searchengine/oss/cl/clucene/src/clucene/highlighter/SimpleHTMLFormatter.h
author Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
Fri, 11 Jun 2010 14:43:47 +0300
changeset 7 a5fbfefd615f
permissions -rw-r--r--
Revision: 201021 Kit: 2010123
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
7
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     1
/**
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     2
 * Copyright 2002-2004 The Apache Software Foundation
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     3
 *
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     4
 * Licensed under the Apache License, Version 2.0 (the "License");
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     5
 * you may not use this file except in compliance with the License.
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     6
 * You may obtain a copy of the License at
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     7
 *
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     8
 *     http://www.apache.org/licenses/LICENSE-2.0
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
     9
 *
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    10
 * Unless required by applicable law or agreed to in writing, software
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    11
 * distributed under the License is distributed on an "AS IS" BASIS,
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    12
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    13
 * See the License for the specific language governing permissions and
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    14
 * limitations under the License.
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    15
 */
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    16
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    17
#ifndef _lucene_search_highlight_simplehtmlformatter_
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    18
#define _lucene_search_highlight_simplehtmlformatter_
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
#if defined(_LUCENE_PRAGMA_ONCE)
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    21
# pragma once
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    22
#endif
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
#include "CLucene/util/StringBuffer.h"
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    25
#include "CLucene/highlighter/Formatter.h"
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
CL_NS_DEF2(search,highlight)
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
/**
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    30
 * Simple {@link Formatter} implementation to highlight terms with a pre and post tag
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    31
 *
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
class SimpleHTMLFormatter :public Formatter
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    34
{
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    35
private:
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    36
	const TCHAR* _preTag;
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    37
	const TCHAR* _postTag;
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    38
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    39
public:
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    40
	~SimpleHTMLFormatter(); 
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    41
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    42
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    43
	SimpleHTMLFormatter(const TCHAR* preTag, const TCHAR* postTag);
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    44
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
	 * Default constructor uses HTML: &lt;B&gt; tags to markup terms
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    47
	 * 
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    48
	 **/
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    49
	SimpleHTMLFormatter();
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    50
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    51
	
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    52
	/**
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    53
	* Returns the original text enclosed in _preTag and _postTag, if the score is greater 
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    54
	* than 0. Otherwise, it returns the original text.
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    55
	* It doesn't use the stemmed text nor the startOffset. 
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    56
	* It allocates memory for the returned text, and it has to be freed by the caller.
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
	TCHAR* highlightTerm(const TCHAR* originalText, const TokenGroup* tokenGroup);
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
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    61
CL_NS_END2
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    62
a5fbfefd615f Revision: 201021
Dremov Kirill (Nokia-D-MSW/Tampere) <kirill.dremov@nokia.com>
parents:
diff changeset
    63
#endif