searchengine/oss/loc/analysisunittest/src/ngramtest.cpp
author hgs
Fri, 15 Oct 2010 12:09:28 +0530
changeset 24 65456528cac2
permissions -rw-r--r--
201041
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
24
hgs
parents:
diff changeset
     1
/*
hgs
parents:
diff changeset
     2
* Copyright (c) 2010 Nokia Corporation and/or its subsidiary(-ies).
hgs
parents:
diff changeset
     3
* All rights reserved.
hgs
parents:
diff changeset
     4
* This component and the accompanying materials are made available
hgs
parents:
diff changeset
     5
* under the terms of "Eclipse Public License v1.0"
hgs
parents:
diff changeset
     6
* which accompanies this distribution, and is available
hgs
parents:
diff changeset
     7
* at the URL "http://www.eclipse.org/legal/epl-v10.html".
hgs
parents:
diff changeset
     8
*
hgs
parents:
diff changeset
     9
* Initial Contributors:
hgs
parents:
diff changeset
    10
* Nokia Corporation - initial contribution.
hgs
parents:
diff changeset
    11
*
hgs
parents:
diff changeset
    12
* Contributors:
hgs
parents:
diff changeset
    13
*
hgs
parents:
diff changeset
    14
* Description: 
hgs
parents:
diff changeset
    15
*
hgs
parents:
diff changeset
    16
*/
hgs
parents:
diff changeset
    17
#include "itk.h"
hgs
parents:
diff changeset
    18
hgs
parents:
diff changeset
    19
#include "thaianalysis.h"
hgs
parents:
diff changeset
    20
hgs
parents:
diff changeset
    21
#include "CLucene.h"
hgs
parents:
diff changeset
    22
hgs
parents:
diff changeset
    23
#include <iostream>
hgs
parents:
diff changeset
    24
hgs
parents:
diff changeset
    25
#include "testutils.h"
hgs
parents:
diff changeset
    26
hgs
parents:
diff changeset
    27
#include "evaluationtool.h"
hgs
parents:
diff changeset
    28
#include "analysisunittest.h"
hgs
parents:
diff changeset
    29
#include "cjkanalyzer.h"
hgs
parents:
diff changeset
    30
#include "ngram.h"
hgs
parents:
diff changeset
    31
hgs
parents:
diff changeset
    32
using namespace analysis; 
hgs
parents:
diff changeset
    33
using namespace evaluationtool; 
hgs
parents:
diff changeset
    34
using namespace lucene::util;
hgs
parents:
diff changeset
    35
hgs
parents:
diff changeset
    36
using namespace lucene::analysis; 
hgs
parents:
diff changeset
    37
hgs
parents:
diff changeset
    38
template<int N>
hgs
parents:
diff changeset
    39
void KoreanNgramAnalyzerTest(Itk::TestMgr* /*testMgr*/) {
hgs
parents:
diff changeset
    40
	CjkNGramAnalyzer analyzer(N);
hgs
parents:
diff changeset
    41
	
hgs
parents:
diff changeset
    42
	Corpus corpus(KOREAN_TEXTCORPUS); 
hgs
parents:
diff changeset
    43
	
hgs
parents:
diff changeset
    44
	for (int i = 0; i < corpus.size(); i++) {
hgs
parents:
diff changeset
    45
		printTokens(analyzer, corpus[i]);
hgs
parents:
diff changeset
    46
	}
hgs
parents:
diff changeset
    47
}
hgs
parents:
diff changeset
    48
hgs
parents:
diff changeset
    49
template<int N>
hgs
parents:
diff changeset
    50
void JamuNgramAnalyzerTest(Itk::TestMgr* /*testMgr*/) {
hgs
parents:
diff changeset
    51
    JamuNGramAnalyzer analyzer(N);
hgs
parents:
diff changeset
    52
    
hgs
parents:
diff changeset
    53
    Corpus corpus(KOREAN_TEXTCORPUS); 
hgs
parents:
diff changeset
    54
    
hgs
parents:
diff changeset
    55
    for (int i = 0; i < corpus.size(); i++) {
hgs
parents:
diff changeset
    56
        printTokens(analyzer, corpus[i]);
hgs
parents:
diff changeset
    57
    }
hgs
parents:
diff changeset
    58
}
hgs
parents:
diff changeset
    59
hgs
parents:
diff changeset
    60
template<int N>
hgs
parents:
diff changeset
    61
void ChinesePrcNgramAnalyzerTest(Itk::TestMgr* /*testMgr*/) {
hgs
parents:
diff changeset
    62
	CjkNGramAnalyzer analyzer(N);
hgs
parents:
diff changeset
    63
	
hgs
parents:
diff changeset
    64
	Corpus corpus(CHINESE_PRC_TEXTCORPUS); 
hgs
parents:
diff changeset
    65
	
hgs
parents:
diff changeset
    66
	for (int i = 0; i < corpus.size(); i++) {
hgs
parents:
diff changeset
    67
		printTokens(analyzer, corpus[i]);
hgs
parents:
diff changeset
    68
	}
hgs
parents:
diff changeset
    69
}
hgs
parents:
diff changeset
    70
hgs
parents:
diff changeset
    71
Itk::TesterBase * CreateNgramAnalyzerUnitTest() 
hgs
parents:
diff changeset
    72
{
hgs
parents:
diff changeset
    73
	using namespace Itk;
hgs
parents:
diff changeset
    74
	SuiteTester
hgs
parents:
diff changeset
    75
		* testSuite = 
hgs
parents:
diff changeset
    76
			new SuiteTester( "ngram" );
hgs
parents:
diff changeset
    77
	
hgs
parents:
diff changeset
    78
	testSuite->add( "korean_1gram", KoreanNgramAnalyzerTest<1>, "korean_1gram" );
hgs
parents:
diff changeset
    79
    testSuite->add( "korean_2gram", KoreanNgramAnalyzerTest<2>, "korean_2gram" );
hgs
parents:
diff changeset
    80
	testSuite->add( "chinese_prc_1gram", ChinesePrcNgramAnalyzerTest<1>, "chinese_prc_1gram" );
hgs
parents:
diff changeset
    81
    testSuite->add( "chinese_prc_2gram", ChinesePrcNgramAnalyzerTest<2>, "chinese_prc_2gram" );
hgs
parents:
diff changeset
    82
    testSuite->add( "jamu_1gram", JamuNgramAnalyzerTest<1>, "jamu_1gram" );
hgs
parents:
diff changeset
    83
    testSuite->add( "jamu_2gram", JamuNgramAnalyzerTest<2>, "jamu_2gram" );
hgs
parents:
diff changeset
    84
    
hgs
parents:
diff changeset
    85
	return testSuite;
hgs
parents:
diff changeset
    86
}