Page MenuHomePhabricator

Special:NewLexeme shows raw wiki markup
Closed, ResolvedPublicBUG REPORT

Description

Steps to replicate the issue (include links if applicable):

  • Open Special:NewLexeme with language set to pnb

What happens?:

Shows raw wiki markup in text

image.png (473×1 px, 95 KB)

What should have happened instead?:

Wiki markup should resolve

Software version (skip for WMF-hosted wikis like Wikipedia):

Other information (browser name/version, screenshots, etc.):

Event Timeline

There’s something strange going on in this message (wikibase-shortcopyrightwarning in pnb). Here’s what its wikitext looks like:

تسیں ”بݨا“ بٹن کریئے، ورتݨ دیاں شرطاں نال تسیں راضی ہندے اے، اتے تہاڈے کم لئی کریاٹیو کامنز سی‌سی 0 لئیسنس ورتوگے سی۔
<!-- جے ایہہ سنیہا تسیں دکھے، وِکی لکھت نال غلت ہو گئی۔ $1 [[$2]] $3 $4 -->

And here it is again, with every character surrounded by U+2068 FIRST STRONG ISOLATE and U+2069 POP DIRECTIONAL ISOLATE:

⁨ت⁩⁨س⁩⁨ی⁩⁨ں⁩⁨ ⁩⁨”⁩⁨ب⁩⁨ݨ⁩⁨ا⁩⁨“⁩⁨ ⁩⁨ب⁩⁨ٹ⁩⁨ن⁩⁨ ⁩⁨ک⁩⁨ر⁩⁨ی⁩⁨ئ⁩⁨ے⁩⁨،⁩⁨ ⁩⁨و⁩⁨ر⁩⁨ت⁩⁨ݨ⁩⁨ ⁩⁨د⁩⁨ی⁩⁨ا⁩⁨ں⁩⁨ ⁩⁨ش⁩⁨ر⁩⁨ط⁩⁨ا⁩⁨ں⁩⁨ ⁩⁨ن⁩⁨ا⁩⁨ل⁩⁨ ⁩⁨ت⁩⁨س⁩⁨ی⁩⁨ں⁩⁨ ⁩⁨ر⁩⁨ا⁩⁨ض⁩⁨ی⁩⁨ ⁩⁨ہ⁩⁨ن⁩⁨د⁩⁨ے⁩⁨ ⁩⁨ا⁩⁨ے⁩⁨،⁩⁨ ⁩⁨ا⁩⁨ت⁩⁨ے⁩⁨ ⁩⁨ت⁩⁨ہ⁩⁨ا⁩⁨ڈ⁩⁨ے⁩⁨ ⁩⁨ک⁩⁨م⁩⁨ ⁩⁨ل⁩⁨ئ⁩⁨ی⁩⁨ ⁩⁨ک⁩⁨ر⁩⁨ی⁩⁨ا⁩⁨ٹ⁩⁨ی⁩⁨و⁩⁨ ⁩⁨ک⁩⁨ا⁩⁨م⁩⁨ن⁩⁨ز⁩⁨ ⁩⁨س⁩⁨ی⁩⁨‌⁩⁨س⁩⁨ی⁩⁨ ⁩⁨0⁩⁨ ⁩⁨ل⁩⁨ئ⁩⁨ی⁩⁨س⁩⁨ن⁩⁨س⁩⁨ ⁩⁨ ⁩⁨و⁩⁨ر⁩⁨ت⁩⁨و⁩⁨گ⁩⁨ے⁩⁨ ⁩⁨س⁩⁨ی⁩⁨۔⁩
⁨<⁩⁨!⁩⁨-⁩⁨-⁩⁨ ⁩⁨ج⁩⁨ے⁩⁨ ⁩⁨ا⁩⁨ی⁩⁨ہ⁩⁨ہ⁩⁨ ⁩⁨س⁩⁨ن⁩⁨ی⁩⁨ہ⁩⁨ا⁩⁨ ⁩⁨ت⁩⁨س⁩⁨ی⁩⁨ں⁩⁨ ⁩⁨د⁩⁨ک⁩⁨ھ⁩⁨ے⁩⁨،⁩⁨ ⁩⁨و⁩⁨ِ⁩⁨ک⁩⁨ی⁩⁨ ⁩⁨ل⁩⁨ک⁩⁨ھ⁩⁨ت⁩⁨ ⁩⁨ن⁩⁨ا⁩⁨ل⁩⁨ ⁩⁨غ⁩⁨ل⁩⁨ت⁩⁨ ⁩⁨ہ⁩⁨و⁩⁨ ⁩⁨گ⁩⁨ئ⁩⁨ی⁩⁨۔⁩⁨ ⁩⁨$⁩⁨1⁩⁨ ⁩⁨[⁩⁨[⁩⁨$⁩⁨2⁩⁨]⁩⁨]⁩⁨ ⁩⁨$⁩⁨3⁩⁨ ⁩⁨$⁩⁨4⁩⁨ ⁩⁨-⁩⁨-⁩⁨>⁩

I think this should show the actual logical character order, all left-to-right; and the result seems to be that all four message arguments, and any link syntax, is wrapped in a comment (<!-- -->). So I have no idea why Special:NewItem?uselang=pnb still shows links, or why the wikitext syntax shows up on Special:NewLexeme – shouldn’t they both just show the first line of the message, all plain text?

Yes, it is strange - it seems specific to this string, in this version of Special:NewLexeme, in this locale. I initially added that commented line when the Special:NewLexeme (old) page and Special:NewLexemeAlpha page were both online, where the comment would be hidden on the first. (The comment saying something along the lines of, if you are seeing this message, it is because there is an error preventing this wikitext and substitutions from resolving.)

I think most of what is on the New Item page is translated through the on-wiki instance of Content Translation, which has not had this issue from what I have seen. It seems like the two places where an issue might be introduced are: 1) the delivery of the string from TranslateWiki to Special:NewLexeme or 2) the handling of the string by the surrounding elements of the Special:NewLexeme front end.

Since Special:NewLexeme is now updated in full, I have just updated the message to have no wikitext besides the substitution variables and a more complete explanation of the issue.

`
تسیں ”بݨا“ بٹن کریئے، ورتݨ دیاں شرطاں نال تسیں راضی ہندے اے، اتے تہاڈے کم لئی کریاٹیو کامنز زیرو لائیسنس ورتوگے سی۔ کوئی کجھ لیݨ دیݨ، عاوم یا نجی دی ورتوں لئی ایہتھوں اِجازت دتی اے۔

وِکیڈیٹے توں غلطی ہوگئی اے جو لوڑے جوڑ نہیں لگادے۔ ایہہ مُلاں ”$1“، ”[[$2]]“ تے ”[$3 $4]“ لئی لائسنس خاص پتے ویکھا چاہیدے۔ ایس غلطی بارے «phabricator.wikimedia.org» تے ”T322946“ مسئلہ لبھ جا سکدے او۔
`

Please just translate the message normally. It’s also used on Special:NewItem, where way more people will see it, so I don’t think it’s appropriate to insert a NewLexeme-specific complaint in your translation.

Also, I think the brackets in the current translatewiki.net version are the wrong way around – here they are wrapped in FSI/PDI again:

$ fsi=$(printf '\u2068'); pdi=$(printf '\u2069'); curl -s 'https://translatewiki.net/w/i.php?title=MediaWiki:Wikibase-shortcopyrightwarning/pnb&action=raw' | sed "s|\\(.\\)|${fsi}\\1${pdi}|g"; printf '\n'

image.png (99×2 px, 38 KB)

The message(s) on the new item page are different as far as I can tell.

The messages are also not legible at all due to being rendered in Italics, which breaks the glyph rendering. The translation is more for other people using TranslateWiki than anything. I do not understand why italics are being used here if users' interpretation of the message is a concern. It is preferable at least have a message rather than a mess of wiki markup in that.

The brackets are in the correct direction - TranslateWiki will actually not let you publish translations without brackets in the correct direction. The markup also resolves in the Translatewiki interface. Using FSI/PDI that way just forces the characters to render that way. They do not behave like quotation marks which can make things confusing.

OK the message and the bottom of the new item page is the same but is getting resolved differently - I figured out the issue by process of elimination looking at the other translations and the new lexeme page source code - the license name variable is hard coded for Special:NewLexeme and substituted in English, something else is happening on Special:NewLexeme. I think if I allow it to insert the English string it will resolve, it was possible to translate the whole thing on the old page

I can make a pull request to remove the italic text.