Mirrors/privacore-open-source-search-engine
1
0
Fork 1
mirror of https://github.com/privacore/open-source-search-engine.git synced 2025-07-15 02:36:08 -04:00
Code Issues Packages Projects Releases Wiki Activity

Remove IndexList.h

Browse Source
This commit is contained in:
Ai Lin Chia
2016-09-05 12:43:11 +02:00
parent cc3ae2b1bc
commit b0fe1fb821
3 changed files with 3 additions and 60 deletions

10
Images.cpp

@ -469,22 +469,16 @@ void Images::gotTermList ( ) {
if ( g_errno ) return;
// check docids in termlist
m_list.resetListPtr();
#if 0
IndexList doesnt know how to deal with modern posdb format, so the check below gives undefined results
// loop over it
for ( ; ! m_list.isExhausted() ; m_list.skipCurrentRecord() ) {
// get the first rec
int64_t d = m_list.getCurrentDocId();
// note it
//log("dup: image is dupped");
// is it us? if so ignore it
if ( d == m_docId ) continue;
if ( Posdb::getDocId(m_list.getCurrentRec()) == m_docId ) continue;
// crap, i guess our image url is not unique. mark it off.
m_errors[i] = EDOCDUP;
// no need to go further
break;
}
#endif
}

3
Images.h

@ -5,7 +5,6 @@
#include "Msg0.h"
#include "Msg13.h"
#include "IndexList.h"
#include "MsgC.h"
#include "SafeBuf.h"
#include "GbFormat.h" // FORMAT_HTML
@ -143,7 +142,7 @@ class Images {
collnum_t m_collnum;
int64_t m_docId;
IndexList m_list;
RdbList m_list;
int32_t m_latestIp;
MsgC m_msgc;

50
IndexList.h

@ -1,50 +0,0 @@
// Matt Wells, copyright Jun 2001
// . IndexList is a list of keys
// . some keys are 12 bytes, some are 6 bytes (compressed)
// . see Indexdb.h for format of the keys
// . you can set from a TitleRec/SiteRec pair
// . you can set from a TermTable
// . we use this class to generate the final indexList for a parsed document
// . we have to do some #include's in the .cpp cuz the TitleRec contains an
// IndexList for holding indexed link Text
// . the TermTable has an addIndexList() function
// . we map 32bit scores from a TermTable to 8bits scores by taking the log
// of the score if it's >= 128*256, otherwise, we keep it as is
// override all funcs in RdbList where m_useShortKeys is true...
// skipCurrentRec() etc need to use m_useHalfKeys in RdbList cuz
// that is needed by many generic routines, like merge_r, RdbMap, Msg1, Msg22..
// We would need to make it a virtual function which would slow things down...
// or make those classes have specialized functions for IndexLists... in
// addition to the RdbLists they already support
#ifndef GB_INDEXLIST_H
#define GB_INDEXLIST_H
#include "RdbList.h"
#include "Titledb.h"
class IndexList : public RdbList {
public:
// these 2 assume 12 and 6 byte keys respectively
int64_t getCurrentDocId () {
if ( isHalfBitOn ( m_listPtr ) ) return getDocId6 (m_listPtr);
else return getDocId12(m_listPtr);
}
int64_t getDocId12 ( char *rec ) {
return ((*(uint64_t *)(rec)) >> 2) & DOCID_MASK; }
int64_t getDocId6 ( char *rec ) {
int64_t docid;
*(int32_t *)(&docid) = *(int32_t *)(void*)rec;
((char *)&docid)[4] = rec[4];
docid >>= 2;
return docid & DOCID_MASK;
}
};
#endif // GB_INDEXLIST_H