jenkins-bot has submitted this change and it was merged.
Change subject: Updated editing support for tpl args with nowiki escaping in
arg wt
......................................................................
Updated editing support for tpl args with nowiki escaping in arg wt
* String fragments in edited params that might interfere with
wikitext parsing are escaped. Tokenizes the arg and tests string
tokens for bad chars.
* TODO: Figure out a way to test this functionality.
* Tested with these and other combinations:
[subbu@earth tests] echo "{{echo | foo}}" | node parse | sed 's/foo/[[
[[ }} {{ bar }} | baz]]/g;' | node parse --html2wt --editMode
{{echo |<nowiki> [[ [[ }} </nowiki>{{ bar }}<nowiki> | baz]]</nowiki>}}
[subbu@earth tests] echo "{{echo | foo}}" | node parse | sed 's/foo/[[
[[ {{ bar }} | baz]]/g;' | node parse --html2wt --editMode
{{echo |<nowiki> [[ </nowiki>[[ {{ bar }} | baz]]}}
[subbu@earth tests] echo "{{echo | foo}}" | node parse | sed 's/foo/[[
{{ bar }} | baz]]/g;' | node parse --html2wt --editMode
{{echo | [[ {{ bar }} | baz]]}}
[subbu@earth tests] echo "{{echo|foo}}" | node parse | sed
's/"foo"/"<s>a}}"/g;' | node parse --html2wt --editMode
{{echo|<nowiki><s>a}}</nowiki>}}
[subbu@earth tests] echo "{{echo|foo}}" | node parse | sed
's/"foo"/"{{a<\/s>"/g;' | node parse --html2wt --editMode
{{echo|<nowiki>{{a</nowiki></s>}}
[subbu@earth tests] echo "{{echo|foo}}" | node parse | sed
's/"foo"/"{{a}}<\/s>"/g;' | node parse --html2wt --editMode
{{echo|{{a}}</s>}}
[subbu@earth tests] echo "{{echo|foo}}" | node parse | sed
's/"foo"/"<s>{{a}}"/g;' | node parse --html2wt --editMode
{{echo|<nowiki><s>{{a}}</nowiki>}}
Change-Id: If1b9c7c1c22dd9a012b5afa48f37c0bc7fd0a165
---
M js/lib/mediawiki.WikitextSerializer.js
1 file changed, 105 insertions(+), 32 deletions(-)
Approvals:
GWicke: Looks good to me, approved
jenkins-bot: Verified
diff --git a/js/lib/mediawiki.WikitextSerializer.js
b/js/lib/mediawiki.WikitextSerializer.js
index 8ff92bf..6db9b41 100644
--- a/js/lib/mediawiki.WikitextSerializer.js
+++ b/js/lib/mediawiki.WikitextSerializer.js
@@ -189,32 +189,8 @@
// console.warn("---HWT---:onl:" + onNewline + ":" + text);
// tokenize the text
- var prefixedText = text;
- if (!onNewline) {
- // Prefix '_' so that no start-of-line wiki syntax matches.
- // Later, strip it from the result.
- // Ex: Consider the DOM: <ul><li> foo</li></ul>
- // We don't want ' foo' to be converted to a <pre>foo</pre>
- // because of the leading space.
- prefixedText = '_' + text;
- }
-
- if ( state.inIndentPre || state.inPHPBlock ) {
- prefixedText = prefixedText.replace(/(\r?\n)/g, '$1_');
- }
-
- var p = new PegTokenizer( state.env ), tokens = [];
- p.on('chunk', function ( chunk ) {
- // Avoid a stack overflow if chunk is large, but still update
token
- // in-place
- for ( var ci = 0, l = chunk.length; ci < l; ci++ ) {
- tokens.push(chunk[ci]);
- }
- });
- p.on('end', function(){ });
-
- // The code below will break if use async tokenization.
- p.processSync( prefixedText );
+ var sol = onNewline && !(state.inIndentPre || state.inPPHPBlock);
+ var tokens = state.serializer.tokenizeStr(state, text, sol);
// If the token stream has a pd.TagTk, pd.SelfclosingTagTk, pd.EndTagTk
or pd.CommentTk
// then this text needs escaping!
@@ -550,6 +526,25 @@
return ["<nowiki>", match[1], "</nowiki>", match[2]].join('');
}
+WSP.tokenizeStr = function(state, str, sol) {
+ var p = new PegTokenizer( state.env ), tokens = [];
+ p.on('chunk', function ( chunk ) {
+ // Avoid a stack overflow if chunk is large,
+ // but still update token in-place
+ for ( var ci = 0, l = chunk.length; ci < l; ci++ ) {
+ tokens.push(chunk[ci]);
+ }
+ });
+ p.on('end', function(){ });
+
+ // Init sol state
+ p.savedSOL = sol;
+
+ // The code below will break if use async tokenization.
+ p.processSync(str);
+ return tokens;
+};
+
WSP.escapeWikiText = function ( state, text, opts ) {
// console.warn("---EWT:ALL1---");
// console.warn("t: " + text);
@@ -692,9 +687,87 @@
}
};
-WSP.escapeTplArgWT = function(arg) {
- // FIXME: to be done
- return arg;
+/**
+ * General strategy:
+ *
+ * Tokenize the arg wikitext. Anything that parses as tags
+ * are good and we need not bother with those. Check for harmful
+ * characters "[]{}|=" in any strings and escape those fragments
+ * since these characters could change semantics of the entire
+ * template transclusion.
+ *
+ * This function makes a couple of assumptions:
+ *
+ * 1. Each arg's wikitext has balanced opening and closing tags.
+ * The code does make some attempt to deal with unclosed opening
+ * and closing tags, but this may not work accurately in all cases.
+ *
+ * 2. The tokenizer sets tsr on all non-string tokens.
+ */
+WSP.escapeTplArgWT = function(state, arg) {
+ function escapeStr(str, buf) {
+ if (str.match(/[\{\[\]\}\|\=]/)) {
+ buf.push("<nowiki>");
+ buf.push(str);
+ buf.push("</nowiki>");
+ } else {
+ buf.push(str);
+ }
+ }
+
+ var tokens = this.tokenizeStr(state, arg, false);
+ var offset = 0, openTags = [], buf = [];
+ for (var i = 0, n = tokens.length; i < n; i++) {
+ var t = tokens[i], da = t.dataAttribs;
+ switch (t.constructor) {
+ case pd.TagTk:
+ if (openTags.length === 0) {
+ offset = da.tsr[0];
+ }
+ openTags.push(t);
+ break;
+
+ case pd.EndTagTk:
+ if (openTags.length > 0) {
+ openTags.pop();
+ }
+ if (openTags.length === 0) {
+ var e = da.tsr[1];
+ buf.push(arg.substring(offset, e));
+ offset = e;
+ }
+ break;
+
+ case pd.NlTk:
+ case pd.Comment:
+ case pd.SelfclosingTagTk:
+ if (openTags.length === 0) {
+ buf.push(arg.substring(da.tsr[0],
da.tsr[1]));
+ offset = da.tsr[1];
+ }
+ break;
+
+ case String:
+ if (openTags.length === 0) {
+ escapeStr(t, buf);
+ offset += t.length;
+ }
+ break;
+
+ case pd.EOFTk:
+ // Deal with any unprocessed fragments (that
should exist
+ // only if we have unbalanced opening/closing
tags).
+ var str = arg.substring(offset, arg.length);
+ if (openTags.length === 0) {
+ buf.push(str);
+ } else {
+ escapeStr(str, buf);
+ }
+ break;
+ }
+ }
+
+ return buf.join('');
};
/**
@@ -2076,7 +2149,7 @@
node, state, cb);
};
-WSP._buildTemplateWT = function(srcParts) {
+WSP._buildTemplateWT = function(state, srcParts) {
var buf = [],
serializer = this;
srcParts.map(function(part) {
@@ -2094,7 +2167,7 @@
if (n > 0) {
for (var i = 0; i < n; i++) {
var k = keys[i],
- v =
serializer.escapeTplArgWT(tpl.params[k].wt);
+ v =
serializer.escapeTplArgWT(state, tpl.params[k].wt);
if (k === (i+1).toString()) {
argBuf.push(v);
} else {
@@ -2168,7 +2241,7 @@
} else {
var dataMW =
JSON.parse(node.getAttribute("data-mw"));
if (dataMW) {
- src =
state.serializer._buildTemplateWT(dataMW.parts || [{ template: dataMW }]);
+ src =
state.serializer._buildTemplateWT(state, dataMW.parts || [{ template: dataMW
}]);
} else {
console.error("ERROR:
No data-mw for: " + node.outerHTML);
src = dp.src;
--
To view, visit https://gerrit.wikimedia.org/r/61635
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings
Gerrit-MessageType: merged
Gerrit-Change-Id: If1b9c7c1c22dd9a012b5afa48f37c0bc7fd0a165
Gerrit-PatchSet: 3
Gerrit-Project: mediawiki/extensions/Parsoid
Gerrit-Branch: master
Gerrit-Owner: Subramanya Sastry <[email protected]>
Gerrit-Reviewer: Cscott <[email protected]>
Gerrit-Reviewer: GWicke <[email protected]>
Gerrit-Reviewer: jenkins-bot
_______________________________________________
MediaWiki-commits mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits
