Possibility to initialize with HTML data

myell0w · Apr 10, 2012 · 8ccb771 · 8ccb771
1 parent dec1e2d
commit 8ccb771
Show file tree

Hide file tree

Showing 2 changed files with 53 additions and 30 deletions.
diff --git a/RaptureXML/RXMLElement.h b/RaptureXML/RXMLElement.h
@@ -52,13 +52,15 @@ typedef void (^RXMLBlock)(RXMLElement *element);
 + (id)elementWithURL:(NSURL *)url;
 + (id)elementWithData:(NSData *)data;
 + (id)elementWithNode:(xmlNodePtr)node;
++ (id)elementWithHTMLData:(NSData *)data;
 
 - (id)initWithString:(NSString *)xmlString encoding:(NSStringEncoding)encoding;
 - (id)initWithFilepath:(NSString *)filename;
 - (id)initWithFilename:(NSString *)filename extension:(NSString *)extension;
 - (id)initWithURL:(NSURL *)url;
 - (id)initWithData:(NSData *)data;
 - (id)initWithNode:(xmlNodePtr)node;
+- (id)initWithHTMLData:(NSData *)data;
 
 - (NSString *)attribute:(NSString *)attributeName;
 - (NSString *)attribute:(NSString *)attributeName inNamespace:(NSString *)xmlNamespace;

diff --git a/RaptureXML/RXMLElement.m b/RaptureXML/RXMLElement.m
@@ -30,6 +30,7 @@
 //
 
 #import "RXMLElement.h"
+#import <libxml/HTMLparser.h>
 #import <libxml/xpath.h>
 #import <libxml/xpathInternals.h>
 
@@ -38,7 +39,7 @@ @interface RXMLElement () {
     xmlNodePtr _node;
 }
 
-- (void)setupWithData:(NSData *)data;
+- (void)setupWithData:(NSData *)data isHTML:(BOOL)html;
 
 @end
 
@@ -48,6 +49,38 @@ @implementation RXMLElement
 @synthesize text = _text;
 @synthesize tagName = _tagName;
 
+////////////////////////////////////////////////////////////////////////
+#pragma mark - Class Methods
+////////////////////////////////////////////////////////////////////////
+
++ (id)elementWithString:(NSString *)attributeXML_ encoding:(NSStringEncoding)encoding {
+    return [[RXMLElement alloc] initWithString:attributeXML_ encoding:encoding];    
+}
+
++ (id)elementWithFilepath:(NSString *)filename {
+    return [[RXMLElement alloc] initWithFilepath:filename];    
+}
+
++ (id)elementWithFilename:(NSString *)filename extension:(NSString *)extension {
+    return [[RXMLElement alloc] initWithFilename:filename extension:extension];
+}
+
++ (id)elementWithURL:(NSURL *)url {
+    return [[RXMLElement alloc] initWithURL:url];
+}
+
++ (id)elementWithData:(NSData *)data {
+    return [[RXMLElement alloc] initWithData:data];
+}
+
++ (id)elementWithNode:(xmlNodePtr)node {
+    return [[RXMLElement alloc] initWithNode:node];
+}
+
++ (id)elementWithHTMLData:(NSData *)data {
+    return [[RXMLElement alloc] initWithHTMLData:data];
+}
+
 ////////////////////////////////////////////////////////////////////////
 #pragma mark - Lifecycle
 ////////////////////////////////////////////////////////////////////////
@@ -56,7 +89,7 @@ - (id)initWithString:(NSString *)xmlString encoding:(NSStringEncoding)encoding {
     if ((self = [super init])) {
         NSData *data = [xmlString dataUsingEncoding:encoding];
 
-        [self setupWithData:data];
+        [self setupWithData:data isHTML:NO];
     }
 
     return self;    
@@ -67,7 +100,7 @@ - (id)initWithFilepath:(NSString *)filename {
         NSString *fullPath = [[[NSBundle bundleForClass:self.class] bundlePath] stringByAppendingPathComponent:filename];
         NSData *data = [NSData dataWithContentsOfFile:fullPath];
 
-        [self setupWithData:data];
+        [self setupWithData:data isHTML:NO];
     }
 
     return self;    
@@ -78,7 +111,7 @@ - (id)initWithFilename:(NSString *)filename extension:(NSString *)extension {
         NSString *fullPath = [[NSBundle bundleForClass:[self class]] pathForResource:filename ofType:extension];
         NSData *data = [NSData dataWithContentsOfFile:fullPath];
 
-        [self setupWithData:data];
+        [self setupWithData:data isHTML:NO];
     }
 
     return self;    
@@ -88,15 +121,15 @@ - (id)initWithURL:(NSURL *)url {
     if ((self = [super init])) {
         NSData *data = [NSData dataWithContentsOfURL:url];
 
-        [self setupWithData:data];
+        [self setupWithData:data isHTML:NO];
     }
 
     return self;    
 }
 
 - (id)initWithData:(NSData *)data {
     if ((self = [super init])) {
-        [self setupWithData:data];
+        [self setupWithData:data isHTML:NO];
     }
 
     return self;    
@@ -111,28 +144,12 @@ - (id)initWithNode:(xmlNodePtr)node {
     return self;        
 }
 
-+ (id)elementWithString:(NSString *)attributeXML_ encoding:(NSStringEncoding)encoding {
-    return [[RXMLElement alloc] initWithString:attributeXML_ encoding:encoding];    
-}
-
-+ (id)elementWithFilepath:(NSString *)filename {
-    return [[RXMLElement alloc] initWithFilepath:filename];    
-}
-
-+ (id)elementWithFilename:(NSString *)filename extension:(NSString *)extension {
-    return [[RXMLElement alloc] initWithFilename:filename extension:extension];
-}
-
-+ (id)elementWithURL:(NSURL *)url {
-    return [[RXMLElement alloc] initWithURL:url];
-}
-
-+ (id)elementWithData:(NSData *)data {
-    return [[RXMLElement alloc] initWithData:data];
-}
-
-+ (id)elementWithNode:(xmlNodePtr)node {
-    return [[RXMLElement alloc] initWithNode:node];
+- (id)initWithHTMLData:(NSData *)data {
+    if ((self = [super init])) {
+        [self setupWithData:data isHTML:YES];
+    }
+
+    return self; 
 }
 
 - (void)dealloc {
@@ -487,8 +504,12 @@ - (void)iterateChildrenMatchingXPathQuery:(NSString *)query usingBlock:(RXMLBloc
 #pragma mark - Private
 ////////////////////////////////////////////////////////////////////////
 
-- (void)setupWithData:(NSData *)data {
-    _document = xmlReadMemory([data bytes], (int)[data length], "", nil, XML_PARSE_RECOVER);
+- (void)setupWithData:(NSData *)data isHTML:(BOOL)html {
+    if (html) {
+        _document = htmlReadMemory(data.bytes, (int)data.length, "", NULL, HTML_PARSE_NOWARNING | HTML_PARSE_NOERROR);
+    } else {
+        _document = xmlReadMemory(data.bytes, (int)data.length, "", nil, XML_PARSE_RECOVER);
+    }
 
     if (_document != NULL) {
         _node = xmlDocGetRootElement(_document);