This file documents LaTeX2RTF , a converter that translates LaTeX to RTF.
In 1994 the first Version of LaTeX2RTF was written by Fernando Dorner and Andreas Granzer of the Viena University supervised by Ralf Schlatterbeck in a one-semester course. They created a simple LaTeX parser and added most of the infrastructure for the program. This was version 1.0 of LaTeX2RTF . In 1995, work on LaTeX2RTF was continued in another one-semester course by Friedrich Polzer and Gerhard Trisko. The result was LaTeX2RTF version 1.5. Ralf Schlatterbeck maintained and extended LaTeX2RTF until 1998.
In 1998 Georg Lehner found the reference to LaTeX2RTF on the TeX Conversion Webpage of Wilfried Hennings and added some functionality and took over the maintainence of the program. The last version release by Georg is 1.8aa. The bulk of development post 1.8aa was done by Scott Prahl. Wilfried Hennings now coordinates the development of the program and maintains the project on SourceForge where there are also (low volume) mailing lists for users latex2rtf-users@lists.sourceforge.net and developers latex2rtf-developers@lists.sourceforge.net. For subscription to these lists: latex2rtf-users-request@lists.sourceforge.net?subject=subscribe or latex2rtf-developers-request@lists.sourceforge.net?subject=subscribe
As of April 2002, version 1.9.13 of LaTeX2RTF is available. One day there shall be a jump to Version 2.0, but this is not history but future ...
The contents of this manual were composed by copying shamelessly what was available in the original sources and documentation.
LaTeX2RTF is a translator program from LaTeX text into "rich text format" files. These files are commonly referred to as RTF files. RTF is a published standard format by Microsoft. This standard can be ambiguous in places and Microsoft ignores parts of the standard it finds inconvenient, but RTF is widely used by many WYSIWIG text editors and is supported by Microsoft Word and most text processors.
LaTeX2RTF translates the text and as much of the formatting information from LaTeX to RTF. Be forewarned that the typeset output is not nearly as good as what you would get from using LaTeX directly. So, why bother translating? Consider,
There are drawbacks to the conversion process. In fact, don't expect any LaTeX file to be converted as you would like, don't expect it to be converted without errors or warnings, and don't be especially surprised when it doesn't convert at all. LaTeX2RTF is known to have many bugs and many missing features. Paradoxically, this number seems to grow more and more with each day. However, we can categorically state that there are some special cases, in which a LaTeX file will be translated to RTF satisfactorily by LaTeX2RTF -- This was sort of disclaimer, ok? OK!
LaTeX is a system for typesetting text and therefore it focuses on the logical structure of a document, whilst RTF is meant to be a transport format for a family of Desktop Publishing Software, dealing mostly with the design of a text.
Although the possible different commands and styles in LaTeX are much more flexible and standardized than in RTF, only a small subset of commands has been implemented to date. See Unimplemented Features.
Some of the capabilities of LaTeX2RTF are restricted in scope or buggy. See Known Bugs.
RTF is a moving target, because Microsoft does not stop inventing new extensions and features for it. So you cannot view newer RTF files with older word processors. LaTeX2RTF eventually will generate RTF output that is unreadable with your Program. The syntax and semantics of RTF are somewhat artistic, i.e., you can generate a syntactically correct RTF file that cannot be displayed by some/most word processors.
For more details on RTF look at the RTF-Tools by Paul DuBois, and the corresponding newsgroups, etc. http://www.primate.wisc.edu/
LaTeX2RTF is available for many Unix Platforms, for the Macintosh, and for MS-DOS, including all versions of MS Windows 95.
The MS-DOS version (1.8aa and up) will also run under all MS Windows versions. It requires an i386 processor or better. The Win32 Version requires Cygnus-Cygwin32. As the MS-DOS version (1.8aa and up) will also run under all MS windows versions, the Win32 version has been discontinued.
Subsequent to 1.8aa, Scott Prahl has fixed many bugs, improved
handling of tabular environments, significantly improved handling of
equations, added macro expansion, improved handling of
cross references, added bitmap rendering of equations,
and added some support for \includegraphics files.
You find the latest version on SourceForge and - with some delay - on CTAN sites: e.g., http://www.dante.de or http://www.ctan.org.
There are a couple of people working on the LaTeX2RTF , coordinated by Wilfried Hennings. Contact the mailing_list for the latest news.
The documentation of the program is found in the doc/ directory in the
file latex2rtf.texi in the GNU TeXInfo format. For your convenience, you
can find HTML and PDF versions of the manual there as well.
To install,
Makefile for your local configuration. In particular, pay attention
to the PREFIX variable. If you do not have root access you might wish
to set the makefile variable PREFIX to be your home directory.
latex2png, found in scripts/.
latex2png requires that
both LaTeX
and ImageMagick are installed.
LaTeX2RTF
will translate documents without a working latex2png, but some
features will be missing.
make
If this is not your first time installation, you may want to preserve
your old configuration (*.cfg) files. Copy them to a safe place before
installing.
make check
[OPTIONAL] This tests LaTeX2RTF on a variety of LaTeX files. Expect a whole lot of warnings, but no outright errors.
make install
If your mkdir doesn't support the -p option, then create
the necessary directories by hand and remove the option from the $MKDIR variable.
If you have other problems, just copy latex2rtf and latex2png to
a binary directory, and move the contents of the cfg/ directory to
the location specified by $CFG_INSTALL.
make install-info
[OPTIONAL] This installs .info files for use with the info
program.
You no longer need to define the environment variable RTFPATH. This
is only necessary if you move the directory containing the .cfg files.
Just define RTFPATH to be the path to the new location.
The UNIX and Mac distributions do not contain an executable for DOS
or Windows. Instead, get the DOS port as file latex2rtf-x.xx_dos.zip
(where x.xx is the version number) from
SourceForge
The DOS distribution contains a precompiled executable which should run under plain DOS and also under any MS Windows system.
To install latex2rtf, extract all files from the zip archive to C:\l2r
or to another folder (i.e. directory) of your choice,
preserving the folder structure (winzip: check "use folder names").
If you extracted them to another folder,
edit the files L2R.BAT and L2RSEM.BAT and change the pathes
to LATEX2RT.EXE and to the cfg files to the folder where you put them.
Make sure that the folder containing the files
L2R.BAT, L2RSEM.BAT and LATEX2PNG.BAT
is in your search path
(e.g. adding to your autoexcec.bat: "PATH=...;...;C:\l2r")
or put these files into a folder which is in your search path.
The file L2R.BAT or L2RSEM.BAT will call
the DJGPP generated version LATEX2RT.EXE.
If you want to use the option to convert equations to bitmaps (-M 12), you must have TeX, ImageMagick and Ghostscript installed. This works only on win32 systems (win95, win98, winNT4, win2000, winXP) because ImageMagick is available only for win32, not for pure DOS.
Either the folders where TeX, ImageMagick and Ghostscript are installed
must be in your search path, or you must edit the file LATEX2PNG.BAT
and specify the folders where TeX, ImageMagick and Ghostscript are installed
in the SET commands, e.g.
... rem Set path to latex here (ending with \) rem if it isn't in your search path: SET TEXDIR=C:\texmf\miktex\bin\ rem Set path to ImageMagick's convert here (ending with \) rem if it isn't in your search path: SET MAGICKDIR=C:\ImageMagick\ rem Set path to Ghostscript's gswin32c.exe here (NOT ending with \) rem if it isn't in your search path: SET GSDIR=C:\Aladdin\gs6.01\bin
If you want a MacOS X version, make sure that you have installed the developer tools CD that is appropriate for your OS version, and then follow the directions above for UNIX installation.
There is a classic MacOS PPC port of the 1.9k of LaTeX2RTF and I have made a binary distribution of this application. Unfortunately, because I (Scott Prahl) do all development under MacOS X, the binaries for the Classic version often lag (far) behind the current UNIX version.
To convert a LaTeX
file, just drag the file onto the LaTeX2RTF
application icon. The
translation is best if there are .aux and .bbl files in the same folder as
the .tex file to be converted. These should be generated using LaTeX
and
bibtex.
The code for LaTeX2RTF is standard ANSI C. Some possible pitfalls are
gcc.
-Wall and -pedantic and
so may generate errors that were not found in a different development
system. Plese report these, but a quick fix is to remove all compiler
options.
RTFPATH (either as environment variable or by
the command line parameter -P).
HAS_NO_STRDUP in the Makefile when your system
lacks this facility.
make checkAll the files in the test directory are converted (with more or less
success) using LaTeX2RTF
and are tested before each CVS check-in and with
all released tarballs. If you have successfully compiled LaTeX2RTF
then
problems are probably caused by
RTFPATH.
HAS_NO_STRDUP in the Makefile when your system
lacks this facility.
LaTeX2RTF
assumes that the .tex file you want to convert is a valid LaTeX
document. The chances of a successful LaTeX2RTF
conversion are slightly better than the
proverbial snowball's if the .tex file doesn't latex properly. Use LaTeX
to find and correct errors before using LaTeX2RTF .
To correctly convert font names you must edit the fonts.cfg configuration file.
This file is used to specify the needed font names and how the LaTeX
default font names
should be converted to RTF. See Font Configuration. LaTeX
variables and user
defined commands are not evaluated. They will be simply ignored. To let LaTeX2RTF
know the
names of variables you can add them in the ignore.cfg file. See Ignore Command.
The environment variable RTFPATH may contain a search path for the support files (all
files ending in .cfg). If no file is found during the search in the search-path or
if the environment variable is not set, the compiled-in default for the configuration-file
directory is used. If the files are not found at all the program aborts.
In the MS-DOS version the search path is separated by `;' in the Unix version by `:'. For the paths themselves apply `\' and `/'. A separator may appear at the beginning or ending of RTFPATH.
Make sure that the configuration files are in the correct directory. LaTeX2RTF
will need at
least fonts.cfg, direct.cfg, ignore.cfg, english.cfg. You may
have to change one ore more of them to suit your needs. See Configuring LaTeX2RTF.
See Missing options, for actual implementations irregularities.
See Reporting Bugs, for information on how to reach the maintainer.
The LaTeX2RTF
command converts a LaTeX
file into RTF text format. The text and
much of the formatting information is translated to RTF making the new file
look similar to the original. The command line syntax is:
latex2rtf [-options] inputfile.[tex]
The -options may consist of one or more of the following
-a auxfile
.aux file (for table and figure references) that
differs from inputfile.aux. If this is omitted, the name of the
inputfile with the suffix replaced .aux'will be taken. You must
provide both files (.tex and the .aux) to be able to
convert cross-references in a LaTeX
file. The .aux is created
by running the inputfile.tex through latex.
-b bblfile
bblfile is specified with the -b option, LaTeX2RTF
uses a
inputfile.bbl. The bblfile file is used for citations and is typically
created by running inputfile.aux through bibtex.
-d
-d option determines the amount of debugging information to send to
stderr while translating. `0' means only Errors, `1' Warning Messages
(default) also. These numbers can go as high as `7'.
-i language
\usepackage[language]{babel} where
language is one of the languages supported by the babel package. All
languages listed in the babel system are supported so far as translations for "Chapter,"
"References," and the like. Specific support for specific options of the german style
package are explictly supported. The french translated names are found in the
french.cfg file in the cfg/ directory. See Language Configuration.
-l
-i latin1 (Note that the default behavior is to use ansinew
which is a superset of latin1). Included for
backwards compatibility.
-o outputfile
outputfile is specified with the -o option, the resulting RTF
filename is formed by removing .tex from the inputfile and appending
.rtf.
-C codepage
\usepackage[codepage]{inputenc} This command
is not needed if the above command is already in the LaTeX2e
file. You may select any
of the following code pages: ansinew, applemac, cp437, cp437de, cp850, cp852, cp865,
decmulti, cp1250, cp1252, latin1, latin2, latin3, latin4, latin5, latin9, and next. The
default behavior is to use ansinew (code page 1252).
-D dots_per_inch
-M#
-M1
-M2
-M4
-M8
-M16
-M3
-M6
-M12
-P /path/to/cfg
.cfg files
-S
-V
-W
-Z#
} to the end of the RTF file. This
is useful for files that are not cleanly converted by LaTeX2RTF
.
With no arguments other than switches starting with a "-",
LaTeX2RTF
acts as a filter, i.e., it reads from stdin and writes to stdout.
In addition, diagnostic messages are sent to stderr. If these standard
channels are not redirected using < and >, then the
input is read from the command line, and both output and error
messages are printed on the screen.
If a non-switch argument is present, LaTeX2RTF assumes it is the name of the input file. The file must have extension ".tex" but the extension is optional. The output file is constructed from the input file name by removing the extension ".tex" and adding ".rtf".
With the -d option you can specify how much processing information
LaTeX2RTF
reports.
If there is a logfile specified the output goes to this file. Nonetheless
Warnings and Errors are logged to stderr always.
Possible values of -d are
In this chapter you find what styles is LaTeX2RTF supposed to translate correctly to RTF.
LaTeX2RTF
understands most of the commands introduced with LaTeX2e
. It supports both
the old 2.09 version of \documentstyle[options]{format#} and the newer
\documentclass[options]{format}.
It is not necesary to specify the -C option if you use \usepackage{isolatin1}
or \documentstyle[isolatin1]{...}. LaTeX2RTF
automagically detects these
packages/style options and switches to processing of ISO-Latin1 codes.
The following languages from the Babel package are supported: afrikaans, german, nynorsk, spanish, bahasa, dutch, icelandic, polish, swedish, basque, english, portuges, turkish, brazil, esperanto, irish, romanian, usorbian, breton, estonian, italian, samin, welsh, catalan, finnish, latin, scottish, croatian, lsorbian, serbian, czech, french, magyar, slovak, danish, galician, norsk, slovene,
The only thing that these files do is to translate various words usually emitted by LaTeX during processing. For example, this ensures that the LaTeX2RTF will provide the correct translation of the word "Chapter" in the converted document.
You can select any of the above languages using the -l option. This
is not needed if your LaTeX
file contains \usepackage[language]{babel}.
Encountering the german package or documentstyle option (by H. Partl of the
Viena University) makes LaTeX2RTF
behave like that: German Quotes, German
Umlauts by "a, etc.... This support is programmed directly
into LaTeX2RTF
and supporting similar features for other languages will require
patching the source code.
See Language Configuration, for details on how to write a language.cfg file
for your language by yourself.
Cross references include everything that you might expect and then some: bibliographic citations, equation references, table references, figure references, and section references. Section, equation, table and figure references are implemented by placing RTF bookmarks around the equation number (or table number or figure number).
Page references work but are implemented as "warm" cross-references. This
means that Word does not automatically update the page references when the file
is opened. To update the page references you must select the entire document
(in Word) and press F9.
Bibliographic references currently require that a valid .aux file be
present. This is where LaTeX2RTF
obtains the reference numbers. It would be
nice if LaTeX2RTF
just automatically numbered the references when there was
no .aux file, but LaTeX2RTF
does not do this yet.
Footnotes are implemented and appear at the bottom of each page.
Indexing is reasonable well supported. The simple mark-up of makeindex
\index{topic!subtopic@\textit{subtopic}}
is supported. The rest of the fancy indexing stuff is not implemented.
The index is created at the location of the \printindex command.
When a file with an index is first opened in Word, you must select the
entire file and update the page references and fields by pressing F9.
Currently, there is no support for \labels of \items in enumerate environments.
The conversion of cross-references is not perfect because of the different
mechanisms in the LaTeX
and Word worlds. In particular, if there are
multiple \label in a figure, table, or section environment then only the first
gets processed. It is also possible to confuse the LaTeX2RTF
in eqnarray environments.
There are four separate levels of equation translation based on the -M switch, See LaTeX2RTF Options. Each equation is now converted either to an EQ field or to a bitmap.
This is an interim solution (for some definition of "interim"). Ideally the equations would become OLE equation objects in the RTF file, but this needs to be implemented.
Some functions in the EQ fields have two or more parameters with a separator between each two. Unfortunately, the interpretation of these separators depends on the country specific settings in the MS Windows system in which the rtf file is opened. E.g. in English versions of MS Windows, the default parameter separator is the comma, in German versions the default is the semicolon. If the parameter in the RTF file does not match the Windows setting, some EQ fields are not interpreted correctly. You can check and set the separator in [Windows control panel - country settings - numbers - list separator]. By default, latex2rtf uses the comma as separator. If latex2rtf is called with the command line parameter -S , the semicolon is inserted as parameter delimiter.
The table code is currently barely working. It needs to be rewritten.
There is now rudimentary support for \includegraphics. Three file types
will be inserted into the RTF file without needing conversion: .pict, .jpeg, and
.png files. EPS files are converted to PNG using convert from the ImageMagick
package.
If there is no \pagestyle command, the RTF output is generated as with plain
pagestyle, i.e. each page get's its page number centered at the bottom.
You must turn this off with the \pagestyle{empty} command in the LaTeX file if you don't want pagenumbers. The headings and myheadings styles are silently ignored by now. The twosided option to the \documentstyle or \documentclass produces the corresponding RTF tokens. Note that these features require RTF Version 1.4.
Hyperlatex support is largely broken at the moment, but continues to improve.
Otfried Schwarzkopf has created the "Hyperlatex Markup Language" which is a "little package that allows you to use LaTeX to prepare documents in HTML." It brings an Emacs lisp program with it to convert the Hyperlatex file to HTML. Hyperlatex can be obtained from the CTAN-sites, See Availability. There are two convenient commands that avoid typing: \link and \xlink that generate an "internal" label which then is used in the following \Ref and \Pageref commands.
LaTeX makes it possible to write `\link{anchor}[ltx]{label}', which typesets: `anchor ltx'. LaTeX2RTF does NOT support this aproach since the optional parameter is thrown away right now, See LaTeX2RTF under Development.
Note that you have to update your .cfg files if you are upgrading, since
there are a lot of HTML oriented commands in Hyperlatex that we simply can
`ignore'.
On processing input LaTeX2RTF
first converts the LaTeX
special characters. If
it encounters one of the standard commands it is converted internally. If a
command is not known to LaTeX2RTF
it is first looked up in
direct.cfg and the RTF code specified there is output. If not found there it
is looked up in the section ignore.cfg. This file includes a lot of LaTeX
commands that do not affect the output (cross reference information and the like), or that we are not able or willing to convert to RTF.
You can use ignore.cfg if you get tired of seeing
WARNING: command: `foo' not found - ignoredand you don't need `foo' in your RTF document. It would be nice to send your additions to the LaTeX2RTF mailing list for inclusion in later distributions.
LaTeX2RTF accepts Unix, MS-DOS, and Macintosh line ending codes (\n, \r\n and \r). The files it creates are the line ending for the platform on which LaTeX2RTF was compiled.
The LaTeX
file may have been created with a wide variety of character
sets. If the LaTeX
lacks the \package[codepage]{inputenc}
definition, then you may need to use the command line switch to manually
select the proper code page. See Input Encoding.
On writing output, LaTeX2RTF
generates \n as line ending code.
Your RTF Reader should accept this on any platform. If you ftp your RTF file
from or to MS-DOS platforms the line ending code can be converted to \r\n. As
this should also be legal to any RTF Reader the resulting RTF rendering should
not be affected.
LaTeX2RTF
does not offer a whole lot of flexibility in how files are translated,
but it does offer some. This flexibility resides in four files direct.cfg,
ignore.cfg, fonts.cfg, and language.cfg. These filese
are documented in the next four sections.
The file direct.cfg is used for converting LaTeX
commands by simple text
replacement. The format consists of lines with a LaTeX
command with backslash
followed by comma. The rest of the line until a . character will be written
to the RTF file when the command is found in the LaTeX
file. Lines
starting with a # character are ignored. After the . everything is ignored
to end of line. To select a specific font use *fontname*, where
fontname be defined in fonts.cfg.
To write the * character use **.
\alpha,{\f*Symbol* a}. #alpha in the Symbol Font
\copyright,\'a9.
The file ignore.cfg is used for defining how to ignore specific commands. This
file is used for recognition of LaTeX
variables, user defined variables, and
some simple commands. All variables are ignored but the converter must know the
names to correctly ignore assignments to variables. Lines in this file consist
of a variable name with backslash, followed by comma and the type of the
variable followed by .. Possible types are
NUMBER
MEASURE
OTHER
= and from there to next
space. e.g., \setbox\bak=\hbox
COMMAND
\ and from there to the occurence of anything
but a letter e.g., \newbox\bak
SINGLE
\noindent
PARAMETER
\foo{bar}
PACKAGE
ENVCMD
eg. `ENVCMD,environ.' Therefore \begin{environ} text
\end{environ}' as `text'.
ENVIRONMENT
\begin{ifhtml} text \end{ifhtml} ignores `text'.
The types are in upper case exactly as above. Do not use spaces. Lines starting
with a # character are ignored. After the . everything is ignored to end of
line. Example:
\pagelength,MEASURE.
The file fonts.cfg contains the font name mapping. For example, this
file determines what font is used to represent \rm characters in the RTF file.
A line consists of a font name in LaTeX
followed by comma and a font
name in RTF. The end is marked by a .. No spaces are allowed. The
LaTeX
font will be converted to the RTF font when it is found
in the LaTeX
file. If multiple translations for the same LaTeX
font are
specified, only the first is used. All fonts in a LaTeX
file that are
not in this file will be mapped to the default font. All RTF fonts
listed in this file will be in every RTF file header whether used or
not. Lines starting with a # character are ignored. After the
. everything is ignored to end of line.
To add a RTF font not
used as substitute for a LaTeX
font -- for example a Symbol font used
in direct.cfg -- use a dummy LaTeX
name like in the following
Dummy3,MathematicalSymbols.
Make sure you use the correct font name. Take care of spaces in font names. The
default fonts are named Roman \rm, Slanted \sl, Sans Serif
\sf, Typewriter \tt, or Calligraphic \cal.
The file(s) language.cfg control the translation of LaTeX
's "hardcoded"
sectioning names.
The standard LaTeX
styles have some fixed Title names like `Part', `Reference'
or `Bibliography' that appeared in English or German in the output with the
original versions of LaTeX2RTF
.
It is unlikely that you will need to create a new language.cfg file.
However, just look at one of the existing files and follow the pattern.
The format is really simple.
As stated in the Debugging section, LaTeX2RTF
provides a means to control
the amount of debugging information through the -d# switch.
By using a debugging level of 4, you can get a pretty good idea of what
LaTeX
command caused the problem and what line that command might be
found on.
Fatal error messages
Error messages
Warning messages
ignore.cfg
file.
Error and Warning messages should follow the GNU Coding standards, i.e. they
have the format
inputfile':line: Error|Warning: message
You can also control the level of debugging output by inserting \verbositylevel{#}
in the LaTeX
file. This is very handy if you have a large LaTeX
file
that is failing in only a small section. For example,
problem free latex file ....
\verbositylevel{5}
problematic code
\verbositylevel{0}
will cause a huge amount of debugging information to be emitted
for the problematic code.
Error reporting and logging still has many inconsistencies, but it gets better with each release. Don't try to make any sense in debugging levels above 4, these are for my own delight only and can change significantly between versions.
The inputfile may be incorrectly identified if it is incorporated
through \input or \include. The line may be also be wrong
at times. See Known Bugs.
\documentstyle
\field commands are not implemented.
\markboth and \markright.
\tableofcontents there would be two approaches: Transfer
sectioning information, title text and then produce page numbers by the rtf-
reader. Scan and label all of the sectioning commands while reading and
then construct the sectioning information using these labels. Needs two
passes on LaTeX
input.
Missing or buggy command line options.
-V
-d
-?
-q
-q (quiet) option, to suppress Warning Messages. By
now this can be achieved by the -d0 option.
-rmajor.minor
--long_names
\input).
Report bugs to to the bug tracking system at SourceForge. Only report bugs for the latest version of LaTeX2RTF that is available. Please provide the following information and observe the following guidelines when reporting a bug in the program:
-V option to LaTeX2RTF
.
Scott's ToDo list
Georg's todo list