iOS 解析 HTML

xml,json都有大量的库来解析，我们如何解析html呢？

TFHpple是一个小型的封装，可以用来解析html，它是对libxml的封装，语法是xpath。

今天我看到一个直接用libxml来解析html，参看：http://www.cocoanetics.com/2011/09/taming-html-parsing-with-libxml-1/#comment-3090 那张图画得一目了然，很值得收藏。这个文章中的源码不能遍历所有的html，我做了一点修改可以将html遍历打印出来

001 // NSData data contains the document data
002 // encoding is the NSStringEncoding of the data
003 // baseURL the documents base URL, i.e. location
004
005 CFStringEncoding cfenc = CFStringConvertNSStringEncodingToEncoding(encoding);
006 CFStringRef cfencstr = CFStringConvertEncodingToIANACharSetName(cfenc);
007 const char*enc = CFStringGetCStringPtr(cfencstr, 0);
008
009 htmlDocPtr _htmlDocument = htmlReadDoc([data bytes],
010       [[baseURL absoluteString] UTF8String],
011       enc,
012       XML_PARSE_NOERROR | XML_PARSE_NOWARNING);
013 if(_htmlDocument)
014 {
015    xmlFreeDoc(_htmlDocument);
016 }
017
018 xmlNodePtr currentNode = (xmlNodePtr)_htmlDocument;
019
020 while(currentNode)
021     {
022         // output node if it is an element
023
024         if(currentNode->type == XML_ELEMENT_NODE)
025         {
026             NSMutableArray *attrArray = [NSMutableArray array];
027
028             for(xmlAttrPtr attrNode = currentNode->properties; attrNode; attrNode = attrNode->next)
029             {
030                 xmlNodePtr contents = attrNode->children;
031
032                 [attrArray addObject:[NSString stringWithFormat:@"%s='%s'", attrNode->name, contents->content]];
033             }
034
035             NSString *attrString = [attrArray componentsJoinedByString:@" "];
036
037             if([attrString length])
038             {
039                 attrString = [@" "stringByAppendingString:attrString];
040             }
041
042             NSLog(@"<%s%@>", currentNode->name, attrString);
043         }
044         else if(currentNode->type == XML_TEXT_NODE)
045         {
046             //NSLog(@"%s", currentNode->content);
047             NSLog(@"%@", [NSString stringWithCString:(const char*)currentNode->content encoding:NSUTF8StringEncoding]);
048         }
049         else if(currentNode->type == XML_COMMENT_NODE)
050         {
051             NSLog(@"/* %s */", currentNode->name);
052         }
053
054
055         if(currentNode && currentNode->children)
056         {
057             currentNode = currentNode->children;
058         }
059         else if(currentNode && currentNode->next)
060         {
061             currentNode = currentNode->next;
062         }
063         else
064         {
065             currentNode = currentNode->parent;
066
067             // close node
068             if(currentNode && currentNode->type == XML_ELEMENT_NODE)
069             {
070                 NSLog(@"</%s>", currentNode->name);
071             }
072
073             if(currentNode->next)
074             {
075                 currentNode = currentNode->next;
076             }
077             else
078             {
079                 while(currentNode)
080                 {
081                     currentNode = currentNode->parent;
082                     if(currentNode && currentNode->type == XML_ELEMENT_NODE)
083                     {
084                         NSLog(@"</%s>", currentNode->name);
085                         if (strcmp((const char*)currentNode->name, "table") == 0)
086                         {
087                             NSLog(@"over");
088                         }
089                     }
090
091                     if(currentNode == nodes->nodeTab[0])
092                     {
093                         break;
094                     }
095
096                     if(currentNode && currentNode->next)
097                     {
098                         currentNode = currentNode->next;
099                         break;
100                     }
101                 }
102             }
103         }
104
105         if(currentNode == nodes->nodeTab[0])
106         {
107             break;
108         }
109     }

 

不过我还是喜欢用TFHpple，因为它很简单，也好用，但是它的功能不是很完完善。比如，不能获取children node,我就写了两个方法，一个是获取children node,一个是获取所有的contents.  还有node的属性content的key与node's content的key一样，都是@"nodeContent", 正确情况下属性的应是@"attributeContent"，

所以我写了这个方法，同时修改node属性的content key.

01 NSDictionary *DictionaryForNode2(xmlNodePtr currentNode, NSMutableDictionary *parentResult)
02 {
03     NSMutableDictionary *resultForNode = [NSMutableDictionary dictionary];
04
05     if(currentNode->name)
06     {
07         NSString *currentNodeContent =
08         [NSString stringWithCString:(const char*)currentNode->name encoding:NSUTF8StringEncoding];
09         [resultForNode setObject:currentNodeContent forKey:@"nodeName"];
10     }
11
12     if(currentNode->content)
13     {
14         NSString *currentNodeContent = [NSString stringWithCString:(const char*)currentNode->content encoding:NSUTF8StringEncoding];
15
16         if(currentNode->type == XML_TEXT_NODE)
17         {
18             if(currentNode->parent->type == XML_ELEMENT_NODE)
19             {
20                 [parentResult setObject:currentNodeContent forKey:@"nodeContent"];
21                 returnnil;
22             }
23
24             if(currentNode->parent->type == XML_ATTRIBUTE_NODE)
25             {
26                 [parentResult
27                  setObject:
28                  [currentNodeContent
29                   stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet]]
30                  forKey:@"attributeContent"];
31                 returnnil;
32
33             }
34         }
35     }
36
37
38
39     xmlAttr *attribute = currentNode->properties;
40     if(attribute)
41     {
42         NSMutableArray *attributeArray = [NSMutableArray array];
43         while(attribute)
44         {
45             NSMutableDictionary *attributeDictionary = [NSMutableDictionary dictionary];
46             NSString *attributeName =
47             [NSString stringWithCString:(const char*)attribute->name encoding:NSUTF8StringEncoding];
48             if(attributeName)
49             {
50                 [attributeDictionary setObject:attributeName forKey:@"attributeName"];
51             }
52
53             if(attribute->children)
54             {
55                 NSDictionary *childDictionary = DictionaryForNode2(attribute->children, attributeDictionary);
56                 if(childDictionary)
57                 {
58                     [attributeDictionary setObject:childDictionary forKey:@"attributeContent"];
59                 }
60             }
61
62             if([attributeDictionary count] > 0)
63             {
64                 [attributeArray addObject:attributeDictionary];
65             }
66             attribute = attribute->next;
67         }
68
69         if([attributeArray count] > 0)
70         {
71             [resultForNode setObject:attributeArray forKey:@"nodeAttributeArray"];
72         }
73     }
74
75     xmlNodePtr childNode = currentNode->children;
76     if(childNode)
77     {
78         NSMutableArray *childContentArray = [NSMutableArray array];
79         while(childNode)
80         {
81             NSDictionary *childDictionary = DictionaryForNode2(childNode, resultForNode);
82             if(childDictionary)
83             {
84                 [childContentArray addObject:childDictionary];
85             }
86             childNode = childNode->next;
87         }
88         if([childContentArray count] > 0)
89         {
90             [resultForNode setObject:childContentArray forKey:@"nodeChildArray"];
91         }
92     }
93
94     returnresultForNode;
95 }

 

TFHppleElement.m里加了两个key 常量

1 NSString * constTFHppleNodeAttributeContentKey  = @"attributeContent";
2 NSString * const TFHppleNodeChildArrayKey        = @"nodeChildArray";

 

并修改获取属性方法为：

1 - (NSDictionary *) attributes
2 {
3   NSMutableDictionary * translatedAttributes = [NSMutableDictionary dictionary];
4   for(NSDictionary * attributeDict in [node objectForKey:TFHppleNodeAttributeArrayKey]) {
5     [translatedAttributes setObject:[attributeDict objectForKey:TFHppleNodeAttributeContentKey]
6                              forKey:[attributeDict objectForKey:TFHppleNodeAttributeNameKey]];
7   }
8   returntranslatedAttributes;
9 }

 

并添加获取children node 方法：

01 - (BOOL) hasChildren
02 {
03     NSArray *childs = [node objectForKey: TFHppleNodeChildArrayKey];
04
05     if(childs)
06     {
07         return YES;
08     }
09
10     return NO;
11 }
12
13 - (NSArray *) children
14 {
15     if([self hasChildren])
16         return[node objectForKey: TFHppleNodeChildArrayKey];
17     returnnil;
18 }

 

最后我还加了一个获取所有content的主法：

1 - (NSString *)contentsAt:(NSString *)xPathOrCss;

 

请看 源码。