## LATEX-L@LISTSERV.UNI-HEIDELBERG.DE

 Options: Use Forum View Use Monospaced Font Show Text Part by Default Show All Mail Headers Message: [<< First] [< Prev] [Next >] [Last >>] Topic: [<< First] [< Prev] [Next >] [Last >>] Author: [<< First] [< Prev] [Next >] [Last >>]

 Subject: Re: latex/3480: Support for UTF-8 missing in inputenc.sty From: Frank Mittelbach <[log in to unmask]> Reply To: Mailing list for the LaTeX3 project <[log in to unmask]> Date: Tue, 7 Jan 2003 21:34:51 +0100 Content-Type: text/plain Parts/Attachments: text/plain (931 lines)
following up on the discussion concering utf-8 support for LaTeX, below is a
package written to provide that support within the inputenc framework.

it is not complete, nor are its tables set up finally, we would need some
volunteers to help us here.

but first i would like to hear comments/suggestions on the approach

cheers
frank

ps the ins file generates a small test document, that generates one error
(deliberately)

-------------- utf8ienc.ins
\input docstrip

\preamble

This is a generated file.

The LaTeX3 Project and any individual authors listed elsewhere
in this file.

It may be distributed and/or modified under the
conditions of the LaTeX Project Public License, either version 1.2
http://www.latex-project.org/lppl.txt
and version 1.2 or later is part of all distributions of LaTeX
version 1999/12/01 or later.

\endpreamble

\keepsilent

\usedir{tex/latex/base}

\generate{\file{utf8.def}{\from{utf8ienc.dtx}{utf8}}
\file{t1enc.dfu}{\from{utf8ienc.dtx}{t1}}
\file{ot1enc.dfu}{\from{utf8ienc.dtx}{ot1}}
\file{omsenc.dfu}{\from{utf8ienc.dtx}{oms}}
\file{utf8-test.tex}{\from{utf8ienc.dtx}{test}}
}

\ifToplevel{
\Msg{***********************************************************}
\Msg{*}
\Msg{* To finish the installation you have to move the following}
\Msg{* files into a directory searched by TeX:}
\Msg{*}
\Msg{* \space\space utf8.def}
\Msg{* \space\space *.dfu}
\Msg{*}
\Msg{* To produce the documentation run the files ending with}
\Msg{* .dtx' through LaTeX.}
\Msg{*}
\Msg{* Happy TeXing}
\Msg{***********************************************************}
}

\endbatchfile
-------------- utf8ienc.ins

-------------- utf8ienc.dtx
% \iffalse
%<*driver>
\documentclass{ltxdoc}
\usepackage[utf8]{inputenc}
\GetFileInfo{utf8.def}
\title{Providing some UTF-8 support via \texttt{inputenc}}
\date{\fileversion\space\filedate{} printed \today}
\author{%
Frank Mittelbach \and Chris Rowley\thanks{Borrowing heavily from
code by David Carlisle and tables by Sebastian Rahtz}}
\begin{document}
\maketitle
\tableofcontents
\DocInput{utf8ienc.dtx}
\end{document}
%</driver>
% \fi
%
% \CheckSum{604}
%
% \section{Introduction}
%
% [The whole section is rather unfinished \ldots\ just like the code.]
%
% \subsection{Background and general stuff}
%
% For many reasons what this package provides is a long way from any
% type of Unicode compliance'.
%
% In stark contrast to 8-bit character sets, with 16 or more bits it can
% easily be very inefficient to support the full range.\footnote{In
%   fact, \LaTeX's current 8-bit support does not go so far as to make
%   all 8-bit characters into valid input.}  Moreover, useful support of
% character input by a typesetting system overwhelmingly means finding
% an acceptable visual representation of a sequence of characters and
% this, for \LaTeX{}, means having available a suitably encoded 8-bit
% font.
%
% Unfortunately it is not possible to predict exactly what valid UTF-8
% octet sequences will appear in a particular file so it is best to
% make all the unsupported but valid sequences produce a reasonably
% clear and noticeable error message.
%
% There are two directions from which to approach the question of what
% to load.  One is to specify the ranges of Unicode characters that will
% result in some sensible typesetting; this requires the provider to
% ensure that suitable fonts are loaded and that these input characters
% generate the correct typesetting via the encodings of those fonts.  The
% other is to inspect the font encodings to be used and use these to
% define which input Unicode characters should be supported.
%
% For Western European languages, at least, going in either direction
% leads to many straightforward decisions and a few that are more
% subjective.  In both cases some of the specifications are \TeX{}
% specific whilst most are independent of the particular typesetting
% software in use.
%
% As we have argued elsewhere, \LaTeX{} needs to refer to characters via
% seven-bit-text' names and, so far, these have been chosen by
% reference to historical sources such as Plain \TeX{} or Adobe encoding
% descriptions.  It is unclear whether this ad hoc naming structure should
% simply be extended or whether it would be useful to
% supplement it with standardised internal Unicode character names such as
% one or more of the following:\footnote{Burkhard und Holger Mittelbach
%   spielen mit mir!  Sie haben etwas hier geschrieben.}
%
% \begin{verbatim}
%   \ltxutwochar <4 hex digits>
%
%   \ltxuchar {<hex digits>}
%     B H U R R R
%
%   \ltxueightchartwo   <2 utf8 octets as 8-bit char tokens>
%   \ltxueightcharthree <3 utf8 octets ...>
%   \ltxueightcharfour  <4 utf8 octets ...>
% \end{verbatim}
%
%
% \subsection{More specific stuff}
%
% In addition to setting up the mechanism for reading UTF-8 characters
% and specifying the \LaTeX-level support available, this package
% contains support for some default historically expected \TeX-related
% characters and some example Unicode definition files' for standard
% font encodings.
%
%
% \subsection{Notes}
%
% No Unicode combining characters.
%
% No attempt to be useful beyond latin and maybe Cyrillic for European
% languages (as of now).
%
%
% \subsection{Basic operation of the code}
%
% The \texttt{inputenc} package makes the upper 8-bit characters active and
%    assigns to all of them an error message. It then waits for the
%    input encoding files to change this set-up.  Similarly, whenever
%    |\inputencoding| is encountered in a document, first the upper
%    8-bit characters are set back to produce an error and then the definitions
%    for the new input encoding are loaded, changing some of the
%    previous settings.
%
%    The 8-bit input encodings currently supported by \texttt{inputenc}
%    all use |\DeclareInputText| and the like to map an
%    8-bit number to some \LaTeX{} internal form, e.g.~to |\"a|.
%
%    The situation when supporting UTF-8 as the input encoding is
%    different, however. Here we only have to set up the actions of
%    those 8-bit numbers that can be the first octet in a UTF-8
%    representation of a Unicode character.  But we cannot simply set
%    this to some internal \LaTeX{} form since the Unicode character
%    consists of more than one octets; instead we have to define this
%    starting octet to parse the right number of further octets that
%    together form the UTF-8 representation of some Unicode character.
%
%    Therefore when switching to \texttt{utf8} within the
%    \texttt{inputenc} framework the characters with numbers (hex)
%    from \texttt{"C3} to \texttt{"E0} are defined to parse for a
%    second octet following, the characters from \texttt{"E1}
%    \texttt{"F0} are defined to parse for two more octets and finally
%    the characters from \texttt{"F1} to \texttt{"F4} are defined to
%    parse for three additional octets.
%
%    Thus when such a character is encountered in the document (so
%    long as expansion is not prohibited) a defined number of
%    unique control sequence name is immediately constructed.
%
%    This control sequence is either defined (good) or undefined
%    (likely); in the latter case the user gets an error message
%    saying that this UTF-8 sequence (or, better, Unicode character)
%    is not supported.
%
%    If it is defined then the definition will expand to a \LaTeX{}
%    internal form: e.g.~for <fill in example> we get |\"a| as the
%    internal form which then, depending on the font encoding,
%    eventually resolves to the single glyph latin-a-umlaut' or to
%    the composite glyph latin-a with an umlaut accent'.
%
%    These mappings from (UTF-8 encoded) Unicode characters to \LaTeX{}
%    internal forms are made indirectly.  The code below provides a
%    declaration |\DeclareUnicodeCharacter| which maps Unicode numbers
%    (as hexadecimal) to \LaTeX{} internal forms.
%
%    This mapping needs to be set up only once so here it is done at
%    |\begin{document}| by looking at the list of font encodings that
%    are loaded by the document and providing mappings related to
%    those font encodings whenever these are available. Thus at most
%    only those Unicode characters that can be represented by the glyphs
%    available in these encodings will be defined.
%
%    if available, that is supposed to provide the necessary mapping
%    information.
%
%
% \StopEventually{}
%
%
%
%
% \section{Coding}
%
% \subsection{Housekeeping}
%
%    The usual introductory bits and pieces:
%
%    \begin{macrocode}
%<utf8>\ProvidesFile{utf8.def}
%<t1>\ProvidesFile{t1enc.dfu}
%<ot1>\ProvidesFile{ot1enc.dfu}
%<oms>\ProvidesFile{omsenc.dfu}
%<test>\ProvidesFile{utf8-test.tex}
[2003/01/07 v1.0a UTF-8 support for inputenc]
%    \end{macrocode}
%
%    \begin{macrocode}
%<*utf8>
\makeatletter
%    \end{macrocode}
%
%
%
% \subsection{Parsing UTF-8 input}
%
% \begin{macro}{\UTFviii@two@octets}
% \begin{macro}{\UTFviii@three@octets}
% \begin{macro}{\UTFviii@four@octets}
%    A UTF-8 char (that is not actually a 7-bit char, i.e.~a
%    single octet) is parsed as
%    follows: each starting octet is an active \TeX{} character token;
%    each of these is defined below to be a macro with one to
%    three arguments nominally (depending on the starting octet). It
%    calls one of |\UTFviii@two@octets|, |\UTFviii@three@octets|, or |\UTFviii@four@octets| which
%    then actually picks up the argument(s).
%
%    From the arguments a control sequence with a name of the form
%    \verb=u8:#1#2..= is constructed where the |#i| ($i>1$) are the
%    arguments and |#1| is the starting octet (as a \TeX{} character
%    token).  Since some or even all of these characters are active
%    (when inputenc is loaded) we need to use |\string| when building
%    the csname.
%
%    The csname thus constructed can of course be undefined but to
%    avoid producing an unhelpful low-level undefined command error we
%    pass it to |\UTFviii@defined| which is responsible for producing
%    a more sensible error message (not yet done!!).  If, however, it is
%    defined we simply execute the thing (which should then expand to
%    an encoding specific internal \LaTeX{} form).
%    \begin{macrocode}
\def\UTFviii@two@octets#1#2{\expandafter
\UTFviii@defined\csname u8:#1\string#2\endcsname}
%    \end{macrocode}
% \end{macro}
%
%    \begin{macrocode}
\def\UTFviii@three@octets#1#2#3{\expandafter
\UTFviii@defined\csname u8:#1\string#2\string#3\endcsname}
%    \end{macrocode}
% \end{macro}
%
%    \begin{macrocode}
\def\UTFviii@four@octets#1#2#3#4{\expandafter
\UTFviii@defined\csname u8:#1\string#2\string#3\string#4\endcsname}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\UTFviii@defined}
%    This tests whether its argument is different from |\relax|: it
%    either calls for a sensible error message (not done), or it gets
%    the |\fi| out of the way (in case the command has arguments) and
%    executes it.
%    \begin{macrocode}
\def\UTFviii@defined#1{%
\ifx#1\relax
\ERRORnotDEFINED#1%
\else\expandafter
#1%
\fi
}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\UTFviii@loop}
%    This wonderful bit of code from Dr Carlisle defines the starting
%    octets to call |\UTFviii@two@octets| etc as appropriate. The starting
%    octet itself is passed directly as the first argument, the others
%    are picked up later en route.
%
%    The |\UTFviii@loop| loops through the numbers starting at
%    |\count@|${}+1$ and ending at |\@tempcnta| each time executing
%
%    All this is done in a group so that temporary catcode changes
%    etc.~vanish after everything is set up.
%    \begin{macrocode}
\begingroup
\catcode\~13
\catcode\"12
%    \end{macrocode}
%
%    \begin{macrocode}
\def\UTFviii@loop{%
\uccode\~\count@
\uppercase\expandafter{\UTFviii@tmp}%
\ifnum\count@<\@tempcnta
\expandafter\UTFviii@loop
\fi}
%    \end{macrocode}
%
%    Setting up 2-byte UTF-8:
%    \begin{macrocode}
\count@"C2
\@tempcnta"E0
\def\UTFviii@tmp{\xdef~{\noexpand\UTFviii@two@octets\string~}}
\UTFviii@loop
%    \end{macrocode}
%    Setting up 3-byte UTF-8:
%    \begin{macrocode}
\count@"E0
\@tempcnta"F0
\def\UTFviii@tmp{\xdef~{\noexpand\UTFviii@three@octets\string~}}
\UTFviii@loop
%    \end{macrocode}
%
%    Setting up 4-byte UTF-8:
%    \begin{macrocode}
\count@"F0
\@tempcnta"F4
\def\UTFviii@tmp{\xdef~{\noexpand\UTFviii@four@octets\string~}}
\UTFviii@loop
\endgroup
%    \end{macrocode}
% \end{macro}
%
%    For this case we must disable the warning generated by
%    \texttt{inputenc} if it doesn't see any new |\DeclareInputText|
%    commands.
%    \begin{macrocode}
\@inpenc@test
%    \end{macrocode}
%
%
%    If this file (\texttt{utf8.def}) is not being read while setting
%    up \texttt{inputenc}, i.e.~in the preamble, but when
%    |\inputencoding| is called somewhere within the document, we do not
%    need to input the specific Unicode mappings again. We therefore
%    stop reading the file at this point.
%    \begin{macrocode}
\ifx\@begindocumenthook\@undefined
\makeatother
%    \end{macrocode}
%    The |\fi| must be on the same line as |\endinput| or else it will
%    never be seen!
%    \begin{macrocode}
\endinput \fi
%    \end{macrocode}
%
%
% \subsection{Mapping Unicode codes to \LaTeX{} internal forms}
%
%
% \begin{macro}{\DeclareUnicodeCharacter}
%    The |\DeclareUnicodeCharacter| declaration defines a mapping from
%    a Unicode character code point to a \LaTeX{} internal form. The first
%    argument is the Unicode number as hexadecimal digits and the second is
%    the actual \LaTeX{} internal form.
%
%    We start by making sure that some characters have the right
%    |\catcode| when they are used in the definitions below.
%    \begin{macrocode}
\begingroup
\catcode\"=12
\catcode\<=12
\catcode\.=12
\catcode\,=12
\catcode\;=12
\catcode\!=12
\catcode\~=13
%    \end{macrocode}
%
%    \begin{macrocode}
\gdef\DeclareUnicodeCharacter#1#2{%
\count@"#1\relax
\typeout{ \space\space defining Unicode char #1 (decimal \the\count@)}%
\begingroup
%    \end{macrocode}
%    Next we do the parsing of the number stored in |\count@| and assign the
%    the macro |\parse@XML@charref| is only there to extend this code
%    to parsing Unicode numbers in other contexts one day (perhaps).
%    \begin{macrocode}
\parse@XML@charref
%    \end{macrocode}
%    Here is an example of what is happening, for 0163 (which is the
%    decimal Unicode number for \textsterling{}). After
%    |\parse@XML@charref| we have, stored in |\UTFviii@tmp|, the sequence:
%    \begin{quote}
%      |\UTFviii@two@octets| \texttt{\^A\textsterling}
%    \end{quote}
%    what we actually need to produce is a definition of the form
%    \begin{quote}
%      |\def\u8:|\texttt{\^A\textsterling} |{|\textit{\LaTeX{} internal form}|}|\,.
%    \end{quote}
%    So first we use the prefix commands |\UTFviii@two@octets|, etc.~to
%    generate the csname that we wish to define \ldots
%    \begin{macrocode}
\def\UTFviii@two@octets##1##2{\csname u8:##1\string##2\endcsname}%
\def\UTFviii@three@octets##1##2##3{\csname u8:##1%
\string##2\string##3\endcsname}%
\def\UTFviii@four@octets##1##2##3##4{\csname u8:##1%
\string##2\string##3\string##4\endcsname}%
%    \end{macrocode}
%    \ldots and then we need to use the right number of |\expandafter|s to
%    finally make the definition: expanding |\UTFviii@tmp| once to get
%    its contents, a second time to replace the prefix command by its
%    |\csname| expansion, and a third time to turn the expansion into
%    a csname after which the |\gdef| finally gets applied.
%    \begin{macrocode}
\expandafter\expandafter\expandafter
\expandafter\expandafter\expandafter
\expandafter
\gdef\UTFviii@tmp{#2}%
\endgroup
}
%    \end{macrocode}
% \end{macro}
%
%
% \begin{macro}{\parse@XML@charref}
%    This macro parses a Unicode number (decimal) and returns its
%    UTF-8 representation as a sequence of \TeX{} car tokens. In the
%    original code it had two argument delimited by \texttt{;} here,
%    however, we supply the Unicode number implicitly.
%    \begin{macrocode}
\gdef\parse@XML@charref{%
%    \end{macrocode}
%    We need to keep a few things local, mainly the |\uccode|'s that
%    are set up below. However, the group originally used here is
%    actually unnecessary since we call this macro only within another
%    group; but it will be important to restore the group if this
%    macro gets used for other purposes.
%    \begin{macrocode}
%  \begingroup
%    \end{macrocode}
%    The original code from David supported the convention that a
%    Unicode slot number could be given either as a decimal or as a
%    hexadecimal (by starting with \texttt{x}).  We do not do this so
%    this code is also removed.  This could be reactivated if one
%    wants to support document commands that accept Unicode numbers
%    (but then the first case needs to be changed from an error
%    message back to something more useful again).
%    \begin{macrocode}
%  \uppercase{\count@\if x\noexpand#1"\else#1\fi#2}\relax
%    \end{macrocode}
%    As |\count@| already contains the right value we make
%    |\parse@XML@charref| work without arguments.
%    \begin{macrocode}
\ifnum\count@<"A0\relax
\ERROR-WE-DONT-DEAL-WITH-THAT
%    \end{macrocode}
%    Do not ask us to provide an explanation for the code below, it is
%    straight borrowed from \texttt{xmltex} by David and we trust him
%    totally (and we are too lazy to reread the Unicode book to see if
%    this is the correct algorithm).\footnote{We were hoping to also
%    find in his work the \TeX{} code for going the other way: from
%    UTF-8 octets to Unicode slot number, but no luck!}
%    \begin{macrocode}
\else\ifnum\count@<"800\relax
\parse@UTFviii@a,%
\else\ifnum\count@<"10000\relax
\parse@UTFviii@a;%
\parse@UTFviii@a,%
\else
\parse@UTFviii@a;%
\parse@UTFviii@a,%
\parse@UTFviii@a!%
\fi
\fi
\fi
%  \endgroup
}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\parse@UTFviii@a}
%    \ldots so somebody else can document this part :-) \ldots~David?:-))))!
%    \begin{macrocode}
\gdef\parse@UTFviii@a#1{%
\@tempcnta\count@
\divide\count@64
\@tempcntb\count@
\multiply\count@64
\uccode#1\@tempcnta
\count@\@tempcntb}
%    \end{macrocode}
% \end{macro}
%
% \begin{macro}{\parse@UTFviii@b}
%    \ldots same here
%    \begin{macrocode}
\gdef\parse@UTFviii@b#1#2#3#4{%
\uccode#3\count@
\uppercase{\gdef\UTFviii@tmp{#2#3#4}}}
%    \end{macrocode}
%
%    \begin{macrocode}
\endgroup
%    \end{macrocode}
% \end{macro}
%
%    \begin{macrocode}
\@onlypreamble\DeclareUnicodeCharacter
%    \end{macrocode}
%    These are preamble only as long as we don't support Unicode
%    charrefs in documents.
%    \begin{macrocode}
\@onlypreamble\parse@XML@charref
\@onlypreamble\parse@UTFviii@a
\@onlypreamble\parse@UTFviii@b
%    \end{macrocode}
%
%
%
%    At begin of document we loop through all defined encodings
%    (stored in |\cdp@list| and for each load a file
%    \textit{name}\texttt{enc.dfu} if it exist. That file is then
%    supposed to contain |\DeclareUnicodeCharacter| declarations.
%    \begin{macrocode}
\AtBeginDocument{%
\begingroup
\def\cdp@elt#1#2#3#4{%
\typeout{Now handling font encoding #1 ...}%
\lowercase{%
\InputIfFileExists{#1enc.dfu}}
{\typeout{... processing Unicode mapping file for font encoding #1}}%
{\typeout{... no Unicode mapping file for font encoding #1}}%
}\cdp@list
\endgroup}
\makeatother
%</utf8>
%    \end{macrocode}
%
%
%
% \section{Mapping characters that existing in font encodings}
%
% This section is a first attempt to provide Unicode definitions for
%    characters whose glyphs are currently provided by the standard \LaTeX{}
%    font-encodings |T1|, |OT1|, etc. They are by no means completed
%    and need checking.
%
% For example, one should check the already existing input encodings
%    for glyphs that may in fact be available and required,
%    e.g.~\texttt{latin4} has a number of glyphs with the |\=|
%    accent. Since the |T1| encoding does not provide such glyphs as
%    these characters are not listed below (yet).
%
% The list below was generated by looking at the current \LaTeX{} font
%    encoding files, e.g., \texttt{t1enc.def} and using the work by
%    Sebastian Rahtz in (\texttt{ucharacters.sty}), with a few
%    modifications.
%
% \subsection{Mappings for T1 glyphs}
%
%    \begin{macrocode}
%<*t1>
\DeclareUnicodeCharacter{00A1}{\textexclamdown }
\DeclareUnicodeCharacter{00A3}{\textsterling}
\DeclareUnicodeCharacter{00AB}{\guillemotleft}
\DeclareUnicodeCharacter{00BB}{\guillemotright}
\DeclareUnicodeCharacter{00BF}{\textquestiondown }
\DeclareUnicodeCharacter{00C0}{\@tabacckludgeA}
\DeclareUnicodeCharacter{00C1}{\@tabacckludge'A}
\DeclareUnicodeCharacter{00C2}{\^A}
\DeclareUnicodeCharacter{00C3}{\~A}
\DeclareUnicodeCharacter{00C4}{\"A}
\DeclareUnicodeCharacter{00C5}{\r A}
\DeclareUnicodeCharacter{00C6}{\AE }
\DeclareUnicodeCharacter{00C7}{\c C}
\DeclareUnicodeCharacter{00C8}{\@tabacckludgeE}
\DeclareUnicodeCharacter{00C9}{\@tabacckludge'E}
\DeclareUnicodeCharacter{00CA}{\^E}
\DeclareUnicodeCharacter{00CB}{\"E}
\DeclareUnicodeCharacter{00CC}{\@tabacckludgeI}
\DeclareUnicodeCharacter{00CD}{\@tabacckludge'I}
\DeclareUnicodeCharacter{00CE}{\^I}
\DeclareUnicodeCharacter{00CF}{\"I}
\DeclareUnicodeCharacter{00D0}{\DH }
\DeclareUnicodeCharacter{00D1}{\~N}
\DeclareUnicodeCharacter{00D2}{\@tabacckludgeO}
\DeclareUnicodeCharacter{00D3}{\@tabacckludge'O}
\DeclareUnicodeCharacter{00D4}{\^O}
\DeclareUnicodeCharacter{00D5}{\~O}
\DeclareUnicodeCharacter{00D6}{\"O}
\DeclareUnicodeCharacter{00D8}{\O }
\DeclareUnicodeCharacter{00D9}{\@tabacckludgeU}
\DeclareUnicodeCharacter{00DA}{\@tabacckludge'U}
\DeclareUnicodeCharacter{00DB}{\^U}
\DeclareUnicodeCharacter{00DC}{\"U}
\DeclareUnicodeCharacter{00DD}{\@tabacckludge'Y}
\DeclareUnicodeCharacter{00DE}{\TH }
\DeclareUnicodeCharacter{00DF}{\ss }
\DeclareUnicodeCharacter{00E0}{\@tabacckludgea}
\DeclareUnicodeCharacter{00E1}{\@tabacckludge'a}
\DeclareUnicodeCharacter{00E2}{\^a}
\DeclareUnicodeCharacter{00E3}{\~a}
\DeclareUnicodeCharacter{00E4}{\"a}
\DeclareUnicodeCharacter{00E5}{\r a}
\DeclareUnicodeCharacter{00E6}{\ae }
\DeclareUnicodeCharacter{00E7}{\c c}
\DeclareUnicodeCharacter{00E8}{\@tabacckludgee}
\DeclareUnicodeCharacter{00E9}{\@tabacckludge'e}
\DeclareUnicodeCharacter{00EA}{\^e}
\DeclareUnicodeCharacter{00EB}{\"e}
\DeclareUnicodeCharacter{00EC}{\@tabacckludge\i}
\DeclareUnicodeCharacter{00EC}{\@tabacckludgei}
\DeclareUnicodeCharacter{00ED}{\@tabacckludge'\i}
\DeclareUnicodeCharacter{00ED}{\@tabacckludge'i}
\DeclareUnicodeCharacter{00EE}{\^\i}
\DeclareUnicodeCharacter{00EE}{\^i}
\DeclareUnicodeCharacter{00EF}{\"\i}
\DeclareUnicodeCharacter{00EF}{\"i}
\DeclareUnicodeCharacter{00F0}{\dh }
\DeclareUnicodeCharacter{00F1}{\~n}
\DeclareUnicodeCharacter{00F2}{\@tabacckludgeo}
\DeclareUnicodeCharacter{00F3}{\@tabacckludge'o}
\DeclareUnicodeCharacter{00F4}{\^o}
\DeclareUnicodeCharacter{00F5}{\~o}
\DeclareUnicodeCharacter{00F6}{\"o}
\DeclareUnicodeCharacter{00F8}{\o }
\DeclareUnicodeCharacter{00F9}{\@tabacckludgeu}
\DeclareUnicodeCharacter{00FA}{\@tabacckludge'u}
\DeclareUnicodeCharacter{00FB}{\^u}
\DeclareUnicodeCharacter{00FC}{\"u}
\DeclareUnicodeCharacter{00FD}{\@tabacckludge'y}
\DeclareUnicodeCharacter{00FE}{\th }
\DeclareUnicodeCharacter{00FF}{\"y}
\DeclareUnicodeCharacter{0102}{\u A}
\DeclareUnicodeCharacter{0103}{\u a}
\DeclareUnicodeCharacter{0104}{\k A}
\DeclareUnicodeCharacter{0105}{\k a}
\DeclareUnicodeCharacter{0106}{\@tabacckludge'C}
\DeclareUnicodeCharacter{0107}{\@tabacckludge'c}
\DeclareUnicodeCharacter{010C}{\v C}
\DeclareUnicodeCharacter{010D}{\v c}
\DeclareUnicodeCharacter{010E}{\v D}
\DeclareUnicodeCharacter{010F}{\v d}
\DeclareUnicodeCharacter{0110}{\DJ }
\DeclareUnicodeCharacter{0111}{\dj }
\DeclareUnicodeCharacter{0118}{\k E}
\DeclareUnicodeCharacter{0119}{\k e}
\DeclareUnicodeCharacter{011A}{\v E}
\DeclareUnicodeCharacter{011B}{\v e}
\DeclareUnicodeCharacter{011E}{\u G}
\DeclareUnicodeCharacter{011F}{\u g}
\DeclareUnicodeCharacter{0130}{\.I}
\DeclareUnicodeCharacter{0131}{\i}
\DeclareUnicodeCharacter{0131}{\i}
\DeclareUnicodeCharacter{0139}{\@tabacckludge'L}
\DeclareUnicodeCharacter{013A}{\@tabacckludge'l}
\DeclareUnicodeCharacter{013D}{\v L}
\DeclareUnicodeCharacter{013E}{\v l}
\DeclareUnicodeCharacter{0141}{\L }
\DeclareUnicodeCharacter{0142}{\l }
\DeclareUnicodeCharacter{0143}{\@tabacckludge'N}
\DeclareUnicodeCharacter{0144}{\@tabacckludge'n}
\DeclareUnicodeCharacter{0147}{\v N}
\DeclareUnicodeCharacter{0148}{\v n}
\DeclareUnicodeCharacter{014A}{\NG }
\DeclareUnicodeCharacter{014B}{\ng }
\DeclareUnicodeCharacter{0150}{\H O}
\DeclareUnicodeCharacter{0151}{\H o}
\DeclareUnicodeCharacter{0152}{\OE }
\DeclareUnicodeCharacter{0153}{\oe }
\DeclareUnicodeCharacter{0154}{\@tabacckludge'R}
\DeclareUnicodeCharacter{0155}{\@tabacckludge'r}
\DeclareUnicodeCharacter{0158}{\v R}
\DeclareUnicodeCharacter{0159}{\v r}
\DeclareUnicodeCharacter{015A}{\@tabacckludge'S}
\DeclareUnicodeCharacter{015B}{\@tabacckludge's}
\DeclareUnicodeCharacter{015E}{\c S}
\DeclareUnicodeCharacter{015F}{\c s}
\DeclareUnicodeCharacter{0160}{\v S}
\DeclareUnicodeCharacter{0161}{\v s}
\DeclareUnicodeCharacter{0162}{\c T}
\DeclareUnicodeCharacter{0163}{\c t}
\DeclareUnicodeCharacter{0164}{\v T}
\DeclareUnicodeCharacter{0165}{\v t}
\DeclareUnicodeCharacter{016E}{\r U}
\DeclareUnicodeCharacter{016F}{\r u}
\DeclareUnicodeCharacter{0170}{\H U}
\DeclareUnicodeCharacter{0171}{\H u}
\DeclareUnicodeCharacter{0178}{\"Y}
\DeclareUnicodeCharacter{0179}{\@tabacckludge'Z}
\DeclareUnicodeCharacter{017A}{\@tabacckludge'z}
\DeclareUnicodeCharacter{017B}{\.Z}
\DeclareUnicodeCharacter{017C}{\.z}
\DeclareUnicodeCharacter{017D}{\v Z}
\DeclareUnicodeCharacter{017E}{\v z}
\DeclareUnicodeCharacter{2013}{\textendash }
\DeclareUnicodeCharacter{2014}{\textemdash}
\DeclareUnicodeCharacter{2018}{\textquoteleft}
\DeclareUnicodeCharacter{2019}{\textquoteright}
\DeclareUnicodeCharacter{201C}{\textquotedblleft }
\DeclareUnicodeCharacter{201D}{\textquotedblright }
\DeclareUnicodeCharacter{2030}{\textperthousand }
\DeclareUnicodeCharacter{2031}{\textpertenthousand }
\DeclareUnicodeCharacter{2039}{\guilsinglleft }
\DeclareUnicodeCharacter{203A}{\guilsinglright }
\DeclareUnicodeCharacter{2423}{\textvisiblespace }
\DeclareUnicodeCharacter{201A}{\quotesinglbase}
\DeclareUnicodeCharacter{201E}{\quotedblbase}
%</t1>
%    \end{macrocode}
%    The following definitions are in the encoding file but have no
%    direct equivalent in Unicode or simply do not make sense in that
%    context (or I couldn't find anything or \ldots :-).
%\begin{verbatim}
%\DeclareTextSymbol{\j}{OT1}{17}
%\DeclareTextSymbol{\SS}{T1}{223}
%\DeclareTextSymbol{\textcompwordmark}{T1}{23}
%
%\DeclareTextAccent{\"}{OT1}{127}
%\DeclareTextAccent{\'}{OT1}{19}
%\DeclareTextAccent{\.}{OT1}{95}
%\DeclareTextAccent{\=}{OT1}{22}
%\DeclareTextAccent{\H}{OT1}{125}
%\DeclareTextAccent{\^}{OT1}{94}
%\DeclareTextAccent{\}{OT1}{18}
%\DeclareTextAccent{\r}{OT1}{23}
%\DeclareTextAccent{\u}{OT1}{21}
%\DeclareTextAccent{\v}{OT1}{20}
%\DeclareTextAccent{\~}{OT1}{126}
%\DeclareTextCommand{\b}{OT1}[1]
%\DeclareTextCommand{\c}{OT1}[1]
%\DeclareTextCommand{\d}{OT1}[1]
%\DeclareTextCommand{\k}{T1}[1]
%\end{verbatim}
%
%
%
% \subsection{Mappings for OT1 glyphs}
%
%    This is even more incomplete as again it covers only the single
%    glyphs from |OT1| plus some that have been explicitly defined for
%    this encoding. Everything that is provided in |T1|, and that
%    could be provided as composite glyphs via |OT1|, could and
%    probably should be set up as well.  Which leaves the many things
%    that are not provided in |T1| but can be provided in |OT1| (and
%    in |T1|) by composite glyphs.

%    \begin{macrocode}
%<*ot1>
\DeclareUnicodeCharacter{00A1}{\textexclamdown }
\DeclareUnicodeCharacter{00A3}{\textsterling}
\DeclareUnicodeCharacter{00BF}{\textquestiondown }
\DeclareUnicodeCharacter{00C5}{\r A}
\DeclareUnicodeCharacter{00C6}{\AE }
\DeclareUnicodeCharacter{00D8}{\O }
\DeclareUnicodeCharacter{00DF}{\ss }
\DeclareUnicodeCharacter{00E6}{\ae }
\DeclareUnicodeCharacter{00EC}{\@tabacckludgei}
\DeclareUnicodeCharacter{00ED}{\@tabacckludge'i}
\DeclareUnicodeCharacter{00EE}{\^i}
\DeclareUnicodeCharacter{00EF}{\"i}
\DeclareUnicodeCharacter{00F8}{\o }
\DeclareUnicodeCharacter{0131}{\i}
\DeclareUnicodeCharacter{0141}{\L }
\DeclareUnicodeCharacter{0142}{\l }
\DeclareUnicodeCharacter{0152}{\OE }
\DeclareUnicodeCharacter{0153}{\oe }
\DeclareUnicodeCharacter{2013}{\textendash }
\DeclareUnicodeCharacter{2014}{\textemdash}
\DeclareUnicodeCharacter{2018}{\textquoteleft}
\DeclareUnicodeCharacter{2019}{\textquoteright}
\DeclareUnicodeCharacter{201C}{\textquotedblleft }
\DeclareUnicodeCharacter{201D}{\textquotedblright }
%</ot1>
%    \end{macrocode}
% Stuff not mapped (note that |\j| ($\jmath$) is not a Unicode character):
%\begin{verbatim}
%\DeclareTextSymbol{\j}{OT1}{17}
%\DeclareTextAccent{\"}{OT1}{127}
%\DeclareTextAccent{\'}{OT1}{19}
%\DeclareTextAccent{\.}{OT1}{95}
%\DeclareTextAccent{\=}{OT1}{22}
%\DeclareTextAccent{\^}{OT1}{94}
%\DeclareTextAccent{\}{OT1}{18}
%\DeclareTextAccent{\~}{OT1}{126}
%\DeclareTextAccent{\H}{OT1}{125}
%\DeclareTextAccent{\u}{OT1}{21}
%\DeclareTextAccent{\v}{OT1}{20}
%\DeclareTextAccent{\r}{OT1}{23}
%\DeclareTextCommand{\b}{OT1}[1]
%\DeclareTextCommand{\c}{OT1}[1]
%\DeclareTextCommand{\d}{OT1}[1]
%\end{verbatim}
%
%
%
% \subsection{Mappings for OMS glyphs}
%
%    Only a few glyphs to set up here.
%    \begin{macrocode}
%<*oms>
\DeclareUnicodeCharacter{00A7}{\textsection}
\DeclareUnicodeCharacter{00B6}{\textparagraph}
\DeclareUnicodeCharacter{02D9}{\textperiodcentered}
\DeclareUnicodeCharacter{2020}{\textdagger}
\DeclareUnicodeCharacter{2021}{\textdaggerdbl}
\DeclareUnicodeCharacter{2022}{\textbullet}
%</oms>
%    \end{macrocode}
%
% Characters like |\textbackslash| are not mapped as they are
%    (primarily) only in the lower 127 and the code here only sets up
%    mappings for UTF-8 characters that are at least 2 octets long.
%\begin{verbatim}
%\DeclareTextSymbol{\textbackslash}{OMS}{110}        % "6E
%\DeclareTextSymbol{\textbar}{OMS}{106}              % "6A
%\DeclareTextSymbol{\textbraceleft}{OMS}{102}        % "66
%\DeclareTextSymbol{\textbraceright}{OMS}{103}       % "67
%\end{verbatim}
%
% But the following (and some others) might actually lurk in Unicode
%    somewhere\ldots
%\begin{verbatim}
%\DeclareTextSymbol{\textasteriskcentered}{OMS}{3}   % "03
%\DeclareTextCommand{\textcircled}{OMS}
%\end{verbatim}
%
%
%
%
% \subsection{Mappings for TS1 glyphs}
%
% Exercise for somebody else.
%
%
% \subsection{Mappings for \texttt{latex.ltx} glyphs}
%
% There is also a collection of characters already set up in the kernel,
% one way or the other. Since these do not clearly relate to any
%    particular font encoding they are mapped when the
% \texttt{utf8} support is first set up.
%
% Also there are a number of |\providecommand|s in the various input
% encoding files which may or may not go into this part.
%    \begin{macrocode}
%<*utf8>
% This space is intentionally empty ...
%</utf8>
%    \end{macrocode}
%
%
% \section{A test document}
%
%    Here is a very small test document which may or may not survive
%    if the current document is transfered from one place to the
%    other.
%    \begin{macrocode}
%<*test>
\documentclass{article}

\usepackage[latin1,utf8]{inputenc}
\usepackage[T1]{fontenc}
\usepackage{trace}

\begin{document}

German umlauts in UTF-8: äöü

\inputencoding{latin1}  % switch to latin1

German umlauts in UTF-8 but read by latin1 (and will produce one
error since \verb=\textcurrency= is not provided): äöü

\inputencoding{utf8}    % switch back to utf8

German umlauts in UTF-8: äöü

\showoutput
\tracingstats=2
\stop
%</test>
%    \end{macrocode}
%
% \Finale
%
\endinput
-------------- utf8ienc.dtx