author | Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com> |
Tue, 06 Jul 2010 15:30:04 +0300 | |
changeset 10 | afe194b6b1cd |
permissions | -rw-r--r-- |
10
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
1 |
/* |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
2 |
* Copyright (c) 2010 Nokia Corporation and/or its subsidiary(-ies). |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
3 |
* All rights reserved. |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
4 |
* This component and the accompanying materials are made available |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
5 |
* under the terms of "Eclipse Public License v1.0" |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
6 |
* which accompanies this distribution, and is available |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
7 |
* at the URL "http://www.eclipse.org/legal/epl-v10.html". |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
8 |
* |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
9 |
* Initial Contributors: |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
10 |
* Nokia Corporation - initial contribution. |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
11 |
* |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
12 |
* Contributors: |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
13 |
* |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
14 |
* Description: |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
15 |
* |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
16 |
*/ |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
17 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
18 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
19 |
#include "evaluationtool.h" |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
20 |
#include "analysisunittest.h" |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
21 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
22 |
#include "testutils.h" |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
23 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
24 |
#include "cpixstrtools.h" |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
25 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
26 |
#define MAX_LINE_LENGTH 512 |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
27 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
28 |
namespace evaluationtool { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
29 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
30 |
using namespace lucene::analysis; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
31 |
using namespace lucene::util; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
32 |
using namespace lucene::index; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
33 |
using namespace lucene::store; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
34 |
using namespace lucene::search; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
35 |
using namespace lucene::document; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
36 |
using namespace lucene::queryParser; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
37 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
38 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
39 |
static const wchar_t HIT_MARK_CHAR = 'X'; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
40 |
static const int HIT_MARK_IDX = 2; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
41 |
static const wchar_t ERROR_MARK_CHAR = '!'; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
42 |
static const int ERROR_MARK_IDX = 0; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
43 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
44 |
static const wchar_t* SUMMARY_STR = L"--- Summary ---"; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
45 |
static const wchar_t* SECTION_END_STR = L"--- Section End ---"; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
46 |
static const wchar_t* SEARCH_SECTION_STR = L"--- Search ---"; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
47 |
static const wchar_t* FILE_END_STR = L"--- File End ---"; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
48 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
49 |
static const wchar_t* ID_FIELD = L"id"; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
50 |
static const wchar_t* CONTENT_FIELD = L"content"; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
51 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
52 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
53 |
Corpus::Corpus(const char* file) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
54 |
: lines_() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
55 |
FileReader reader(file, "UTF-8"); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
56 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
57 |
wchar_t line[MAX_LINE_LENGTH]; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
58 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
59 |
while (readLine(reader, line, MAX_LINE_LENGTH)) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
60 |
if (wcslen(line)) lines_.push_back( std::wstring( line ) ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
61 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
62 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
63 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
64 |
const wchar_t* Corpus::operator[](int i) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
65 |
return lines_[i].c_str(); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
66 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
67 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
68 |
int Corpus::size() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
69 |
return lines_.size(); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
70 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
71 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
72 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
73 |
#define MAX_ID_LENGTH 10 |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
74 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
75 |
PreparedCorpus::PreparedCorpus(Corpus& corpus, |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
76 |
Analyzer& analyzer, |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
77 |
Analyzer* queryAnalyzer, |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
78 |
Analyzer* prefixAnalyzer) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
79 |
: size_( corpus.size() ), |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
80 |
prefixAnalyzer_( prefixAnalyzer ), |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
81 |
dir_() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
82 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
83 |
dir_.reset( FSDirectory::getDirectory( INDEX_DIRECTORY, true ) ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
84 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
85 |
IndexWriter writer(dir_.get(), &analyzer, true, false); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
86 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
87 |
wchar_t id[MAX_ID_LENGTH]; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
88 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
89 |
for (int i = 0; i < corpus.size(); i++) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
90 |
Document doc; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
91 |
snwprintf(id, MAX_ID_LENGTH, L"%d", i); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
92 |
doc.add(*new Field( ID_FIELD, id, Field::INDEX_NO | Field::STORE_YES)); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
93 |
doc.add(*new Field( CONTENT_FIELD, corpus[i], Field::INDEX_TOKENIZED | Field::STORE_NO)); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
94 |
writer.addDocument(&doc); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
95 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
96 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
97 |
writer.optimize(); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
98 |
writer.close(); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
99 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
100 |
queryParser_.reset(new QueryParser(CONTENT_FIELD, queryAnalyzer ? queryAnalyzer : &analyzer)); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
101 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
102 |
searcher_.reset(new IndexSearcher(dir_.get())); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
103 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
104 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
105 |
int PreparedCorpus::size() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
106 |
return size_; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
107 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
108 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
109 |
int PreparedCorpus::indexSize() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
110 |
std::vector<std::string> v; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
111 |
dir_->list(&v); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
112 |
int ret = 0; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
113 |
for (int i = 0; i < v.size(); i++) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
114 |
ret += dir_->fileLength(v[i].c_str()); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
115 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
116 |
return ret; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
117 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
118 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
119 |
void PreparedCorpus::search(const wchar_t* query, std::bitset<MAXLINES>& hits ) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
120 |
int qlen = wcslen( query ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
121 |
while (qlen > 0 && iswspace(query[qlen-1])) qlen--; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
122 |
auto_ptr<Query> q; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
123 |
if ( query[qlen-1] == '*' && prefixAnalyzer_ ) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
124 |
// Simplified prefix query parser |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
125 |
wchar_t buf[512]; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
126 |
memcpy(buf, query, sizeof(wchar_t)*(qlen-1)); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
127 |
buf[qlen-1] = '\0'; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
128 |
// Assume, that prefix query contains only one word |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
129 |
auto_ptr<TokenStream> t( prefixAnalyzer_->tokenStream(NULL, new StringReader(buf)) ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
130 |
Token token; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
131 |
t->next(&token); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
132 |
Term* term = new Term( CONTENT_FIELD, token.termText() ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
133 |
q.reset( new PrefixQuery( term ) ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
134 |
_CLDECDELETE( term ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
135 |
} else { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
136 |
q.reset( queryParser_->parse(query) ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
137 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
138 |
if ( q.get() ) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
139 |
auto_ptr<Hits> h( searcher_->search( q.get() ) ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
140 |
for (int i = 0; i < h->length(); i++) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
141 |
int id; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
142 |
Cpt::wconvertInteger(&id, h->doc(i).get(ID_FIELD)); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
143 |
hits[id] = true; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
144 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
145 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
146 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
147 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
148 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
149 |
Results::Results(std::bitset<MAXLINES>& hits, int lines) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
150 |
: hits_(hits), lines_(lines) {} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
151 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
152 |
Results::Results() |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
153 |
: hits_(), lines_(0) {} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
154 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
155 |
Results::Results(PreparedCorpus& corpus, |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
156 |
const wchar_t* query) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
157 |
: lines_(corpus.size()) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
158 |
corpus.search(query, hits_); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
159 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
160 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
161 |
bool Results::hit(int i) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
162 |
return hits_[i]; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
163 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
164 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
165 |
void Results::append(bool hit) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
166 |
hits_[lines_++] = hit; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
167 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
168 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
169 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
170 |
int Results::length() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
171 |
return lines_; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
172 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
173 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
174 |
EvaluationRecordEntry::EvaluationRecordEntry( |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
175 |
const wchar_t* query, |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
176 |
Results& ideal, |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
177 |
Results& measured) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
178 |
: query_( query ), |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
179 |
ideal_( ideal ), |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
180 |
measured_( measured ) {} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
181 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
182 |
EvaluationRecordEntry::EvaluationRecordEntry(Reader& reader) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
183 |
wchar_t line[MAX_LINE_LENGTH]; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
184 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
185 |
readLine(reader, line, MAX_LINE_LENGTH); // corpusName |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
186 |
readLine(reader, line, MAX_LINE_LENGTH); // analyzerName |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
187 |
readLine(reader, line, MAX_LINE_LENGTH); // query |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
188 |
wchar_t* cut = line; while (*cut && *cut != ':') cut++; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
189 |
cut++; while (*cut == ' ') cut++; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
190 |
query_ = cut; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
191 |
readLine(reader, line, MAX_LINE_LENGTH); // status |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
192 |
readLine(reader, line, MAX_LINE_LENGTH); // hits |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
193 |
readLine(reader, line, MAX_LINE_LENGTH); // errors |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
194 |
readLine(reader, line, MAX_LINE_LENGTH); // false positives |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
195 |
readLine(reader, line, MAX_LINE_LENGTH); // false negatives |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
196 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
197 |
while (readLine(reader, line, MAX_LINE_LENGTH)) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
198 |
if (wcscmp(line, SECTION_END_STR) == 0) break; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
199 |
bool found = (line[HIT_MARK_IDX] == HIT_MARK_CHAR); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
200 |
bool error = (line[ERROR_MARK_IDX] == ERROR_MARK_CHAR); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
201 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
202 |
measured_.append(found); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
203 |
ideal_.append((!error)?found:!found); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
204 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
205 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
206 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
207 |
EvaluationRecordEntry::EvaluationRecordEntry() |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
208 |
: query_(), ideal_(), measured_() {} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
209 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
210 |
EvaluationRecord::EvaluationRecord(const char* file) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
211 |
: entries_() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
212 |
FileReader reader(file, "UTF-8"); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
213 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
214 |
wchar_t line[MAX_LINE_LENGTH]; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
215 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
216 |
while (readLine(reader, line, MAX_LINE_LENGTH)) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
217 |
// Skip summary |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
218 |
if (wcscmp(line, SUMMARY_STR) == 0) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
219 |
while (readLine(reader, line, MAX_LINE_LENGTH) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
220 |
&& wcscmp(line, SECTION_END_STR) != 0); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
221 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
222 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
223 |
// Eof |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
224 |
if (wcscmp(line, FILE_END_STR) == 0) break; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
225 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
226 |
// Search section |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
227 |
if (wcscmp(line, SEARCH_SECTION_STR) == 0) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
228 |
entries_.push_back( EvaluationRecordEntry( reader ) ); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
229 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
230 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
231 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
232 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
233 |
int EvaluationRecord::length() { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
234 |
return entries_.size(); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
235 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
236 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
237 |
const wchar_t* EvaluationRecord::query(int i) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
238 |
return entries_[i].query_.c_str(); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
239 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
240 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
241 |
Results& EvaluationRecord::ideal(int i) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
242 |
return entries_[i].ideal_; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
243 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
244 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
245 |
Results& EvaluationRecord::measured(int i) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
246 |
return entries_[i].measured_; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
247 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
248 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
249 |
Evaluation::Evaluation(Results& ideal, Results& measured) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
250 |
: ideal_( ideal ), |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
251 |
measured_( measured ) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
252 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
253 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
254 |
bool Evaluation::falsePositive(int line) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
255 |
return (!ideal_.hit(line))&&measured_.hit(line); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
256 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
257 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
258 |
bool Evaluation::falseNegative(int line) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
259 |
return ideal_.hit(line)&&(!measured_.hit(line)); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
260 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
261 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
262 |
bool Evaluation::error(int line) |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
263 |
{ |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
264 |
return (ideal_.hit(line)!=measured_.hit(line)?1:0); |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
265 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
266 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
267 |
int Evaluation::errors() |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
268 |
{ |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
269 |
int ret = 0; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
270 |
for (int i = 0; i < ideal_.length(); i++) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
271 |
if (error(i)) ret++; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
272 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
273 |
return ret; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
274 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
275 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
276 |
int Evaluation::falsePositives() |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
277 |
{ |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
278 |
int ret = 0; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
279 |
for (int i = 0; i < ideal_.length(); i++) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
280 |
if (falsePositive(i)) ret++; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
281 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
282 |
return ret; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
283 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
284 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
285 |
int Evaluation::falseNegatives() |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
286 |
{ |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
287 |
int ret = 0; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
288 |
for (int i = 0; i < ideal_.length(); i++) { |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
289 |
if (falseNegative(i)) ret++; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
290 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
291 |
return ret; |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
292 |
} |
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
293 |
|
afe194b6b1cd
Revision: 201025
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff
changeset
|
294 |
} |