László Németh
e668dc84eb
Add new paragraph justification option "Split compound words" for
custom compound-based "smart" hyphenation. Add compound-based "smart"
hyphenation support for Danish, Dutch, Esperanto, Estonian, German,
Norwegian, Swedish (and potentially, Icelandic, Finnish and
Latvian) languages by fixing Hunspell morphological analysis and the
associated lingucomponent code. (Previously compound-based "smart"
hyphenation was enabled only for Hungarian.)
With the new option it's possible to keep MSO/MS-DOCX interoperability,
also introducing a better hyphenation algorithm for several languages.
Details:
* Add check box to Alignment in paragraph formatting dialog window;
* Store properties in paragraph model:
css::style::ParagraphProperties::ParaIsCompoundBased
* Implement it by checking the new option in the previously
added commit f9f694f5d9f39c5801bc5fec45867252c293bbcb;
* Add ODF import/export (loext:compound-based);
* Add ODF unit tests by the new tdf173389_smart_hyphenation_nl.odt
and tdf173389_smart_hyphenation_hu.odt. Now use the previous
tdf119908_smart_hyphenation_hu for disabled compound-based "smart"
hyphenation, also the new tdf119908_smart_hyphenation_nl.odt.
Note: the Dutch test document results better compound-based
hyphenation in two positions.
* Add Hunspell patch to fix Danish, Dutch, Esperanto, Estonian,
German, Norwegian, Swedish (and future Icelandic, Latvian,
Finnish) morphological analysis for compound word constituents;
* hyphenimp.cxx hyphenate(): fix tokenization for consecutive
pa: elements, like resulted in the fixed Hunspell dictionaries.
Follow-up to commit f9f694f5d9
"tdf#173389 sw: add compound-based “smart” hyphenation for readability".
Change-Id: I4414bd588646ecf6dc86b77642833ac9c7820a98
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/211341
Tested-by: Jenkins
Reviewed-by: László Németh <nemeth@numbertext.org>
Schemas for Validating ODF Files
Schemas that can be used for validating ODF files
subdirs:
mathml2: W3C MathML 2.0 XML Schema (needed for Math embedded objects)odf1.0,odf1.1,odf1.2: official OASIS RelaxNG schemasodf1.3: current OASIS draft ODF 1.3 RelaxNG schemalibreoffice: draft ODF schema, with additional LO extensions
The extension schema in libreoffice/ is used by all unit tests if
--with-export-validation is given, which is the default.
Notably this means that if you add a new feature to the ODF filters and you add the required unit test for the new feature, then most likely the test will fail with a complaint from the validator; in this case the schema needs to be updated to contain the new elements and attributes.
The extension schema uses the RelaxNG "include" feature to refer to the ODF
schema; this means that it only contains those parts of the schema that
actually need to be changed - this works well in many cases because the ODF
schema is quite well structured with many named patterns, but unfortunately
there are a few places where that isn't the case and large chunks needed to be
copied to override them.
In the easy case, to add an attribute you just want to search for the
corresponding element, which will have a "foo-attlist" named pattern, and then
add another attribute like this:
<rng:define name="draw-enhanced-geometry-attlist" combine="interleave"> rng:optional <rng:attribute name="drawooo:sub-view-size"> <rng:ref name="string"/> </rng:attribute> </rng:optional> </rng:define>
Currently only the features that are actually exported in the unit tests have been added to the schema; there is still some work to do here to add everything; the crashtesting script also does ODF validation of all files and now also uses the custom schema.
Unfortunately it turned out that there are a lot of extensions already for which no proposal exists [1], and in many cases not even an entry on the Wiki [2], so clearly something like this extension schema is needed.
[1] git grep TODO schema/libreoffice