HTML利用第三方框架HTML Parser解析

最新推荐文章于 2023-01-06 22:53:52 发布

gaozhenyusky

最新推荐文章于 2023-01-06 22:53:52 发布

阅读量779

点赞数

分类专栏：数据解析文章标签： html 框架 github

本文链接：https://blog.csdn.net/gaozhenyusky/article/details/38589501

版权

数据解析专栏收录该内容

3 篇文章 0 订阅

订阅专栏

先导入HTML parser三方框架

添加libxml2包和路径

#import "ViewController.h"

#import "HTMLParser.h"
#import "HTMLNode.h"

//#import "Student.h"
//#import "Book.h"

@interface ViewController ()
            
@property (weak, nonatomic) IBOutlet UIWebView *webView;
@property (weak, nonatomic) IBOutlet UIImageView *imageView;

@end

@implementation ViewController
            
- (void)viewDidLoad {
    [super viewDidLoad];
    
    NSString *path = [[NSBundle mainBundle] pathForResource:@"test" ofType:@"html"];
    
//    NSString *htmlStr = [NSString stringWithContentsOfFile:path encoding:NSUTF8StringEncoding error:nil];
    NSURL *url = [NSURL URLWithString:@"http://www.baidu.com"];
    //基础框架里面的方法
    //NSNumber/NSString/NSArray/NSDictionary
    //NSMutableString/NSMutableArray/NSMutableDictionary
    NSString *htmlStr = [NSString stringWithContentsOfURL:url encoding:NSUTF8StringEncoding error:nil];
    NSLog(@"%@", htmlStr);
    
    NSData *data = [NSData dataWithContentsOfURL:url];
    NSString *str10 = [[NSString alloc] initWithData:data encoding:NSUTF8StringEncoding];
    NSLog(@"%@", str10);
    
    NSData *imgData = [NSData dataWithContentsOfURL:[NSURL URLWithString:@"http://www.baidu.com/img/bd_logo.png"]];
//    NSLog(@"%@", imgData);
    [imgData writeToFile:@"/Users/apple/Desktop/bd_logo.png" atomically:YES];
    
    self.imageView.image = [UIImage imageWithData:imgData];
    
//    NSString *imgStr = [[NSString alloc] initWithData:imgData encoding:NSUTF8StringEncoding];
//    NSLog(@"<----->%@", imgStr);
    
    HTMLParser *parser = [[HTMLParser alloc] initWithString:htmlStr error:nil];
    HTMLNode *bodyNode = [parser body];

//    parser = nil;
    
    //根据class找到一个子节点
    HTMLNode *abcNode = [bodyNode findChildOfClass:@"abc"];
    if (abcNode) {
        //获取节点里的第一个内容
        NSLog(@"%@", [abcNode contents]);
        //获取节点里的所有内容
        NSLog(@"%@", [abcNode allContents]);
    }
    
    NSLog(@"--------");
    //根据类属性查找子节点
    NSArray *array = [bodyNode findChildrenOfClass:@"abc"];
    for (HTMLNode *node in array) {
        NSLog(@"%@", [node contents]);
    }
    
    //根据属性查找子节点
    HTMLNode *aNode = [bodyNode findChildWithAttribute:@"target" matchingName:@"_blank" allowPartial:YES];
    NSLog(@"tag name: %@: %@", [aNode tagName], [aNode contents]);
    
    //根据节点名查找子节点
    HTMLNode *imgNode = [bodyNode findChildTag:@"img"];
    NSLog(@"%@", [imgNode rawContents]);//获取该节点的原始内容
    
    Student *stu = [[Student alloc] init];
    stu.book = [[Book alloc] init];
    
    NSURLRequest *request = [NSURLRequest requestWithURL:url];
    [self.webView loadRequest:request];
}