* Sebastian Hammer, Adam Dickmeiss
*
* $Log: extract.c,v $
- * Revision 1.25 1995-11-16 15:34:54 adam
+ * 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
+ * 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
* Uses new record management system in both indexer and server.
*
* Revision 1.24 1995/11/15 19:13:08 adam
#include <recctrl.h>
#include "index.h"
-#if RECORD_BASE
#include "recindex.h"
-#endif
-
-static Dict file_idx;
+static Dict matchDict;
-#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;
+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 !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)))
+ 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);
}
-#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);
+ 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;
}
p->which = Word_String;
}
-static void wordAdd (const RecWord *p)
+struct recKeys {
+ int buf_used;
+ int buf_max;
+ char *buf;
+} reckeys;
+
+static void addRecordKey (const RecWord *p)
{
- struct it_key key;
+ char *dst;
+ char attrSet;
+ short attrUse;
size_t i;
- if (kused + 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,
- p->attrSet, p->attrUse,
- key_databaseName);
+ 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++)
- ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
- ((char*)key_buf) [kused++] = '\0';
+ *dst++ = p->u.string[i];
+ *dst++ = '\0';
+
+ memcpy (dst, &p->seqno, sizeof(p->seqno));
+ dst += sizeof(p->seqno);
+
break;
default:
- return ;
+ 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++);
+ 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;
}
- ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
- key.sysno = key_sysno;
- key.seqno = p->seqno;
- memcpy ((char*)key_buf + kused, &key, sizeof(key));
- kused += sizeof(key);
+ assert (off == reckeys->buf_used);
}
-static void wordAddAny (const RecWord *p)
+static const char **searchRecordKey (struct recKeys *reckeys,
+ int attrSetS, int attrUseS)
+{
+ static const char *ws[32];
+ int off = 0;
+ int startSeq = -1;
+ int i;
+
+ 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);
+
+ 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)
+ {
+ int woff;
+
+
+ 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;
+}
+
+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)
+
+static int atois (const char **s)
{
- return read (fd, buf, count);
+ int val = 0, c;
+ while ( (c=**s) >= '0' && c <= '9')
+ {
+ val = val*10 + c - '0';
+ ++(*s);
+ }
+ return val;
}
-#endif
-void file_extract (int cmd, const char *fname, const char *kname,
- char *databaseName)
+
+static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
+ const char *fname,
+ const char *recordType,
+ const char *spec)
+{
+ static char dstBuf[2048];
+ char *dst = dstBuf;
+ const char *s = spec;
+ static const char **w;
+ int i;
+
+ 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 (*s == '$')
+ {
+ int spec_len;
+ char special[32];
+ const char *spec_src = NULL;
+ const char *s1 = ++s;
+ while (*s1 && *s1 != ' ' && *s1 != '\t')
+ s1++;
+
+ spec_len = s1 - s;
+ if (spec_len > 31)
+ spec_len = 31;
+ 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 = recordType;
+ else
+ spec_src = NULL;
+ if (spec_src)
+ {
+ strcpy (dst, spec_src);
+ dst += strlen(spec_src);
+ }
+ }
+ 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, "No match criteria for record %s in group %s",
+ fname, rGroup->groupName ? rGroup->groupName : "none");
+ return NULL;
+ }
+ return dstBuf;
+}
+
+int fileExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup,
+ int deleteFlag)
{
+ SYSNO sysnotmp;
int i, r;
+ char gprefix[128];
char ext[128];
- SYSNO sysno;
char ext_res[128];
const char *file_type;
- void *file_info;
+ const char *file_match;
struct recExtractCtrl extractCtrl;
- RecType rt;
+ RecType recType;
+ Record rec;
+ char *matchStr;
+
+ if (!rGroup->groupName || !*rGroup->groupName)
+ *gprefix = '\0';
+ else
+ sprintf (gprefix, "%s.", rGroup->groupName);
- key_databaseName = databaseName;
+ logf (LOG_DEBUG, "fileExtractAdd %s", fname);
+
+ /* determine file extension */
for (i = strlen(fname); --i >= 0; )
if (fname[i] == '/')
{
strcpy (ext, fname+i+1);
break;
}
- sprintf (ext_res, "fileExtension.%s", ext);
+ /* determine file type - depending on extension */
+ sprintf (ext_res, "%sfileExtension.%s", gprefix, ext);
if (!(file_type = res_get (common_resource, ext_res)))
- return;
- if (!(rt = recType_byName (file_type)))
- return;
- logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
- file_info = dict_lookup (file_idx, kname);
- if (!file_info)
+ return 0;
+ if (!(recType = recType_byName (file_type)))
+ return 0;
+
+ /* determine match criteria */
+ sprintf (ext_res, "%sfileMatch.%s", gprefix, ext);
+ file_match = res_get (common_resource, ext_res);
+ if (!file_match)
{
-#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
+ sprintf (ext_res, "%sfileMatch", gprefix);
+ file_match = res_get (common_resource, ext_res);
}
- else
- memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
+ /* 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";
+
+ /* open input file */
if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
{
logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
- return;
+ return 0;
}
+
+ /* extract keys */
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
+ r = (*recType->extract)(&extractCtrl);
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 0;
+ }
+
+ /* perform match if sysno not known and if match criteria is specified */
+
+ matchStr = NULL;
+ if (!sysno && file_match)
+ {
+ char *rinfo;
+
+ sysno = &sysnotmp;
+ matchStr = fileMatchStr(&reckeys, rGroup, fname, file_type,
+ file_match);
+ if (matchStr)
+ {
+ rinfo = dict_lookup (matchDict, matchStr);
+ if (rinfo)
+ memcpy (sysno, rinfo+1, sizeof(*sysno));
+ else
+ *sysno = 0;
+ }
+ else
+ {
+ logf (LOG_WARN, "Record not inserted");
+ return 0;
+ }
+ }
+
+ /* new record ? */
+ if (! *sysno)
+ {
+ if (deleteFlag)
+ {
+ logf (LOG_LOG, "? record %s", fname);
+ return 1;
+ }
+ logf (LOG_LOG, "add record %s", 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++;
+ }
+ 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]);
+ if (deleteFlag)
+ {
+ logf (LOG_LOG, "delete record %s", fname);
+ records_deleted++;
+ rec_del (records, &rec);
+ return 1;
+ }
+ else
+ {
+ logf (LOG_LOG, "update record %s", fname);
+ flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
+ records_updated++;
+ }
+ }
+ 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 (rGroup->databaseName, &rec->size[3]);
+
+ rec_put (records, &rec);
+ return 1;
}
projects / idzebra-moved-to-github.git / blobdiff