Cscott has uploaded a new change for review.
https://gerrit.wikimedia.org/r/156013
Change subject: WIP: WTS: handle absolute URLs (/wiki/...) better.
......................................................................
WIP: WTS: handle absolute URLs (/wiki/...) better.
This helps us with html2wt of parser tests written for PHP which use
<a> tags with absolute links and without `typeof` attributes.
Change-Id: I9a535d07d19b4a9c3af677c301d217b889f8441f
---
M lib/wts.LinkHandler.js
1 file changed, 40 insertions(+), 4 deletions(-)
git pull ssh://gerrit.wikimedia.org:29418/mediawiki/services/parsoid
refs/changes/13/156013/1
diff --git a/lib/wts.LinkHandler.js b/lib/wts.LinkHandler.js
index b08c503..5b31603 100644
--- a/lib/wts.LinkHandler.js
+++ b/lib/wts.LinkHandler.js
@@ -1,6 +1,8 @@
"use strict";
require('./core-upgrade.js');
+var url = require('url');
+
var Util = require('./mediawiki.Util.js').Util,
DU = require('./mediawiki.DOMUtils.js').DOMUtils,
pd = require('./mediawiki.parser.defines.js'),
@@ -34,6 +36,36 @@
}
};
+// Helper function for munging protocol-less absolute URLs:
+// If this URL is absolute, but doesn't contain a protocol,
+// try to find a localinterwiki protocol that would work.
+var getHref = function( env, node ) {
+ var href = node.getAttribute('href') || '';
+ if (/^\/[^\/]/.test(href)) {
+ // protocol-less but absolute. let's find a base href
+ var i, bases = [], nhref;
+ env.conf.wiki.interwikiMap.forEach(function(interwikiInfo,
prefix) {
+ if (interwikiInfo.localinterwiki !== undefined &&
+ interwikiInfo.url !== undefined) {
+ // this is a possible base href
+ bases.push(interwikiInfo.url);
+ }
+ });
+ for (i=0; i<bases.length; i++) {
+ // evaluate the url relative to this base
+ nhref = url.resolve(bases[i], href);
+ // can this match the pattern?
+ var re = '^' +
+
bases[i].split('$1').map(Util.escapeRegExp).join('[\\s\\S]*') +
+ '$';
+ if (new RegExp(re).test(nhref)) {
+ return nhref;
+ }
+ }
+ }
+ return href;
+}
+
// Helper function for getting RT data from the tokens
var getLinkRoundTripData = function( env, node, state ) {
var dp = DU.getDataParsoid( node );
@@ -54,7 +86,7 @@
}
}
- var href = node.getAttribute('href') || '';
+ var href = getHref( env, node );
// Save the token's "real" href for comparison
rtData.href = href.replace( /^(\.\.?\/)+/, '' );
@@ -244,6 +276,10 @@
// Get the rt data from the token and tplAttrs
linkData = getLinkRoundTripData(env, node, state);
+ if (linkData.type === null && !node.querySelector('IMG')) {
+ linkData.type = 'mw:ExtLink';
+ }
+
if ( linkData.type !== null && linkData.target.value !== null ) {
// We have a type and target info
@@ -257,7 +293,7 @@
}
var target = linkData.target,
- href = node.getAttribute('href') || '';
+ href = getHref( env, node );
if
(/\b(mw:ExtLink|mw:PageProp\/Language)\b/.test(linkData.type)) {
var targetVal = target.fromsrc || true ? target.value :
Util.decodeURI(target.value);
// Check if the href matches any of our interwiki URL
patterns
@@ -479,7 +515,7 @@
if ( contentStr &&
// Can we minimize this?
( target.value === contentStr ||
- node.getAttribute('href') ===
contentStr) &&
+ getHref( env, node ) === contentStr) &&
// But preserve non-minimal encoding
(target.modified ||
linkData.contentModified || dp.stx === 'url'))
{
@@ -560,7 +596,7 @@
} else {
// href is already percent-encoded, etc., but it might
contain
// spaces or other wikitext nasties. escape the
nasties.
- var hrefStr =
escapeExtLinkURL(node.getAttribute('href'));
+ var hrefStr = escapeExtLinkURL(getHref( env, node ));
cb( '[' + hrefStr + ' ' +
state.serializeLinkChildrenToString(node,
this.wteHandlers.aHandler, false) +
']', node );
--
To view, visit https://gerrit.wikimedia.org/r/156013
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: newchange
Gerrit-Change-Id: I9a535d07d19b4a9c3af677c301d217b889f8441f
Gerrit-PatchSet: 1
Gerrit-Project: mediawiki/services/parsoid
Gerrit-Branch: master
Gerrit-Owner: Cscott <canan...@wikimedia.org>
_______________________________________________
MediaWiki-commits mailing list
MediaWiki-commits@lists.wikimedia.org
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
