(Bug 63642): Cleanup treebuilder fixup of formatting elts.

* First pass refixing treebuilder-fixed DOM where formatting elts
  (<small>, <b>, etc.) wrap <figures> inside tables. This patch
  moves the formatting elt inside the caption.

* Added a test snippet that is tested in wt2html, wt2wt, and
  selser modes. wt2wt will always fail, but will enable selser
  tests (which if passing will get blacklisted since results
  will differ from wt2wt) and will let us catch regressions in
  those results.

Change-Id: Idaaf46565ad277289b84a4711da38ee79f2d483f

lib/dom.markTreeBuilderFixups.js

@@ -164,7 +164,6 @@ function findAutoInsertedTags(env, node) {
 		sibling, expectedName, reg;
 
 	while (c !== null) {
-
 		// Skip over template/extension content
 		if (DU.isTplElementNode( env, node )) {
 			var about = node.getAttribute( 'about' );
@@ -267,9 +266,41 @@ function findAutoInsertedTags(env, node) {
 	}
 }
 
+function cleanupFormattingTagFixup(env, node) {
+	node = node.firstChild;
+	while (node !== null) {
+		if (DU.isGeneratedFigure(node)) {
+			var c = node.firstChild;
+			// <a> tags are formatting elts as well
+			if (DU.isFormattingElt(c) && !DU.hasNodeName(c, "a") && !c.nextSibling) {
+				// Fix up DOM appropriately
+				var fig = node;
+				var formattingElt = c;
+				DU.migrateChildren(formattingElt, fig);
+				var caption = fig.lastChild;
+				DU.migrateChildren(caption, formattingElt);
+				caption.appendChild(formattingElt);
+
+				// If both the start and end tags are auto-inserted,
+				// DSR algo will automatically ignore the tag.
+				// Otherwise, TSR needs clearing. However,
+				// for simpler logic and code readability reasons,
+				// we are unconditionally clearing it out.
+				DU.getDataParsoid(formattingElt).tsr = null;
+			}
+		} else if (DU.isElt(node)) {
+			cleanupFormattingTagFixup(env, node);
+		}
+		node = node.nextSibling;
+	}
+}
+
 function markTreeBuilderFixups(body, env) {
 	findAutoInsertedTags(env, body);
 	findDeletedStartTags(env, body);
+
+	// Bug 63642 and friend
+	cleanupFormattingTagFixup(env, body);
 }
 
 if (typeof module === "object") {

lib/mediawiki.DOMUtils.js

@@ -566,6 +566,10 @@ DOMUtils = DU = {
 		return this.isList(n) || this.isListItem(n);
 	},
 
+	isGeneratedFigure: function(n) {
+		return this.isElt(n) && (/(^|\s)mw:Image(\s|$)/).test(n.getAttribute("typeof"));
+	},
+
 	/**
 	 * Get the first preceding sibling of 'node' that is an element,
 	 * or return `null` if there is no such sibling element.

tests/parserTests.txt

@@ -18038,7 +18038,7 @@ Lead
 
 
 ###
-### Parsoids-specific tests
+### Parsoid-specific tests
 ### Parsoid-PHP parser incompatibilities
 ###
 !!test
@@ -18061,6 +18061,35 @@ parsoid=wt2html,wt2wt
 </dl>
 !!end
 
+#### -----------------------------------------------------------------
+#### Parsoid-specific functionality tests
+#### -----------------------------------------------------------------
+
+# Bug 63642: Formatting elt fixup is cleaned up.
+# We know wt2wt will fail, but we expect selser to pass.
+# Due to the nature of our testing, wt2wt and selser tests will enter the
+# blacklist and we'll catch selser regressions based on changes to the
+# blacklist entries for selser tests.
+!! test
+Bad treebuilder fixup of formatting elt is cleaned up
+!! options
+parsoid=wt2html,wt2wt
+!! wikitext
+{|
+|
+<small>
+[[Image:Foobar.jpg|right|Test]]
+</small>
+|}
+!! html/parsoid
+<table>
+<tbody><tr><td>
+<p><small></small></p>
+<figure class="mw-default-size mw-halign-right" typeof="mw:Image"><a href="./File:Foobar.jpg"><img resource="./File:Foobar.jpg" src="//example.com/images/3/3a/Foobar.jpg" height="220" width="1941"></a><figcaption><small>Test</small></figcaption></figure>
+<p></p></td></tr>
+</tbody></table>
+!! end
+
 #### ----------------------------------------------------------------
 #### Parsoid-only testing of Parsoid's impl of <ref> and <references>
 #### tags. Parsoid's output for these tags differs from that of the