Get twitter:image:src from HTML metadata.

Frameworks/RSParser/HTML/RSHTMLMetadata.h

@@ -13,6 +13,7 @@
 @class RSHTMLOpenGraphProperties;
 @class RSHTMLOpenGraphImage;
 @class RSHTMLTag;
+@class RSHTMLTwitterProperties;
 
 @interface RSHTMLMetadata : NSObject
 
@@ -26,6 +27,8 @@
 @property (nonatomic, readonly) NSArray <RSHTMLMetadataFeedLink *> *feedLinks;
 
 @property (nonatomic, readonly) RSHTMLOpenGraphProperties *openGraphProperties;
+@property (nonatomic, readonly) RSHTMLTwitterProperties *twitterProperties;
+
 @end
 
 
@@ -67,3 +70,14 @@
 @property (nonatomic, readonly) NSString *altText;
 
 @end
+
+@interface RSHTMLTwitterProperties : NSObject
+
+// TODO: the rest. At this writing (Nov. 26, 2017) I just care about twitter:image:src.
+
+- (instancetype)initWithURLString:(NSString *)urlString tags:(NSArray <RSHTMLTag *> *)tags;
+
+@property (nonatomic, readonly) NSString *imageURL; // twitter:image:src
+
+@end
+

Frameworks/RSParser/HTML/RSHTMLMetadata.m

@@ -71,6 +71,7 @@ static NSString *kTypeKey = @"type";
 	_feedLinks = objectsOfClassWithTags([RSHTMLMetadataFeedLink class], feedLinkTags, urlString);
 
 	_openGraphProperties = [[RSHTMLOpenGraphProperties alloc] initWithURLString:urlString tags:tags];
+	_twitterProperties = [[RSHTMLTwitterProperties alloc] initWithURLString:urlString tags:tags];
 	
 	return self;
 }
@@ -382,4 +383,38 @@ static NSString *ogContentKey = @"content";
 
 @end
 
+@implementation RSHTMLTwitterProperties
+
+static NSString *twitterNameKey = @"name";
+static NSString *twitterContentKey = @"content";
+static NSString *twitterImageSrc = @"twitter:image:src";
+
+- (instancetype)initWithURLString:(NSString *)urlString tags:(NSArray <RSHTMLTag *> *)tags {
+
+	self = [super init];
+	if (!self) {
+		return nil;
+	}
+
+	for (RSHTMLTag *tag in tags) {
+
+		if (tag.type != RSHTMLTagTypeMeta) {
+			continue;
+		}
+		NSString *name = tag.attributes[twitterNameKey];
+		if (!name || ![name isEqualToString:twitterImageSrc]) {
+			continue;
+		}
+		NSString *content = tag.attributes[twitterContentKey];
+		if (!content || content.length < 1) {
+			continue;
+		}
+		_imageURL = content;
+		break;
+	}
+
+	return self;
+}
+
+@end
 

Frameworks/RSParser/RSParserTests/HTMLMetadataTests.swift

@@ -83,6 +83,15 @@ class HTMLMetadataTests: XCTestCase {
 		}
 	}
 
+	func testCocoPerformance() {
+
+		// 0.004 sec on my 2012 iMac
+		let d = parserData("coco", "html", "https://www.theatlantic.com/entertainment/archive/2017/11/coco-is-among-pixars-best-movies-in-years/546695/")
+		self.measure {
+			let _ = RSHTMLMetadataParser.htmlMetadata(with: d)
+		}
+	}
+
 	func testSixColors() {
 
 		let d = parserData("sixcolors", "html", "http://sixcolors.com/")
@@ -120,4 +129,13 @@ class HTMLMetadataTests: XCTestCase {
 		let image = openGraphData.images.first!
 		XCTAssert(image.url == "https://cdn.theatlantic.com/assets/media/img/mt/2017/11/1033101_first_full_length_trailer_arrives_pixars_coco/facebook.jpg?1511382177")
 	}
+
+	func testCocoTwitterImage() {
+
+		let d = parserData("coco", "html", "https://www.theatlantic.com/entertainment/archive/2017/11/coco-is-among-pixars-best-movies-in-years/546695/")
+		let metadata = RSHTMLMetadataParser.htmlMetadata(with: d)
+		let twitterData = metadata.twitterProperties!
+		let imageURL = twitterData.imageURL!
+		XCTAssert(imageURL == "https://cdn.theatlantic.com/assets/media/img/mt/2017/11/1033101_first_full_length_trailer_arrives_pixars_coco/facebook.jpg?1511382177")
+	}
 }