Merge pull request #96 from SWTI2014/parsing/html-entities

Author: rs22

packages/HTML.package/DHtmlFormatterTest.class/instance/test07HtmlEntitiesShouldBeParsedEverywhere.st

@@ -0,0 +1,10 @@
+testing
+test07HtmlEntitiesShouldBeParsedEverywhere
+	| document |
+	document := HtmlDocument new.
+	document parseContents: 
+		(ReadStream on: '<title>Test &auml;, &ouml; and &uuml;</title>',
+			'<img src=''http:&#47;&#47;'' alt=''&#248; nothing''> </img>').
+	self assert: '<title>Test ä, ö and ü</title>' equals: document children first asString.
+	self assert: 'http://' equals: document children second src.
+	self assert: 'ø nothing' equals: document children second alt.

packages/HTML.package/DHtmlFormatterTest.class/methodProperties.json

@@ -10,4 +10,5 @@
 		"test03HtmlEntitiesShouldBeParsedCorrectly" : "SS 5/27/2014 22:08",
 		"test04DecimalHtmlEntitiesShouldBeParsedCorrectly" : "SS 5/27/2014 21:24",
 		"test05HexadecimalHtmlEntitiesShouldBeParsedCorrectly" : "SS 5/27/2014 22:09",
-		"test06NonHtmlEntitiesShouldBeReturned" : "SS 5/27/2014 22:09" } }
+		"test06NonHtmlEntitiesShouldBeReturned" : "SS 5/27/2014 22:09",
+		"test07HtmlEntitiesShouldBeParsedEverywhere" : "SS 6/5/2014 19:04" } }

packages/HTML.package/HtmlCDATANode.class/instance/innerContents.st

@@ -1,4 +1,4 @@
 accessing
 innerContents
 
-	^self rawContent asUnHtml
+	^self rawContent

packages/HTML.package/HtmlCDATANode.class/methodProperties.json

@@ -6,7 +6,7 @@
 		"asString" : "tb 1/3/2006 17:36",
 		"checkForUnescapedCharacters" : "tb 3/1/2006 15:36",
 		"initialize" : "j.c 6/6/2007 14:02",
-		"innerContents" : "SS 5/27/2014 20:35",
+		"innerContents" : "SS 6/5/2014 11:33",
 		"isCDATA" : "tb 12/2/2005 09:13",
 		"prefix" : "tb 12/8/2005 12:02",
 		"suffix" : "tb 12/8/2005 12:02",

packages/HTML.package/HtmlDOMNode.class/instance/attributes..st

@@ -3,8 +3,10 @@ attributes: anObject
 	"Set the value of attributes"
 
 	| cstr |
-
-	attributes := anObject.
+	
+	attributes := Dictionary new.
+	anObject keysAndValuesDo: 
+		[ :eachKey :eachValue | attributes at: eachKey put: eachValue asUnHtml].
 	id := (attributes at: 'id' ifAbsent: [nil]).
 	cstr := attributes at: 'class' ifAbsent: [nil].
 	classes := cstr isNil ifTrue: [#()] ifFalse: [cstr findTokens: ' '].

packages/HTML.package/HtmlDOMNode.class/methodProperties.json

@@ -9,7 +9,7 @@
 		"addToHtmlMorph:" : "rs 5/28/2014 12:40:28.418",
 		"associatedForm" : "rs 5/29/2014 23:10:34.728",
 		"attributes" : "tb 11/30/2005 12:26",
-		"attributes:" : "tb 7/12/2007 11:04",
+		"attributes:" : "SS 6/7/2014 10:27",
 		"childDTDTags" : "tb 12/1/2005 09:11",
 		"childTags" : "tb 12/1/2005 09:12",
 		"classes" : "tb 1/4/2006 13:52",

packages/HTML.package/HtmlIMGNode.class/methodProperties.json

@@ -3,7 +3,7 @@
 		 },
 	"instance" : {
 		"addToFormatter:" : "SN 5/25/2014 15:55",
-		"alt" : "pf 5/18/2014 13:54",
+		"alt" : "SS 6/5/2014 11:34",
 		"imageExtent" : "pf 5/18/2014 13:52",
 		"src" : "pf 5/18/2014 13:53",
 		"tag" : "" } }

packages/HTML.package/HtmlParseNode.class/instance/rawContent.st

@@ -5,4 +5,4 @@ rawContent
 
 	in := ReadStream on: self document content.
 	in position: self start.	
-	^in next: (self end - self start)
+	^ (in next: (self end - self start)) asUnHtml

packages/HTML.package/HtmlParseNode.class/methodProperties.json

@@ -36,7 +36,7 @@
 		"parseContents:" : "tb 1/3/2006 17:15",
 		"prefix" : "tb 1/3/2006 16:47",
 		"prefixId" : "tb 1/3/2006 16:47",
-		"rawContent" : "tb 1/3/2006 16:47",
+		"rawContent" : "SS 6/5/2014 11:33",
 		"start" : "tb 1/3/2006 16:47",
 		"start:" : "sebastian.sastre 9/20/2010 11:52",
 		"startForNextElement" : "tb 1/3/2006 16:47",

packages/HTML.package/HtmlTITLENode.class/instance/innerContents.st

@@ -0,0 +1,4 @@
+accessing
+innerContents
+
+	^self rawContent

packages/HTML.package/HtmlTITLENode.class/methodProperties.json

@@ -2,4 +2,5 @@
 	"class" : {
 		 },
 	"instance" : {
+		"innerContents" : "SS 6/5/2014 11:33",
 		"tag" : "" } }

packages/HTML.package/monticello.meta/version

@@ -1 +1 @@
-395d3b09-0758-43c6-b996-b0bde1577d11
+429e44ef-e21e-4f8a-9b16-0d1f975279c6

packages/HTML.package/monticello.meta/version.d/HTML-SS.73_265cd527-7860-4ed2-9ef0-84086c072947

@@ -0,0 +1 @@
+(name 'HTML-SS.73'message 'Add test to check HTML entities are parsed everywhere correctly'id '265cd527-7860-4ed2-9ef0-84086c072947'date '5 June 2014'time '7:06:27.091 pm'author 'SS'ancestors ((id '395d3b09-0758-43c6-b996-b0bde1577d11'))stepChildren ())

packages/HTML.package/monticello.meta/version.d/HTML-SS.74_429e44ef-e21e-4f8a-9b16-0d1f975279c6

@@ -0,0 +1 @@
+(name 'HTML-SS.74'message 'Run asUnHtml only once when attributes are set.'id '429e44ef-e21e-4f8a-9b16-0d1f975279c6'date '7 June 2014'time '10:29:46.122 am'author 'SS'ancestors ((id '265cd527-7860-4ed2-9ef0-84086c072947'))stepChildren ())