latexml 0.8.0-1 / usr / share / perl5 / LaTeXML / Package / babel.def.ltxml

# -*- CPERL -*-
# /=====================================================================\ #
# |  babel.def                                                          | #
# | Implementation for LaTeXML                                          | #
# |=====================================================================| #
# | Part of LaTeXML:                                                    | #
# |  Public domain software, produced as part of work done by the       | #
# |  United States Government & not subject to copyright in the US.     | #
# |---------------------------------------------------------------------| #
# | Thanks to the arXMLiv group for initial implementation              | #
# |    http://arxmliv.kwarc.info/                                       | #
# | Released to the Public Domain                                       | #
# |---------------------------------------------------------------------| #
# | Bruce Miller <bruce.miller@nist.gov>                        #_#     | #
# | http://dlmf.nist.gov/LaTeXML/                              (o o)    | #
# \=========================================================ooo==U==ooo=/ #
package LaTeXML::Package::Pool;
use strict;
use warnings;
use LaTeXML::Package;

# \foreignlanguage{language}{phrase}

# MORE USER COMMANDS!!!!

#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
# Implementation

#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
# Language Definitions

#======================================================================
# User level

#======================================================================
# Internal

DefMacro('\user@group',     'user');
DefMacro('\language@group', 'english');
DefMacro('\system@group',   'system');

# NOTE: even though we don't need hyphenation, we need (I think)
# to have the various \l@english, \l@german, etc bound to numbers
# (which would be used by TeX's \language register!)
AssignValue('babel@languages' => 0);
DefMacro('\l@english', '0');

# \ProvidesLanguage{language}[release information]
DefPrimitive('\ProvidesLanguage{}[]', sub {
    my ($stomach, $lang, $version) = @_;
    # $lang = ToString(Expand($lang));
    # my $n = 1+LookupValue('babel@languages');
    # AssignValue('babel@languages'=>$n);
    # DefMacro('\l@'.$lang, "$n");
    return; });

# \LdfInit{language}{something odd about captions?}
DefMacro('\LdfInit{}{}',
  '\chardef\atcatcode=\catcode`\@'
    . '\catcode`\@=11\relax'
    . '\chardef\eqcatcode=\catcode`\='
    . '\catcode`\==12\relax');

# do these 2 really need the catcode stuff?
DefMacro('\ldf@finish{}',
  # \loadlocalcfg{#1}%
  '\expandafter\main@language\expandafter{#1}'
    . '\catcode`\@=\atcatcode \let\atcatcode\relax'
    . '\catcode`\==\eqcatcode \let\eqcatcode\relax');
DefMacro('\ldf@quit{}',
  '\expandafter\main@language\expandafter{#1}'
    . '\catcode`\@=\atcatcode \let\atcatcode\relax'
    . '\catcode`\==\eqcatcode \let\eqcatcode\relax'
    . '\endinput');

DefMacro('\loadlocalcfg{}', '');

#======================================================================
# Various state saving stuff.

Let('\originalTeX', '\@empty');

DefMacro('\addto DefToken {}', sub {
    my ($gullet, $cs, $tokens) = @_;
    # Needs error checking!
    my $defn = LookupDefinition($cs);
    $defn = undef if $defn && $defn eq LookupDefinition(T_CS('\relax'));    # ?
    if ($defn && !$defn->isExpandable) {
      Warn('unexpected', $cs, $gullet, "expected an expandable token, got '" . ToString($cs) . "'");
      $defn = undef; }
    DefMacroI($cs, undef, ($defn ? Tokens($defn->getExpansion->unlist, $tokens->unlist) : $tokens)); });

DefMacro('\babel@savevariable {}', sub {
    my ($gullet, $token) = @_;
    AddToMacro(T_CS('\originalTeX'), $token, T_OTHER('='), Expand(Tokens(T_CS('\the'), $token))); });

DefMacro('\babel@beginsave', sub { AssignValue('babel@savecnt' => 0); });

DefMacro('\babel@save DefToken', sub {
    my ($gullet, $token) = @_;
    my $n = 1 + LookupValue('babel@savecnt');
    AssignValue('babel@savecnt' => $n);
    my $v = T_CS('\babel@' . $n);
    Let($v, $token);
    AddToMacro(T_CS('\originalTeX'), T_CS('\let'), $token, $v); });

#======================================================================
# language setting stuff

# \selectlanguage{language}
DefMacro('\selectlanguage{}', '\bbl@push@language\aftergroup\bbl@pop@language\bbl@set@language{#1}');

DefMacro('\bbl@push@language', sub {
    PushValue(bbl_language_stack => ToString(Expand(T_CS('\languagename')))); });
DefMacro('\bbl@pop@language', sub {
    my $lang = PopValue('bbl_language_stack');
    DefMacro('\languagename', $lang);
    Invocation(T_CS('\bbl@set@language'), Tokens(Explode($lang))); });

DefMacro('\bbl@set@language{}', '\edef\languagename{#1}\select@language{\languagename}');
DefMacro('\select@language{}', sub {
    my ($gullet, $language) = @_;
    $language = ToString(Expand($language));

   #  print STDERR "SELECTING $language ".$gullet->getLocator." at ".$_[0]->getLocator."\n";
   #  print STDERR "ORIGINAL: ".ToString(LookupMeaning(T_CS('\originalTeX'))->getExpansion)."\n";
   #  print STDERR "CAPTIONS: ".ToString(LookupMeaning(T_CS('\captions'.$language))->getExpansion)."\n";
   #  print STDERR "DATE: ".ToString(LookupMeaning(T_CS('\date'.$language))->getExpansion)."\n";
   #  print STDERR "EXTRAS: ".ToString(LookupMeaning(T_CS('\extras'.$language))->getExpansion)."\n";

    (T_CS('\originalTeX'),
      T_CS('\def'), T_CS('\originalTeX'),
      T_BEGIN, T_CS('\noextras' . $language), T_CS('\let'), T_CS('\originalTeX'), T_CS('\@empty'), T_END,

      T_CS('\languageshorthands'), T_BEGIN, T_OTHER('none'), T_END,
      T_CS('\babel@beginsave'),

      T_CS('\captions' . $language),
      T_CS('\date' . $language),
      T_CS('\extras' . $language)); });

DefMacro('\main@language{}',
  '\def\bbl@main@language{#1}'
    . '\let\languagename\bbl@main@language'
    . '\language=\csname l@\languagename\endcsname\relax');

DefMacro('\foreignlanguage{}{}',
  '\begingroup\originalTeX\foreign@language{#1}#2\csname noextras#1\endcsname\endgroup');
DefMacro('\foreign@language{}',
  '\def\languagename{#1}'
    . '\language=\csname l@#1\endcsname\relax'
    . '\languageshorthands{none}'
    . '\csname extras#1\endcsname'
    . '\expandafter\ifx\csname #1hyphenmins\endcsname\relax'
    . '\set@hyphenmins\tw@\thr@@\relax'
    . '\else'
    . '\expandafter\expandafter\expandafter\set@hyphenmins'
    . '\csname #1hyphenmins\endcsname\relax'
    . '\fi');

DefMacro('\otherlanguage{}',  '\selectlanguage{#1}');    # \ignorespaces ???
DefMacro('\endotherlanguage', '\originalTeX');

DefMacro('\iflanguage{}',
  '\let\@tempa\@secondoftwo'
    . '\expandafter\ifnum\csname l@#1\endcsname=\language\let\@tempa\@firstoftwo\fi'
    . '\@tempa');

#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
# Shorthands

#======================================================================
# User-level

# Shorthands
#\useshorthands{char}
DefMacro('\useshorthands{}', '\def\user@group{user}\initiate@active@char{#1}');

#\defineshorthand{charseq}{expansion}
DefMacro('\defineshorthand', '\declare@shorthand{user}');

#\languageshorthands{language}
DefMacro('\languageshorthands{}', '\def\language@group{#1}');

#\aliasshorthand{char1}{char2}
# ??? DefMacro('\aliasshorthand{}{}','\initiate@active@char{#2}

#\shorthandon{chars}
DefMacro('\shorthandon{}', sub {
    #print STDERR "SHORTHANDON :".ToString($_[1])." at ".$_[0]->getLocator."\n";
    map { AssignCatcode($_, CC_ACTIVE) } map { $_->getCSName } $_[1]->unlist; });
#\shorthandoff{chars}
DefMacro('\shorthandoff{}', sub {
    #print STDERR "SHORTHANDOFF :".ToString($_[1])." at ".$_[0]->getLocator."\n";
    map { AssignCatcode($_, CC_OTHER) } map { $_->getCSName } $_[1]->unlist; });

#======================================================================
# internal
RawTeX('\newif\if@safe@actives\@safe@activesfalse');
DefMacro('\bbl@restore@actives', '\if@safe@actives\@safe@activesfalse\fi');

DefPrimitive('\bbl@activate{}', sub {
    my ($stomach, $char) = @_;
    $char = ToString($char);
    #  print STDERR "ACTIVATING $char"." at ".$_[0]->getGullet->getLocator."\n";
    DefMacroI($char, undef, T_CS('\active@char' . $char)); });

DefPrimitive('\bbl@deactivate{}', sub {
    my ($stomach, $char) = @_;
    $char = ToString($char);
    #  print STDERR "DEACTIVATING $char"." at ".$_[0]->getGullet->getLocator."\n";
    DefMacroI($char, undef, T_CS('\normal@char' . $char)); });

# Not clear... Some chars stand alone, others are a prefix..!?!
DefPrimitive('\initiate@active@char DefToken', sub {
    my ($stomach, $token) = @_;
    my $char = ToString($token);
    return if LookupDefinition(T_CS('\active@char' . $char));
#  print STDERR "INITIATING $char (catcode ".$token->getCatcode.") at ".$_[0]->getGullet->getLocator."\n";
    my $user = ToString(Expand(T_CS('\user@group')));
###???
#### AssignCatcode($char,CC_ACTIVE);
    # Define the "normal" expansion of $token, noting that it may ALREADY be active!
    if ($token->getCatcode == CC_ACTIVE) {    # Already active?
      if (!LookupDefinition($token)) {
        Let('\normal@char' . $char, $token);    # !!
        DefMacroI($char, undef, T_CS('\normal@char' . $char)); } }
    else {
      if ($char eq "'") {
        Let('\bbl@primes@@@', '\'');
        # This is probably not enough since ' is changed.... ?
        DefMacroI('\normal@char' . $char, undef, '\textormath{\'}{\bbl@primes@@@}'); }
      elsif ($char eq '^') {
        DefMacroI('\bbl@act@caret', undef,
          '\ifmmode\csname normal@char\string^\endcsname'
            . '\else\bbl@afterfi{\if@safe@actives'
            . '\bbl@afterelse\csname normal@char\string' . $char . '\endcsname'
            . '\else'
            . '\bbl@afterfi\csname user@active\string' . $char . '\endcsname'
            . '\fi}\fi'); }
      else {
        $stomach->bgroup;
        AssignCatcode($char, CC_ACTIVE);
        DefMacroI($char, undef, T_CS('\active@char' . $char), scope => 'global');
        $stomach->egroup;
        AtBeginDocument(sub { AssignCatcode($char, CC_ACTIVE); });
        PushValue('SPECIALS', $char);
        DefMacroI('\normal@char' . $char, undef, T_OTHER($char)); } }
    # Now define a chain of expansions for user, language and system.
    if ($char eq '^') {
      DefMacroI('\active@char' . $char, undef, '\bbl@act@caret'); }
    else {
      DefMacroI('\active@char' . $char, undef,
        '\if@safe@actives'
          . '\bbl@afterelse\csname normal@char\string' . $char . '\endcsname'
          . '\else'
          . '\bbl@afterfi\csname user@active\string' . $char . '\endcsname'
          . '\fi'); }
    DefMacroI('\user@active' . $char, undef,
      '\expandafter\ifx\csname \user@group @sh@\string' . $char . '@\endcsname\relax'
        . '\bbl@afterelse\bbl@sh@select\user@group' . $char
        . '{user@active@arg\string' . $char . '}{language@active\string' . $char . '}'
        . '\else'
        . '\bbl@afterfi\csname \user@group @sh@\string' . $char . '@\endcsname'
        . '\fi');
    DefMacroI('\user@active@arg' . $char, 'Token',
      '\expandafter\ifx\csname \user@group @sh@\string' . $char . '@\string#1@\endcsname\relax'
        . '\bbl@afterelse\csname language@active\string' . $char . '\endcsname#1'
        . '\else'
        . '\bbl@afterfi\csname \user@group @sh@\string' . $char . '@\string#1@\endcsname'
        . '\fi');
    DefMacroI('\\' . $user . '@sh@' . $char . '@@', undef, '\csname normal@char\string' . $char . '\endcsname');
    DefMacroI('\language@active' . $char, undef,
      '\expandafter\ifx\csname \language@group @sh@\string' . $char . '@\endcsname\relax'
        . '\bbl@afterelse\bbl@sh@select\language@group' . $char
        . '{language@active@arg\string' . $char . '}{system@active\string' . $char . '}'
        . '\else'
        . '\bbl@afterfi\csname \language@group @sh@\string' . $char . '@\endcsname'
        . '\fi');
    DefMacroI('\language@active@arg' . $char, 'Token',
      '\expandafter\ifx\csname \language@group @sh@\string' . $char . '@\string#1@\endcsname\relax'
        . '\bbl@afterelse\csname system@active\string' . $char . '\endcsname#1'
        . '\else'
        . '\bbl@afterfi\csname \language@group @sh@\string' . $char . '@\string#1@\endcsname'
        . '\fi');
    DefMacroI('\system@active' . $char, undef,
      '\expandafter\ifx\csname \system@group @sh@\string' . $char . '@\endcsname\relax'
        . '\bbl@afterelse\bbl@sh@select\system@group' . $char
        . '{system@active@arg\string' . $char . '}{normal@char\string' . $char . '}'
        . '\else'
        . '\bbl@afterfi\csname \system@group @sh@\string' . $char . '@\endcsname'
        . '\fi');
    DefMacroI('\system@active@arg' . $char, 'Token',
      '\expandafter\ifx\csname \system@group @sh@\string' . $char . '@\string#1@\endcsname\relax'
        . '\bbl@afterelse\csname normal@char\string' . $char . '\endcsname'
        . '\else'
        . '\bbl@afterfi\csname \system@group @sh@\string' . $char . '@\string#1@\endcsname'
        . '\fi');
    DefMacroI('\user@sh@' . $char . '@\protect@', undef,
      '\csname user@active\string' . $char . '\endcsname');
});

RawTeX('\long\def\bbl@afterelse#1\else#2\fi{\fi#1}'
    . '\long\def\bbl@afterfi#1\fi{\fi#1}');

DefPrimitive('\declare@shorthand{}{}{}', sub {
    my ($stomach, $lang, $c, $expansion) = @_;
    $lang = ToString($lang);
    $c    = ToString($c);
    # Recognize 1 or 2 char shorthands
    if ($c =~ /^(.)(.)$/) {
      my ($c1, $c2) = ($1, $2);
      Let('\\' . $lang . '@sh@' . $c1 . '@sel', '\bbl@firstcs');
      DefMacroI('\\' . $lang . '@sh@' . $c1 . '@' . $c2 . '@', undef, $expansion); }
    else {
      Let('\\' . $lang . '@sh@' . $c . '@sel', '\bbl@scndcs');
      DefMacroI('\\' . $lang . '@sh@' . $c . '@', undef, $expansion); } });
DefMacro('\bbl@sh@select{}{}',
  '\expandafter\ifx\csname#1@sh@\string#2@sel\endcsname\relax'
    . '\bbl@afterelse\bbl@scndcs'
    . '\else'
    . '\bbl@afterfi\csname#1@sh@\string#2@sel\endcsname'
    . '\fi');
DefMacro('\bbl@firstcs{}{}', '\csname#1\endcsname');
DefMacro('\bbl@scndcs{}{}',  '\csname#2\endcsname');

DefMacro('\textormath{}{}', '\ifmmode\bbl@afterelse#2\else\bbl@afterfi#1\fi');

RawTeX('\declare@shorthand{system}{"}{\csname normal@char\string"\endcsname}'
    . '\declare@shorthand{system}{\'}{\textormath{\csname normal@char\string\'\endcsname}{\sp\bgroup\prim@s}}'
    . '\declare@shorthand{system}{`}{\csname normal@char\string`\endcsname}'
    . '\initiate@active@char{~}'
    . '\declare@shorthand{system}{~}{\leavevmode\nobreak\ }'
    . '\bbl@activate{~}'
);

DefMacro('\bbl@disc{}{}', '#1');
#======================================================================
DefMacro('\umlautlow',  '');    # ?
DefMacro('\umlauthigh', '');
#======================================================================
# \bbl@declare@ttribute{lang}{attribute}{code}
DefMacro('\bbl@declare@ttribute{}{}{}', '\expandafter\def\csname#1@attr@#2\endcsname{#3}');

# language attributes
# \languageattribute{language}{langattrs}
DefPrimitive('\languageattribute{}{}', sub {
    my ($gullet, $language, $attributes) = @_;
    $language = ToString(Expand($language));
    my @attributes = split(/\s*,\s*/, ToString(Expand($attributes)));
    #    print STDERR "ATTRIBUTES $language: ".join(', ',@attributes)."\n";

    foreach my $t (map { T_CS('\\' . $language . '@attr@' . $_) } @attributes) {
      Digest($t); }
    return; });
#======================================================================
# Hyphenation stuff --- ignroable for LaTeXML

DefPrimitive('\@nopatterns{}', '');

DefPrimitive('\allowhyphens', '');
DefMacro('\providehyphenmins{}{}',
  '\expandafter\ifx\csname #1hyphenmins\endcsname\relax'
    . '\@namedef{#1hyphenmins}{#2}'
    . '\fi');
DefMacro('\set@hyphenmins{}{}', '\lefthyphenmin#1\righthyphenmin#2');

DefPrimitive('\bbl@frenchspacing',    '');
DefPrimitive('\bbl@nonfrenchspacing', '');

# \adddialect{dialect}{language}
# notes that dialect's hyphenation is equivalent to language's (ignorable)
DefMacro('\adddialect{}{}', '\global\chardef#1#2\relax');

# \begin{hyphenrules}{language}
#DefEnvironment('{hyphenrules}{}','#body');
DefMacro('\hyphenrules{}', '\language=\csname l@#1\endcsname\relax\languageshorthands{none}');

#======================================================================
DefMacro('\latintext', '\fontencoding{\latinencoding}\selectfont\def\encodingdefault{\latinencoding}');
RawTeX('\edef\latinencoding{\cf@encoding}');
#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
RawTeX('\main@language{english}');
#AtBeginDocument('\expandafter\selectlanguage\expandafter{\bbl@main@language}');

#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
DefPrimitiveI('\dj', undef, "\x{0111}");
DefPrimitiveI('\DJ', undef, "\x{0110}");
DefPrimitiveI('\ij', undef, "ij");
DefPrimitiveI('\IJ', undef, "IJ");

DefPrimitiveI('\flq',  undef, "\x{2039}");
DefPrimitiveI('\frq',  undef, "\x{203A}");
DefPrimitiveI('\flqq', undef, UTF(0xAB));
DefPrimitiveI('\frqq', undef, UTF(0xBB));

DefPrimitiveI('\glq', undef, "\x{201A}");
DefPrimitiveI('\grq', undef, "\x{2019}");
#DefPrimitiveI('\glqq',undef,"\x{2018}\x{2018}");
#DefPrimitiveI('\grqq',undef,"\x{2019}\x{2019}");
DefPrimitiveI('\glqq', undef, "\x{201E}");
DefPrimitiveI('\grqq', undef, "\x{201D}");

DefPrimitiveI('\SS', undef, "SS");

DefPrimitiveI('\guilsinglleft',  undef, "\x{2039}");
DefPrimitiveI('\guilsinglright', undef, "\x{203A}");
DefPrimitiveI('\guillemotleft',  undef, UTF(0xAB));
DefPrimitiveI('\guillemotright', undef, UTF(0xBB));

#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
# It seems that we can successfully (?) read the ldf,
# but there may be cases where we can't out-think the ldf writer.
# Eg. in french, there're lots of \up{\lowercase{o}}\kern ...
# But there's a U+00AA (but it's underlined; for Spanish?). Wouldn't that be better?

# The point is, we might want to add some ltxml code to override what's in the ldf.
#%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1;