3_14_GLib库入门与实践_xml解析

简介

一个简单的xml如下：

<xml>
<node attr1="value1" attr2="value2">aaa</node>
</xml>

其中：

• <xml>和<node>分别是开始标签（或者叫开始节点）
• <xml>没有属性，<node>有两个属性
• node的attr1属性值为value1
• node的attr2属性值为value2
• xml标签没有值，但它包含子标签
• node标签的值为aaa
• </xml>和</node>是结束标签

xml有多种解析方式，最常见的是DOM方式和SAX方式。下面是两种解析方式的对比。

SAX（Simple API for XML）

• 只能读，不能修改，只能顺序访问，适合解析大型XML，解析速度快
• 常应用于处理大量数据的XML，实现异构系统的数据访问，实现跨平台
• 从文档的开始通过每一节点移动，定位一个特定的节点

DOM（Document Object Model）

• 不仅能读，还能修改，而且能够实现随机访问，缺点是解析速度慢，适合解析小型文档
• 一般应用与小型的配置XML，方便操作
• 为载入到内存的文档节点建立类型描述，呈现可横向移动、潜在巨大的树型结构
• 在内存中生成节点树操作代价昂贵

GLib库提供的解析方式为SAX，官方文档建议在解析简单xml时使用，如果解析非常复杂的xml时，可选用别的xml解析器。

GLib对xml数据格式的约定如下：

• 只允许UTF-8编码
• 没有用户定义的实体
• 处理指令、注释和doctype声明是“传递的”，但不会以任何方式进行解释
• 没有DTD或验证

GLib库支持的xml数据格式如下：

• 元素
• 属性
• 5个标准实体：&<>“'
• 字符引用
• 标记为CDATA的部分

数据结构

GLib库的SAX解析器的核心是GMarkupParser结构体，该结构体内有五个函数指针

struct GMarkupParser {
  void (*start_element)(xxx);
  void (*end_element)(xxx);
  void (*text)(xxx);
  void (*passthrough)(xxx);
  void (*error)(xxx);
}

• start_element：解析到开始标签时调用本回调函数
• end_element：解析到结束标签时调用本回调函数
• text：解析开始标签和结束标签中的值（如果标签下没有值或只有子标签，则text为空）
• passthrough：解析到注释时调用本回调函数
• error：解析出错或者前面解析步骤设置error变量时调用本回调函数

函数列表

gchar * 	g_markup_escape_text ()
gchar * 	g_markup_printf_escaped ()
gchar * 	g_markup_vprintf_escaped ()
GMarkupParseContext * 	g_markup_parse_context_new ()
gboolean 	g_markup_parse_context_parse ()
gboolean 	g_markup_parse_context_end_parse ()
void 	g_markup_parse_context_free ()
void 	g_markup_parse_context_get_position ()
const gchar * 	g_markup_parse_context_get_element ()
const GSList * 	g_markup_parse_context_get_element_stack ()
gpointer 	g_markup_parse_context_get_user_data ()
void 	g_markup_parse_context_push ()
gpointer 	g_markup_parse_context_pop ()
GMarkupParseContext * 	g_markup_parse_context_ref ()
void 	g_markup_parse_context_unref ()
gboolean 	g_markup_collect_attributes ()

函数功能分类

// 创建解析器
GMarkupParseContext * 	g_markup_parse_context_new ()

// 释放解析器
void 	g_markup_parse_context_free ()

// 解析
gboolean 	g_markup_parse_context_parse ()
gboolean 	g_markup_parse_context_end_parse ()
void 	g_markup_parse_context_push ()
gpointer 	g_markup_parse_context_pop ()

// 引用解引用
GMarkupParseContext * 	g_markup_parse_context_ref ()
void 	g_markup_parse_context_unref ()

// 属性收集
gboolean 	g_markup_collect_attributes ()

// 解析位置获取
void 	g_markup_parse_context_get_position ()

// 标签获取
const gchar * 	g_markup_parse_context_get_element ()

// 标签栈获取
const GSList * 	g_markup_parse_context_get_element_stack ()

// 用户数据获取
gpointer 	g_markup_parse_context_get_user_data ()

// 转义字符处理
gchar * 	g_markup_escape_text ()
gchar * 	g_markup_printf_escaped ()
gchar * 	g_markup_vprintf_escaped ()

函数功能说明及综合演示

基本用法

待解析xml内容：

<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text100</element_name>
<element_name attribute_name3="attribute_value3" attribute_name4="attribute_value4">text200</element_name>
</elements>
</xml>

本xml的elements节点下有两个子节点，每个节点又有两个属性及属性值，使用GLib对该xml解析，程序如下。
解析程序：
源码见glib_examples\glib_markup\glib_markup_basic

#include <glib.h>

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;

    g_printf("(start)element_name: %s \n", element_name);
    while (*name_cursor) {
        g_printf("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(end)element_name: %s \n", element_name);
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(text)text:%s \n", text);
} 

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = NULL, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 

    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_basic.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf) ; 

    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_basic]# ./glib_markup_basic
<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text100</element_name>
<element_name attribute_name3="attribute_value3" attribute_name4="attribute_value4">text200</element_name>
</elements>
</xml>
(start)element_name: xml
(text)text:

(start)element_name: head
(end)element_name: head
(text)text:

(start)element_name: element_num
(text)text:2
(end)element_name: element_num
(text)text:

(start)element_name: elements
(text)text:

(start)element_name: element_name
(start)name_cursor:attribute_name1, value_cursor:attribute_value1
(start)name_cursor:attribute_name2, value_cursor:attribute_value2
(text)text:text100
(end)element_name: element_name
(text)text:

(start)element_name: element_name
(start)name_cursor:attribute_name3, value_cursor:attribute_value3
(start)name_cursor:attribute_name4, value_cursor:attribute_value4
(text)text:text200
(end)element_name: element_name
(text)text:

(end)element_name: elements
(text)text:

(end)element_name: xml

以简单节点<element_num>2</element_num>为例，说明一下SXA方式解析xml的过程。

1. 出现element_num开始标签（<element_num>），调用start回调函数开始解析；
2. 由于element_num标签下有值，因此会走到text函数，并解析出标签的值为2；
3. 出现element_num结束标签（</element_num>），调用end回调函数，解析结束。

elements下面有两个子节点，节点名均为element_name，每个子节点都会被解析到，当解析到子节点时，会取出子节点的属性标签（attribute_namex）、属性值（attribute_value1）和子节点值（两个子节点的值均为text）。

不同节点下子节点名称相同的解析方法

待解析xml内容：

<Glib>
        <CoreApplications>
                <Module version="2.48" name="Main Event Loop">manages all available sources of events</Module>
                <Module version="2.48" name="Thread Pools">pools of threads to execute work concurrently</Module>
        </CoreApplications>
        <Utilities>
                <Module version="2.48" name="File Utilities">various file-related functions</Module>
                <Module version="2.48" name="Simple XML Subset Parser">parses a subset of XML</Module>
        </Utilities>
        <DataTypes>
                <Module version="2.48" name="Sequences">scalable lists</Module>
                <Module version="2.48" name="Byte Arrays">arrays of bytes</Module>
        </DataTypes>
</Glib>

上面xml，CoreApplications、Utilities和DataTypes三个节点下面的子节点名字一模一样，只有name属性的值不一样，最终想要的结果是解析各Module子节点的name属性值，然后解析各Module子节点的节点值，将两者一一对应。下面是对该xml解析的程序示例。
解析程序：
源码见glib_examples\glib_markup\glib_markup_same_node_name

#include <stdio.h>
#include <glib.h>

#define TEST_GMARKUP_CORE_APPS "CoreApplications"
#define TEST_GMARKUP_UTILITIES "Utilities"
#define TEST_GMARKUP_DATATYPES "DataTypes"

#define TEST_GMARKUP_MODULE "Module"

gboolean g_core_flag = FALSE;
gboolean g_util_flag = FALSE;
gboolean g_data_flag = FALSE;

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;
#if 0
    g_printf("(start)element_name: %s \n", element_name);
    while (*name_cursor) {
        g_printf("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
#endif
#if 1
    if(0 == strcmp(element_name, TEST_GMARKUP_CORE_APPS)) {
        g_print("(start)start parse of core \n");
        g_core_flag = TRUE;
    }

    if(0 == strcmp(element_name, TEST_GMARKUP_UTILITIES)) {
        g_print("(start)start parse of util \n");
        g_util_flag = TRUE;
    }

    if(0 == strcmp(element_name, TEST_GMARKUP_DATATYPES)) {
        g_print("(start)start parse of data \n");
        g_data_flag = TRUE;
    }

    if(0 == strcmp(element_name, TEST_GMARKUP_MODULE)){
        if(g_core_flag) {
            while (*name_cursor) {
                g_print("(start)core: name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
                name_cursor++;
                value_cursor++;
            }
        }

        if(g_util_flag) {
            while (*name_cursor) {
                g_print("(start)util: name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
                name_cursor++;
                value_cursor++;
            }
        }

        if(g_data_flag) {
            while (*name_cursor) {
                g_print("(start)data: name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
                name_cursor++;
                value_cursor++;
            }
        }
        
    }
#endif
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
#if 0
    g_printf("(end)element_name: %s \n", element_name);
#endif
#if 1
    if(0 == strcmp(element_name, TEST_GMARKUP_CORE_APPS)) {
        g_print("(end)end parse of core \n");
        g_core_flag = FALSE;
    }
    if(0 == strcmp(element_name, TEST_GMARKUP_UTILITIES)) {
        g_print("(end)end parse of util \n");
        g_util_flag = FALSE;
    }
    if(0 == strcmp(element_name, TEST_GMARKUP_DATATYPES)) {
        g_print("(end)end parse of data \n");
        g_data_flag = FALSE;
    }
#endif
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
#if 0
    g_printf("(text)text:%s \n", text);
#endif
#if 1
    if(g_core_flag) {
        g_print("(text)core text:%s \n", text);
    }
    if(g_util_flag) {
        g_print("(text)util text:%s \n", text);
    }
    if(g_data_flag) {
        g_print("(text)data text:%s \n", text);
    }
#endif
} 

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = NULL, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 
    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_same_node_name.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf) ; 
    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_same_node_name]# ./glib_markup_same_node_name
<Glib>
        <CoreApplications>
                <Module version="2.48" name="Main Event Loop">manages all available sources of events</Module>
                <Module version="2.48" name="Thread Pools">pools of threads to execute work concurrently</Module>
        </CoreApplications>
        <Utilities>
                <Module version="2.48" name="File Utilities">various file-related functions</Module>
                <Module version="2.48" name="Simple XML Subset Parser">parses a subset of XML</Module>
        </Utilities>
        <DataTypes>
                <Module version="2.48" name="Sequences">scalable lists</Module>
                <Module version="2.48" name="Byte Arrays">arrays of bytes</Module>
        </DataTypes>
</Glib>
(start)start parse of core
(text)core text:

(start)core: name_cursor:version, value_cursor:2.48
(start)core: name_cursor:name, value_cursor:Main Event Loop
(text)core text:manages all available sources of events
(text)core text:

(start)core: name_cursor:version, value_cursor:2.48
(start)core: name_cursor:name, value_cursor:Thread Pools
(text)core text:pools of threads to execute work concurrently
(text)core text:

(end)end parse of core
(start)start parse of util
(text)util text:

(start)util: name_cursor:version, value_cursor:2.48
(start)util: name_cursor:name, value_cursor:File Utilities
(text)util text:various file-related functions
(text)util text:

(start)util: name_cursor:version, value_cursor:2.48
(start)util: name_cursor:name, value_cursor:Simple XML Subset Parser
(text)util text:parses a subset of XML
(text)util text:

(end)end parse of util
(start)start parse of data
(text)data text:

(start)data: name_cursor:version, value_cursor:2.48
(start)data: name_cursor:name, value_cursor:Sequences
(text)data text:scalable lists
(text)data text:

(start)data: name_cursor:version, value_cursor:2.48
(start)data: name_cursor:name, value_cursor:Byte Arrays
(text)data text:arrays of bytes
(text)data text:

(end)end parse of data

一个节点的解析被分为了三个回调函数（start、end和text），因此，本程序使用全局变量的方式在start和text之间传递消息，使节点的属性值和节点自身值可以正确关联。

三种注释及解析方法

xml有三种注释方法，单行注释，多行注释和CDATA注释，其中最后一种，CDATA，里面可以是任意内容，可以包含数据、注释、甚至是转义字符。

待解析xml内容：

<xml>
<head/>
<element_num>2</element_num>
<!-- single line comment  -->
<elements>
<!--
  multiple lines comment 1
  multiple lines comment 2
-->
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
<![CDATA[
        <!-- include other comment format -->
  multiple cdata comment 1
  multiple cdata comment 2
  <!--
    comment line 1
    comment line 2
  -->
]]>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
</elements>
</xml>

GLib的xml注释解析在passthrough函数完成。
解析程序：
源码见glib_examples\glib_markup\glib_markup_comment

#include <glib.h>

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;

    g_printf("(start)element_name: %s \n", element_name);
    while (*name_cursor) {
        g_printf("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(end)element_name: %s \n", element_name);
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(text)text:%s \n", text);
}

static void passthrough( GMarkupParseContext *context, 
        const gchar *passthrough_text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(passthrough)passthrough:%s \n", passthrough_text);
}

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = passthrough, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 
    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_comment.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf);

    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_comment]# ./glib_markup_comment
<xml>
<head/>
<element_num>2</element_num>
<!-- single line comment  -->
<elements>
<!--
  multiple lines comment 1
  multiple lines comment 2
-->
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
<![CDATA[
        <!-- include other comment format -->
  multiple cdata comment 1
  multiple cdata comment 2
  <!--
    comment line 1
    comment line 2
  -->
]]>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
</elements>
</xml>
(start)element_name: xml
(text)text:

(start)element_name: head
(end)element_name: head
(text)text:

(start)element_name: element_num
(text)text:2
(end)element_name: element_num
(text)text:

(passthrough)passthrough:<!-- single line comment  -->
(text)text:

(start)element_name: elements
(text)text:

(passthrough)passthrough:<!--
  multiple lines comment 1
  multiple lines comment 2
-->
(text)text:

(start)element_name: element_name
(start)name_cursor:attribute_name1, value_cursor:attribute_value1
(start)name_cursor:attribute_name2, value_cursor:attribute_value2
(text)text:text
(end)element_name: element_name
(text)text:

(passthrough)passthrough:<![CDATA[
        <!-- include other comment format -->
  multiple cdata comment 1
  multiple cdata comment 2
  <!--
    comment line 1
    comment line 2
  -->
]]>
(text)text:

(start)element_name: element_name
(start)name_cursor:attribute_name1, value_cursor:attribute_value1
(start)name_cursor:attribute_name2, value_cursor:attribute_value2
(text)text:text
(end)element_name: element_name
(text)text:

(end)element_name: elements
(text)text:

(end)element_name: xml

解析器出错演示

解析过程中可能会出现解析失败的情况，此时可以在error回调函数里打印出错信息。
待解析xml内容：

<xml>
<head/>
<element_num>2</element>
<elements>
<element_name>text1</element_name>
<element_name>text2</element_name>
</elements>
</xml>

解析程序：
源码见glib_examples\glib_markup\glib_markup_error

#include <glib.h>

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;

    g_printf("(start)element_name: %s \n", element_name);
    while (*name_cursor) {
        g_printf("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(end)element_name: %s \n", element_name);
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(text)text:%s \n", text);
} 

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = NULL, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 
    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_error.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf) ; 

    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_error]# ./glib_markup_error
<xml>
<head/>
<element_num>2</element>
<elements>
<element_name>text1</element_name>
<element_name>text2</element_name>
</elements>
</xml>
(start)element_name: xml
(text)text:

(start)element_name: head
(end)element_name: head
(text)text:

(start)element_name: element_num
(text)text:2
(err)parse xml failed: Error on line 3 char 25: Element 'element' was closed, but the currently open element is 'element_num'
context parse failed!

error回调函数有xml解析失败的详细原因，如果不设置，则解析失败时只提示失败，不会给解析错误原因。

将上面的例子，parser变量的error成员值置空，再次运行，可以看到，没有解析错误原因。

[root@centos7_6 glib_markup_error]# ./glib_markup_error
<xml>
<head/>
<element_num>2</element>
<elements>
<element_name>text1</element_name>
<element_name>text2</element_name>
</elements>
</xml>
(start)element_name: xml
(text)text:

(start)element_name: head
(end)element_name: head
(text)text:

(start)element_name: element_num
(text)text:2
context parse failed!

子解析器

解析器支持嵌套，使用g_markup_parse_context_push和g_markup_parse_context_pop函数可以将当前解析器入栈，此时就可以开始新的解析器了。以下面xml为例，当解析到<Utilities>节点时，在start回调函数将当前解析器上下文压栈，使用新的解析器开始解析Utilities的子节点。Utilities解析完，在end回调函数将当前解析器上下文出栈。待解析xml和程序见下。

待解析xml内容：

<Glib>
        <CoreApplications>
                <Module version="2.48" name="Main Event Loop">manages all available sources of events</Module>
                <Module version="2.48" name="Thread Pools">pools of threads to execute work concurrently</Module>
        </CoreApplications>
        <Utilities>
                <Module version="2.48" name="File Utilities">various file-related functions</Module>
                <Module version="2.48" name="Simple XML Subset Parser">parses a subset of XML</Module>
        </Utilities>
        <DataTypes>
                <Module version="2.48" name="Sequences">scalable lists</Module>
                <Module version="2.48" name="Byte Arrays">arrays of bytes</Module>
        </DataTypes>
</Glib>

解析程序：
源码见glib_examples\glib_markup\glib_markup_sub_parse

#include <stdio.h>
#include <glib.h>

#define TEST_GMARKUP_CORE_APPS "CoreApplications"
#define TEST_GMARKUP_UTILITIES "Utilities"
#define TEST_GMARKUP_DATATYPES "DataTypes"

#define TEST_GMARKUP_MODULE "Module"

gboolean g_core_flag = FALSE;
gboolean g_util_flag = FALSE;
gboolean g_data_flag = FALSE;

static void sub_start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;
#if 1
    g_print("(sub_start)element_name: %s \n", element_name);
    while (*name_cursor) {
        g_print("(sub_start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
#endif
} 

static void sub_end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
#if 1
    g_print("(sub_end)element_name: %s \n", element_name);
#endif
} 

static void sub_text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
#if 1
    g_print("(sub_text)text:%s \n", text);
#endif
}

GMarkupParser sub_parser = { 
    . start_element = sub_start, 
    . end_element = sub_end, 
    . text = sub_text, 
    . passthrough = NULL , 
    . error = NULL 
}; 


static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;
#if 0
    g_printf("(start)element_name: %s \n", element_name);
    while (*name_cursor) {
        g_printf("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
#endif
#if 1
    if(0 == strcmp(element_name, TEST_GMARKUP_CORE_APPS)) {
        g_print("(start)start parse of core \n");
        g_core_flag = TRUE;
    }

    if(0 == strcmp(element_name, TEST_GMARKUP_UTILITIES)) {
        g_print("(start)start parse of util \n");
        g_util_flag = TRUE;

        g_markup_parse_context_push (context, &sub_parser, NULL);
    }

    if(0 == strcmp(element_name, TEST_GMARKUP_DATATYPES)) {
        g_print("(start)start parse of data \n");
        g_data_flag = TRUE;
    }

    if(0 == strcmp(element_name, TEST_GMARKUP_MODULE)){
        if(g_core_flag) {
            while (*name_cursor) {
                g_print("(start)core: name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
                name_cursor++;
                value_cursor++;
            }
        }

        if(g_util_flag) {
            while (*name_cursor) {
                g_print("(start)util: name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
                name_cursor++;
                value_cursor++;
            }
        }

        if(g_data_flag) {
            while (*name_cursor) {
                g_print("(start)data: name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
                name_cursor++;
                value_cursor++;
            }
        }
        
    }
#endif
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
#if 0
    g_printf("(end)element_name: %s \n", element_name);
#endif
#if 1
    if(0 == strcmp(element_name, TEST_GMARKUP_CORE_APPS)) {
        g_print("(end)end parse of core \n");
        g_core_flag = FALSE;
    }
    if(0 == strcmp(element_name, TEST_GMARKUP_UTILITIES)) {
        g_print("(end)end parse of util \n");
        g_util_flag = FALSE;

        g_markup_parse_context_pop(context);
    }
    if(0 == strcmp(element_name, TEST_GMARKUP_DATATYPES)) {
        g_print("(end)end parse of data \n");
        g_data_flag = FALSE;
    }
#endif
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
#if 0
    g_printf("(text)text:%s \n", text);
#endif
#if 1
    if(g_core_flag) {
        g_print("(text)core text:%s \n", text);
    }

    if(g_util_flag) {
        g_print("(text)util text:%s \n", text);
    }
    if(g_data_flag) {
        g_print("(text)data text:%s \n", text);
    }
#endif
} 

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = NULL, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 
    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_sub_parser.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf) ; 
    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_sub_parse]# ./glib_markup_sub_parse
<Glib>
        <CoreApplications>
                <Module version="2.48" name="Main Event Loop">manages all available sources of events</Module>
                <Module version="2.48" name="Thread Pools">pools of threads to execute work concurrently</Module>
        </CoreApplications>
        <Utilities>
                <Module version="2.48" name="File Utilities">various file-related functions</Module>
                <Module version="2.48" name="Simple XML Subset Parser">parses a subset of XML</Module>
        </Utilities>
        <DataTypes>
                <Module version="2.48" name="Sequences">scalable lists</Module>
                <Module version="2.48" name="Byte Arrays">arrays of bytes</Module>
        </DataTypes>
</Glib>
(start)start parse of core
(text)core text:

(start)core: name_cursor:version, value_cursor:2.48
(start)core: name_cursor:name, value_cursor:Main Event Loop
(text)core text:manages all available sources of events
(text)core text:

(start)core: name_cursor:version, value_cursor:2.48
(start)core: name_cursor:name, value_cursor:Thread Pools
(text)core text:pools of threads to execute work concurrently
(text)core text:

(end)end parse of core
(start)start parse of util
(sub_text)text:

(sub_start)element_name: Module
(sub_start)name_cursor:version, value_cursor:2.48
(sub_start)name_cursor:name, value_cursor:File Utilities
(sub_text)text:various file-related functions
(sub_end)element_name: Module
(sub_text)text:

(sub_start)element_name: Module
(sub_start)name_cursor:version, value_cursor:2.48
(sub_start)name_cursor:name, value_cursor:Simple XML Subset Parser
(sub_text)text:parses a subset of XML
(sub_end)element_name: Module
(sub_text)text:

(end)end parse of util
(start)start parse of data
(text)data text:

(start)data: name_cursor:version, value_cursor:2.48
(start)data: name_cursor:name, value_cursor:Sequences
(text)data text:scalable lists
(text)data text:

(start)data: name_cursor:version, value_cursor:2.48
(start)data: name_cursor:name, value_cursor:Byte Arrays
(text)data text:arrays of bytes
(text)data text:

(end)end parse of data

以上结果可以看出，在解析到Utilities节点时，该节点内部的子节点使用的是子解析器。

属性收集

g_markup_collect_attributes是一个变参函数，可以根据属性名获取到对应的属性值。当然，遍历name_cursor和value_cursor也可以得到属性名和属性值，但如果属性值有很多又知道属性名的情况下，可以使用此函数快速获取到属性值。
属性收集有一个重要的参数为GMarkupCollectType，当设置G_MARKUP_COLLECT_OPTIONAL时，参数为可选，即使找不到该参数，也不会报错，只会返回NULL。如果不设置该参数，则会报解析失败。

待解析xml内容：

<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
</elements>
</xml>

解析程序：
源码见glib_examples\glib_markup\glib_markup_attributes

#include <glib.h>

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;

    char *value1 = NULL;
    char *value2 = NULL;

    g_print("(start)element_name: %s \n", element_name);

    if (g_markup_collect_attributes (element_name,
            attribute_names,
            attribute_values,
            error,
            G_MARKUP_COLLECT_STRING | G_MARKUP_COLLECT_OPTIONAL, "attribute_name1", &value1,
            G_MARKUP_COLLECT_STRING | G_MARKUP_COLLECT_OPTIONAL, "attribute_name2", &value2,
            G_MARKUP_COLLECT_INVALID) ) {
        g_print("value1: %s \n", value1);
        g_print("value2: %s \n", value2);
    } else {
        g_print("no such attribute \n");
    }

    while (*name_cursor) {
        g_print("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(end)element_name: %s \n", element_name);
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(text)text:%s \n", text);
}

static void passthrough( GMarkupParseContext *context, 
        const gchar *passthrough_text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(passthrough)passthrough:%s \n", passthrough_text);
}

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = passthrough, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 
    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_collect_attributes.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf) ; 

    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_attributes]# ./glib_markup_attributes
<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
</elements>
</xml>
(start)element_name: xml
value1: (null)
value2: (null)
(text)text:

(start)element_name: head
value1: (null)
value2: (null)
(end)element_name: head
(text)text:

(start)element_name: element_num
value1: (null)
value2: (null)
(text)text:2
(end)element_name: element_num
(text)text:

(start)element_name: elements
value1: (null)
value2: (null)
(text)text:

(start)element_name: element_name
value1: attribute_value1
value2: attribute_value2
(start)name_cursor:attribute_name1, value_cursor:attribute_value1
(start)name_cursor:attribute_name2, value_cursor:attribute_value2
(text)text:text
(end)element_name: element_name
(text)text:

(end)element_name: elements
(text)text:

(end)element_name: xml

获取解析位置

使用g_markup_parse_context_get_position函数可以获取到当前解析到的位置信息。

待解析xml内容：

<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name1 attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name1>
<element_name2 attribute_name3="attribute_value3" attribute_name4="attribute_value4">text</element_name2>
</elements>
</xml>

解析程序：
源码见glib_examples\glib_markup\glib_markup_get_position

#include <glib.h>

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;

    gint line_number = 0;
    gint char_number = 0;

    g_print("(start)element_name: %s \n", element_name);

    g_markup_parse_context_get_position(context, &line_number, &char_number);
    g_print("(get_position)line_number:%d, char_number:%d \n", line_number, char_number);


    while (*name_cursor) {
        g_print("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(end)element_name: %s \n", element_name);
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(text)text:%s \n", text);
}

static void passthrough( GMarkupParseContext *context, 
        const gchar *passthrough_text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(passthrough)passthrough:%s \n", passthrough_text);
}

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = passthrough, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 
    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_get_position.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf) ; 
    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 build]# ./glib_markup_get_position
<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name1 attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name1>
<element_name2 attribute_name3="attribute_value3" attribute_name4="attribute_value4">text</element_name2>
</elements>
</xml>
(start)element_name: xml
(get_position)line_number:2, char_number:1
(text)text:

(start)element_name: head
(get_position)line_number:2, char_number:8
(end)element_name: head
(text)text:

(start)element_name: element_num
(get_position)line_number:3, char_number:15
(text)text:2
(end)element_name: element_num
(text)text:

(start)element_name: elements
(get_position)line_number:5, char_number:1
(text)text:

(start)element_name: element_name1
(get_position)line_number:5, char_number:87
(start)name_cursor:attribute_name1, value_cursor:attribute_value1
(start)name_cursor:attribute_name2, value_cursor:attribute_value2
(text)text:text
(end)element_name: element_name1
(text)text:

(start)element_name: element_name2
(get_position)line_number:6, char_number:87
(start)name_cursor:attribute_name3, value_cursor:attribute_value3
(start)name_cursor:attribute_name4, value_cursor:attribute_value4
(text)text:text
(end)element_name: element_name2
(text)text:

(end)element_name: elements
(text)text:

(end)element_name: xml

获取标签

g_markup_parse_context_get_element函数可以获取当前正在解析的标签。在start回调函数中，element_name本身就是正在处理的标签。

待解析xml内容：

<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name1 attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name1>
<element_name2 attribute_name3="attribute_value3" attribute_name4="attribute_value4">text</element_name2>
</elements>
</xml>

解析程序：
源码见glib_examples\glib_markup\glib_markup_element

#include <glib.h>

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;

    gchar *element;

    g_print("(start)element_name: %s \n", element_name);

    element = (gchar *)g_markup_parse_context_get_element(context);
    if(NULL != element) {
        g_print("(get_element)get: %s \n", element);
    }

    while (*name_cursor) {
        g_print("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(end)element_name: %s \n", element_name);
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(text)text:%s \n", text);
}

static void passthrough( GMarkupParseContext *context, 
        const gchar *passthrough_text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(passthrough)passthrough:%s \n", passthrough_text);
}

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = passthrough, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 
    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_element.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf) ; 

    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_element]# ./glib_markup_element
<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name1 attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name1>
<element_name2 attribute_name3="attribute_value3" attribute_name4="attribute_value4">text</element_name2>
</elements>
</xml>
(start)element_name: xml
(get_element)get: xml
(text)text:

(start)element_name: head
(get_element)get: head
(end)element_name: head
(text)text:

(start)element_name: element_num
(get_element)get: element_num
(text)text:2
(end)element_name: element_num
(text)text:

(start)element_name: elements
(get_element)get: elements
(text)text:

(start)element_name: element_name1
(get_element)get: element_name1
(start)name_cursor:attribute_name1, value_cursor:attribute_value1
(start)name_cursor:attribute_name2, value_cursor:attribute_value2
(text)text:text
(end)element_name: element_name1
(text)text:

(start)element_name: element_name2
(get_element)get: element_name2
(start)name_cursor:attribute_name3, value_cursor:attribute_value3
(start)name_cursor:attribute_name4, value_cursor:attribute_value4
(text)text:text
(end)element_name: element_name2
(text)text:

(end)element_name: elements
(text)text:

(end)element_name: xml

获取标签栈

返回一个标签链表，该链表为当前标签及其所有的父标签，一直递归到根标签。
举例：<html><h1><text>aaa</text></h1></html>，当从text标签获取标签栈时，链表中的元素依次为text->h1->html。

待解析xml内容：

<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
<element_name attribute_name3="attribute_value3" attribute_name4="attribute_value4">text</element_name>
</elements>
</xml>

解析程序：
源码见glib_examples\glib_markup\glib_markup_element_stack

#include <glib.h>

static void _slist_print(gpointer data, gpointer user_data)
{
    g_print("(stack)%s: %s \n", user_data, data);
    return;
}

static void start( GMarkupParseContext *context, 
        const gchar *element_name, 
        const gchar **attribute_names, 
        const gchar **attribute_values, 
        gpointer user_data, 
        GError ** error ) 
{ 
    const gchar **name_cursor = attribute_names; 
    const gchar **value_cursor = attribute_values;

    GSList *stack;

    g_print("(start)element_name: %s \n", element_name);

    stack = (GSList *) g_markup_parse_context_get_element_stack (context);
    if(g_slist_length(stack) > 0) {
        g_slist_foreach(stack, _slist_print, (gpointer)element_name);
    } else {
        g_print("(stack)%s: no element stack \n", element_name);
    }


    while (*name_cursor) {
        g_print("(start)name_cursor:%s, value_cursor:%s \n", *name_cursor, *value_cursor);
        name_cursor++;
        value_cursor++;
    }
} 

static void end( GMarkupParseContext *context, 
        const gchar *element_name, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(end)element_name: %s \n", element_name);
} 

static void text( GMarkupParseContext *context, 
        const gchar *text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_printf("(text)text:%s \n", text);
}

static void passthrough( GMarkupParseContext *context, 
        const gchar *passthrough_text, 
        gsize text_len, 
        gpointer user_data, 
        GError **error ) 
{
    g_print("(passthrough)passthrough:%s \n", passthrough_text);
}

static void err(GMarkupParseContext *context, 
        GError *error,
        gpointer user_data)
{
    g_print("(err)parse xml failed: %s \n", error->message);

}

GMarkupParser parser = { 
    . start_element = start, 
    . end_element = end, 
    . text = text, 
    . passthrough = passthrough, 
    . error = err 
}; 

int main( ) 
{
    GMarkupParseContext *context; 

    gchar *buf = NULL;
    gsize length = 0;

    g_file_get_contents("xml_get_element_stack.xml" , &buf, &length, NULL); 
    g_print( "%s \n" , buf);

    context = g_markup_parse_context_new(&parser, 0, NULL, NULL);
    if(NULL == context) {
        g_print("context new failed! \n");
        return -1;
    }

    if(FALSE == g_markup_parse_context_parse(context, buf, length, NULL)) {
        g_print("context parse failed! \n");
    }
    
    g_markup_parse_context_free(context);

    return 0;
}

运行结果：

[root@centos7_6 glib_markup_element_stack]# ./glib_markup_element_stack
<xml>
<head/>
<element_num>2</element_num>
<elements>
<element_name attribute_name1="attribute_value1" attribute_name2="attribute_value2">text</element_name>
<element_name attribute_name3="attribute_value3" attribute_name4="attribute_value4">text</element_name>
</elements>
</xml>
(start)element_name: xml
(stack)xml: xml
(text)text:

(start)element_name: head
(stack)head: head
(stack)head: xml
(end)element_name: head
(text)text:

(start)element_name: element_num
(stack)element_num: element_num
(stack)element_num: xml
(text)text:2
(end)element_name: element_num
(text)text:

(start)element_name: elements
(stack)elements: elements
(stack)elements: xml
(text)text:

(start)element_name: element_name
(stack)element_name: element_name
(stack)element_name: elements
(stack)element_name: xml
(start)name_cursor:attribute_name1, value_cursor:attribute_value1
(start)name_cursor:attribute_name2, value_cursor:attribute_value2
(text)text:text
(end)element_name: element_name
(text)text:

(start)element_name: element_name
(stack)element_name: element_name
(stack)element_name: elements
(stack)element_name: xml
(start)name_cursor:attribute_name3, value_cursor:attribute_value3
(start)name_cursor:attribute_name4, value_cursor:attribute_value4
(text)text:text
(end)element_name: element_name
(text)text:

(end)element_name: elements
(text)text:

(end)element_name: xml

xml字符转义

当xml内包含&、<、>、'、"这几个字符时，xml无法解析，需要转换成相应的实体。
下面是五个在XML文件中预定义好的实体：

• &--& 和
• <--&lt; 小于号
• >--> 大于号
• '--' 单引号
• "--" 双引号