iOS开发之html解析

from: http://blog.sina.com.cn/s/blog_833996210100yyt4.html

使用XPath解析html

可以从此处 https://github.com/topfunky/hpple下载工程，将TFHpple.h,TFHpple.m,TFHppleElement.h,TFHppleElement.m,XPathQuery.h,XPathQuery.m加到自己的项目中，在Frameworks中导入libxml2.x

在项目中找到Other Linker Flags，加入-libxml2

在项目中找到Header Search Paths，加入/usr/include/libxml2

代码如下：

NSString *urlString = nil;
 
urlString = @"http://www.xiyou.edu.cn/new/lm.jsp?urltype=tree.TreeTempUrl&wbtreeid=724";
 
NSData *htmlData = [[NSData alloc] initWithContentsOfURL:[NSURL URLWithString:urlString]];
 
NSData *toHtmlData = [self toUTF8:htmlData];
 
TFHpple *xpathParser = [[TFHpple alloc] initWithHTMLData:toHtmlData];
 
 
NSArray *aArray = [xpathParser searchWithXPathQuery:@"//a"];
 
if ([span count] > 0) {
            
            for (int i = 87; i < 102; i++) {
                            //从<a>的第82个开始取值，共获取15个值
                TFHppleElement *aElement = [aArray objectAtIndex:i];    
                NSArray *aArr = [aElement children];
                TFHppleElement *aEle = [aArr objectAtIndex:0];
                NSArray *aChild = [aEle children];
                TFHppleElement *aChildEle = [aChild objectAtIndex:0];
                NSArray *aChildren = [aChildEle children];
                NSString *aStr = [[aChildren objectAtIndex:0] content];
                NSLog(@"aStr:%@",aStr);
                NSDictionary *aAttributeDict = [aElement attributes];
                NSLog(@"aAttributeDict:%@",aAttributeDict);
                
                            //获取a中的属性值
                NSString *hrefStr = [NSString stringWithFormat:@"http://www.xiyou.edu.cn%@",[aAttributeDict objectForKey:@"href"]];
                NSLog(@"hrefStr:%@",hrefStr);
                
                [currentNewsArr addObject:aStr];
                [currentHrefArr addObject:hrefStr];
                
            }
[htmlData release];
[xpathParser release];
}

//如果解析的网页不是utf8编码，如gbk编码，可以先将其转换为utf8编码再对其进行解析

 

-(NSData *) toUTF8:(NSData *)sourceData {  
    CFStringRef gbkStr = CFStringCreateWithBytes(NULL, [sourceData bytes], [sourceData length],kCFStringEncodingGB_18030_2000, false);  
    
    if (gbkStr == NULL) {  
        return nil;  
    } else {  
        NSString *gbkString = (NSString *)gbkStr; 
        //根据网页源代码中编码方式进行修改，此处为从gbk转换为utf8
             NSString *utf8_String = [gbkString stringByReplacingOccurrencesOfString:@"META http-equiv="Content-Type" content="text/html; charset=GBK""  
                                                                      withString:@"META http-equiv="Content-Type" content="text/html; charset=UTF-8""];  
        
        return [utf8_String dataUsingEncoding:NSUTF8StringEncoding];                             
    }                                     
}