-/* $Id: extract.c,v 1.209 2006-05-10 08:13:21 adam Exp $
- Copyright (C) 1995-2005
+/* $Id: extract.c,v 1.213 2006-05-17 17:46:45 adam Exp $
+ Copyright (C) 1995-2006
Index Data ApS
This file is part of the Zebra server.
#include <direntz.h>
#include <charmap.h>
-#if _FILE_OFFSET_BITS == 64
-#define PRINTF_OFF_T "%Ld"
+#ifdef WIN32
+#define PRINTF_OFF_T "%I64d"
+#else
+/* !WIN32 */
+#if SIZEOF_OFF_T == SIZEOF_LONG_LONG
+#define PRINTF_OFF_T "%lld"
#else
#define PRINTF_OFF_T "%ld"
#endif
+#endif
+
+
#define USE_SHELLSORT 0
#if USE_SHELLSORT
}
}
+static void extract_add_index_string (RecWord *p, const char *str, int length);
+
static void extract_set_store_data_prepare(struct recExtractCtrl *p);
static void extract_init (struct recExtractCtrl *p, RecWord *w)
ctrl->flagShowRecords = !zh->m_flag_rw;
}
+static void all_matches_add(struct recExtractCtrl *ctrl)
+{
+ RecWord word;
+ extract_init(ctrl, &word);
+ word.index_name = "allrecords";
+ word.index_type = 'w';
+ word.seqno = 1;
+ extract_add_index_string (&word, "", 0);
+}
+
static ZEBRA_RES file_extract_record(ZebraHandle zh,
SYSNO *sysno, const char *fname,
int deleteFlag,
RecType recType,
void *recTypeClientData)
{
+ const char *match_str_to_print = "";
RecordAttr *recordAttr;
int r;
const char *matchStr = 0;
}
return ZEBRA_FAIL;
}
+ all_matches_add(&extractCtrl);
if (extractCtrl.match_criteria[0])
- matchStr = extractCtrl.match_criteria;
+ matchStr = extractCtrl.match_criteria;
}
- /* perform match if sysno not known and if match criteria is specified */
+ /* if matchStr is set now - we assume it's printable .
+ For internal matchStr (see below) we don't print */
+ if (matchStr)
+ match_str_to_print = matchStr;
+
+ /* perform internal match if sysno not known and if match criteria is
+ specified already */
if (!sysno)
{
sysnotmp = 0;
*sysno = rec->sysno;
if (zh->records_processed < zh->m_file_verbose_limit)
- if (matchStr)
+ {
yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T
" " ZINT_FORMAT " %s" ,
zh->m_record_type,
- fname, recordOffset, *sysno, matchStr);
- else
- yaz_log(YLOG_LOG, "add %s %s " PRINTF_OFF_T
- " " ZINT_FORMAT ,
- zh->m_record_type,
- fname, recordOffset, *sysno);
-
+ fname, recordOffset, *sysno, match_str_to_print);
+ }
recordAttr = rec_init_attr (zh->reg->zei, rec);
recordAttr->staticrank = extractCtrl.staticrank;
else
{
if (zh->records_processed < zh->m_file_verbose_limit)
- if (matchStr)
+ {
yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T
" " ZINT_FORMAT " %s" ,
zh->m_record_type,
- fname, recordOffset, *sysno, matchStr);
- else
- yaz_log(YLOG_LOG, "delete %s %s " PRINTF_OFF_T
- " " ZINT_FORMAT ,
- zh->m_record_type,
- fname, recordOffset, *sysno);
-
-
-
+ fname, recordOffset, *sysno, match_str_to_print);
+ }
zh->records_deleted++;
if (matchStr)
{
{
/* flush new keys for sort&search etc */
if (zh->records_processed < zh->m_file_verbose_limit)
- if (matchStr)
- yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T
- " " ZINT_FORMAT " %s" ,
- zh->m_record_type,
- fname, recordOffset, *sysno, matchStr);
- else
- yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T
- " " ZINT_FORMAT ,
- zh->m_record_type,
- fname, recordOffset, *sysno);
-
+ {
+ yaz_log(YLOG_LOG, "update %s %s " PRINTF_OFF_T
+ " " ZINT_FORMAT " %s" ,
+ zh->m_record_type,
+ fname, recordOffset, *sysno, match_str_to_print);
+ }
recordAttr->staticrank = extractCtrl.staticrank;
#if NATTR
extract_flushSortKeys (zh, *sysno, 1, zh->reg->sortKeys);
return ZEBRA_FAIL;
}
+ all_matches_add(&extractCtrl);
+
if (extractCtrl.match_criteria[0])
match_criteria = extractCtrl.match_criteria;
return 0;
}
+void extract_rec_keys_adjust(ZebraHandle zh, int is_insert,
+ zebra_rec_keys_t reckeys)
+{
+ ZebraExplainInfo zei = zh->reg->zei;
+ struct ord_stat {
+ int no;
+ int ord;
+ struct ord_stat *next;
+ };
+
+ if (zebra_rec_keys_rewind(reckeys))
+ {
+ struct ord_stat *ord_list = 0;
+ struct ord_stat *p;
+ size_t slen;
+ const char *str;
+ struct it_key key_in;
+ while(zebra_rec_keys_read(reckeys, &str, &slen, &key_in))
+ {
+ int ord = key_in.mem[0];
+
+ for (p = ord_list; p ; p = p->next)
+ if (p->ord == ord)
+ {
+ p->no++;
+ break;
+ }
+ if (!p)
+ {
+ p = xmalloc(sizeof(*p));
+ p->no = 1;
+ p->ord = ord;
+ p->next = ord_list;
+ ord_list = p;
+ }
+ }
+
+ p = ord_list;
+ while (p)
+ {
+ struct ord_stat *p1 = p;
+
+ if (is_insert)
+ zebraExplain_ord_adjust_occurrences(zei, p->ord, p->no, 1);
+ else
+ zebraExplain_ord_adjust_occurrences(zei, p->ord, - p->no, -1);
+ p = p->next;
+ xfree(p1);
+ }
+ }
+}
+
void extract_flushRecordKeys (ZebraHandle zh, SYSNO sysno,
int cmd,
zebra_rec_keys_t reckeys,
{
ZebraExplainInfo zei = zh->reg->zei;
+ extract_rec_keys_adjust(zh, cmd, reckeys);
+
if (!zh->reg->key_buf)
{
int mem= 1024*1024* atoi( res_get_def( zh->res, "memmax", "8"));
zh->reg->key_buf_used +=
key_SU_encode(ch, (char*)zh->reg->key_buf +
zh->reg->key_buf_used);
-
+
/* copy the 0-terminated stuff from str to output */
memcpy((char*)zh->reg->key_buf + zh->reg->key_buf_used, str, slen);
zh->reg->key_buf_used += slen;
}
}
-void extract_add_index_string (RecWord *p, const char *str, int length)
+void extract_add_index_string(RecWord *p, const char *str, int length)
{
struct it_key key;