【C++】使用libxml解析XML文件

1.     构建xmlDocPtr对象，xmlParseMemory(str,str.len) 字符串转为XML文档 , 从文件读取内容构建xmlReadFile(szDocName,"GB2312",XML_PARSE_RECOVER)

2.     xmlDocGetRootElement()，获取XML文档根节点

3.     xmlStrcmp，比较XML字符串，与strcmp差不多 ，

4.     curr = curr->xmlChildrenNode，XML节点指针指向第一个子节点；curr->properties 节点属性

5.     curr = curr->next，XML节点指针指向下一个兄弟节点

6.     xmlNodeGetContent，获取XML节点的内容；xmlGetProp获取属性

7.     xmlFreeDoc，释放节点，与free差不多

OpenCV中使用FileStorage创建和解析XML文件。但FileStorage不能处理节点为空的情况，且对属性也不友好。例子见这里。

libxml是C接口的处理xml文件的开源代码库。可以在这里下载。

对于一段如下的xml文件：

 xml version="1.0"?>
 <Message Version="1.0">
   <Info evaluateType="2" mediaFile="1.avi">1Info>
   <Items>
     <Item frameNum="0" scale="2" />
     <Item frameNum="1" scale="2" />
     <Item frameNum="2" scale="2" />
     <Item frameNum="2913" scale="2">
       <Label type="2" l="564" t="178" r="594" b="280" id="15" />
       <Label type="2" l="528" t="184" r="560" b="288" id="11" />
       <Label type="2" l="490" t="182" r="522" b="288" id="16" />
       <Label type="2" l="614" t="358" r="704" b="600" id="4" />
     Item>
   Items>
 Message>

libxml处理的函数如下：

 string getLabes(char *szDocName, mapint,int>,Rect > &labels){
     xmlDocPtr doc;
     xmlNodePtr curNode;
     xmlChar *szKey;
     string filename="";

     doc = xmlReadFile(szDocName,"GB2312",XML_PARSE_RECOVER); //解析文件

     if (doc==NULL){
         fprintf(stderr,"Document not parsed successfully.\n");
         return filename;

     }

     curNode = xmlDocGetRootElement(doc); //确定文档根元素
     if (NULL == curNode){
         fprintf(stderr,"empty document\n");
         xmlFreeDoc(doc);
         return filename;
     }

     if (xmlStrcmp(curNode->name, BAD_CAST "Message")) {
         fprintf(stderr,"document of the wrong type, root node != Message");
         xmlFreeDoc(doc);
         return filename;
     }

     curNode = curNode->xmlChildrenNode;
     xmlNodePtr propNodePtr = curNode;
     xmlNodePtr itemNodePtr;

     while(curNode != NULL){
         //取出文件名称
         if (xmlHasProp(curNode,BAD_CAST "mediaFile")) {
             propNodePtr = curNode;

         }
         if (!xmlStrcmp(curNode->name, BAD_CAST "Items")) {
             itemNodePtr = curNode->xmlChildrenNode;
         }
         curNode = curNode->next;
     }

     xmlAttrPtr attrPtr = propNodePtr->properties;
     while (attrPtr != NULL){
         if (!xmlStrcmp(attrPtr->name, BAD_CAST "mediaFile")){
             xmlChar* szAttr = xmlGetProp(propNodePtr,BAD_CAST "mediaFile");
             char* szAttrG = u2g((char*)szAttr);
             filename = string(szAttrG);
             //cout<<"get filename: "<
             xmlFree(szAttr);
         }
         attrPtr = attrPtr->next;
     }

     // Item
     while (itemNodePtr != NULL){
         int frameNum = 0;
         int peopleID = 0;
         Rect rect;

         //
         xmlAttrPtr attrPtr = itemNodePtr->properties;
         while (attrPtr != NULL){
             if (!xmlStrcmp(attrPtr->name, BAD_CAST "frameNum")){
                 xmlChar* szAttr = xmlGetProp(itemNodePtr,BAD_CAST "frameNum");
                 frameNum = atoi((char*)szAttr);
                 //cout<<"get frameNum: "<
                 xmlFree(szAttr);
             }
             attrPtr = attrPtr->next;
         }

         // Label
         xmlNodePtr childNodePtr = itemNodePtr->xmlChildrenNode;
         while (childNodePtr != NULL){
             int l=0,t=0,r=0,b=0;
             attrPtr = childNodePtr->properties;
             while(attrPtr != NULL){
                 if (!xmlStrcmp(attrPtr->name, BAD_CAST "l")){
                     xmlChar* szAttr = xmlGetProp(childNodePtr,BAD_CAST "l");
                     l=atoi((char*)szAttr);
                     xmlFree(szAttr);
                 }
                 if (!xmlStrcmp(attrPtr->name, BAD_CAST "t")){
                     xmlChar* szAttr = xmlGetProp(childNodePtr,BAD_CAST "t");
                     t=atoi((char*)szAttr);
                     xmlFree(szAttr);
                 }
                 if (!xmlStrcmp(attrPtr->name, BAD_CAST "r")){
                     xmlChar* szAttr = xmlGetProp(childNodePtr,BAD_CAST "r");
                     r=atoi((char*)szAttr);
                     xmlFree(szAttr);
                 }
                 if (!xmlStrcmp(attrPtr->name, BAD_CAST "b")){
                     xmlChar* szAttr = xmlGetProp(childNodePtr,BAD_CAST "b");
                     b=atoi((char*)szAttr);
                     xmlFree(szAttr);
                 }
                 if (!xmlStrcmp(attrPtr->name, BAD_CAST "id")){
                     xmlChar* szAttr = xmlGetProp(childNodePtr,BAD_CAST "id");
                     peopleID=atoi((char*)szAttr);
                     xmlFree(szAttr);
                 }
                 attrPtr = attrPtr->next;
             }
             if(!(l==0&&t==0&&b==0&&r==0)){
                 rect.x=l;rect.y=t;rect.width=r-l;rect.height=b-t;
                 labels[pair<int,int>(frameNum,peopleID)] = rect;
                 //cout<<"get label: "<
             }
             childNodePtr = childNodePtr->next;
         }

         itemNodePtr = itemNodePtr->next;
     }

     xmlFreeDoc(doc);
     return filename;
 }

int GetSituation(const std::string&response) const
{
        int situation = -1;
        if(true)
        {
                const char* xmlContent = response.c_str();

                xmlDocPtr doc;
                xmlNodePtr cur;
                doc = xmlParseMemory(xmlContent, strlen(xmlContent));
                if(doc == NULL) {
                        LOG(ERROR) << "document not parsed successully.";
                        return situation;
                }
                cur = xmlDocGetRootElement(doc);
                if(cur == NULL) {
                        LOG(ERROR) << "document is empty.";
                        return situation;
                }
                if(xmlStrcmp(cur->name, (const xmlChar*) "result")) {
                        LOG(ERROR) << "document of wrong type.";
                        return situation;
                }
                cur = cur->xmlChildrenNode;
                while(cur != NULL)
                {
                        if(!xmlStrcmp(cur->name, (const xmlChar*) "situation")) {
                          char* situStr = (char*)xmlNodeGetContent(cur);
                                LOG(INFO) << "situation is: " << situStr;
                                situation = atoi(situStr);
                                xmlFree(situStr);
                                return situation;
                        }
                        cur = cur->next;
                }
        }
        return situation;
}