% This is the macro package used by OpTeX, see http://petr.olsak.net/optex

% vlna.opm, Petr Olšák <petr@olsak.net>, 2021

% Vlna prevents breaks in spaces after single-letter prepositions, after (or
% before) degrees, and between value and SI units. It enables or disables
% split hyphens too. All folks are done at the postprocessing phase before TeX splits
% the typesetting material into the paragraph lines.

% vlna.opm is based on Lua code from the package `luavlna` (by Michal Hoftich
% and Miro Hrončok). See `texdoc luavlna` or end of this file for more information.

\_codedecl \singlechars {Preventing single-chars prepositions and more <2023-06-23>}

\_namespace{vlna}

   \_doc -----------------------------
   We declare new attribute, register callbacks, and define Lua functions that
   point to Lua functions from `luavlna.lua` package.
   We need this package installed.

   We need not `require ltluatex` because `luatexbase.callbaks` are defined
   in the `optex.lua` file and it is part of the OpTeX format.
   \_cod -----------------------------

\_newattribute \.preventsinglestatus

\_directlua{%
   luavlna = require "luavlna"

   luatexbase.add_to_callback("pre_linebreak_filter", luavlna.preventsingle,"LuaVlna")
   luatexbase.add_to_callback("hyphenate", luavlna.split_hyphens, "allow hyphen breaks")
   luavlna.preventsingleid = \_the\_attributealloc

   set_singlechars = function(lang,chars)
     if (lang == "") then return nil end
     local chars = chars  or ""
     local singlechars = {}
     for pos, char in utf8.codes(chars) do
       singlechars[utf8.char(char)] = true
     end
     luavlna.singlechars(lang,singlechars)
   end

   set_compounds = function(lang, compounds)
     if (lang == "") then return nil end
     local c = compounds:explode(",+") % match multiple colons as one
     local  compoundstable = {}
     for _, compound in pairs(c) do
       compoundstable[compound] = true
     end
     luavlna.initials(lang, compoundstable)
   end

   set_main_language = function(lang)
     if (lang == "") then return nil end
     luavlna.set_main_language(tonumber(lang))
   end	

   enable_split_hyphens = function(lang)
     if (lang == "") then return nil end
     luavlna.split_hyphen_langs[tonumber(lang)] = true
   end

   disable_split_hyphens = function(lang)
     if (lang == "") then return nil end
     luavlna.split_hyphen_langs[tonumber(lang)] = nil
   end
}

   \_doc -----------------------------
   We need not `require luavlna-langno` because of the conversion from
   language name to the `\language` number is done by the macro
   `\.langtonum``<sequence>={<langname>}` defined here.
   It does `\def<sequence>{<langnumber>}` or `\def\sequence{}` if <langname>
   is unknown. The <langname> is case insensitive and these names are
   declared in the hyphen-lan.opm file, section 2.37.2 in OpTeX documentation.
   \_cod -----------------------------

\_def\.langtonum#1=#2{\_lowercase{\_edef#1{\_trycs{_ulan:#2}{}}}%
   \_ifx#1\_empty \_opwarning{vlna: unknown language "#2", ignored}%
   \_else \_edef#1{\_the\_cs{_#1Patt}}\_fi}

   \_doc -----------------------------
   `\singlechars``{<langname>}{<prepositions>}` sets <prepositions> for
   given language.\nl
   `\compoundinitials``{<langname>}{<comma-separated-list>}` sets
   compound characters from the list for given language. 
   \_cod -----------------------------

\_def\.singlechars#1#2{\.langtonum\.l={#1}\_directlua{set_singlechars("\.l","#2")}}
\_def\.compoundinitials#1#2{\.langtonum\.l={#1}\_directlua{set_compounds("\.l","#2")}}

   \_doc -----------------------------
   `\enablesplithyphens` and `\disablesplithyphens` are dependent on the language.
   \_cod -----------------------------

\_def\.enablesplithyphens#1{\.langtonum\.l={#1}\_directlua{enable_split_hyphens("\.l")}}
\_def\.disablesplithyphens#1{\.langtonum\.l={#1}\_directlua{disable_split_hyphens("\.l")}}

   \_doc -----------------------------
   By default, the language-dependent features of the `luavlna` depend on
   the current `\language` value, i.e. on the currently selected hyphenation.
   `\preventsinglelang``{<langname>}` activates the language-dependent
   features regardless of the current `\language` value.
   \_cod -----------------------------

\_def\.preventsinglelang#1{\.langtonum\.l={#1}\_directlua{set_main_language("\.l")}}

   \_doc -----------------------------
   Switching on/off `luavlna` preposition features should be done using attribute
   setting. It will allow you to switch them on/off in the middle of the paragraph.
   \_cod -----------------------------

\_def\.preventsingleon{\.preventsinglestatus=2 }
\_def\.preventsingleoff{\.preventsinglestatus=1 }

   \_doc -----------------------------
   Switching on/off the other `luavlna` features is implemented only by
   setting a special `luavlna` variable.
   \_cod -----------------------------

\_def\.preventsingledebugon{\_directlua{luavlna.debug(true)}}
\_def\.preventsingledebugoff{\_directlua{luavlna.debug(false)}}

\_def\.nopredegrees{\_directlua{luavlna.no_predegrees = true}}
\_def\.nosufdegrees{\_directlua{luavlna.no_sufdegrees = true}}
\_def\.nounits{\_directlua{luavlna.no_unit = true}}
\_def\.noinitials{\_directlua{luavlna.no_initials = true}}

\_nspublic
   \singlechars            % declare single-letter prepositions per language
   \compoundinitials       % declare compound characters per language
   \enablesplithyphens     % enable split hyphens per language
   \disablesplithyphens    % disable split hyphens per language
   \preventsinglelang      % declare main language independent of current hyphenation
   \preventsingleon        % switch on the prepositions feature (default)
   \preventsingleoff       % switch off the prepositions feature
   \nopredegrees           % switch off pre-degrees feature (default is on)
   \nosufdegrees           % switch off post-degrees feature (default is on)
   \nounits                % switch off SI units feature (default is on)
   \noinitials             % switch off initials feature (default is on)
   \preventsingledebugon   % switch on debug printing (purple squares)
   \preventsingledebugoff  % switch off debug printing (default)
   ;

\_endnamespace

% Default values:
\singlechars{Czech}{AIiVvOoUuSsZzKk}
\singlechars{Slovak}{AIiVvOoUuSsZzKk}
\compoundinitials{Czech}{Ch,CH}  % doesn't work in version v0.1g of luavlna
\enablesplithyphens{Czech}
\enablesplithyphens{Slovak}
% + predegrees, sufdegrees, units, initials (they are independent of language value)

\_endcode

What should be done automatically after `\load[vlna]`:

   Prepositions feature: v lese -> v~lese,  (A taky zde) -> (A~taky zde).

   Pre-degrees feature: Mgr. Bown -> Mgr.~Brown.

   Post-degrees feature: John Brown, PhD. -> John Brown,~PhD.

   SI units feature: 42 cm -> 42~cm (with thin space)

   Initials feature:  J. Brown -> J.~Brown

   Split hyphens feature: propan-butan -> propan\discretionary{-}{-}{-}butan.

The prepositions and split-hyphens features are language-dependent, the other
features are language-independent.

A testing file:

\begtt
\load[vlna]
\fontfam[LMfonts]
\cslang
\preventsingledebugon

Mgr. J. Brown, PhD., zkusí v lese (A taky v závorce) zkontrolovat rozměr 42 m.
Obejde se bez propan-butanových lahví.
\bye
\endtt

Note, that versinon v0.1g 2020-06-02 of the luavlna package is an
experimental code. It seems that `\compoundinitials` does not work in this
version and when you use `\hbox{-}` in a paragraph when
`\enablesplithyphens` then LuaTeX hangs.