LaTeX writer/template: Improve babel support.

Previously we used the `.ini` files for every language, but
for European languages these tend to provide inferior results
to the `.ldf` files used by classic Babel. Currently Babel
documentation recommends using the classic system for European
languages written in Latin and Cyrillic scripts and Vietnamese.
So the LaTeX writer and template now follow this guidance.

Main languages in the list of languages with good "classic" support
are added to global documentclass options and will be automatically
handled by Babel using the `.ldf` files.

If the main language is not in this list, the `babeloptions` variable
will be set to `provide=*`, which will cause support to be loaded from
the `.ini` file rather than an `.ldf`. So, for example, setting
`-V babeloptions=''` with a polytonic Greek document will cause the
`.ldf` support to be used instead of the `.ini`.

The default setting of this variable can be overwritten, but in most
cases the default should give good results.

Closes #8283.

1 files changed, 100 insertions, 1 deletions

diff --git a/src/Text/Pandoc/Writers/LaTeX.hs b/src/Text/Pandoc/Writers/LaTeX.hs
index 77d40c3e7..93601a72e 100644
--- a/src/Text/Pandoc/Writers/LaTeX.hs
+++ b/src/Text/Pandoc/Writers/LaTeX.hs
@@ -265,9 +265,15 @@ pandocToLaTeX options (Pandoc meta blocks) = do
                       (literal $ renderLang l)) mblang
         $ maybe id (\l -> defField "babel-lang"
                       (literal l)) babelLang
+        $ (case babelLang of -- see #8283
+                Just l | l `notElem` ldfLanguages
+                         -> defField "babeloptions" ("provide=*" :: Text)
+                _ -> id)
         $ defField "babel-otherlangs"
              (map literal
-               (nubOrd . catMaybes . filter (/= babelLang)
+               (filter (`elem` ldfLanguages) .
+                nubOrd . catMaybes .
+                filter (/= babelLang)
                 $ map toBabel docLangs))
         $ defField "selnolig-langs"
              (literal . T.intercalate "," $
@@ -1187,3 +1193,96 @@ inSoulCommand pa = do
   result <- pa
   modify $ \st -> st{ stInSoulCommand = oldInSoulCommand }
   pure result
+
+-- Babel languages with a .ldf that works well with all engines (see #8283).
+-- We follow the guidance from the Babel documentation:
+-- "In general, you should do this for European languages written in Latin
+-- and Cyrillic scripts, as well as for Vietnamese."
+ldfLanguages :: [Text]
+ldfLanguages =
+  [ "magyar"
+  , "croatian"
+  , "ngerman"
+  , "germanb"
+  , "german"
+  , "austrian"
+  , "ngermanb"
+  , "naustrian"
+  , "nswissgerman"
+  , "swissgerman"
+  , "italian"
+  , "greek"
+  , "azerbaijani"
+  , "american"
+  , "newzealand"
+  , "UKenglish"
+  , "USenglish"
+  , "australian"
+  , "british"
+  , "canadian"
+  , "english"
+  , "bahasa"
+  , "slovak"
+  , "finnish"
+  , "occitan"
+  , "swedish"
+  , "brazil"
+  , "portuguese"
+  , "portuges"
+  , "brazilian"
+  , "spanish"
+  , "norwegian"
+  , "norsk"
+  , "nynorsk"
+  , "bulgarian"
+  , "breton"
+  , "belarusian"
+  , "piedmontese"
+  , "esperanto"
+  , "lithuanian"
+  , "ukraineb"
+  , "scottishgaelic"
+  , "scottish"
+  , "dutch"
+  , "afrikaans"
+  , "czech"
+  , "serbian"
+  , "latvian"
+  , "catalan"
+  , "basque"
+  , "albanian"
+  , "irish"
+  , "serbianc"
+  , "interlingua"
+  , "bosnian"
+  , "friulan"
+  , "romanian"
+  , "icelandic"
+  , "classiclatin"
+  , "ecclesiasticlatin"
+  , "medievallatin"
+  , "latin"
+  , "georgian"
+  , "macedonian"
+  , "welsh"
+  , "vietnamese"
+  , "romansh"
+  , "danish"
+  , "lsorbian"
+  , "usorbian"
+  , "polish-compat"
+  , "polish"
+  , "estonian"
+  , "french"
+  , "frenchb"
+  , "canadien"
+  , "acadian"
+  , "francais"
+  , "turkish"
+  , "hindi"
+  , "northernsami"
+  , "samin"
+  , "russianb"
+  , "galician"
+  , "slovene"
+  ]