* Sebastian Hammer, Adam Dickmeiss
*
* $Log: extract.c,v $
- * Revision 1.28 1995-11-21 09:20:30 adam
+ * 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
+ * More record fields - they are enumerated now.
+ * New options: flagStoreData flagStoreKey.
+ *
+ * Revision 1.31 1995/11/24 11:31:35 adam
+ * Commands add & del read filenames from stdin if source directory is
+ * empty.
+ * Match criteria supports 'constant' strings.
+ *
+ * Revision 1.30 1995/11/22 17:19:16 adam
+ * Record management uses the bfile system.
+ *
+ * Revision 1.29 1995/11/21 15:01:14 adam
+ * New general match criteria implemented.
+ * New feature: document groups.
+ *
+ * Revision 1.28 1995/11/21 09:20:30 adam
* Yet more work on record match.
*
* Revision 1.27 1995/11/20 16:59:45 adam
#include "recindex.h"
-static Dict file_idx;
+static Dict matchDict;
static Records records = NULL;
static size_t key_buf_used;
static int key_file_no;
+static int records_inserted = 0;
+static int records_updated = 0;
+static int records_deleted = 0;
+
+#define MATCH_DICT "match"
+
void key_open (int mem)
{
if (mem < 50000)
key_buf_used = 0;
key_file_no = 0;
- if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
+ if (!(matchDict = dict_open (MATCH_DICT, 20, 1)))
{
- logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
+ logf (LOG_FATAL, "dict_open fail of %s", MATCH_DICT);
exit (1);
}
assert (!records);
key_flush ();
xfree (key_buf);
rec_close (&records);
- dict_close (file_idx);
+ 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);
return key_file_no;
}
{
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;
attrSet, attrUse, databaseName);
while (*src)
((char*)key_buf) [key_buf_used++] = index_char_cvt (*src++);
+ src++;
((char*)key_buf) [key_buf_used++] = '\0';
((char*) key_buf)[key_buf_used++] = cmd;
key_buf_used += sizeof(key);
off = src - reckeys->buf;
}
- assert (off = reckeys->buf_used);
+ assert (off == reckeys->buf_used);
}
-#if 0
-static int key_cmd;
-static int key_sysno;
-static const char *key_databaseName;
-static int key_del_max;
-static int key_del_used;
-static char *key_del_buf;
-
-static void wordAdd (const RecWord *p)
+static const char **searchRecordKey (struct recKeys *reckeys,
+ int attrSetS, int attrUseS)
{
- struct it_key key;
- size_t i;
-
- 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,
- p->attrSet, p->attrUse,
- key_databaseName);
- switch (p->which)
- {
- case Word_String:
- for (i = 0; p->u.string[i]; i++)
- ((char*)key_buf) [key_buf_used++] =
- index_char_cvt (p->u.string[i]);
- ((char*)key_buf) [key_buf_used++] = '\0';
- break;
- default:
- return ;
- }
- ((char*) key_buf)[key_buf_used++] = ((key_cmd == 'a') ? 1 : 0);
- key.sysno = key_sysno;
- key.seqno = p->seqno;
- memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
- key_buf_used += sizeof(key);
+ static const char *ws[32];
+ int off = 0;
+ int startSeq = -1;
+ int i;
- if (key_cmd == 'a' && key_del_used >= 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;
- if (key_del_used + 1024 > key_del_max)
- {
- char *kbn;
-
- if (!(kbn = malloc (key_del_max += 64000)))
- {
- logf (LOG_FATAL, "malloc");
- exit (1);
- }
- if (key_del_buf)
- memcpy (kbn, key_del_buf, key_del_used);
- free (key_del_buf);
- key_del_buf = kbn;
- }
- switch (p->which)
+ int seqno;
+ const char *wstart;
+
+ memcpy (&attrSet, src, sizeof(attrSet));
+ src += sizeof(attrSet);
+
+ 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 (attrUseS == attrUse && attrSetS == attrSet)
{
- case Word_String:
- for (i = 0; p->u.string[i]; i++)
- ((char*)key_del_buf) [key_del_used++] = p->u.string[i];
- ((char*)key_del_buf) [key_del_used++] = '\0';
- break;
- default:
- return ;
- }
- attrSet = p->attrSet;
- memcpy (key_del_buf + key_del_used, &attrSet, sizeof(attrSet));
- key_del_used += sizeof(attrSet);
+ int woff;
- attrUse = p->attrUse;
- memcpy (key_del_buf + key_del_used, &attrUse, sizeof(attrUse));
- key_del_used += sizeof(attrUse);
- memcpy (key_del_buf + key_del_used, &p->seqno, sizeof(p->seqno));
- key_del_used += sizeof(p->seqno);
+ if (startSeq == -1)
+ startSeq = seqno;
+ woff = seqno - startSeq;
+ if (woff >= 0 && woff < 31)
+ ws[woff] = wstart;
+ }
+
+ off = src - reckeys->buf;
}
+ assert (off == reckeys->buf_used);
+ return ws;
}
-#endif
-
static void addRecordKeyAny (const RecWord *p)
{
if (p->attrSet != 1 || p->attrUse != 1016)
addRecordKey (p);
}
+#define FILE_READ_BUFSIZE 4096
+
static char *file_buf;
static int file_offset;
static int file_bufsize;
+static int file_noread;
static void file_read_start (int fd)
{
file_offset = 0;
- file_buf = xmalloc (4096);
- file_bufsize = read (fd, file_buf, 4096);
+ file_buf = xmalloc (FILE_READ_BUFSIZE);
+ file_bufsize = read (fd, file_buf, FILE_READ_BUFSIZE);
+ file_noread = 0;
}
static void file_read_stop (int fd)
{
xfree (file_buf);
+ file_buf = NULL;
}
static int file_read (int fd, char *buf, size_t count)
}
file_bufsize = 0;
file_offset = 0;
+ file_noread += r;
return r;
}
- file_bufsize = r = read (fd, file_buf, 4096);
+ file_bufsize = r = read (fd, file_buf, FILE_READ_BUFSIZE);
if (r == -1)
{
logf (LOG_FATAL|LOG_ERRNO, "read");
{
file_offset = r;
memcpy (buf + l, file_buf, r);
+ file_noread += (l+r);
return l + r;
}
else
{
file_offset = count;
memcpy (buf + l, file_buf, count - l);
+ file_noread += count;
return count;
}
}
memcpy (buf, file_buf + file_offset, count);
file_offset += count;
+ file_noread += count;
return count;
}
-int fileExtract (SYSNO *sysno, const char *fname, const char *databaseName,
- int deleteFlag)
+static int atois (const char **s)
{
- int i, r;
- char ext[128];
- char ext_res[128];
- const char *file_type;
- struct recExtractCtrl extractCtrl;
- RecType rt;
- Record rec;
+ int val = 0, c;
+ while ( (c=**s) >= '0' && c <= '9')
+ {
+ val = val*10 + c - '0';
+ ++(*s);
+ }
+ return val;
+}
- logf (LOG_DEBUG, "fileExtractAdd %s", fname);
+static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
+ const char *fname,
+ const char *spec)
+{
+ static char dstBuf[2048];
+ char *dst = dstBuf;
+ const char *s = spec;
+ static const char **w;
+ int i;
- for (i = strlen(fname); --i >= 0; )
- if (fname[i] == '/')
- {
- strcpy (ext, "");
+ while (1)
+ {
+ while (*s == ' ' || *s == '\t')
+ s++;
+ if (!*s)
break;
+ if (*s == '(')
+ {
+ char matchFlag[32];
+ int attrSet, attrUse;
+ int first = 1;
+
+ s++;
+ attrSet = atois (&s);
+ if (*s != ',')
+ {
+ logf (LOG_WARN, "Missing , in match criteria %s in group %s",
+ spec, rGroup->groupName ? rGroup->groupName : "none");
+ return NULL;
+ }
+ s++;
+ attrUse = atois (&s);
+ w = searchRecordKey (reckeys, attrSet, attrUse);
+ assert (w);
+
+ if (*s == ')')
+ {
+ for (i = 0; i<32; i++)
+ matchFlag[i] = 1;
+ }
+ else
+ {
+ logf (LOG_WARN, "Missing ) in match criteria %s in group %s",
+ spec, rGroup->groupName ? rGroup->groupName : "none");
+ return NULL;
+ }
+ s++;
+
+ for (i = 0; i<32; i++)
+ if (matchFlag[i] && w[i])
+ {
+ if (first)
+ {
+ *dst++ = ' ';
+ first = 0;
+ }
+ strcpy (dst, w[i]);
+ dst += strlen(w[i]);
+ }
+ if (first)
+ {
+ logf (LOG_WARN, "Record in file %s didn't contain match"
+ " fields in (%d,%d)", fname, attrSet, attrUse);
+ return NULL;
+ }
}
- else if (fname[i] == '.')
+ else if (*s == '$')
{
- strcpy (ext, fname+i+1);
- break;
+ int spec_len;
+ char special[64];
+ const char *spec_src = NULL;
+ const char *s1 = ++s;
+ while (*s1 && *s1 != ' ' && *s1 != '\t')
+ s1++;
+
+ spec_len = s1 - s;
+ if (spec_len > 63)
+ spec_len = 63;
+ memcpy (special, s, spec_len);
+ special[spec_len] = '\0';
+ s = s1;
+
+ if (!strcmp (special, "group"))
+ spec_src = rGroup->groupName;
+ else if (!strcmp (special, "database"))
+ spec_src = rGroup->databaseName;
+ else if (!strcmp (special, "filename"))
+ spec_src = fname;
+ else if (!strcmp (special, "type"))
+ spec_src = rGroup->recordType;
+ else
+ spec_src = NULL;
+ if (spec_src)
+ {
+ strcpy (dst, spec_src);
+ dst += strlen (spec_src);
+ }
}
- sprintf (ext_res, "fileExtension.%s", ext);
- if (!(file_type = res_get (common_resource, ext_res)))
- return 0;
- if (!(rt = recType_byName (file_type)))
- return 0;
+ else if (*s == '\"' || *s == '\'')
+ {
+ int stopMarker = *s++;
+ char tmpString[64];
+ int i = 0;
- if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
+ while (*s && *s != stopMarker)
+ {
+ if (i < 63)
+ tmpString[i++] = *s++;
+ }
+ if (*s)
+ s++;
+ tmpString[i] = '\0';
+ strcpy (dst, tmpString);
+ dst += strlen (tmpString);
+ }
+ else
+ {
+ logf (LOG_WARN, "Syntax error in match criteria %s in group %s",
+ spec, rGroup->groupName ? rGroup->groupName : "none");
+ return NULL;
+ }
+ *dst++ = 1;
+ }
+ if (dst == dstBuf)
{
- logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
- return 0;
+ logf (LOG_WARN, "No match criteria for record %s in group %s",
+ fname, rGroup->groupName ? rGroup->groupName : "none");
+ return NULL;
}
+ return dstBuf;
+}
- extractCtrl.subType = "";
- extractCtrl.init = wordInit;
- extractCtrl.add = addRecordKeyAny;
+static int recordExtract (SYSNO *sysno, const char *fname,
+ struct recordGroup *rGroup, int deleteFlag,
+ int fd,
+ RecType recType)
+{
+ struct recExtractCtrl extractCtrl;
+ int r;
+ char *matchStr;
+ SYSNO sysnotmp;
+ Record rec;
- reckeys.buf_used = 0;
- file_read_start (extractCtrl.fd);
- extractCtrl.readf = file_read;
- r = (*rt->extract)(&extractCtrl);
- file_read_stop (extractCtrl.fd);
- close (extractCtrl.fd);
+ if (fd != -1)
+ {
+ 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 (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 */
+
+ matchStr = NULL;
+ if (!sysno)
{
- logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
- return 0;
+ sysnotmp = 0;
+ sysno = &sysnotmp;
+ if (rGroup->recordId && *rGroup->recordId)
+ {
+ char *rinfo;
+
+ matchStr = fileMatchStr (&reckeys, rGroup, fname,
+ rGroup->recordId);
+ if (matchStr)
+ {
+ rinfo = dict_lookup (matchDict, matchStr);
+ if (rinfo)
+ memcpy (sysno, rinfo+1, sizeof(*sysno));
+ }
+ else
+ {
+ logf (LOG_WARN, "Record not inserted");
+ return 0;
+ }
+ }
}
- if (! *sysno) /* match criteria */
+
+ /* new record ? */
+ if (! *sysno)
{
- logf (LOG_LOG, "add record %s", fname);
+ if (deleteFlag)
+ {
+ logf (LOG_LOG, "? %s", fname);
+ return 1;
+ }
+ logf (LOG_LOG, "add %s %s", rGroup->recordType, fname);
rec = rec_new (records);
*sysno = rec->sysno;
- flushRecordKeys (*sysno, 1, &reckeys, databaseName);
+ if (matchStr)
+ {
+ dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
+ }
+ flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
+
+ records_inserted++;
}
else
{
struct recKeys delkeys;
-
+
rec = rec_get (records, *sysno);
- delkeys.buf_used = rec->size[2];
- delkeys.buf = rec->info[2];
- flushRecordKeys (*sysno, 0, &delkeys, rec->info[3]);
- flushRecordKeys (*sysno, 1, &reckeys, databaseName);
+ delkeys.buf_used = rec->size[recInfo_delKeys];
+ delkeys.buf = rec->info[recInfo_delKeys];
+ flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]);
+ if (deleteFlag)
+ {
+ if (!delkeys.buf_used)
+ {
+ logf (LOG_WARN, "cannot delete %s - no delete keys", fname);
+ }
+ else
+ {
+ 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
+ {
+ if (!delkeys.buf_used)
+ {
+ logf (LOG_WARN, "cannot update %s - no delete keys",
+ fname);
+ }
+ else
+ {
+ logf (LOG_LOG, "update %s %s", rGroup->recordType,
+ fname);
+ flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
+ records_updated++;
+ }
+ }
}
- free (rec->info[0]);
- rec->info[0] = rec_strdup (file_type, &rec->size[0]);
+ xfree (rec->info[recInfo_fileType]);
+ rec->info[recInfo_fileType] =
+ rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
- free (rec->info[1]);
- rec->info[1] = rec_strdup (fname, &rec->size[1]);
+ xfree (rec->info[recInfo_filename]);
+ rec->info[recInfo_filename] =
+ rec_strdup (fname, &rec->size[recInfo_filename]);
- free (rec->info[2]);
- if (reckeys.buf_used > 0)
+ xfree (rec->info[recInfo_delKeys]);
+ if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
{
- rec->info[2] = malloc (reckeys.buf_used);
- rec->size[2] = reckeys.buf_used;
- memcpy (rec->info[2], reckeys.buf, rec->size[2]);
+ 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]);
}
else
{
- rec->info[2] = NULL;
- rec->size[2] = 0;
+ rec->info[recInfo_delKeys] = NULL;
+ rec->size[recInfo_delKeys] = 0;
}
- free (rec->info[3]);
- rec->info[3] = rec_strdup (databaseName, &rec->size[3]);
+
+ xfree (rec->info[recInfo_storeData]);
+ if (rGroup->flagStoreData == 1)
+ {
+ rec->size[recInfo_storeData] = file_noread;
+ rec->info[recInfo_storeData] = xmalloc (file_noread);
+ if (file_noread < FILE_READ_BUFSIZE)
+ memcpy (rec->info[recInfo_storeData], file_buf, file_noread);
+ else
+ {
+ if (lseek (fd, 0L, SEEK_SET) < 0)
+ {
+ logf (LOG_ERRNO|LOG_FATAL, "seek to 0 in %s", fname);
+ exit (1);
+ }
+ if (read (fd, rec->info[recInfo_storeData], file_noread)
+ < file_noread)
+ {
+ logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
+ file_noread, fname);
+ exit (1);
+ }
+ }
+ }
+ else
+ {
+ rec->info[recInfo_storeData] = NULL;
+ rec->size[recInfo_storeData] = 0;
+ }
+ xfree (rec->info[recInfo_databaseName]);
+ rec->info[recInfo_databaseName] =
+ rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]);
rec_put (records, &rec);
return 1;
}
+
+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];
+ RecType recType;
+ struct recordGroup rGroupM;
+ struct recordGroup *rGroup = &rGroupM;
+
+ memcpy (rGroup, rGroupP, sizeof(*rGroupP));
+
+ if (!rGroup->groupName || !*rGroup->groupName)
+ *gprefix = '\0';
+ else
+ sprintf (gprefix, "%s.", rGroup->groupName);
+
+ logf (LOG_DEBUG, "fileExtract %s", fname);
+
+ /* determine file extension */
+ for (i = strlen(fname); --i >= 0; )
+ if (fname[i] == '/')
+ {
+ strcpy (ext, "");
+ break;
+ }
+ else if (fname[i] == '.')
+ {
+ strcpy (ext, fname+i+1);
+ break;
+ }
+ /* determine file type - depending on extension */
+ 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 (rGroup->recordType)))
+ {
+ logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
+ return 0;
+ }
+
+ /* determine match criteria */
+ if (!rGroup->recordId)
+ {
+ sprintf (ext_res, "%srecordId.%s", gprefix, ext);
+ rGroup->recordId = res_get (common_resource, ext_res);
+ }
+
+ /* determine database name */
+ if (!rGroup->databaseName)
+ {
+ sprintf (ext_res, "%sdatabase.%s", gprefix, ext);
+ if (!(rGroup->databaseName = res_get (common_resource, ext_res)))
+ {
+ sprintf (ext_res, "%sdatabase", gprefix);
+ rGroup->databaseName = res_get (common_resource, ext_res);
+ }
+ }
+ if (!rGroup->databaseName)
+ rGroup->databaseName = "Default";
+
+ if (rGroup->flagStoreData == -1)
+ {
+ const char *sval;
+ sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
+ if (!(sval = res_get (common_resource, ext_res)))
+ {
+ sprintf (ext_res, "%sstoreData", gprefix);
+ sval = res_get (common_resource, ext_res);
+ }
+ if (sval)
+ rGroup->flagStoreData = atoi (sval);
+ }
+ if (rGroup->flagStoreData == -1)
+ rGroup->flagStoreData = 0;
+
+ if (rGroup->flagStoreKeys == -1)
+ {
+ const char *sval;
+
+ sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
+ if (!(sval = res_get (common_resource, ext_res)))
+ {
+ sprintf (ext_res, "%sstoreKeys", gprefix);
+ sval = res_get (common_resource, ext_res);
+ }
+ if (sval)
+ rGroup->flagStoreKeys = atoi (sval);
+ }
+ if (rGroup->flagStoreKeys == -1)
+ rGroup->flagStoreKeys = 0;
+
+ if (sysno && deleteFlag)
+ fd = -1;
+ else
+ {
+ 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, recType);
+ file_read_stop (fd);
+ if (fd != -1)
+ close (fd);
+ return 1;
+}
+
projects / idzebra-moved-to-github.git / blobdiff