704671 - Add L20n bindings for HTML5

Status: RESOLVED INCOMPLETE

(Core :: Internationalization: Localization, defect)

Comment 1

[Zibi Braniecki [:zbraniecki][:gandalf]]

Attachment: WIP 1

WIP that works with code from bug 704500 to make HTML5 localizable

Comment 2

[Zibi Braniecki [:zbraniecki][:gandalf]]

Attachment: WIP 1

d'uh, that was just hg stat, not hg diff. Proper WIP patch

Comment 3

[Henri Sivonen (:hsivonen) (not reading bugmail until 2026-07-27)]

Comment on attachment 576344 [details] [diff] [review]
WIP 1

>         } else if (nsHtml5Atoms::link == aName) {
>           nsString* rel = aAttributes->getValue(nsHtml5AttributeName::ATTR_REL);
>+          nsString* type = aAttributes->getValue(nsHtml5AttributeName::ATTR_TYPE);
>           // Not splitting on space here is bogus but the old parser didn't even
>           // do a case-insensitive check.
>           if (rel && rel->LowerCaseEqualsASCII("stylesheet")) {
>             nsString* url = aAttributes->getValue(nsHtml5AttributeName::ATTR_HREF);
>             if (url) {
>               nsString* charset = aAttributes->getValue(nsHtml5AttributeName::ATTR_CHARSET);
>               mSpeculativeLoadQueue.AppendElement()->InitStyle(*url,
>                                                     (charset) ? *charset : EmptyString());
>             }
>+          } else if (type && type->LowerCaseEqualsASCII("intl/l20n")) {

Instead of dispatching on the type attribute and minting a new mediatype (if you do mint a new media type, it should be text/l20n, not intl/l20n), it would make more sense to mint a new rel token, such as "l10n" and check for that. (Note that in principle, you should check if any space-separated token in the rel value matches your token. The stylesheet check above is only an optimization and, therefore, cheats and doesn't deal with rel values that have multiple tokens.)

>+        if (localName->Equals(NS_LITERAL_STRING("l10n-id"))) {

You should add l10n-id to nsGkAtomList and then do a pointer comparison between localName and nsGkAtoms::l10nid here.

>+    case eTreeOpL20nResourceAdded: {
>+      PRUnichar* buffer = mOne.unicharPtr;

You need to add 
case eTreeOpL20nResourceAdded:
to the part of the nsHtml5TreeOperation's destructor that delete[]s mOne.unicharPtr;

Also, since this isn't ready to be exposed to Web content without the shadow tree stuff, this implementation needs a check to see if the document is a chrome doc so that the l20n stuff only runs for chrome docs until we are ready to expose this to Web content in general.

Comment 4

[Zibi Braniecki [:zbraniecki][:gandalf]]

Seven years later, we're making another attempt to refactor our l10n layer.

The new tracking bug is bug 1365426 and I'll mark the previous effort as "INCOMPLETE".