/*
- * 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 1.47 1996-01-17 14:57:48 adam
+ * 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
+ * New fileVerboseFlag in record group control.
+ *
+ * Revision 1.49 1996/02/05 12:29:57 adam
+ * Logging reduced a bit.
+ * The remaining running time is estimated during register merge.
+ *
+ * Revision 1.48 1996/02/01 20:53:26 adam
+ * The temporary per-record keys are compacted a little, and duplication
+ * of the per-records keys are avoided when they are saved in the record
+ * information buffer.
+ *
+ * Revision 1.47 1996/01/17 14:57:48 adam
* Prototype changed for reader functions in extract/retrieve. File
* is identified by 'void *' instead of 'int.
*
#include <assert.h>
#include <unistd.h>
#include <fcntl.h>
-#include <ctype.h>
#include <alexutil.h>
#include <recctrl.h>
#include "index.h"
-#include "recindex.h"
+#include "zinfo.h"
static Dict matchDict;
static int records_inserted = 0;
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)
+ ++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);
+ }
+}
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)
{
key_flush ();
xfree (key_buf);
+#if 1
+ zebTargetInfo_close (zti, 1);
+#endif
rec_close (&records);
dict_close (matchDict);
- 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;
}
int buf_used;
int buf_max;
char *buf;
+ char prevAttrSet;
+ short prevAttrUse;
} reckeys;
static void addRecordKey (const RecWord *p)
char attrSet;
short attrUse;
size_t i;
+ int lead = 0;
if (reckeys.buf_used+1024 > reckeys.buf_max)
{
char *b;
- b = xmalloc (reckeys.buf_max += 65000);
+ b = xmalloc (reckeys.buf_max += 128000);
if (reckeys.buf_used > 0)
memcpy (b, reckeys.buf, reckeys.buf_used);
xfree (reckeys.buf);
reckeys.buf = b;
}
dst = reckeys.buf + reckeys.buf_used;
- switch (p->which)
+
+ attrSet = p->attrSet;
+ if (reckeys.buf_used > 0 && reckeys.prevAttrSet == attrSet)
+ lead |= 1;
+ else
+ reckeys.prevAttrSet = attrSet;
+ attrUse = p->attrUse;
+ if (reckeys.buf_used > 0 && reckeys.prevAttrUse == attrUse)
+ lead |= 2;
+ else
+ reckeys.prevAttrUse = attrUse;
+
+ *dst++ = lead;
+
+ if (!(lead & 1))
{
- case Word_String:
- attrSet = p->attrSet;
memcpy (dst, &attrSet, sizeof(attrSet));
dst += sizeof(attrSet);
-
- attrUse = p->attrUse;
+ }
+ 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';
-
- memcpy (dst, &p->seqno, sizeof(p->seqno));
- dst += sizeof(p->seqno);
-
- break;
- default:
- return;
}
+ 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++ = '\0';
+
+ memcpy (dst, &p->seqno, sizeof(p->seqno));
+ dst += sizeof(p->seqno);
reckeys.buf_used = dst - reckeys.buf;
}
static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
const char *databaseName)
{
+ char attrSet = -1;
+ short attrUse = -1;
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;
- char attrSet;
- short attrUse;
struct it_key key;
-
- memcpy (&attrSet, src, sizeof(attrSet));
- src += sizeof(attrSet);
-
- memcpy (&attrUse, src, sizeof(attrUse));
- src += sizeof(attrUse);
+ 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);
+ }
if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
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);
+
+ lead = zebTargetInfo_lookupSU (zti, attrSet, attrUse);
+ if (lead < 0)
+ lead = zebTargetInfo_addSU (zti, attrSet, attrUse);
+ assert (lead > 0);
+ ((char*) key_buf) [key_buf_used++] = lead;
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';
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)
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 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)
}
if (first)
{
- logf (LOG_WARN, "Record in file %s didn't contain match"
- " fields in (%d,%d)", fname, attrSet, attrUse);
+ logf (LOG_WARN, "Record didn't contain match"
+ " fields in (%d,%d)", attrSet, attrUse);
return NULL;
}
}
return dstBuf;
}
+struct recordLogInfo {
+ const char *fname;
+ char *op;
+ struct recordGroup *rGroup;
+};
+
+static void recordLogPreamble (int level, const char *msg, void *info)
+{
+ struct recordLogInfo *p = info;
+ FILE *outf = log_file ();
+
+ 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);
+ log_event_start (NULL, NULL);
+}
+
static int recordExtract (SYSNO *sysno, const char *fname,
struct recordGroup *rGroup, int deleteFlag,
struct file_read_info *fi, RecType recType,
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;
reckeys.buf_used = 0;
+ reckeys.prevAttrUse = -1;
+ reckeys.prevAttrSet = -1;
+
+ recordOffset = fi->file_moffset;
+ extractCtrl.offset = recordOffset;
extractCtrl.readf = file_read;
+ extractCtrl.seekf = file_seek;
+ extractCtrl.endf = file_end;
+ extractCtrl.map_chrs_input = map_chrs_input;
+ extractCtrl.flagShowRecords = rGroup->flagShowRecords;
r = (*recType->extract)(&extractCtrl);
-
+
if (r)
{
+ /* error occured during extraction ... */
logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, 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);
+ return 0;
+ }
}
/* perform match if sysno not known and if match criteria is specified */
}
else
{
- logf (LOG_WARN, "Record not inserted");
+ logf (LOG_WARN, "Bad match criteria");
return 0;
}
}
}
- /* new record ? */
if (! *sysno)
{
+ /* new record */
if (deleteFlag)
{
- logf (LOG_LOG, "? %s", fname);
+ logf (LOG_LOG, "Cannot delete new record");
return 1;
}
- logf (LOG_LOG, "add %s %s", rGroup->recordType, fname);
+ logInfo.op = "add";
+ if (rGroup->fileVerboseFlag)
+ logf (LOG_LOG, "add %s %s+%ld", rGroup->recordType,
+ fname, (long) recordOffset);
rec = rec_new (records);
*sysno = rec->sysno;
}
else
{
+ /* record already exists */
struct recKeys delkeys;
rec = rec_get (records, *sysno);
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 %s: storeKeys false",
- fname);
+ logf (LOG_WARN, "cannot delete; storeKeys false");
}
else
{
- logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname);
+ if (rGroup->fileVerboseFlag)
+ 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
{
+ /* record going to be updated */
+ logInfo.op = "update";
if (!delkeys.buf_used)
{
- logf (LOG_WARN, "cannot update %s: storeKeys false",
- fname);
+ logf (LOG_WARN, "cannot update; storeKeys false");
}
else
{
- logf (LOG_LOG, "update %s %s", rGroup->recordType,
- fname);
+ if (rGroup->fileVerboseFlag)
+ 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)
{
+#if 1
+ rec->size[recInfo_delKeys] = reckeys.buf_used;
+ rec->info[recInfo_delKeys] = reckeys.buf;
+ reckeys.buf = NULL;
+ reckeys.buf_max = 0;
+#else
rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used);
rec->size[recInfo_delKeys] = reckeys.buf_used;
memcpy (rec->info[recInfo_delKeys], reckeys.buf,
rec->size[recInfo_delKeys]);
+#endif
}
else
{
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
+ rec->size[recInfo_storeData] = fi->file_max;
+ rec->info[recInfo_storeData] = xmalloc (fi->file_max);
+ 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], fi->file_max)
+ < fi->file_max)
+ {
+ logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
+ fi->file_max, fname);
+ exit (1);
}
}
else
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];
sprintf (ext_res, "%srecordType", gprefix);
if (!(rGroup->recordType = res_get (common_resource, ext_res)))
{
- logf (LOG_LOG, "? %s", fname);
+ if (rGroup->fileVerboseFlag)
+ logf (LOG_LOG, "? %s", fname);
return 0;
}
}
}
if (!rGroup->recordType)
{
- logf (LOG_LOG, "? record %s", fname);
+ if (rGroup->fileVerboseFlag)
+ logf (LOG_LOG, "? record %s", fname);
return 0;
}
if (!(recType = recType_byName (rGroup->recordType, subType)))
}
}
fi = file_read_start (fd);
- recordExtract (sysno, fname, rGroup, deleteFlag, fi, recType, subType);
+ do
+ {
+ file_begin (fi);
+ 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);
- return 1;
+ return r;
}
projects / idzebra-moved-to-github.git / blobdiff