* Sebastian Hammer, Adam Dickmeiss
*
* $Log: extract.c,v $
- * Revision 1.26 1995-11-20 11:56:24 adam
+ * 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
+ * New update method: the 'old' keys are saved for each records.
+ *
+ * Revision 1.26 1995/11/20 11:56:24 adam
* Work on new traversal.
*
* Revision 1.25 1995/11/16 15:34:54 adam
#include <recctrl.h>
#include "index.h"
-#if RECORD_BASE
#include "recindex.h"
-#endif
static Dict file_idx;
-
-#if RECORD_BASE
static Records records = NULL;
-#else
-static int sys_idx_fd = -1;
-static SYSNO sysno_next;
-#endif
-static int key_cmd;
-static int key_sysno;
-static char *key_databaseName;
static char **key_buf;
static size_t ptr_top;
static size_t ptr_i;
-static size_t kused;
+static size_t key_buf_used;
static int key_file_no;
void key_open (int mem)
{
-#if !RECORD_BASE
- void *file_key;
-#endif
if (mem < 50000)
mem = 50000;
key_buf = xmalloc (mem);
ptr_top = mem/sizeof(char*);
ptr_i = 0;
- kused = 0;
+
+ key_buf_used = 0;
key_file_no = 0;
if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
exit (1);
}
-#if RECORD_BASE
assert (!records);
records = rec_open (1);
-#else
- file_key = dict_lookup (file_idx, ".");
- if (file_key)
- memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
- else
- sysno_next = 1;
- if ((sys_idx_fd = open (FNAME_SYS_IDX, O_RDWR|O_CREAT, 0666)) == -1)
- {
- logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
- exit (1);
- }
-#endif
}
struct encode_info {
}
logf (LOG_LOG, "finished section %d", key_file_no);
ptr_i = 0;
- kused = 0;
+ key_buf_used = 0;
}
int key_close (void)
{
key_flush ();
xfree (key_buf);
-#if RECORD_BASE
rec_close (&records);
-#else
- close (sys_idx_fd);
- dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
-#endif
dict_close (file_idx);
+
return key_file_no;
}
p->which = Word_String;
}
+struct recKeys {
+ int buf_used;
+ int buf_max;
+ char *buf;
+} reckeys;
+
+static void addRecordKey (const RecWord *p)
+{
+ char *dst;
+ char attrSet;
+ short attrUse;
+ size_t i;
+
+ if (reckeys.buf_used+1024 > reckeys.buf_max)
+ {
+ char *b;
+
+ b = malloc (reckeys.buf_max += 65000);
+ if (reckeys.buf_used > 0)
+ memcpy (b, reckeys.buf, reckeys.buf_used);
+ free (reckeys.buf);
+ reckeys.buf = b;
+ }
+ dst = reckeys.buf + reckeys.buf_used;
+ switch (p->which)
+ {
+ case Word_String:
+ attrSet = p->attrSet;
+ memcpy (dst, &attrSet, sizeof(attrSet));
+ dst += sizeof(attrSet);
+
+ attrUse = p->attrUse;
+ 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;
+ }
+ reckeys.buf_used = dst - reckeys.buf;
+}
+
+static void flushRecordKeys (SYSNO sysno, int cmd, struct recKeys *reckeys,
+ const char *databaseName)
+{
+ int off = 0;
+ 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);
+
+ 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);
+ while (*src)
+ ((char*)key_buf) [key_buf_used++] = index_char_cvt (*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);
+ key.sysno = sysno;
+ memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
+ key_buf_used += sizeof(key);
+ off = src - reckeys->buf;
+ }
+ 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)
{
struct it_key key;
size_t i;
- if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
+ if (key_buf_used + 1024 > (ptr_top-ptr_i)*sizeof(char*))
key_flush ();
++ptr_i;
- key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
- kused += index_word_prefix ((char*)key_buf + kused,
+ 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) [kused++] = index_char_cvt (p->u.string[i]);
- ((char*)key_buf) [kused++] = '\0';
+ ((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)[kused++] = ((key_cmd == 'a') ? 1 : 0);
+ ((char*) key_buf)[key_buf_used++] = ((key_cmd == 'a') ? 1 : 0);
key.sysno = key_sysno;
key.seqno = p->seqno;
- memcpy ((char*)key_buf + kused, &key, sizeof(key));
- kused += sizeof(key);
+ memcpy ((char*)key_buf + key_buf_used, &key, sizeof(key));
+ key_buf_used += sizeof(key);
+
+ if (key_cmd == 'a' && key_del_used >= 0)
+ {
+ 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)
+ {
+ 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);
+
+ 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);
+ }
}
-static void wordAddAny (const RecWord *p)
+#endif
+
+static void addRecordKeyAny (const RecWord *p)
{
if (p->attrSet != 1 || p->attrUse != 1016)
{
memcpy (&w, p, sizeof(w));
w.attrSet = 1;
w.attrUse = 1016;
- wordAdd (&w);
+ addRecordKey (&w);
}
- wordAdd (p);
+ addRecordKey (p);
}
-
-#define FILE_READ_BUF 1
-#if FILE_READ_BUF
static char *file_buf;
static int file_offset;
static int file_bufsize;
file_offset += count;
return count;
}
-#else
-static int file_read (int fd, char *buf, size_t count)
-{
- return read (fd, buf, count);
-}
-#endif
-SYSNO file_extract (int cmd, const char *fname, const char *kname,
- char *databaseName)
+
+int fileExtract (SYSNO *sysno, const char *fname, const char *databaseName,
+ int deleteFlag)
{
int i, r;
char ext[128];
- SYSNO sysno;
char ext_res[128];
const char *file_type;
- void *file_info;
struct recExtractCtrl extractCtrl;
RecType rt;
+ Record rec;
+
+ logf (LOG_DEBUG, "fileExtractAdd %s", fname);
- key_databaseName = databaseName;
for (i = strlen(fname); --i >= 0; )
if (fname[i] == '/')
{
return 0;
if (!(rt = recType_byName (file_type)))
return 0;
- logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
- file_info = dict_lookup (file_idx, kname);
- if (!file_info)
- {
-#if RECORD_BASE
- Record rec = rec_new (records);
-
- sysno = rec->sysno;
- dict_insert (file_idx, kname, sizeof(sysno), &sysno);
- rec->info[0] = rec_strdup (file_type);
- rec->info[1] = rec_strdup (kname);
- rec_put (records, rec);
- rec_rm (rec);
-#else
- sysno = sysno_next++;
- dict_insert (file_idx, kname, sizeof(sysno), &sysno);
- lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
- write (sys_idx_fd, file_type, strlen (file_type)+1);
- write (sys_idx_fd, kname, strlen(kname)+1);
-#endif
- }
- else
- memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
{
logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
return 0;
}
+
extractCtrl.subType = "";
extractCtrl.init = wordInit;
- extractCtrl.add = wordAddAny;
-#if FILE_READ_BUF
+ extractCtrl.add = addRecordKeyAny;
+
+ reckeys.buf_used = 0;
file_read_start (extractCtrl.fd);
-#endif
extractCtrl.readf = file_read;
- key_sysno = sysno;
- key_cmd = cmd;
r = (*rt->extract)(&extractCtrl);
-#if FILE_READ_BUF
file_read_stop (extractCtrl.fd);
-#endif
close (extractCtrl.fd);
- if (r)
+
+ if (r)
+ {
logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
- return sysno;
+ return 0;
+ }
+ if (! *sysno) /* match criteria */
+ {
+ logf (LOG_LOG, "add record %s", fname);
+ rec = rec_new (records);
+ *sysno = rec->sysno;
+
+ flushRecordKeys (*sysno, 1, &reckeys, databaseName);
+ }
+ 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);
+ }
+ free (rec->info[0]);
+ rec->info[0] = rec_strdup (file_type, &rec->size[0]);
+
+ free (rec->info[1]);
+ rec->info[1] = rec_strdup (fname, &rec->size[1]);
+
+ free (rec->info[2]);
+ if (reckeys.buf_used > 0)
+ {
+ rec->info[2] = malloc (reckeys.buf_used);
+ rec->size[2] = reckeys.buf_used;
+ memcpy (rec->info[2], reckeys.buf, rec->size[2]);
+ }
+ else
+ {
+ rec->info[2] = NULL;
+ rec->size[2] = 0;
+ }
+ free (rec->info[3]);
+ rec->info[3] = rec_strdup (databaseName, &rec->size[3]);
+
+ rec_put (records, &rec);
+ return 1;
}
projects / idzebra-moved-to-github.git / blobdiff