Tinyxml解析过程源码分析

原文链接：Tinyxml解析过程源码分析
发布时间：2016-01-09 09:02:26

tinyxml是一个优秀的，易用的，开源的xml解析库，xml解析的最关键之处，就是如何将xml文件内容解析成内存中的可用、易用的程序数据—DOM(Document Object Model)树。DOM其实就是多叉树，每个节点只需知道自己的第一个子节点（first child）和下一个兄弟节点（next sibling），即可实现元素数据的解析。

有关tinyxml内部的结构设计，本文不详述，网上的已有很多分析，请参见 http://www.cnblogs.com/kex1n/archive/2010/09/23/1833468.html 。本文重点分析tinyxml是如何生成DOM树的过程，完成这个任务的函数就是 TiXmlDocument::LoadFile() 函数，下面分析其代码实现。

1.TiXmlDocument::LoadFile()

打开xml文件

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
/**
 * @brief TiXmlDocument::LoadFile
 * @param _filename  xml文件名
 * @param encoding   文件编码类型
 * @return
 *
 * 这个方法只是打开xml文件，然后调用另一个LoadFile()方法
 *
 */
bool TiXmlDocument::LoadFile( const char* _filename, TiXmlEncoding encoding )
{
	TIXML_STRING filename( _filename );
	value = filename;

	// reading in binary mode so that tinyxml can normalize the EOL
	FILE* file = TiXmlFOpen( value.c_str (), "rb" );

	if ( file )
	{
		bool result = LoadFile( file, encoding );
		fclose( file );
		return result;
	}
	else
	{
		SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
		return false;
	}
}

2.TiXmlDocument::LoadFile() 读取xml到数组，统一换行符为\n

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
/**
 * @brief TiXmlDocument::LoadFile
 * @param file
 * @param encoding
 * @return
 *
 * 1.将文件内容读到一个字符数组中
 * 2.换行符统一替换成\n，文件换行符在不同的系统实现不同，有\n，\r\n，\r三种形式。
 * 3.调用Parse()方法
 */

bool TiXmlDocument::LoadFile( FILE* file, TiXmlEncoding encoding )
{
	if ( !file ) 
	{
		SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
		return false;
	}

	// Delete the existing data:
	Clear();
	location.Clear();

	// Get the file size, so we can pre-allocate the string. HUGE speed impact.
	long length = 0;
	fseek( file, 0, SEEK_END );
	length = ftell( file );
	fseek( file, 0, SEEK_SET );

	// Strange case, but good to handle up front.
	if ( length <= 0 )
	{
		SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
		return false;
	}

	// Subtle bug here. TinyXml did use fgets. But from the XML spec:
	// 2.11 End-of-Line Handling
	// ...the XML processor MUST behave as if it normalized all line breaks in external 
	// parsed entities (including the document entity) on input, before parsing, by translating 
	// both the two-character sequence #xD #xA and any #xD that is not followed by #xA to 
	// a single #xA character.
	//
	// It is not clear fgets does that, and certainly isn't clear it works cross platform. 
	// Generally, you expect fgets to translate from the convention of the OS to the c/unix
	// convention, and not work generally.

	/*
	while( fgets( buf, sizeof(buf), file ) )
	{
		data += buf;
	}
	*/

	char* buf = new char[ length+1 ];
	buf[0] = 0;

	if ( fread( buf, length, 1, file ) != 1 ) {
		delete [] buf;
		SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
		return false;
	}

	// Process the buffer in place to normalize new lines.
	const char* p = buf;	// the read head
	char* q = buf;			// the write head
	const char CR = 0x0d;
	const char LF = 0x0a;

	buf[length] = 0;
	while( *p ) {
		assert( p < (buf+length) );
		assert( q <= (buf+length) );
		assert( q <= p );

		if ( *p == CR ) {
			*q++ = LF;
			p++;
			if ( *p == LF ) {		// check for CR+LF (and skip LF)
				p++;
			}
		}
		else {
			*q++ = *p++;
		}
	}
	assert( q <= (buf+length) );
	*q = 0;

	Parse( buf, 0, encoding ); //解析xml

	delete [] buf;
	return !Error();
}

3.TiXmlDocument::Parse() 解析整个xml文档，生成DOM树

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
/**
 * @brief TiXmlDocument::Parse
 * @param p
 * @param prevData
 * @param encoding
 * @return
 *
 * 完成DOM的建立
 *
 */
const char* TiXmlDocument::Parse( const char* p, TiXmlParsingData* prevData, TiXmlEncoding encoding )
{
	ClearError();

	// Parse away, at the document level. Since a document
	// contains nothing but other tags, most of what happens
	// here is skipping white space.
	if ( !p || !*p )
	{
		SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
		return 0;
	}

	// Note that, for a document, this needs to come
	// before the while space skip, so that parsing
	// starts from the pointer we are given.
	location.Clear();
	if ( prevData )
	{
		location.row = prevData->cursor.row;
		location.col = prevData->cursor.col;
	}
	else
	{
		location.row = 0;
		location.col = 0;
	}
	TiXmlParsingData data( p, TabSize(), location.row, location.col );
	location = data.Cursor();

	if ( encoding == TIXML_ENCODING_UNKNOWN )
	{
		// Check for the Microsoft UTF-8 lead bytes.
		const unsigned char* pU = (const unsigned char*)p;
		if (	*(pU+0) && *(pU+0) == TIXML_UTF_LEAD_0
			 && *(pU+1) && *(pU+1) == TIXML_UTF_LEAD_1
			 && *(pU+2) && *(pU+2) == TIXML_UTF_LEAD_2 )
		{
			encoding = TIXML_ENCODING_UTF8;
			useMicrosoftBOM = true;
		}
	}

    /*这个方法的功能是判断的当前的指针p指向的字符
     *是不是空白字符（即空格或换行符），如果是，则指针
     *前移，找到一个不是空白字符的字符，返回当前的指针位置
     *如果不是，还返回这个指针
     *
     *由此可见这个方法也非常重要，不断地跳过空白字符，不停地解析数据
     */
    p = SkipWhiteSpace( p, encoding );
	if ( !p )
	{
		SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
		return 0;
	}

    /*
     *重点在这里
     *解析xml字符串，直到结束'\0'
     *
     */
	while ( p && *p )
	{
        /*根据头部判断当前的指针指向哪种节点，然后new一个
         *相应的节点，并返回该节点指针，并且设置该节点的父节点为this
         *TiXmlNode是一个基类
         *是对xml的元素、注释、文本、文档声明的抽象
         */
		TiXmlNode* node = Identify( p, encoding );
		if ( node )
		{
            /*下面是多态执行的，不同的节点类型，实现是不同的
             *假设node是一个元素节点，那么这个元素就会有属性
             *就会有子元素等信息，所以要继续解析，因此这个node也有子节点，
             *直到这个节点，这就是多叉树形成的原因
             *直到这个节点内容结束，返回当前位置指针。
             */
			p = node->Parse( p, &data, encoding );

            /*
             *将这个节点，连接到父节点树上
             */
            LinkEndChild( node );
		}
		else
		{
			break;
		}

		// Did we get encoding info?
		if (    encoding == TIXML_ENCODING_UNKNOWN
			 && node->ToDeclaration() )
		{
			TiXmlDeclaration* dec = node->ToDeclaration();
			const char* enc = dec->Encoding();
			assert( enc );

			if ( *enc == 0 )
				encoding = TIXML_ENCODING_UTF8;
			else if ( StringEqual( enc, "UTF-8", true, TIXML_ENCODING_UNKNOWN ) )
				encoding = TIXML_ENCODING_UTF8;
			else if ( StringEqual( enc, "UTF8", true, TIXML_ENCODING_UNKNOWN ) )
				encoding = TIXML_ENCODING_UTF8;	// incorrect, but be nice
			else 
				encoding = TIXML_ENCODING_LEGACY;
		}

		p = SkipWhiteSpace( p, encoding );
	}

	// Was this empty?
	if ( !firstChild ) {
		SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, encoding );
		return 0;
	}

	// All is well.
	return p;
}

4.TiXmlElement::Parse() 解析元素，生成元素多叉树

还有注释类、文档声明类实现的 Parse() 方法，在此省略不述。

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
/**
 * @brief TiXmlElement::Parse
 * @param p
 * @param data
 * @param encoding
 * @return
 *
 *是基类TiXmlNode::Parse()的一种实现，用来解析元素类型的多叉树
 * 
 */
const char* TiXmlElement::Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding )
{
	p = SkipWhiteSpace( p, encoding );
	TiXmlDocument* document = GetDocument();

	if ( !p || !*p )
	{
		if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, 0, 0, encoding );
		return 0;
	}

	if ( data )
	{
		data->Stamp( p, encoding );
		location = data->Cursor();
	}

	if ( *p != '<' )
	{
		if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, p, data, encoding );
		return 0;
	}

	p = SkipWhiteSpace( p+1, encoding );

	// Read the name.
	const char* pErr = p;

    //获取元素名 （value是类成员）
    p = ReadName( p, &value, encoding );
	if ( !p || !*p )
	{
		if ( document )	document->SetError( TIXML_ERROR_FAILED_TO_READ_ELEMENT_NAME, pErr, data, encoding );
		return 0;
	}

    TIXML_STRING endTag ("</");
    //获取这个元素的结束标记
    endTag += value;

	// Check for and read attributes. Also look for an empty
	// tag or an end tag.
	while ( p && *p )
	{
		pErr = p;
		p = SkipWhiteSpace( p, encoding );
		if ( !p || !*p )
		{
			if ( document ) document->SetError( TIXML_ERROR_READING_ATTRIBUTES, pErr, data, encoding );
			return 0;
		}
		if ( *p == '/' )
		{
			++p;
			// Empty tag.
			if ( *p  != '>' )
			{
				if ( document ) document->SetError( TIXML_ERROR_PARSING_EMPTY, p, data, encoding );		
				return 0;
			}
			return (p+1);
		}

        // 读取元素的值
		else if ( *p == '>' )
		{
			// Done with attributes (if there were any.)
			// Read the value -- which can include other
			// elements -- read the end tag, and return.
			++p;
            //有可能这个元素没有值，接着又是子元素，如<Person><Boy>Jim</Boy></Person>
			p = ReadValue( p, data, encoding );		// Note this is an Element method, and will set the error if one happens.
			if ( !p || !*p ) {
				// We were looking for the end tag, but found nothing.
				// Fix for [ 1663758 ] Failure to report error on bad XML
				if ( document ) document->SetError( TIXML_ERROR_READING_END_TAG, p, data, encoding );
				return 0;
			}

			// We should find the end tag now
			// note that:
			// </foo > and
			// </foo> 
			// are both valid end tags.
			if ( StringEqual( p, endTag.c_str(), false, encoding ) )
			{
				p += endTag.length();
				p = SkipWhiteSpace( p, encoding );
				if ( p && *p && *p == '>' ) {
					++p;
					return p;
				}
				if ( document ) document->SetError( TIXML_ERROR_READING_END_TAG, p, data, encoding );
				return 0;
			}
			else
			{
				if ( document ) document->SetError( TIXML_ERROR_READING_END_TAG, p, data, encoding );
				return 0;
			}
		}

        //读取元素的属性
		else
		{
			// Try to read an attribute:
			TiXmlAttribute* attrib = new TiXmlAttribute();
			if ( !attrib )
			{
				return 0;
			}

			attrib->SetDocument( document );
			pErr = p;
			p = attrib->Parse( p, data, encoding );

			if ( !p || !*p )
			{
				if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, pErr, data, encoding );
				delete attrib;
				return 0;
			}

			// Handle the strange case of double attributes:
			#ifdef TIXML_USE_STL
			TiXmlAttribute* node = attributeSet.Find( attrib->NameTStr() );
			#else
			TiXmlAttribute* node = attributeSet.Find( attrib->Name() );
			#endif
			if ( node )
			{
				if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, pErr, data, encoding );
				delete attrib;
				return 0;
			}

			attributeSet.Add( attrib );
		}
	}
	return p;
}

5.TiXmlElement::ReadValue() 读取元素的值，解析子元素

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
/**
 * @brief TiXmlElement::ReadValue
 * @param p
 * @param data
 * @param encoding
 * @return
 * 读取元素的值和解析子元素
 *
 */

const char* TiXmlElement::ReadValue( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding )
{
	TiXmlDocument* document = GetDocument();

	// Read in text and elements in any order.
	const char* pWithWhiteSpace = p;
	p = SkipWhiteSpace( p, encoding );

	while ( p && *p )
	{
		if ( *p != '<' )
		{
			// Take what we have, make a text element.
			TiXmlText* textNode = new TiXmlText( "" );

			if ( !textNode )
			{
			    return 0;
			}

			if ( TiXmlBase::IsWhiteSpaceCondensed() )
			{
				p = textNode->Parse( p, data, encoding );
			}
			else
			{
				// Special case: we want to keep the white space
				// so that leading spaces aren't removed.
				p = textNode->Parse( pWithWhiteSpace, data, encoding );
			}

			if ( !textNode->Blank() )
				LinkEndChild( textNode );
			else
				delete textNode;
		} 

        //一个子元素标签的开始，解析子元素
		else 
		{
			// We hit a '<'
			// Have we hit a new element or an end tag? This could also be
			// a TiXmlText in the "CDATA" style.
			if ( StringEqual( p, "</", false, encoding ) )
			{
				return p;
			}
			else
			{   
				TiXmlNode* node = Identify( p, encoding );
				if ( node )
				{
					p = node->Parse( p, data, encoding );
					LinkEndChild( node );
				}				
				else
				{
					return 0;
				}
			}
		}
		pWithWhiteSpace = p;
		p = SkipWhiteSpace( p, encoding );
	}

	if ( !p )
	{
		if ( document ) document->SetError( TIXML_ERROR_READING_ELEMENT_VALUE, 0, 0, encoding );
	}	
	return p;
}