,31 +1146,63 @@ static int recordExtract (SYSNO *sysno, const char *fname, int r; char *matchStr; SYSNO sysnotmp; + off_t recordOffset = 0; 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; - /* extract keys */ extractCtrl.subType = subType; extractCtrl.init = wordInit; - extractCtrl.add = addRecordKeyAny; + extractCtrl.add = addRecordKey; + extractCtrl.dh = rGroup->dh; 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.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) { - logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r); + /* 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); + } + return 0; + } + if (reckeys.buf_used == 0) + { + /* the extraction process returned no information - the record + 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; } } @@ -861,17 +1234,19 @@ static int recordExtract (SYSNO *sysno, const char *fname, } } - /* new record ? */ if (! *sysno) { + /* 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", rGroup->recordType, fname); + if (records_processed < rGroup->fileVerboseLimit) + logf (LOG_LOG, "add %s %s %ld", rGroup->recordType, + fname, (long) recordOffset); rec = rec_new (records); *sysno = rec->sysno; @@ -880,60 +1255,73 @@ 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++; } else { + /* record already exists */ struct recKeys delkeys; rec = rec_get (records, *sysno); 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) { - logInfo.op = "delete"; + /* record going to be deleted */ 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) - logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname); + if (records_processed < rGroup->fileVerboseLimit) + logf (LOG_LOG, "delete %s %s %ld", rGroup->recordType, + fname, (long) recordOffset); records_deleted++; if (matchStr) dict_delete (matchDict, matchStr); rec_del (records, &rec); } + logRecord (0); return 1; } else { - logInfo.op = "update"; + /* record going to be updated */ 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) - logf (LOG_LOG, "update %s %s", rGroup->recordType, fname); + if (records_processed < rGroup->fileVerboseLimit) + logf (LOG_LOG, "update %s %s %ld", rGroup->recordType, + fname, (long) recordOffset); flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); records_updated++; } } } + /* update file type */ xfree (rec->info[recInfo_fileType]); rec->info[recInfo_fileType] = rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]); + /* update filename */ xfree (rec->info[recInfo_filename]); rec->info[recInfo_filename] = rec_strdup (fname, &rec->size[recInfo_filename]); + /* update delete keys */ xfree (rec->info[recInfo_delKeys]); if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1) { @@ -955,30 +1343,26 @@ static int recordExtract (SYSNO *sysno, const char *fname, rec->size[recInfo_delKeys] = 0; } + /* update store data */ xfree (rec->info[recInfo_storeData]); if (rGroup->flagStoreData == 1) { - rec->size[recInfo_storeData] = fi->file_noread; - rec->info[recInfo_storeData] = xmalloc (fi->file_noread); -#if FILE_READ_BUFSIZE - if (fi->file_noread < FILE_READ_BUFSIZE) - memcpy (rec->info[recInfo_storeData], fi->file_buf, - fi->file_noread); - else -#endif + 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) { - if (lseek (fi->fd, 0L, SEEK_SET) < 0) - { - logf (LOG_ERRNO|LOG_FATAL, "seek to 0 in %s", fname); - exit (1); - } - if (read (fi->fd, rec->info[recInfo_storeData], fi->file_noread) - < fi->file_noread) - { - logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s", - fi->file_noread, fname); - exit (1); - } + logf (LOG_ERRNO|LOG_FATAL, "seek to %ld in %s", fname, + (long) recordOffset); + exit (1); + } + if (read (fi->fd, rec->info[recInfo_storeData], size) < size) + { + logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s", + fi->file_max, fname); + exit (1); } } else @@ -986,18 +1370,28 @@ static int recordExtract (SYSNO *sysno, const char *fname, rec->info[recInfo_storeData] = NULL; rec->size[recInfo_storeData] = 0; } + /* update database name */ xfree (rec->info[recInfo_databaseName]); rec->info[recInfo_databaseName] = rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]); + /* update offset */ + xfree (rec->info[recInfo_offset]); + + rec->size[recInfo_offset] = sizeof(recordOffset); + rec->info[recInfo_offset] = xmalloc (sizeof(recordOffset)); + memcpy (rec->info[recInfo_offset], &recordOffset, sizeof(recordOffset)); + + /* commit this record */ rec_put (records, &rec); + logRecord (0); return 1; } int fileExtract (SYSNO *sysno, const char *fname, const struct recordGroup *rGroupP, int deleteFlag) { - int i, fd; + int r, i, fd; char gprefix[128]; char ext[128]; char ext_res[128]; @@ -1037,7 +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; } @@ -1045,7 +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; } @@ -1110,18 +1504,22 @@ 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; } } fi = file_read_start (fd); - recordExtract (sysno, fname, rGroup, deleteFlag, fi, recType, subType); - log_event_start (NULL, NULL); + do + { + file_begin (fi); + r = recordExtract (sysno, fname, rGroup, deleteFlag, fi, + recType, subType); + } while (r && !sysno && fi->file_more); file_read_stop (fi); if (fd != -1) close (fd); - return 1; + return r; }