https://bugs.documentfoundation.org/show_bug.cgi?id=39935

--- Comment #22 from Christophe Strobbe <[email protected]> ---
(In reply to Heiko Tietze from comment #21)
> I wonder what the "primary language" (if not to say only) of a document
> would be that is written in Arab with English citations. And why do you need
> this, given that style:language* is an attribute at the paragraph?

Even though LibreOffice Writer's UI allows you to set a language "For all text"
(see Tools > Language > For All Text ...), I have no idea where this is
reflected in the document (XML, manifest ...).

However, at the paragraph level, elements such as text:p and text:span have the
language or languages identified via a a style (e.g. text:style-name="T6"). If
the value of each language category (Western, Asian, CTL) is set to a language,
the style element (more specifically, the element style:text-properties inside
style:style) will identify three languages, e.g. English for Western,
Simplified Chinese for Asian and Hindi for CTL. For a system that only
processes markup, e.g. an XSLT conversion, and does not perform language
detection on top of that, the language markup is ambiguous.

However, if, for example, for a paragraph in English, you explicitly set Asia
and CTL to "None", the style:text-properties elements will reflect this by
means of the following attributes:

style:language-asian="zxx" style:country-asian="none"
style:language-complex="zxx" style:country-complex="none"

This is unambiguous for systems that process just the XML markup, for example
for conversion to DAISY (the original use case in 2011), where correct language
markup is important for text-to-speech conversion.

-- 
You are receiving this mail because:
You are the assignee for the bug.

Reply via email to