author | hgs |
Fri, 15 Oct 2010 12:09:28 +0530 | |
changeset 24 | 65456528cac2 |
parent 7 | a5fbfefd615f |
permissions | -rw-r--r-- |
7
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
1 |
#include "clucene/stdheader.h" |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
2 |
#include "queryscorer.h" |
24 | 3 |
#include <wchar.h> |
7
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
4 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
5 |
CL_NS_DEF2(search,highlight) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
6 |
CL_NS_USE(index) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
7 |
CL_NS_USE(analysis) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
8 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
9 |
QueryScorer::QueryScorer(const Query * query): |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
10 |
_uniqueTermsInFragment(true), |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
11 |
_termsToFind(false,true) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
12 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
13 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
14 |
WeightedTerm** _weighted_terms = QueryTermExtractor::getTerms(query); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
15 |
initialize(_weighted_terms); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
16 |
_CLDELETE_ARRAY(_weighted_terms); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
17 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
18 |
QueryScorer::~QueryScorer() |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
19 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
20 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
21 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
22 |
/* QueryScorer(Query* query, CL_NS(index)::IndexReader* reader, const TCHAR* fieldName) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
23 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
24 |
WeightedTerm** _weighted_terms = QueryTermExtractor.getIdfWeightedTerms(query, reader, fieldName); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
25 |
initialize(_weighted_terms); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
26 |
}*/ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
27 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
28 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
29 |
QueryScorer::QueryScorer(WeightedTerm** weightedTerms) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
30 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
31 |
initialize(weightedTerms); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
32 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
33 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
34 |
void QueryScorer::initialize(WeightedTerm** weightedTerms) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
35 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
36 |
_currentTextFragment = NULL; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
37 |
_totalScore = 0; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
38 |
_maxTermWeight = 0; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
39 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
40 |
// Copy external weighted terms |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
41 |
int i=0; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
42 |
while ( weightedTerms[i] != NULL ){ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
43 |
const WeightedTerm* existingTerm=_termsToFind.get(weightedTerms[i]->getTerm()); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
44 |
if( (existingTerm==NULL) ||(existingTerm->getWeight()<weightedTerms[i]->getWeight()) ) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
45 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
46 |
//if a term is defined more than once, always use the highest scoring weight |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
47 |
WeightedTerm* term = weightedTerms[i]; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
48 |
_termsToFind.put(term->getTerm(), term); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
49 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
50 |
_maxTermWeight=max(_maxTermWeight,weightedTerms[i]->getWeight()); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
51 |
}else |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
52 |
_CLDELETE(weightedTerms[i]); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
53 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
54 |
i++; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
55 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
56 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
57 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
58 |
void QueryScorer::startFragment(TextFragment * newFragment) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
59 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
60 |
_uniqueTermsInFragment.clear(); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
61 |
_currentTextFragment=newFragment; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
62 |
_totalScore=0; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
63 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
64 |
} |
24 | 65 |
/* |
66 |
* Compares the Query term to Field text token if match, returns 1 |
|
67 |
* |
|
68 |
*/ |
|
69 |
||
70 |
int QueryScorer::matchQuryText(const TCHAR* termText) |
|
71 |
{ |
|
72 |
CL_NS(util)::LHashMap<const TCHAR*, const WeightedTerm *, |
|
73 |
CL_NS(util)::Compare::TChar, |
|
74 |
CL_NS(util)::Equals::TChar, |
|
75 |
CL_NS(util)::Deletor::tcArray, |
|
76 |
CL_NS(util)::Deletor::Object<const WeightedTerm> >::iterator i = _termsToFind.begin(); |
|
77 |
for(; i != _termsToFind.end() ; i ++) |
|
78 |
{ |
|
79 |
const TCHAR * temp = i->first; |
|
80 |
int tempLen = wcslen(temp); |
|
81 |
int matchFlag = 1; |
|
82 |
if(tempLen <= wcslen(termText)) |
|
83 |
{ |
|
84 |
for(int j = tempLen - 1 ; j >=0 ; j--) |
|
85 |
{ |
|
86 |
if(temp[j] != termText[j]) |
|
87 |
{ |
|
88 |
matchFlag = 0; |
|
89 |
break; |
|
90 |
} |
|
91 |
} |
|
92 |
} |
|
93 |
else |
|
94 |
{ |
|
95 |
matchFlag = 0; |
|
96 |
} |
|
97 |
if(matchFlag) |
|
98 |
return 1; |
|
99 |
||
100 |
} |
|
101 |
return 0; |
|
102 |
} |
|
103 |
||
7
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
104 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
105 |
float_t QueryScorer::getTokenScore(Token * token) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
106 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
107 |
const TCHAR* termText=token->termText(); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
108 |
|
24 | 109 |
// const WeightedTerm* queryTerm = _termsToFind.get(termText); |
110 |
// Instead of checking for weighted terms directly match qurey text to field text. |
|
111 |
int isQueryTerm = matchQuryText(termText); |
|
112 |
if(isQueryTerm==0) |
|
7
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
113 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
114 |
//not a query term - return |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
115 |
return 0; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
116 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
117 |
//found a query term - is it unique in this doc? |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
118 |
if(_uniqueTermsInFragment.find(termText)==_uniqueTermsInFragment.end()) |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
119 |
{ |
24 | 120 |
//_totalScore+=queryTerm->getWeight(); |
121 |
/* |
|
122 |
* Keeping the Score value to 1 |
|
123 |
* |
|
124 |
*/ |
|
125 |
_totalScore+=1; |
|
7
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
126 |
TCHAR* owned_term = stringDuplicate(termText); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
127 |
_uniqueTermsInFragment.insert(owned_term); |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
128 |
} |
24 | 129 |
//return queryTerm->getWeight(); |
130 |
return 1; |
|
7
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
131 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
132 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
133 |
/** |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
134 |
* |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
135 |
* @return The highest weighted term (useful for passing to GradientFormatter to set |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
136 |
* top end of coloring scale. |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
137 |
*/ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
138 |
float_t QueryScorer::getMaxTermWeight() |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
139 |
{ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
140 |
return _maxTermWeight; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
141 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
142 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
143 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
144 |
float_t QueryScorer::getFragmentScore(){ |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
145 |
return _totalScore; |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
146 |
} |
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
147 |
|
a5fbfefd615f
Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
148 |
CL_NS_END2 |