Marcoil has uploaded a new change for review.
https://gerrit.wikimedia.org/r/160698
Change subject: WIP: Don't emit data-parsoid attributes for template content
......................................................................
WIP: Don't emit data-parsoid attributes for template content
(For 70786)
Change-Id: I695da6e4485ba62f3a1a4cc9e9fdd1a39637f327
---
M lib/dom.cleanup.js
M lib/dom.markFosteredContent.js
A lib/dom.removeTransclusionDataParsoid.js
M lib/mediawiki.DOMPostProcessor.js
M lib/mediawiki.DOMUtils.js
5 files changed, 84 insertions(+), 9 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/mediawiki/services/parsoid
refs/changes/98/160698/1
diff --git a/lib/dom.cleanup.js b/lib/dom.cleanup.js
index 701b1e6..4105389 100644
--- a/lib/dom.cleanup.js
+++ b/lib/dom.cleanup.js
@@ -55,7 +55,7 @@
return next;
}
- dp.tagId = undefined;
+ delete dp.tagId;
// Remove data-parsoid.src from templates and
extensions that have
// valid data-mw and dsr. This should reduce
data-parsoid bloat.
@@ -65,16 +65,16 @@
(!dp.tsr ||
node.getAttribute("data-mw") && dp.dsr &&
dp.dsr[0] && dp.dsr[1]))
{
- dp.src = undefined;
+ delete dp.src;
}
// Remove tsr
if (dp.tsr) {
- dp.tsr = undefined;
+ delete dp.tsr;
}
// Remove temporary information
- dp.tmp = undefined;
+ delete dp.tmp;
// Make dsr zero-range for fostered content
// to prevent selser from duplicating this content
diff --git a/lib/dom.markFosteredContent.js b/lib/dom.markFosteredContent.js
index 903df5c..9c723fd 100644
--- a/lib/dom.markFosteredContent.js
+++ b/lib/dom.markFosteredContent.js
@@ -137,7 +137,7 @@
DU.deleteNode( c );
} else if ( DU.isElt( c ) ) {
dp = DU.getDataParsoid( c );
- dp.inTransclusion = undefined;
+ delete dp.inTransclusion;
if ( c.childNodes.length > 0 ) {
markFosteredContent( c, env );
}
diff --git a/lib/dom.removeTransclusionDataParsoid.js
b/lib/dom.removeTransclusionDataParsoid.js
new file mode 100644
index 0000000..010a63e
--- /dev/null
+++ b/lib/dom.removeTransclusionDataParsoid.js
@@ -0,0 +1,57 @@
+"use strict";
+
+var DU = require('./mediawiki.DOMUtils.js').DOMUtils;
+
+function removeTransclusionDataParsoid(node, env, options, atTopLevel,
tplInfo) {
+ // Dont bother with this on sub-pipelines
+ if (!atTopLevel) {
+ return;
+ }
+
+ var c = node.firstChild;
+ while (c) {
+ var next = c.nextSibling;
+
+ if (DU.isElt(c)) {
+ // Identify template/extension content (not interested
in "mw:Param" nodes).
+ // We are interested in the very first node.
+ if (DU.isTplOrExtToplevelNode(env, c) &&
+
/(^|\s)mw:(Extension|Transclusion)/.test(c.getAttribute("typeof")))
+ {
+ // We know that tplInfo will be null here since
we don't
+ // mark up nested transclusions.
+ var about = c.getAttribute('about');
+ tplInfo = { first: c, last:
DU.getAboutSiblings(c, about).last() };
+ }
+
+ // Process subtree first
+ removeTransclusionDataParsoid(c, env, options,
atTopLevel, tplInfo);
+
+ // Cannot delete the first node since that carries the
transclusion
+ // information (typeof, data-mw). We could delete and
migrate
+ // the info over, but more pain than worth it. We can
reconsider if
+ // this ever becomes an issue.
+ if (tplInfo) {
+ if (c !== tplInfo.first) {
+ // TODO: We can't remove dp from nodes
with stx information
+ // right now or wt2wt fails
+ var dp = DU.getDataParsoid(c);
+ if (!dp.stx) {
+ DU.removeDataParsoid(c);
+ }
+ }
+
+ // Clear tpl info
+ if (c === tplInfo.last) {
+ tplInfo = null;
+ }
+ }
+ }
+
+ c = next;
+ }
+}
+
+if (typeof module === "object") {
+ module.exports.removeTransclusionDataParsoid =
removeTransclusionDataParsoid;
+}
diff --git a/lib/mediawiki.DOMPostProcessor.js
b/lib/mediawiki.DOMPostProcessor.js
index 636017c..2c36041 100644
--- a/lib/mediawiki.DOMPostProcessor.js
+++ b/lib/mediawiki.DOMPostProcessor.js
@@ -27,7 +27,8 @@
unpackDOMFragments =
require('./dom.t.unpackDOMFragments.js').unpackDOMFragments,
wrapTemplates = require('./dom.wrapTemplates.js').wrapTemplates,
stripEmptyElements =
require('./dom.stripEmptyElements.js').stripEmptyElements,
- lintWikitextFixup = require('./dom.linter.js').logWikitextFixups;
+ lintWikitextFixup = require('./dom.linter.js').logWikitextFixups,
+ removeTransclusionDataParsoid =
require('./dom.removeTransclusionDataParsoid.js').removeTransclusionDataParsoid;
// map from mediawiki metadata names to RDFa property names
var metadataMap = {
@@ -206,9 +207,15 @@
domVisitor2.addHandler( 'td', tableFixer.stripDoubleTDs.bind(
tableFixer, env ) );
domVisitor2.addHandler( 'td',
tableFixer.reparseTemplatedAttributes.bind( tableFixer, env ) );
domVisitor2.addHandler( 'th',
tableFixer.reparseTemplatedAttributes.bind( tableFixer, env ) );
- // 4. Save data.parsoid into data-parsoid html attribute.
- domVisitor2.addHandler( null, cleanupAndSaveDataParsoid.bind( null, env
) );
this.processors.push(domVisitor2.traverse.bind(domVisitor2));
+
+ // Remove data-parsoid from transcluded content
+ this.processors.push(removeTransclusionDataParsoid);
+
+ var domVisitor3 = new DOMTraverser(env);
+ // 4. Save data.parsoid into data-parsoid html attribute.
+ domVisitor3.addHandler( null, cleanupAndSaveDataParsoid.bind( null, env
) );
+ this.processors.push(domVisitor3.traverse.bind(domVisitor3));
}
// Inherit from EventEmitter
diff --git a/lib/mediawiki.DOMUtils.js b/lib/mediawiki.DOMUtils.js
index d901ab3..5021578 100644
--- a/lib/mediawiki.DOMUtils.js
+++ b/lib/mediawiki.DOMUtils.js
@@ -265,7 +265,8 @@
var val = data[key];
if ( val && val.constructor === String ) {
node.setAttribute('data-' + key, val);
- } else if (val instanceof Object) {
+ } else if (val instanceof Object &&
+ Object.keys(val).length > 0) {
this.setJSONAttribute(node, 'data-' + key, val);
}
// Else: throw error?
@@ -308,6 +309,16 @@
return this.setJSONAttribute( node, "data-parsoid", dpObj );
},
+ /**
+ * Remove the data-parsoid attribute on a node.
+ *
+ * @returns {Node} `node`, without data-parsoid
+ */
+ removeDataParsoid: function(node) {
+ delete this.getNodeData(node).parsoid;
+ return node;
+ },
+
getNodeData: function ( node ) {
if ( !node.dataobject ) {
node.dataobject = {};
--
To view, visit https://gerrit.wikimedia.org/r/160698
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: I695da6e4485ba62f3a1a4cc9e9fdd1a39637f327
Gerrit-PatchSet: 1
Gerrit-Project: mediawiki/services/parsoid
Gerrit-Branch: master
Gerrit-Owner: Marcoil <marc...@wikimedia.org>
_______________________________________________
MediaWiki-commits mailing list
MediaWiki-commits@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
