通过tinyxml解析百度热词榜单

时间：2011-06-12 来源：westfly

百度热词榜单

初学TinyXml，盯上了解析百度热词榜单。通过curl获取rss地址上的数据。

由于TinyXml不支持，所以要转换为UTF-8格式，命令为

iconv -f gb2312 -t utf-8 baidutop.xml > baidutop.xml

写得很烂，写的时候，心里对内存管理部分总是瞻前顾后的，真是奇怪啊。

View Code

#include <iostream>   
#include <tinyxml/tinyxml.h>
#include <tinyxml/tinystr.h>
  
#include <vector>

using namespace std; 

struct CHrefLink
{
    string word;
    string url;
};

struct CBaiDuTopKey
{
    int rank;
    long account;
    
    string keyWord;
    string wordUrl;
    vector <CHrefLink> href;
};



void GetHotRank(vector <CBaiDuTopKey> & hot,const char *xml);
void ExtraHrefLink(TiXmlElement *hrefElement,CHrefLink &link);
void ExtraHrefLink(TiXmlElement *hrefElement,string  & word,string & url);
void ExtraKeyWords(TiXmlElement *tableColumn,vector <CHrefLink>    &vKey);
int main()
{
 //创建一个XML的文档对象。   
    TiXmlDocument *myDocument = new TiXmlDocument("baidutop.xml");   
    myDocument->LoadFile();   
       
    //获得根元素
    TiXmlElement *RootElement = myDocument->RootElement();   
  
    //输出根元素名称   
    cout << RootElement->Value() << endl;   
   
    //解析获得有用的信息
    TiXmlElement *channelElement = RootElement->FirstChildElement();
    string title = channelElement->FirstChildElement()->GetText();
    TiXmlElement *itemElement = (TiXmlElement *)channelElement->LastChild();
    
    
    int ttl = atoi(((TiXmlElement *)itemElement->PreviousSibling())->GetText());
    
    TiXmlElement * descriptionElement = (TiXmlElement *)(itemElement->LastChild());
    
    vector <CBaiDuTopKey> hotword;
    
    GetHotRank(hotword,descriptionElement->GetText());

    
    
    cout<<"Name :"<<title<<"\t Life Time:"<<ttl<<endl;
    for(vector<CBaiDuTopKey>::iterator iter = hotword.begin();
            iter!=hotword.end();++iter)
    {
        cout<<"\nrank \t"<<(*iter).rank<<"\t amount:"<<(*iter).account;
        cout<<"\nkeyword\t"<<(*iter).keyWord<<"\nurl\t"<<(*iter).wordUrl;
    }
    cout<<endl;
    delete myDocument;
    return 0;
}


void GetHotRank(vector <CBaiDuTopKey> & hotVector,const char *xml)
{
    TiXmlDocument *doc = new TiXmlDocument();
    
    int row    =    0;
    
    
    doc->Parse(xml);
    TiXmlElement *tableElement = doc->RootElement();
    //
    TiXmlElement *tableRoot = tableElement->FirstChildElement();
    if(!tableRoot)
    {
        cout<<"errro !"<<endl;
        return ;
    }
    //第一个元素为table的表头，剔除
    for(TiXmlElement *tableRow = tableRoot->FirstChildElement()->NextSiblingElement();
            tableRow;
            tableRow = tableRow->NextSiblingElement())
    {
        CBaiDuTopKey    rankedword;
        TiXmlElement *tableColumn = tableRow->FirstChildElement();
        
        //关键词排名
        rankedword.rank=atoi(tableColumn->GetText());
        tableColumn = tableColumn->NextSiblingElement();
        //关键词
        ExtraHrefLink(tableColumn->FirstChildElement(),rankedword.keyWord,rankedword.wordUrl);
        tableColumn = tableColumn->NextSiblingElement();
        //搜索量
        rankedword.account=atol(tableColumn->GetText());
        //链接
        tableColumn = tableColumn->NextSiblingElement();
        
        hotVector.push_back(rankedword);
    }
    delete doc;
}
void ExtraHrefLink(TiXmlElement *hrefElement,CHrefLink &link)
{
        link.url = hrefElement->FirstAttribute()->Value();
        link.word = hrefElement->GetText();
        
}
void ExtraHrefLink(TiXmlElement *hrefElement,string  & word,string & url)
{
        url = hrefElement->FirstAttribute()->Value();
        word = hrefElement->GetText();
        
}
void ExtraKeyWords(TiXmlElement *tableColumn,vector <CHrefLink>    &vKey)
{
    TiXmlElement *hrefElement = tableColumn->FirstChildElement();
    CHrefLink    links;
    while(hrefElement)
    {
        ExtraHrefLink(hrefElement,links);
        vKey.push_back(links);
        hrefElement = hrefElement->NextSiblingElement();
    }
}