Merge pull request knockout#1880 from knockout/1880-revert-to-jquery-…

…parsing Removing jQuery HTML parsing by default is breaking
lucksuper · Sep 28, 2015 · a2d8d56 · a2d8d56
2 parents 8decc43 + b85e327
commit a2d8d56
Show file tree

Hide file tree

Showing 3 changed files with 13 additions and 21 deletions.
diff --git a/spec/lib/loadDependencies.js b/spec/lib/loadDependencies.js
@@ -40,7 +40,7 @@
         if (url) {
             var shouldInclude = getParam(name);
             if ((dependency.include || shouldInclude) && shouldInclude !== "0" && shouldInclude !== "false") {
-                if (shouldInclude && /^[0-9]+\.[0-9.]+$/.test(shouldInclude)) {
+                if (shouldInclude && shouldInclude !== "1" && shouldInclude !== "true") {
                     url = url.replace(dependency.versionString || 'latest', shouldInclude);
                 }
                 jasmine.addScriptReference(url);

diff --git a/spec/parseHtmlFragment.js b/spec/parseHtmlFragment.js
@@ -11,10 +11,10 @@ describe('Parse HTML fragment', function() {
 
     ko.utils.arrayForEach(
     [
-        { html: '<tr-component></tr-component>', parsed: ['<tr-component></tr-component>'] },
+        { html: '<tr-component></tr-component>', parsed: ['<tr-component></tr-component>'], jQueryRequiredVersion: "3.0" },
         { html: '<thead><tr><th><thcomponent>hello</thcomponent></th></tr></thead>', parsed: ['<thead><tr><th><thcomponent>hello</thcomponent></th></tr></thead>'], ignoreRedundantTBody: true },
-        { html: '<tbody-component>world</tbody-component>', parsed: ['<tbody-component>world</tbody-component>'], minSupportedIEVersion: 8 },
-        { html: '<tfoot-component>foo</tfoot-component>', parsed: ['<tfoot-component>foo</tfoot-component>'] },
+        { html: '<tbody-component>world</tbody-component>', parsed: ['<tbody-component>world</tbody-component>'], jQueryRequiredVersion: "3.0" },
+        { html: '<tfoot-component>foo</tfoot-component>', parsed: ['<tfoot-component>foo</tfoot-component>'], jQueryRequiredVersion: "3.0" },
         { html: '<div></div>', parsed: ['<div></div>'] },
         { html: '<custom-component></custom-component>', parsed: ['<custom-component></custom-component>'] },
         { html: '<tr></tr>', parsed: ['<tr></tr>'] },
@@ -27,6 +27,12 @@ describe('Parse HTML fragment', function() {
         { html: '<optgroup label=x><option>text</option></optgroup>', parsed: ['<optgroup label=x><option>text</option></optgroup>'] },
         { html: '<option>text</option>', parsed: [ '<option>text</option>' ] }
     ], function (data) {
+        // jQuery's HTML parsing fails on element names like tr-* (but this is fixed in jQuery 3.x).
+        if (window.jQuery && data.jQueryRequiredVersion && jQuery.fn.jquery < data.jQueryRequiredVersion) {
+            it('unsupported environment for parsing ' + data.html, function () { });
+            return;
+        }
+
         it('should parse ' + data.html + ' correctly', function () {
             // IE 6-8 has a lot of trouble with custom elements. We have several strategies for dealing with
             // this, each involving different (awkward) requirements for the application.
@@ -55,14 +61,6 @@ describe('Parse HTML fragment', function() {
                 if (window.innerShiv) {
                     window.innerShiv.reset();
                 }
-
-                // Out of all the combinations above, there is still one edge case we can't support without
-                // dropping jQuery HTML parsing altogether. That is, if you're using jQuery, its parser fails
-                // on elements named 'tbody-*', on IE 6 and 7 (but it works on IE 8+). This is such an extreme
-                // edge case that it's preferable to leave this element name unsupported.
-                if (jasmine.ieVersion < data.minSupportedIEVersion) {
-                    return;
-                }
             }
 
             var parsedNodes = ko.utils.parseHtmlFragment(data.html, document);

diff --git a/src/utils.domManipulation.js b/src/utils.domManipulation.js
@@ -16,13 +16,7 @@
         },
 
         // This is needed for old IE if you're *not* using either jQuery or innerShiv. Doesn't affect other cases.
-        mayRequireCreateElementHack = ko.utils.ieVersion <= 8,
-
-        // We prefer not to use jQuery's HTML parsing, because it fails on element names like tr-*, even
-        // on the latest browsers (not even just on IE). But we retain use of jQuery HTML parsing for old
-        // IE, to avoid breaking compatibility with parsing edge-cases. Strangely, jQuery's HTML parsing
-        // works OK on elements named tr-* on old IE browsers.
-        allowJQueryHtmlParsing = ko.utils.ieVersion <= 8;
+        mayRequireCreateElementHack = ko.utils.ieVersion <= 8;
 
     function getWrap(tags) {
         var m = tags.match(/^<([a-z]+)[ >]/);
@@ -101,7 +95,7 @@
     }
 
     ko.utils.parseHtmlFragment = function(html, documentContext) {
-        return allowJQueryHtmlParsing && jQueryInstance ?
+        return jQueryInstance ?
             jQueryHtmlParse(html, documentContext) :   // As below, benefit from jQuery's optimisations where possible
             simpleHtmlParse(html, documentContext);  // ... otherwise, this simple logic will do in most common cases.
     };
@@ -119,7 +113,7 @@
             // jQuery contains a lot of sophisticated code to parse arbitrary HTML fragments,
             // for example <tr> elements which are not normally allowed to exist on their own.
             // If you've referenced jQuery we'll use that rather than duplicating its code.
-            if (allowJQueryHtmlParsing && jQueryInstance) {
+            if (jQueryInstance) {
                 jQueryInstance(node)['html'](html);
             } else {
                 // ... otherwise, use KO's own parsing logic.