mw.libs.ve

Decode a URI component into a mediawiki article title

N.B. Illegal article titles can result from fairly reasonable input (e.g. "100%25beef"); see https://phabricator.wikimedia.org/T137847 T137847} .

De-duplicate TemplateStyles, like Parsoid does.

Disable redundant copies of TemplateStyles in a rendered surface.

#reduplicateStyles expands Parsoid's deduplicated TemplateStyles into one copy per occurrence, and ve.ce.MWTransclusionNode#filterRenderedDomElements keeps them in rendered content, so a template-heavy article has hundreds of copies of a handful of stylesheets all taking part in style resolution. They are identical, so only the first for each key has any effect.

Disabling rather than removing keeps the cascade unchanged and leaves the rendered DOM intact. Safe to re-run: the first surviving copy of each key is enabled and the rest disabled, so a survivor is promoted if the enabled copy is deleted mid-edit.

Encode a page title into a Parsoid resource name.

Fix fragment links which should be relative to the current document

This prevents these links from trying to navigate to another page, or open in a new window.

Call this after ve.targetLinksToNewWindow, as it removes the target attribute. Call this after LinkCache.styleParsoidElements, as it breaks that method by including the query string.

Convert a URL pattern with a $1 placeholder into a RegExp

Parse URL to get title it points to.

Extract the page title from a Parsoid resource name.

Split Parsoid resource name into the href prefix and the page title.

Re-duplicate deduplicated TemplateStyles, for correct rendering when editing a section or when templates are removed during the edit.

Take a string that might be a URL or a page title and return a URL

Smoothly scroll the window to a specific vertical position

Strip legacy (non-HTML5) IDs; typically found as section IDs inside headings.

Remove Parsoid section wrappers from a Parsoid HTML string

This produces the same document as #unwrapParsoidSections, minus the section IDs that method copies onto headings, so only use it where those are not needed, e.g. diffing.

It is much cheaper. #unwrapParsoidSections moves every child out of every section, and the cost of a DOM move is proportional to the subtree moved, so a long article pays for its whole tree. Doing it on the string instead costs nothing beyond the parse that follows.

Wikitext cannot produce a <section> element, so every closing tag belongs to one of Parsoid's wrappers. Should that ever stop holding, this returns null and the caller can unwrap the DOM.

Unwrap Parsoid sections

data-mw-section-id attributes are copied to the first child (the heading) during this step so that we can place the cursor in the correct place when section editing. These attributes must be removed before being sent back to Parsoid to avoid unnecessary re-serialization.