From f9c8743106009c428504c3523fceab1a69769a80 Mon Sep 17 00:00:00 2001
From: iska <iska.gp@gmail.com>
Date: Sat, 21 Feb 2015 14:30:44 +0100
Subject: [PATCH] Add "tree construction dispatcher" and token handling stub

https://html.spec.whatwg.org/multipage/syntax.html#tree-construction
---
 HTMLKit.xcodeproj/project.pbxproj |  2 ++
 HTMLKit/HTMLElement.h             |  2 ++
 HTMLKit/HTMLElementTypes.h        | 27 ++++++++++++++
 HTMLKit/HTMLParser.m              | 60 +++++++++++++++++++++++++++++++
 4 files changed, 91 insertions(+)
 create mode 100644 HTMLKit/HTMLElementTypes.h

diff --git a/HTMLKit.xcodeproj/project.pbxproj b/HTMLKit.xcodeproj/project.pbxproj
index 869af18..3a4744d 100644
--- a/HTMLKit.xcodeproj/project.pbxproj
+++ b/HTMLKit.xcodeproj/project.pbxproj
@@ -98,6 +98,7 @@
 /* End PBXCopyFilesBuildPhase section */
 
 /* Begin PBXFileReference section */
+		6223211D1A969B9300BACED5 /* HTMLElementTypes.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = HTMLElementTypes.h; sourceTree = "<group>"; };
 		624493A419CCC54100BCDDF4 /* HTMLTokenizer.h */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.h; path = HTMLTokenizer.h; sourceTree = "<group>"; };
 		624493A519CCC54100BCDDF4 /* HTMLTokenizer.m */ = {isa = PBXFileReference; fileEncoding = 4; lastKnownFileType = sourcecode.c.objc; path = HTMLTokenizer.m; sourceTree = "<group>"; };
 		624493A919CCE84A00BCDDF4 /* HTMLTokenizerStates.h */ = {isa = PBXFileReference; lastKnownFileType = sourcecode.c.h; path = HTMLTokenizerStates.h; sourceTree = "<group>"; };
@@ -242,6 +243,7 @@
 				62F873E919E088C90062683C /* HTMLParser.h */,
 				62F873EA19E088C90062683C /* HTMLParser.m */,
 				628B7CE61A080E1000602C87 /* HTMLNamespaces.h */,
+				6223211D1A969B9300BACED5 /* HTMLElementTypes.h */,
 				6279F87119E17DC700F12EE5 /* HTMLParserInsertionModes.h */,
 				6279F87219E1808D00F12EE5 /* HTMLElement.h */,
 				6279F87319E1808D00F12EE5 /* HTMLElement.m */,
diff --git a/HTMLKit/HTMLElement.h b/HTMLKit/HTMLElement.h
index 4b71e68..012b88d 100644
--- a/HTMLKit/HTMLElement.h
+++ b/HTMLKit/HTMLElement.h
@@ -7,9 +7,11 @@
 //
 
 #import <Foundation/Foundation.h>
+#import "HTMLNamespaces.h"
 
 @interface HTMLElement : NSObject
 
 @property (nonatomic, strong, readonly) NSString *tagName;
+@property (nonatomic, assign, readonly) HTMLNamespace namespace;
 
 @end
diff --git a/HTMLKit/HTMLElementTypes.h b/HTMLKit/HTMLElementTypes.h
new file mode 100644
index 0000000..3cc9c95
--- /dev/null
+++ b/HTMLKit/HTMLElementTypes.h
@@ -0,0 +1,27 @@
+//
+//  HTMLElementTypes.h
+//  HTMLKit
+//
+//  Created by Iska on 19/02/15.
+//  Copyright (c) 2015 BrainCookie. All rights reserved.
+//
+
+#import	"HTMLElement.h"
+#import "HTMLNamespaces.h"
+
+#define matches(x, ...)  [[NSSet setWithObjects:__VA_ARGS__, nil] containsObject:x]
+
+NS_INLINE BOOL IsNodeMathMLTextIntegrationPoint(HTMLElement *node)
+{
+	return (node.namespace == HTMLNamespaceMathML && matches(node.tagName, @"mi", @"mo", @"mn", @"ms", @"mtext"));
+}
+
+NS_INLINE BOOL IsNodeHTMLIntegrationPoint(HTMLElement *node)
+{
+	if (node.namespace == HTMLNamespaceMathML && [node.tagName isEqualToString:@"annotation-xml"]) {
+#warning Implement HTML Element Attributes
+	} else if (node.namespace == HTMLNamespaceSVG) {
+		return matches(node.tagName, @"foreignObject", @"desc", @"title");
+	}
+	return NO;
+}
diff --git a/HTMLKit/HTMLParser.m b/HTMLKit/HTMLParser.m
index d2a08cc..a1b6f01 100644
--- a/HTMLKit/HTMLParser.m
+++ b/HTMLKit/HTMLParser.m
@@ -24,4 +24,64 @@
 
 @implementation HTMLParser
 
+#pragma mark - Parse
+
+- (id)parse
+{
+	for (HTMLToken *token in _tokenizer) {
+		[self handleToken:token];
+	}
+	return nil;
+}
+
+- (void)handleToken:(HTMLToken *)token
+{
+	BOOL (^ treeConstructionDispatcher)(HTMLElement *node) = ^BOOL(HTMLElement *node){
+
+		if (node == nil) {
+			return YES;
+		}
+		if (node.namespace == HTMLNamespaceHTML) {
+			return YES;
+		}
+		if (IsNodeMathMLTextIntegrationPoint(node)) {
+			if (token.type == HTMLTokenTypeStartTag) {
+				return !matches([(HTMLStartTagToken *)token tagName], @"mglyph", @"malignmark");
+			}
+			if (token.type == HTMLTokenTypeCharacter) {
+				return YES;
+			}
+		}
+		if (node.namespace == HTMLNamespaceMathML && [node.tagName isEqualToString:@"annotation-xml"]) {
+			if (token.type == HTMLTokenTypeStartTag) {
+				return [[(HTMLStartTagToken *)token tagName] isEqualToString:@"svg"];
+			}
+		}
+		if (IsNodeHTMLIntegrationPoint(node)) {
+			return token.type == HTMLTokenTypeStartTag || token.type == HTMLTokenTypeCharacter;
+		}
+		if (token.type == HTMLTokenTypeEOF) {
+			return YES;
+		}
+
+		return NO;
+	};
+
+	if (treeConstructionDispatcher(self.adjustedCurrentNode)) {
+		[self handleToken:token byApplyingRulesForInsertionMode:_insertionMode];
+	} else {
+		[self handleTokenByApplyingRulesForParsingTokensInForeignContent:token];
+	}
+}
+
+- (void)handleToken:(HTMLToken *)token byApplyingRulesForInsertionMode:(HTMLInsertionMode)insertionMode
+{
+	
+}
+
+- (void)handleTokenByApplyingRulesForParsingTokensInForeignContent:(HTMLToken *)token
+{
+
+}
+
 @end