* Sebastian Hammer, Adam Dickmeiss
*
* $Log: extract.c,v $
- * Revision 1.33 1995-11-27 09:56:20 adam
+ * Revision 1.40 1995-12-05 16:57:40 adam
+ * More work on regular patterns.
+ *
+ * Revision 1.39 1995/12/05 13:20:18 adam
+ * Bug fix: file_read sometimes returned early EOF.
+ *
+ * Revision 1.38 1995/12/04 17:59:21 adam
+ * More work on regular expression conversion.
+ *
+ * Revision 1.37 1995/12/04 14:22:27 adam
+ * Extra arg to recType_byName.
+ * Started work on new regular expression parsed input to
+ * structured records.
+ *
+ * Revision 1.36 1995/11/30 08:34:29 adam
+ * Started work on commit facility.
+ * Changed a few malloc/free to xmalloc/xfree.
+ *
+ * Revision 1.35 1995/11/28 14:26:21 adam
+ * Bug fix: recordId with constant wasn't right.
+ * Bug fix: recordId dictionary entry wasn't deleted when needed.
+ *
+ * Revision 1.34 1995/11/28 09:09:38 adam
+ * Zebra config renamed.
+ * Use setting 'recordId' to identify record now.
+ * Bug fix in recindex.c: rec_release_blocks was invokeded even
+ * though the blocks were already released.
+ * File traversal properly deletes records when needed.
+ *
+ * Revision 1.33 1995/11/27 09:56:20 adam
* Record info elements better enumerated. Internal store of records.
*
* Revision 1.32 1995/11/25 10:24:05 adam
{
char *b;
- b = malloc (reckeys.buf_max += 65000);
+ b = xmalloc (reckeys.buf_max += 65000);
if (reckeys.buf_used > 0)
memcpy (b, reckeys.buf, reckeys.buf_used);
- free (reckeys.buf);
+ xfree (reckeys.buf);
reckeys.buf = b;
}
dst = reckeys.buf + reckeys.buf_used;
addRecordKey (p);
}
-#define FILE_READ_BUFSIZE 4096
+#define FILE_READ_BUFSIZE 4096
+
+static int file_noread;
+#if FILE_READ_BUFSIZE
static char *file_buf;
static int file_offset;
static int file_bufsize;
-static int file_noread;
+#endif
static void file_read_start (int fd)
{
+ file_noread = 0;
+#if FILE_READ_BUFSIZE
file_offset = 0;
file_buf = xmalloc (FILE_READ_BUFSIZE);
file_bufsize = read (fd, file_buf, FILE_READ_BUFSIZE);
- file_noread = 0;
+#endif
}
static void file_read_stop (int fd)
{
+#if FILE_READ_BUFSIZE
xfree (file_buf);
file_buf = NULL;
+#endif
}
static int file_read (int fd, char *buf, size_t count)
{
+#if FILE_READ_BUFSIZE
int l = file_bufsize - file_offset;
if (count > l)
if (l > 0)
memcpy (buf, file_buf + file_offset, l);
count = count-l;
- if (count > file_bufsize)
+ if (count > FILE_READ_BUFSIZE)
{
if ((r = read (fd, buf + l, count)) == -1)
{
}
file_bufsize = 0;
file_offset = 0;
- file_noread += r;
- return r;
+ file_noread += l+r;
+ return l+r;
}
file_bufsize = r = read (fd, file_buf, FILE_READ_BUFSIZE);
if (r == -1)
{
file_offset = r;
memcpy (buf + l, file_buf, r);
- file_noread += (l+r);
- return l + r;
+ file_noread += l+r;
+ return l+r;
}
else
{
file_offset += count;
file_noread += count;
return count;
+#else
+ int r;
+ r = read (fd, buf, count);
+ if (r > 0)
+ file_noread += r;
+ return r;
+#endif
}
static int atois (const char **s)
static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
const char *fname,
- const char *recordType,
const char *spec)
{
static char dstBuf[2048];
special[spec_len] = '\0';
s = s1;
- if (strcmp (special, "group"))
+ if (!strcmp (special, "group"))
spec_src = rGroup->groupName;
- else if (strcmp (special, "database"))
+ else if (!strcmp (special, "database"))
spec_src = rGroup->databaseName;
- else if (strcmp (special, "filename"))
+ else if (!strcmp (special, "filename"))
spec_src = fname;
- else if (strcmp (special, "type"))
- spec_src = recordType;
+ else if (!strcmp (special, "type"))
+ spec_src = rGroup->recordType;
else
spec_src = NULL;
if (spec_src)
while (*s && *s != stopMarker)
{
if (i < 63)
- tmpString[i++] = *s;
+ tmpString[i++] = *s++;
}
if (*s)
s++;
static int recordExtract (SYSNO *sysno, const char *fname,
struct recordGroup *rGroup, int deleteFlag,
- int fd,
- const char *file_type,
- RecType recType)
+ int fd, RecType recType, char *subType)
{
struct recExtractCtrl extractCtrl;
int r;
SYSNO sysnotmp;
Record rec;
- extractCtrl.fd = fd;
-
- /* extract keys */
- extractCtrl.subType = "";
- extractCtrl.init = wordInit;
- extractCtrl.add = addRecordKeyAny;
-
- reckeys.buf_used = 0;
- extractCtrl.readf = file_read;
- r = (*recType->extract)(&extractCtrl);
-
- if (r)
+ if (fd != -1)
{
- logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
- return 0;
+ extractCtrl.fd = fd;
+ /* extract keys */
+ extractCtrl.subType = subType;
+ extractCtrl.init = wordInit;
+ extractCtrl.add = addRecordKeyAny;
+
+ reckeys.buf_used = 0;
+ extractCtrl.readf = file_read;
+ r = (*recType->extract)(&extractCtrl);
+
+ if (r)
+ {
+ logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
+ return 0;
+ }
}
/* perform match if sysno not known and if match criteria is specified */
{
sysnotmp = 0;
sysno = &sysnotmp;
- if (rGroup->fileMatch)
+ if (rGroup->recordId && *rGroup->recordId)
{
char *rinfo;
- matchStr = fileMatchStr(&reckeys, rGroup, fname, file_type,
- rGroup->fileMatch);
+ matchStr = fileMatchStr (&reckeys, rGroup, fname,
+ rGroup->recordId);
if (matchStr)
{
rinfo = dict_lookup (matchDict, matchStr);
{
if (deleteFlag)
{
- logf (LOG_LOG, "? record %s", fname);
+ logf (LOG_LOG, "? %s", fname);
return 1;
}
- logf (LOG_LOG, "add record %s", fname);
+ logf (LOG_LOG, "add %s %s", rGroup->recordType, fname);
rec = rec_new (records);
*sysno = rec->sysno;
if (matchStr)
+ {
dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
+ }
flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
records_inserted++;
{
if (!delkeys.buf_used)
{
- logf (LOG_WARN, "cannot delete %s - no delete keys",
- fname);
+ logf (LOG_WARN, "cannot delete %s - no delete keys", fname);
}
else
- logf (LOG_LOG, "delete record %s", fname);
- records_deleted++;
- rec_del (records, &rec);
+ {
+ SYSNO sysnoz = 0;
+ logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname);
+ records_deleted++;
+ if (matchStr)
+ dict_insert (matchDict, matchStr, sizeof(sysnoz), &sysnoz);
+ rec_del (records, &rec);
+ }
return 1;
}
else
}
else
{
- logf (LOG_LOG, "update record %s", fname);
+ logf (LOG_LOG, "update %s %s", rGroup->recordType,
+ fname);
flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
records_updated++;
}
}
}
- free (rec->info[recInfo_fileType]);
+ xfree (rec->info[recInfo_fileType]);
rec->info[recInfo_fileType] =
- rec_strdup (file_type, &rec->size[recInfo_fileType]);
+ rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
- free (rec->info[recInfo_filename]);
+ xfree (rec->info[recInfo_filename]);
rec->info[recInfo_filename] =
rec_strdup (fname, &rec->size[recInfo_filename]);
- free (rec->info[recInfo_delKeys]);
+ xfree (rec->info[recInfo_delKeys]);
if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
{
- rec->info[recInfo_delKeys] = malloc (reckeys.buf_used);
+ 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]);
rec->size[recInfo_delKeys] = 0;
}
- free (rec->info[recInfo_storeData]);
+ xfree (rec->info[recInfo_storeData]);
if (rGroup->flagStoreData == 1)
{
rec->size[recInfo_storeData] = file_noread;
- rec->info[recInfo_storeData] = malloc (file_noread);
+ rec->info[recInfo_storeData] = xmalloc (file_noread);
+#if FILE_READ_BUFSIZE
if (file_noread < FILE_READ_BUFSIZE)
memcpy (rec->info[recInfo_storeData], file_buf, file_noread);
else
+#endif
{
if (lseek (fd, 0L, SEEK_SET) < 0)
{
rec->info[recInfo_storeData] = NULL;
rec->size[recInfo_storeData] = 0;
}
- free (rec->info[recInfo_databaseName]);
+ xfree (rec->info[recInfo_databaseName]);
rec->info[recInfo_databaseName] =
rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
return 1;
}
-int fileExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup,
- int deleteFlag)
+int fileExtract (SYSNO *sysno, const char *fname,
+ const struct recordGroup *rGroupP, int deleteFlag)
{
int i, fd;
char gprefix[128];
char ext[128];
char ext_res[128];
- const char *file_type;
+ char subType[128];
RecType recType;
+ struct recordGroup rGroupM;
+ struct recordGroup *rGroup = &rGroupM;
+ memcpy (rGroup, rGroupP, sizeof(*rGroupP));
+
if (!rGroup->groupName || !*rGroup->groupName)
*gprefix = '\0';
else
break;
}
/* determine file type - depending on extension */
- sprintf (ext_res, "%sfileExtension.%s", gprefix, ext);
- if (!(file_type = res_get (common_resource, ext_res)))
+ if (!rGroup->recordType)
+ {
+ sprintf (ext_res, "%srecordType.%s", gprefix, ext);
+ if (!(rGroup->recordType = res_get (common_resource, ext_res)))
+ {
+ sprintf (ext_res, "%srecordType", gprefix);
+ if (!(rGroup->recordType = res_get (common_resource, ext_res)))
+ {
+ logf (LOG_LOG, "? %s", fname);
+ return 0;
+ }
+ }
+ }
+ if (!rGroup->recordType)
+ {
+ logf (LOG_LOG, "? record %s", fname);
return 0;
- if (!(recType = recType_byName (file_type)))
+ }
+ if (!(recType = recType_byName (rGroup->recordType, subType)))
+ {
+ logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
return 0;
+ }
/* determine match criteria */
- if (rGroup->fileMatch)
+ if (!rGroup->recordId)
{
- sprintf (ext_res, "%sfileMatch.%s", gprefix, ext);
- rGroup->fileMatch = res_get (common_resource, ext_res);
- if (!rGroup->fileMatch)
- {
- sprintf (ext_res, "%sfileMatch", gprefix);
- rGroup->fileMatch = res_get (common_resource, ext_res);
- }
+ sprintf (ext_res, "%srecordId.%s", gprefix, ext);
+ rGroup->recordId = res_get (common_resource, ext_res);
}
/* determine database name */
if (rGroup->flagStoreKeys == -1)
rGroup->flagStoreKeys = 0;
- /* open input file */
- if ((fd = open (fname, O_RDONLY)) == -1)
+ if (sysno && deleteFlag)
+ fd = -1;
+ else
{
- logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
- return 0;
+ if ((fd = open (fname, O_RDONLY)) == -1)
+ {
+ logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
+ return 0;
+ }
}
file_read_start (fd);
- recordExtract (sysno, fname, rGroup, deleteFlag, fd,
- file_type, recType);
+ recordExtract (sysno, fname, rGroup, deleteFlag, fd, recType, subType);
file_read_stop (fd);
- close (fd);
+ if (fd != -1)
+ close (fd);
return 1;
}