%D \module
%D   [       file=lang-mis,
%D        version=1997.03.20, % used to be supp-lan.tex
%D          title=\CONTEXT\ Language Macros,
%D       subtitle=Compounds,
%D         author=Hans Hagen,
%D           date=\currentdate,
%D      copyright={PRAGMA ADE \& \CONTEXT\ Development Team}]
%C
%C This module is part of the \CONTEXT\ macro||package and is
%C therefore copyrighted by \PRAGMA. See mreadme.pdf for
%C details.

%D This one will be updated stepwise to \LMTX. See lang-mis.mkiv for previous
%D implementations and removed code.

\writestatus{loading}{ConTeXt Language Macros / Compounds}

%D More or less replaced.

%D \gdef\starttest#1\stoptest{\starttabulate[|l|l|p|]#1\stoptabulate}
%D \gdef\test     #1{\NC\detokenize{#1}\NC\hyphenatedword{#1}\NC#1\NC\NR}

\unprotect

%D One of \TEX's strong points in building paragraphs is the way hyphenations are
%D handled. Although for real good hyphenation of non||english languages some
%D extensions to the program are needed, fairly good results can be reached with the
%D standard mechanisms and an additional macro, at least in Dutch.
%D
%D \CONTEXT\ originates in the wish to typeset educational materials, especially in
%D a technical environment. In production oriented environments, a lot of compound
%D words are used. Because the Dutch language poses no limits on combining words, we
%D often favor putting dashes between those words, because it facilitates reading,
%D at least for those who are not that accustomed to it.
%D
%D In \TEX\ compound words, separated by a hyphen, are not hyphenated at all. In
%D spite of the multiple pass paragraph typesetting this can lead to parts of words
%D sticking into the margin. The solution lays in saying \type
%D {spoelwater||terugwinunit} instead of \type {spoelwater-terugwinunit}. By using a
%D one character command like \type {|}, delimited by the same character \type {|},
%D we get ourselves both a decent visualization (in \TEXEDIT\ and colored verbatim
%D we color these commands yellow) and an efficient way of combining words.
%D
%D The sequence \type{||} simply leads to two words connected by a hyphen. Because
%D we want to distinguish such a hyphen from the one inserted when \TEX\ hyphenates
%D a word, we use a bit longer one.
%D
%D \hyphenation {spoel-wa-ter te-rug-win-unit}
%D
%D \starttest
%D \test {spoelwater||terugwinunit}
%D \stoptest
%D
%D As we already said, the \type{|} is a command. This commands accepts an optional
%D argument before it's delimiter, which is also a \type{|}.
%D
%D \hyphenation {po-ly-meer che-mie}
%D
%D \starttest
%D \test {polymeer|*|chemie}
%D \stoptest
%D
%D Arguments like \type{*} are not interpreted and inserted directly, in contrary to
%D arguments like:
%D
%D \starttest
%D \test {polymeer|~|chemie}
%D \test {|(|polymeer|)|chemie}
%D \test {polymeer|(|chemie|)| }
%D \stoptest
%D
%D Although such situations seldom occur |<|we typeset thousands of pages before we
%D encountered one that forced us to enhance this mechanism|>| we also have to take
%D care of comma's.
%D
%D  \hyphenation {uit-stel-len}
%D
%D  \starttest
%D  \test {op||, in|| en uitstellen}
%D  \stoptest
%D
%D The next special case (concerning quotes) was brought to my attention by Piet
%D Tutelaers, one of the driving forces behind rebuilding hyphenation patterns for
%D the dutch language.\footnote{In 1996 the spelling of the dutch language has been
%D slightly reformed which made this topic actual again.} We'll also take care of
%D this case.
%D
%D \starttest
%D \test {AOW|'|er}
%D \test {cd|'|tje}
%D \test {ex|-|PTT|'|er}
%D \test {rock|-|'n|-|roller}
%D \stoptest
%D
%D Tobias Burnus pointed out that I should also support something like
%D
%D \starttest
%D \test {well|_|known}
%D \stoptest
%D
%D to stress the compoundness of hyphenated words.
%D
%D Of course we also have to take care of the special case:
%D
%D \starttest
%D \test {text||color and ||font}
%D \stoptest

%D \macros
%D   {installdiscretionaries}
%D
%D The mechanism described here is one of the older inner parts of \CONTEXT. The
%D most recent extensions concerns some special cases as well as the possibility to
%D install other characters as delimiters. The prefered way of specifying compound
%D words is using \type{||}, which is installed by:
%D
%D \starttyping
%D \installdiscretionary | -
%D \stoptyping
%D
%D We used to have an installable mechanism but in the perspective of \MKIV\ and
%D especialy \LMTX\ it no longer makes sense to complicate the code, so from now on
%D we only deal with the active bar. Older code can be seen in the archives. It also
%D means that we now just hardcode the bar. We also deal with math differently.

%D \macros
%D   {compoundhyphen}
%D
%D Now let's go to the macros. First we define some variables. In the main \CONTEXT\
%D modules these can be tuned by a setup command. Watch the (maybe) better looking
%D compound hyphen.

% hm why ex

\ifdefined\compoundhyphen \else
    \permanent\protected\def\compoundhyphen{\hbox{-\kern-.10775\emwidth-}} % .25\exheight
\fi

%D The last two variables are needed for subsentences |<|like this one|>| which we
%D did not yet mention. We want to enable breaking but at the same time don't want
%D compound characters like |-| or || to be separated from the words. \TEX\ hackers
%D will recognise the next two macro's:

\ifdefined\prewordbreak \else \permanent\protected\def\prewordbreak    {\penalty\plustenthousand\hskip\zeroskip\relax} \fi
\ifdefined\postwordbreak\else \permanent\protected\def\postwordbreak   {\penalty\zerocount      \hskip\zeroskip\relax} \fi
\ifdefined\hspaceamount \else                     \def\hspaceamount#1#2{.16667\emwidth}                                 \fi % will be overloaded

\permanent\protected\def\permithyphenation{\ifhmode\wordboundary\fi} % doesn't remove spaces

%D \macros
%D   {beginofsubsentence,endofsubsentence,
%D    beginofsubsentencespacing,endofsubsentencespacing}
%D
%D In the previous macros we provided two hooks which can be used to support nested
%D sub||sentences. In \CONTEXT\ these hooks are used to insert a small space when
%D needed.
%D
%D The following piece of code is a torture test compound handling. The \type
%D {\relax} before the \type {\ifmmode} is needed because of the alignment scanner
%D (in \ETEX\ this problem is not present because there a protected macro is not
%D expanded. Thanks to Tobias Burnus for providing this example.
%D
%D \startformula
%D   \left|f(x_n)-{1\over2}\right| =
%D      {\cases{|{1\over2}-x_n| &for $0\le x_n < {1\over2}$\cr
%D              |x_n-{1\over2}| &for ${1\over2}<x_n\le1$   \cr}}
%D \stopformula

\installcorenamespace{discretionaryaction}
\installcorenamespace{discretionarymode}

\aliased\let\installdiscretionaries\gobbletwoarguments % this alias will go
\aliased\let\installdiscretionary  \gobbletwoarguments % this alias will go

\setnewconstant\discretionarymode\plusone

\permanent\protected\def\ignorediscretionaries{\discretionarymode\zerocount}
\permanent\protected\def\obeydiscretionaries  {\discretionarymode\plusone}

\def\lang_discretionaries_command
  {\begincsname\??discretionarymode
     \ifcase\discretionarymode
       n%
     \else
       t%
     \fi
   \endcsname}

% \catcode\barasciicode\activecatcode
% \amcode \barasciicode\othercatcode

\letcatcodecommand\ctxcatcodes\barasciicode\lang_discretionaries_command

%D The macro \type {\lang_discretionaries_check_before} takes care of loners like
%D \type {||word}, while it counterpart \type {\lang_discretionaries_check_after} is
%D responsible for handling the comma.

\newconditional\punctafterdiscretionary
\newconditional\spaceafterdiscretionary

\def\lang_discretionaries_check_before %is used grouped
  {\ifvmode
     \dontleavehmode
   \fi
   \ifhmode
    %\begingroup
    %\setbox\scratchbox\lastbox
    %\ifzeropt\wd\scratchbox
    %  \box\scratchbox\relax
    %  \endgroup
    %  \let\postwordbreak\prewordbreak
    %\else
    %  \box\scratchbox\relax
    %  \endgroup
    %\fi
   \fi}

\def\lang_discretionaries_check_after
  {\punctafterdiscretionary\conditionalfalse
   \spaceafterdiscretionary\conditionalfalse
   \ifx\blankspace\nexttoken \spaceafterdiscretionary\conditionaltrue \orelse
   \ifx\space     \nexttoken \spaceafterdiscretionary\conditionaltrue \orelse
   \ifx          .\nexttoken \punctafterdiscretionary\conditionaltrue \orelse
   \ifx          ,\nexttoken \punctafterdiscretionary\conditionaltrue \orelse
   \ifx          :\nexttoken \punctafterdiscretionary\conditionaltrue \orelse
   \ifx          ;\nexttoken \punctafterdiscretionary\conditionaltrue \fi}

\edefcsname\??discretionarymode n\endcsname
  {\detokenize{|}}

\defcsname\??discretionarymode t\endcsname#1|%
  {\bgroup
   \def\next{\lang_discretionaries_handle{#1}}%
   \futurelet\nexttoken\next}%

\permanent\protected\def\lang_discretionaries_handle#1%
  {\edef\lang_discretionaries_token{\detokenize{#1}}%
   \lang_discretionaries_check_after
   \ifempty\lang_discretionaries_token
     \ifx|\nexttoken % takes care of |||
       \ifcsname\??discretionaryaction\string|\endcsname
         \lastnamedcs
       \orelse\ifconditional\spaceafterdiscretionary
         \wordboundary\hbox{\letterbar}\relax
       \orelse\ifconditional\punctafterdiscretionary
         \wordboundary\hbox{\letterbar}\wordboundary
       \else
         \wordboundary\hbox{\letterbar}\wordboundary
       \fi
       \def\next{\afterassignment\egroup\let\next=}%
     \else
       \lang_discretionaries_check_before
       \ifcsname\??discretionaryaction\endcsname
         \lastnamedcs
       \orelse\ifconditional\spaceafterdiscretionary
         \wordboundary\defaultdiscretionaryhyphen\relax
       \orelse\ifconditional\punctafterdiscretionary
         \wordboundary\defaultdiscretionaryhyphen\relax
       \else
         \wordboundary\defaultdiscretionaryhyphen\wordboundary
       \fi
       \let\next\egroup
     \fi
   \orelse\ifcsname\??discretionaryaction\lang_discretionaries_token\endcsname
     \lastnamedcs
     \let\next\egroup
   \else
     \lang_discretionaries_check_before
     \ifconditional\spaceafterdiscretionary
       \wordboundary\hbox{#1}\relax
     \orelse\ifconditional\punctafterdiscretionary
       \wordboundary\hbox{#1}\relax
     \else
       \wordboundary\discretionary{\hbox{#1}}{}{\hbox{#1}}\wordboundary
      %\discretionary options \plusthree{\hbox{#1}}{}{\hbox{#1}}%
     \fi
     \let\next\egroup
   \fi
   \next}

%D \macros
%D   {directdiscretionary}
%D
%D In those situations where the nature of characters is less predictable, we can
%D use the more direct approach:

\permanent\protected\def\directdiscretionary
  {\begincsname\??discretionarymode
     \ifcase\discretionarymode
       n%
     \else
       d%
     \fi
   \endcsname}

\permanent\protected\def\indirectdiscretionary
  {\begincsname\??discretionarymode
     \ifcase\discretionarymode
       n%
     \else
       i%
     \fi
   \endcsname}

\protected\defcsname\??discretionarymode d\endcsname#1%
  {\ifcsname\??discretionaryaction\detokenize{#1}\endcsname
     \expandafter\lastnamedcs
   \else
     \expandafter\indirectdiscretionary
   \fi{#1}}

\protected\defcsname\??discretionarymode i\endcsname#1%
  {\wordboundary\discretionary{\hbox{#1}}{}{\hbox{#1}}\wordboundary}
 %{\discretionary options \plusthree{\hbox{#1}}{}{\hbox{#1}}}

\permanent\protected\def\definetextmodediscretionary #1
  {\defcsname\??discretionaryaction\detokenize{#1}\endcsname}

% \start \hsize 1mm
% test |||test test|||, test\blank
% test test|-|, test|-| and test|-|test\blank
% test test|_|, test|_| and test|_|test\blank
% test cd|'|tje\blank
% test |(|test test|)|, test\blank
% test test test|x|, test\blank
% test|~|test
% test|^|test
% \stop

% x\discretionary{1}{2}{3}xxxxxxx
% xxxxxxx\discretionary{1}{2}{3}x
%
% xxx3xxx
% xxx1<newline>2xxx

\def\lang_discretionaries_hyphen_like#1#2%
  {\ifconditional\spaceafterdiscretionary
     \wordboundary\hbox{#1}\relax
   \orelse\ifconditional\punctafterdiscretionary
     \wordboundary\hbox{#1}\relax
   \else
     \wordboundary#2\wordboundary
   \fi}

\definetextmodediscretionary {}
  {\lang_discretionaries_hyphen_like\textmodehyphen\textmodehyphendiscretionary}

\definetextmodediscretionary -
  {\lang_discretionaries_hyphen_like\normalhyphen\normalhyphendiscretionary}

\definetextmodediscretionary _
  {\lang_discretionaries_hyphen_like\composedhyphen\composedhyphendiscretionary}

\definetextmodediscretionary )
  {\lang_discretionaries_hyphen_like{)}{\discretionary{-)}{}{)}}}

\definetextmodediscretionary (
  {\ifdim\lastskip>\zeropoint
     (\wordboundary
   \else
     \wordboundary\discretionary{}{(-}{(}\wordboundary
    %\discretionary options \plusthree{}{(-}{(}%
   \fi}

\definetextmodediscretionary ~
  {\wordboundary\discretionary{-}{}{\thinspace}\wordboundary}
 %{\discretionary options \plusthree{-}{}{\thinspace}}

\definetextmodediscretionary '
  {\wordboundary\discretionary{-}{}{'}\wordboundary}
 %{\discretionary options \plusthree{-}{}{'}}

\definetextmodediscretionary ^
  {\wordboundary
   \discretionary{\hbox{\normalstartimath|\normalstopimath}}{}{\hbox{\normalstartimath|\normalstopimath}}%
   \wordboundary} % bugged
 %{\discretionary options \plusthree{\hbox{\normalstartimath|\normalstopimath}}{}{\hbox{\normalstartimath|\normalstopimath}}}

\definetextmodediscretionary <
  {\beginofsubsentence\wordboundary\beginofsubsentencespacing
   \aftergroup\ignorespaces} % tricky, we need to go over the \nextnextnext

\definetextmodediscretionary >
  {\removeunwantedspaces
   \endofsubsentencespacing\wordboundary\endofsubsentence}

\definetextmodediscretionary =
  {\removeunwantedspaces
   \wordboundary\midsentence\wordboundary
   \aftergroup\ignorespaces}

% french

\definetextmodediscretionary : {\removeunwantedspaces\wordboundary\kern\hspaceamount\empty{:}:}
\definetextmodediscretionary ; {\removeunwantedspaces\wordboundary\kern\hspaceamount\empty{;};}
\definetextmodediscretionary ? {\removeunwantedspaces\wordboundary\kern\hspaceamount\empty{?}?}
\definetextmodediscretionary ! {\removeunwantedspaces\wordboundary\kern\hspaceamount\empty{!}!}

\definetextmodediscretionary * {\wordboundary\discretionary{-}{}{\kern.05\emwidth}\wordboundary}

% spanish

\definetextmodediscretionary ?? {\wordboundary\questiondown}
\definetextmodediscretionary !! {\wordboundary\exclamdown}

\permanent\protected\def\defaultdiscretionaryhyphen{\compoundhyphen}

%D \macros
%D   {fakecompoundhyphen}
%D
%D In headers and footers as well as in active pieces of text we need a dirty hack.
%D Try to imagine what is needed to savely break the next text across a line and at
%D the same time make the words interactive.
%D
%D \starttyping
%D \goto{Some||Long||Word}
%D \stoptyping

\permanent\protected\def\fakecompoundhyphen
  {\enforced\permanent\protected\def\|{\mathortext\vert\lang_compounds_fake_hyphen}}

\def\lang_compounds_fake_hyphen
  {\enforced\permanent\protected\def##1|%
     {\ifempty{##1}\compoundhyphen\else##1\fi
      \wordboundary % was a signal
      \allowbreak}}

%D \macros
%D   {midworddiscretionary}
%D
%D If needed, one can add a discretionary hyphen using \type
%D {\midworddiscretionary}. This macro does the same as \PLAIN\ \TEX's \type {\-},
%D but, like the ones implemented earlier, this one also looks ahead for spaces and
%D grouping tokens.

\permanent\protected\def\midworddiscretionary
  {\futurelet\nexttoken\lang_discretionaries_mid_word}

\def\lang_discretionaries_mid_word
  {\ifx\nexttoken\blankspace\orelse
   \ifx\nexttoken\bgroup    \orelse
   \ifx\nexttoken\egroup    \orelse
     \discretionary{-}{}{}%
   \fi}

% \aliased\let\ignorecompoundcharacter\relax

%D \macros
%D   {disablediscretionaries,disablecompoundcharacter}
%D
%D Occasionally we need to disable this mechanism. For the moment we assume that
%D \type {|} is used.

\aliased\let\disablediscretionaries   \ignorediscretionaries
%aliased\let\disablecompoundcharacters\ignorecompoundcharacter

%D \macros
%D   {normalcompound}
%D
%D Handy in for instance XML. (Kind of obsolete)

\ifdefined\normalcompound \else \aliased\let\normalcompound=| \fi

%D \macros
%D   {compound}
%D
%D We will overload the already active \type {|} so we have to save its meaning in
%D order to be able to use this handy macro.
%D
%D \starttyping
%D so test\compound{}test can be used instead of test||test
%D \stoptyping

\permanent\protected\gdef\compound#1{|#1|}

\appendtoks
    \enforced\permanent\protected\def|#1|{\ifx#1\empty\empty-\else#1\fi}%
\to \everysimplifycommands

%D Here we hook some code into the clean up mechanism needed for verbatim data.

\appendtoks
    %disablecompoundcharacters
    \disablediscretionaries
\to \everycleanupfeatures

%D Here:
%D
%D \startbuffer
%D {\red somelongword}{\blue \compounddiscretionary}{\green somelongword}
%D \stopbuffer
%D
%D \typebuffer \blank {\hsize3mm\getbuffer\par} \blank

\permanent\protected\def\compounddiscretionary
   {\discretionary
      options \plusthree
      {\ifnum\prehyphenchar >\zerocount\char\prehyphenchar \fi}%
      {\ifnum\posthyphenchar>\zerocount\char\posthyphenchar\fi}%
      {\ifnum\posthyphenchar>\zerocount\char\posthyphenchar\fi}}

% \setcatcodetable\prtcatcodes % because we activated the bar

\protect \endinput