From 36b74b57662e8392533d09eebd5f2081dca148d9 Mon Sep 17 00:00:00 2001
From: iska <iska.gp@gmail.com>
Date: Tue, 31 Mar 2015 01:01:41 +0200
Subject: [PATCH] Add HTML fragment serialization (inner & outer HTML)

https://html.spec.whatwg.org/multipage/syntax.html#serialising-html-fragments
---
 HTMLKit/HTMLComment.m      |  7 +++++++
 HTMLKit/HTMLDocumentType.m |  7 +++++++
 HTMLKit/HTMLElement.m      | 39 ++++++++++++++++++++++++++++++++++++++
 HTMLKit/HTMLNode.h         |  2 ++
 HTMLKit/HTMLNode.m         | 16 ++++++++++++++++
 HTMLKit/HTMLText.m         | 20 +++++++++++++++++++
 6 files changed, 91 insertions(+)

diff --git a/HTMLKit/HTMLComment.m b/HTMLKit/HTMLComment.m
index 0f5aac0..23194bc 100644
--- a/HTMLKit/HTMLComment.m
+++ b/HTMLKit/HTMLComment.m
@@ -38,6 +38,13 @@
 	return copy;
 }
 
+#pragma mark - Serialization
+
+- (NSString *)outerHTML
+{
+	return [NSString stringWithFormat:@"<!--%@-->", self.data];
+}
+
 #pragma mark - Description
 
 - (NSString *)description
diff --git a/HTMLKit/HTMLDocumentType.m b/HTMLKit/HTMLDocumentType.m
index a9b20de..ce6687f 100644
--- a/HTMLKit/HTMLDocumentType.m
+++ b/HTMLKit/HTMLDocumentType.m
@@ -133,6 +133,13 @@ NS_INLINE BOOL nilOrEqual(id first, id second) {
 	return copy;
 }
 
+#pragma mark - Serialization
+
+- (NSString *)outerHTML
+{
+	return [NSString stringWithFormat:@"<!DOCTYPE %@>", self.name];
+}
+
 #pragma mark - Description
 
 - (NSString *)description
diff --git a/HTMLKit/HTMLElement.m b/HTMLKit/HTMLElement.m
index 58fb0b6..4425dcd 100644
--- a/HTMLKit/HTMLElement.m
+++ b/HTMLKit/HTMLElement.m
@@ -8,6 +8,8 @@
 
 #import "HTMLElement.h"
 #import "HTMLOrderedDictionary.h"
+#import "NSString+HTMLKit.h"
+#import "HTMLText.h"
 
 @interface HTMLElement ()
 {
@@ -96,6 +98,43 @@
 	return copy;
 }
 
+#pragma mark - Serialization
+
+- (NSString *)outerHTML
+{
+	NSMutableString *result = [NSMutableString string];
+
+	[result appendFormat:@"<%@", self.tagName];
+	[self.attributes enumerateKeysAndObjectsUsingBlock:^(NSString *key, NSString *value, BOOL *stop) {
+		NSRange range = NSMakeRange(0, value.length);
+		NSMutableString *escaped = [value mutableCopy];
+		[escaped replaceOccurrencesOfString:@"&" withString:@"&amp;" options:0 range:range];
+		[escaped replaceOccurrencesOfString:@"\00A0" withString:@"&nbsp;" options:0 range:range];
+		[escaped replaceOccurrencesOfString:@"\"" withString:@"&quot;" options:0 range:range];
+
+		[result appendFormat:@" %@=\"%@\"", key, escaped];
+	}];
+
+	[result appendString:@">"];
+
+	if ([self.tagName isEqualToAny:@"area", @"base", @"basefont", @"bgsound", @"br", @"col", @"embed",
+		 @"frame", @"hr", @"img", @"input", @"keygen", @"link", @"menuitem", @"meta", @"param", @"source",
+		 @"track", @"wbr", nil]) {
+		return result;
+	}
+
+	if ([self.tagName isEqualToAny:@"pre", @"textarea", @"listing", nil] && self.firstChiledNode.type == HTMLNodeText) {
+		HTMLText *textNode = (HTMLText *)self.firstChiledNode;
+		if ([textNode.data hasPrefix:@"\n"]) {
+			[result appendString:@"\n"];
+		}
+	}
+	[result appendString:self.innerHTML];
+	[result appendFormat:@"</%@>", self.tagName];
+
+	return result;
+}
+
 #pragma mark - Description
 
 - (NSString *)description
diff --git a/HTMLKit/HTMLNode.h b/HTMLKit/HTMLNode.h
index dddbd4f..45f2950 100644
--- a/HTMLKit/HTMLNode.h
+++ b/HTMLKit/HTMLNode.h
@@ -79,6 +79,8 @@ typedef NS_ENUM(short, HTMLNodeType)
 - (NSEnumerator *)treeEnumerator;
 - (NSEnumerator *)reverseTreeEnumerator;
 
+- (NSString *)outerHTML;
+- (NSString *)innerHTML;
 - (NSString *)treeDescription;
 
 @end
diff --git a/HTMLKit/HTMLNode.m b/HTMLKit/HTMLNode.m
index 5b790a4..94fd4d1 100644
--- a/HTMLKit/HTMLNode.m
+++ b/HTMLKit/HTMLNode.m
@@ -8,7 +8,10 @@
 
 #import "HTMLNode.h"
 #import "HTMLDocument.h"
+#import "HTMLDocumentType.h"
 #import "HTMLElement.h"
+#import "HTMLText.h"
+#import "HTMLComment.h"
 #import "HTMLKitExceptions.h"
 #import "HTMLNodeTreeEnumerator.h"
 
@@ -353,6 +356,19 @@ NS_INLINE void CheckInvalidCombination(HTMLNode *parent, HTMLNode *node, NSStrin
 	return copy;
 }
 
+#pragma mark - Serialization
+
+- (NSString *)outerHTML
+{
+	[self doesNotRecognizeSelector:_cmd];
+	return nil;
+}
+
+- (NSString *)innerHTML
+{
+	return [[_childNodes.array valueForKey:@"outerHTML"] componentsJoinedByString:@""];
+}
+
 #pragma mark - Description
 
 - (NSString *)treeDescription
diff --git a/HTMLKit/HTMLText.m b/HTMLKit/HTMLText.m
index 10ae1dd..36bb035 100644
--- a/HTMLKit/HTMLText.m
+++ b/HTMLKit/HTMLText.m
@@ -7,6 +7,8 @@
 //
 
 #import "HTMLText.h"
+#import "HTMLElement.h"
+#import "NSString+HTMLKit.h"
 
 @implementation HTMLText
 
@@ -48,6 +50,24 @@
 	return copy;
 }
 
+#pragma mark - Serialization
+
+- (NSString *)outerHTML
+{
+	if ([self.parentElement.tagName isEqualToAny:@"style", @"script", @"xmp", @"iframe", @"noembed", @"noframes",
+		 @"plaintext", @"noscript", nil]) {
+		return self.data;
+	} else {
+		NSRange range = NSMakeRange(0, self.data.length);
+		NSMutableString *escaped = [self.data mutableCopy];
+		[escaped replaceOccurrencesOfString:@"&" withString:@"&amp;" options:0 range:range];
+		[escaped replaceOccurrencesOfString:@"\00A0" withString:@"&nbsp;" options:0 range:range];
+		[escaped replaceOccurrencesOfString:@"<" withString:@"&lt;" options:0 range:range];
+		[escaped replaceOccurrencesOfString:@">" withString:@"&gt;" options:0 range:range];
+		return escaped;
+	}
+}
+
 #pragma mark - Description
 
 - (NSString *)description