tinyxml文档.docVIP

自己所在的项目组中，读取 xml 配置文件使用的是 tinyxml，所以对 tinyxml 用法熟些。当为自己的代码选择 xml 解析库时，搜索得知 rapidxml 是一个不错的选择，小巧，快速，作者还和 strlen 作过对比： Platform Compiler strlen() RapidXml pugixml 0.3 pugxml TinyXml Pentium 4 MSVC 8.0 2.5 5.4 7.0 61.7 298.8 Pentium 4 gcc 4.1.1 0.8 6.1 9.5 67.0 413.2 Core 2 MSVC 8.0 1.0 4.5 5.0 24.6 154.8 Core 2 gcc 4.1.1 0.6 4.6 5.4 28.3 229.3 Athlon XP MSVC 8.0 3.1 7.7 8.0 25.5 182.6 Athlon XP gcc 4.1.1 0.9 8.2 9.2 33.7 265.2 Pentium 3 MSVC 8.0 2.0 6.3 7.0 30.9 211.9 Pentium 3 gcc 4.1.1 1.0 6.7 8.9 35.3 316.0 可见确实不错。关于如何保持这么高的效率，RAPIDXML Manual中提到： In-situ parsing. When building DOM tree, RapidXml does not make copies of string data, such as node names and values. Instead, it stores pointers to interior of the source text. Use of template metaprogramming techniques. This allows it to move much of the work to compile time. Through magic of the templates, C++ compiler generates a separate copy of parsing code for any combination of parser flags you use. In each copy, all possible decisions are made at compile time and all unused code is omitted. Extensive use of lookup tables for parsing. Hand-tuned C++ with profiling done on several most popular CPUs. 包含文件：rapidxml.hpp：实现内存池 解析string 异常处理，包含： xml_base 基数节点类 xml_attribute 文本类 xml_node 节点类 xml_document 文档类 rapidxml_iterators.hpp：提供两个迭代器类：node_iterator， attribute_iteratorrapidxml_print.hpp：提供跟字符串，流对象的装换函数rapidxml_utils.hpp：提供一个file用来读取文件使用，跟两个计数函数 count_children，count_attributes 随便写个文件测试一下，xml 文件： [xml]?xml version=1.0 encoding=GB2312?!--test--TestUnitPointPospos xy=100,200 xz=200,300/name name=三角形上的点//PointPositem strName=hello world!//TestUnit[/xml] 代码： [c]#include iostream#include rapidxml.hpp#include rapidxml_print.hppusing namespace std;using namespace rapidxml;int main(){FILE* fp = fopen(test.xml, r);if (! fp)return 1;//获取文件长度fseek(fp, 0, SEEK_END);size_t len = ftell(fp);fseek(fp, 0, SEEK_SET);//读取文件char* pBuf = new char[len+1];if (! pBuf)return 1;memset(pBuf, 0, len+1);fr