fig2vect

The fig2vect program converts Fig drawings to MetaPost, EPS, PDF, TeX and SVG files.
It is intended for LaTeX users which want to use their Fig drawings with both LaTeX and pdfLaTeX.

Here is a list of differences between fig2vect and transfig:

• fig2vect is specialized for use with LaTeX/pdfLaTeX.
  The transfig package provides a large number of output drivers for general purpose usage. The fig2vect program is intended for those who want to use Fig drawings with LaTeX/pdfLaTeX. At this time there are only five output drivers (MetaPost, EPS, PDF, TeX and SVG).
• X-Spline to Bezier spline conversion in fig2vect.
  In fig2dev X-splines are converted to a polyline when reading the Fig file. The fig2vect program calculates Bezier splines. Each X-spline segment can be drawn by using a number of Bezier spline segments.
• Fill patterns are implemented as vector graphics.
  Fig2vect uses vector drawing operations to fill objects, not bitmap images.
• Support for UTF-8 encoded text.
  As more and more Linux/Unix installations use a ``lang.UTF-8'' languages setting, support for UTF-8 encoded text becomes more and more important.
  Fig2vect can handle Fig files containing UTF-8 encoded texts.
  See the UTF-8 support FAQ section for details
• Both fig2vect and transfig are designed to be extensible.
  Both fig2vect and transfig allow programmers to add new output drivers. In transfig an output driver is a set of functions obtaining a pointer to a drawing primitive. For each drawing primitive one function from the set is invoked, the function has only access to the drawing primitive's data.
  In fig2vect an output driver is one function obtaining a pointer to the entire drawing and the conversion options. The driver function can traverse the list of drawing primitives as often as necessary. This can be used to collect style information in a first pass and to write output using the style information in a second pass.
  
• Different configuration mechanisms.
  The fig2dev program is configured via command line options only.
  The fig2vect program uses a configuration file (${prefix}/etc/fig2vect/fig2vect.cfg or a users own copy) to define different output configurations for different purposes. Command line options are used to choose an output configuration and/or to overwrite settings from the output configuration.
• No rotation of embedded images
  XFig and jFig behave differently if the first point of an image polygon is not the upper left point. XFig rotates the image by multiples of 90 degree depending on which point was specified first. jFig shows the image without any rotation.
  The fig2vect program does not rotate the image. If you really want a rotated image, use a graphics program to do the rotation and embed the rotated image specifying the upper left point first. This allows correct processing by XFig, jFig, transfig and fig2vect.

Yes, you should. There are no binary installation packages out there.

The following libraries are needed/recommended to install fig2vect:

• zlib (recommended)
  http://www.gzip.org/zlib
• libbzip2 (recommended)
  http://sources.redhat.com/bzip2/
• dklibs (required)
  http://dklibs.sourceforge.net
• libpng (recommended)
  http://www.libpng.org/pub/png/libpng.html
• jpeglib (recommended)
  ftp://ftp.uu.net/graphics/jpeg/jpegsrc.v6b.tar.gz
• NetPBM (recommended)
  ftp://ftp.metalab.unc.edu/pub/Linux/apps/graphics/convert/netpbm-10.11.10.tgz

The fig2vect project is hosted on SourceForge, you can get the sources from the project's download area via http://sourceforge.net/projects/fig2vect.

• Unpack the archive fig2vect-*.tar.gz, i.e.

  gzip -dc fig2vect-0.9.8.tar.gz | tar xf -
  

• Change into the fig2vect directory

  cd fig2vect
  

• Run the configure script

  ./configure
  

• Build the software

  make
  

• Install the software

  make install
  

For the wide majority of users I recommend to obtain the setup executable from http://sourceforge.net/projects/dklibs.
The dklibs-win32 package contains executable setup to install the dklibs library set and some of the applications using it (fig2vect is one of them). There are two files for each version: The dklibs-win32-*-user.exe installs binaries, libraries, header files and documentation. The dklibs-win32-*-dev.exe installs all that, the sources for the projects and the sources for the libraries needed for the projects. The *dev.exe is significantly larger than the *user.exe.

Installing from source is recommended for experienced programmers only. You need a good working knowledge of your development system, your C compiler and linker's options and makefiles.
You should also read http://dklibs.sourceforge.net/inst_w32.pdf. This document gives some hints about installing the dklibs library set and applications using it from source.

A file ${prefix}/etc/fig2vect/fig2vect.cfg is created when the fig2vect package is installed the first time. Additionally a file fig2vect.cfg.sample is installed.
During update installation the fig2vect.cfg file is not changed as it might be modified. The current version of the fig2vect.cfg file from the distribution is installed as fig2vect.cfg.sample.

The configuration file is prepared to have three section:

• Section 1: the pre-defined configurations from the fig2vect distribution.
• Section 2: planned for site-wide configurations.
• Section 3: planned for user-specific configurations in the users private copy of the configuration file.

• Replace section 1 of the existing configuration file ${prefix}/etc/fig2vect/fig2vect.cfg by the section 1 of ${prefix}/etc/fig2vect/fig2vect.cfg.sample. This guarantees that all predefined configurations are up to date.
• Inform all users that ${prefix}/etc/fig2vect/fig2vect.cfg was updated. Users can now update their private copies and replace the sections 1 and 2 of ${HOME}/.defaults/fig2vect.cfg by the contents of the ${prefix}/etc/fig2vect/fig2vect.cfg file.

Use
  fig2vect options input-file output-file
  fig2vect options directory
to run the program. When using the MetaPost driver you can also use
  fig2vect options input-file
  fig2vect options
If there is no file name for the output file specified output goes to standard input. If the file name for the input file is omitted the program processes standard input.

Yes, it's possible if support for zlib and/or the bzip2 library was enabled during build. Simply specify a file name "*.fig.gz" or "*.fig.bz2" for input files or "*.mp.gz"..."*.eps.bz2" for output files.
To find out whether or not zlib/bzip2 support is enabled run

fig2vect --version

zlib         General compression library
libbz2       Library for compressing/decompressing data in the bzip2 format

The fig2vect program uses a configuration file fig2vect.cfg. This file consists of serveral sections, each section describes one configuration. A section consists of a configuration name and several key/value pairs. Keys can consist of multi-part strings. The keys are also referred to as options.

The name for a configuration consists of the base driver name ("mp", "eps", "tex" or "svg"), a dot (".") and a unique name of your choice containing numbers, digits, underscore and minus.
In the configuration file the name is surrounded by "[" and "]".

If a user chooses a configuration "mp.my-conf" by specifying

fig2vect -l mp.my-conf ...

By default the configuration file "${prefix}/etc/fig2vect/fig2vect.cfg" is used. If a file "$HOME/.defaults/fig2vect.cfg" is found, this file is used instead of the default configuration file.

The fig2vect program inspects all drawing elements except the texts and calculates the elements dimensions. The element dimensions of all elements are combined to calculate the dimensions for the entire drawing. The output file dimensions are choosen to match the borders of the outermost elements.
Texts are skipped for two reasons:

• Width and height of a text depend on the output driver used to print that text. Image dimension calucalation is done before the output driver is run.
• The methods for calculating text width and height would be driver-specific. I.e. for the SVG output driver we would need to be able to read and process SVG fonts, for the EPS output driver we would need to read *.pfb and *.afm files, for the MetaPost driver we would need to read *.dvi and *.tfm files...

If your Fig file contains text on the outermost left/right/top/bottom position you can use either

• a white filled rectangle or
• a background rectangle (see below)

Use either

• a white filled rectangle or
• a background rectangle (see below)

A background rectangle is an additional rectangle you add to the Fig file. This rectangle is not drawn to the output file, it is only used to calculate the output file dimensions. A background rectangle must have the following features assigned:

• object type 2 (polygon), subtype 2 (rectangle/box),
• layer (depth) 999,
• stroke color -1 (default),
• line width 0,
• area fill -1 (no fill), fill color does not matter.

remove background rectangle = yes

Produce a PNG image first (see below) and convert the PNG image to the bitmap type you need, i.e. using NetPBM, ImageMagick, Xnview (the nconvert program) or Gimp.

There are four choices:

• Convert the image to SVG and convert the SVG file to PNG using either the Squiggle (the Batik SVG viewer) or the Batik rasterizer.
  Using the Batik rasterizer is the recommended method if your *.fig file does not contain special text.
  All ranges of the output file not covered by graphics elements will be transparent, so this method is well-suited to create graphics for web sites...
  This method will only work if the Fig file does not contain special text.

  fig2vect -lsvg image.fig image.svg
  java -jar /.../batik-rasterizer.jar image.svg
  

  The Batik software is available from xml.apache.org. A java program is available in the Java Runtime Environment (JRE) from java.sun.com.
• Convert the image into a standalone PDF and convert the PDF file to PNG using pdftoppm (form the xpdf package) and pnmtopng (from the NetPBM package).
  This is the recommended method if your *.fig file contains special text.
  

  fig2vect -l pdf.tex image.fig imagei.pdf
  fig2vect -l tex.full -i imagei image.fig image.tex
  pdflatex image && pdflatex image
  pdftoppm -t1lib yes -freetype yes -aa yes -aaVector yes image.pdf image
  pnmtopng < image-000001.ppm > image.png
  

• Convert the image into a standalone PDF file and use Ghostscript to convert the PDF to PNG.
  

  fig2vect -lpdf.tex image.fig imagei.pdf
  fig2vect -ltex.full -iimagei image.fig image.tex
  pdflatex image && pdflatex image && pdflatex image
  gs ... -sOutputFile=image.png image.pdf
  

  Use the following options for Ghostscript:

  -dBATCH	run in batch mode
  -dNOPAUSE	do not pause after each page
  -dNOCACHE	do not cache rendered characters
  -sDEVICE=png16m	choose output file type PNG
  -dGraphicsAlphaBits=4	Anti-Aliasing for graphics uses 4 bits (1, 2 or 4 bits, or omit this option)
  -dTextAlphaBits=4	Anti-Aliasing for text uses 4 bits (1, 2 or 4 bits, or omit this option)
  -dNOCACHE	do not cache rendered characters
  -r300	use 300 dpi (you may choose another resolution)

  You might want to try different combinations of graphics alpha bits and text alpha bits.
• Convert the image into a standalone EPS file and use Ghostscript to convert the EPS to PNG.

  fig2vect -leps image.fig image.eps
  gs ... -sOutputFile=image.png image.eps
  

  Use the following options for Ghostscript:

  -dBATCH	run in batch mode
  -dNOPAUSE	do not pause after each page
  -dEPSCrop	choose output image size to fit EPS BoundingBox
  -dNOCACHE	do not cache rendered characters
  -sDEVICE=png16m	choose output file type PNG
  -dGraphicsAlphaBits=4	Anti-Aliasing for graphics uses 4 bits (1, 2 or 4 bits, or omit this option)
  -dTextAlphaBits=4	Anti-Aliasing for text uses 4 bits (1, 2 or 4 bits, or omit this option)
  -dNOCACHE	do not cache rendered characters
  -r300	use 300 dpi (you may choose another resolution)

  You might want to try different combinations of graphics alpha bits and text alpha bits.

Use the optipng program, this program can be found on SourceForge at http://sourceforge.net/projects/optipng.
For normal compression run:

optipng image.png

optipng -q image.png

optipng -i1 image.png

optipng -o7 image.png

optipng -zc1-9 -zm1-9 -zs0-3 -f0-5 image.png

The MetaPost program can handle LaTeX special text, so you can use special text in your Fig files when using this driver. The output created by MetaPost can be used with both LaTeX and pdfLateX. As I nearly always use the MetaPost driver, this driver is better tested than the other drivers.
The MetaPost language does not provide a possibility to include any images, so you can not use included images when using this driver.

• Choose the correct TeX command
  If your document is processed by LaTeX or pdfLaTeX the configuration file should contain

  tex command = latex
  

  Only if the document is processed by plain TeX or pdfTeX you should use

  tex command = tex
  

  Remember to either set the TEX environment output variable to "latex" before running mpost or to invoke mpost using

  mpost -tex=latex
  

  or

  mpost --tex=latex
  

  to process the *.mp file. The number of minuses may differ depending on your LaTeX distribution.
• Use the same fonts in the image as in the document text
  Using the same fonts in images and in the document text looks more professional than using different fonts. Create a minimal preamble for font setup (strip down the document preamble) and use

  preamble file = ...
  

  to refer to this minimal preamble.
  If -- for example -- your document has a preamble like

  \documentclass[12pt]{book}
  \usepackage{ifpdf}
  \usepackage[latin1]{inputenc}
  \usepackage[T1]{fontenc}
  \usepackage{textcomp}
  \usepackage[intlimits]{amsmath}
  \usepackage{amssymb}
  \usepackage{mathptmx}
  \usepackage[scaled=.92]{helvet}
  \usepackage{courier}
  \usepackage{array}
  \usepackage{enumerate}
  \usepackage{longtable}
  \usepackage{latexsym}
  \usepackage{varioref}
  \usepackage{makeidx}
  \usepackage{listings}
  \lstset{numbers=left, numberstyle=\tiny, numbersep=5pt, basicstyle=\small}
  \usepackage{color}
  \ifpdf
  \usepackage[activate=normal]{pdfcprot}
  \usepackage[pdftex]{graphicx}
  \usepackage{epstopdf}
  \pdfcompresslevel=9
  \usepackage[
      pdftex,
      a4paper=true,
      pdftitle={The fig2vect manual},
      pdfsubject={fig2vect},
      pdfauthor={Dipl.-Ing. D. Krause},
      colorlinks=true,
      bookmarks,
      bookmarksnumbered,
      linkcolor=linkgreen,
      pdfpagemode=None,
      pdfstartview=FitH
  ]{hyperref}
  \else
  \usepackage[dvips]{graphicx}
  \DeclareGraphicsRule{.png}{eps}{.bb}{`bmeps #1}
  \usepackage[dvips]{hyperref}
  \fi
  \definecolor{linkgreen}{rgb}{0,0.5,0}
  \def\theyear{2002}
  \author{Dipl.-Ing.~D.~Krause}
  \title{The fig2vect manual}
  \include{cover}
  \DeclareMathOperator{\element}{\in}
  \DeclareMathOperator{\notelement}{!\in}
  \newcommand{\ora}[1]{\overrightarrow{#1}}
  \DeclareMathOperator{\grad}{grad}
  \DeclareMathOperator{\divergence}{div}
  \DeclareMathOperator{\rot}{rot}
  \newcommand{\atpos}[2]{{\left.#1\right|}_{#2}}
  \newcommand{\inrange}[3]{{\left.#1\right|}_{#2}^{#3}}
  \newcommand{\intd}[0]{\mathrm{d}}
  \DeclareMathOperator{\arcsinh}{arcsinh}
  \DeclareMathOperator{\card}{card}
  \DeclareMathOperator{\Res}{Res}
  \newcommand{\imagelw}[2]{
      \begin{figure}[ht]
      {\centering
      \includegraphics[width=\linewidth]{#1.mps}
      \caption{#2}
      \label{fig:#1}
      }
      \end{figure}
  }
  \newcommand{\ximagelw}[3]{
      \begin{figure}[ht]
      {\centering
      \includegraphics[width=#1\linewidth]{#2.mps}
      \caption{#3}
      \label{fig:#2}
      }
      \end{figure}
  }
  \DeclareMathOperator{\oplaplace}{\triangle}
  \DeclareMathOperator{\opLaplace}{{\mathfrak L}}
  \DeclareMathOperator{\iopLaplace}{{\opLaplace}^{-1}}
  \DeclareMathOperator{\definedas}{\overset{=}{\text{\tiny{Def}}}}
  \DeclareMathOperator{\lcorrelates}{\laplace}
  \DeclareMathOperator{\ilcorrelates}{\Laplace}
  \DeclareMathOperator{\mustbe}{\overset{\text{\tiny !}}{=}}
  \makeindex
  

  create a stripped down file

  \documentclass[12pt]{book}
  \usepackage[latin1]{inputenc}
  \usepackage[T1]{fontenc}
  \usepackage{textcomp}
  \usepackage[intlimits]{amsmath}
  \usepackage{amssymb}
  \usepackage{mathptmx}
  \usepackage[scaled=.92]{helvet}
  \usepackage{courier}
  

  and save it as "/home/jim/mypreamble.tex".
  In the configuration file use

  preamble file = /home/jim/mypreamble.tex
  

• Change fonts automatically if necessary
  Use

  normal text	= handling:tex,font:similar,size:fig
  special text	= handling:tex,font:tex,size:tex,mbox
  

  to allow automatic font changing for normal text. If -- for example -- a Fig file uses "Palatino-Bold" for a text and fig2vect was configured to use the preamble file "/home/jim/mypreamble.tex" the text is written in roman font configured by that preamble, fontweight bold, upright style (not italic).
• No embedded fonts in the *.mps image
  When translating the LaTeX source to PS/PDF the document fonts are embedded into the output file by dvips or pdflatex once for the document. Our images should not attempt to embed the same fonts again, they should re-use the fonts already embedded.
  Use

  embed fonts = no
  

• Use MetaPost arrowheads
  Use

  use metapost arrowheads = yes
  

  to allow usage of MetaPost arrowheads. MetaPost produces curved arrowheads at the end of curved lines.

• Avoid text handling by TeX/LaTeX
  When creating the Fig file avoid use of special text.
  In the configuration file use

  normal text	= handling:none,font:fig,size:fig
  

  to make sure that only the 35 well-known PostScript fonts are used.
• Embed the fonts
  Use

  embed fonts = yes
  

  to make sure the mpost output file contains all the fonts needed to show the texts.

When using this driver you can use included images (PNG, JPEG, NetPBM and EPS) in the Fig file.
You can not use LaTeX special text in conjunction with this driver. In theory we could do the same things as MetaPost: write special text to a LaTeX file, process the file using LaTeX, read the DVI file and convert to EPS. Practically this is a lot of work, requiring a good knowledge of DVI file structure, PS font encodings, font metrics... and a team of volunteers contributing code.

Use

ps level = 1

Use

dsc comments = yes

dsc comments = no

This operator is only allowed in PS files, not in EPS files. PS images included into documents should not contain the showpage operator, use

ps showpage = no

ps showpage = yes

This operator is only allowed in PS files, not in EPS files. It can be used to restrict media size to match the graphics size. Turn it of for EPS files and for those PS files embedded into documents using

ps setpagedevice = no

ps setpagedevice = yes

• remove bitmap border
  The image box for an inluced image is specified as a polygon in the Fig file. There is no GUI-based mechanism to change the line width of this polygon. To draw the included image without stroking the image border use

  remove bitmap border = yes
  

  To draw the image box border use

  remove bitmap border = no
  

• keep bitmap aspect ratio
  In many cases the width/height relation of the image will differ from the relation of the image box.
  Use

  keep bitmap aspect ratio = yes
  

  to keep the original ratio leaving some parts of the image box unused.
  Using

  keep bitmap aspect ratio = no
  

  you can fill the entire polygon using different scale factors for width and height.
• fill bitmap background
  There is no GUI-based way to change the background color / fill pattern of a polygon used for image inclusion.
  Use

  fill bitmap background = no
  

  to ignore any background filling information from the Fig file or

  fill bitmap background = yes
  

  to fill the polygon before adding the bitmap.
• ps run-length encoding
  allows or denies usage of run-length compression.
• separated rgb channels
  PS level 2 and above provides two choices for structuring color image data:
  • Combined data source
    For each pixel the data stream contains the red-, green- and blue value.

    separated rgb channels = no
    

  • Separated data sources
    The data stream first contains the red-values for all the pixels followed by all the green-values, followed again by all the blue-values.

    separated rgb channels = yes
    

  Separated data sources allow to run-length compress colored areas and runs. In a combined data source run-length compression is only possible for gray areas and runs (red=green=blue). Separated data sources will produce smaller EPS files when containing colores areas.
  Unfortunately the PostScript code must use 3 strings of width*height bytes to uncompress image data from separated sources while only one string of 3*width bytes is necessary to uncompress data from a combined data source. Separated data sources are not recommended to produce print jobs for printers with minimal memory equipment.

• bitmap image dictionary
  Use

  bitmap image dictionary = yes
  

  to encapsulate the drawing code in a

  ... dict begin
    % drawing code here
  end
  

  structure. All the strings required to uncompress bitmap image data are now defined in this new dictionary. After finishing all drawing operations the dictionary is removed from the dict stack. As there is no reference to the dictionary it is unusable now, the garbage collection is allowed to remove the dictionary. The only references to the strings are contained in the dictionary, as no access to the dictionary is possible there is also no access to the strings possible. So the strings can be removed by the garbage collection too.
  Note: Do not use this option if the Fig file includes EPS images! The dictionary size is calculated to match the number of strings needed for PNG/JPEG/NetPBM images. There is no place in the dictionary for additional definitions from included EPS files.
• force garbage collection
  Use

  force garbage collection = yes
  

  to add a line

  1 vmreclaim
  

  to the end of EPS output. This tells the PostScript interpreter to attempt to run the garbage collection after finishing the drawing operations.

This driver produces PDF images which can be included in a LaTeX source processed with pdflatex.
If there are alpha channels in embedded PNG images, the alpha channel is converted into PDF transparency data. This allows to place partially transparent images over other content.
The PDF driver has only minimal text processing capabilities (only non-special left-aligned text, all fonts are converted into the 14 PDF base fonts). It is recommended to use the PDF driver in conjunction with the TeX driver, the TeX driver should be used for both special and non-special text.

Use the configuration entry

plain text streams = yes

-o plain.text.streams=yes

Use the configuration entry

large pattern cell = yes

-o large.pattern.cell=yes

Use the configuration entry

interpolate images = yes

-o interpolate.images=yes

Yes, you can. But if you use PDFs containing alpha channels with pdfLaTeX you need a special line

\pdfpageattr {/Group << /S /Transparency /I true /CS /DeviceRGB>>}

The message

At least one alpha channel was converted to PDF.
  When referencing PDFs containing alpha data from your *.tex sources, you should use
  \ifpdf\pdfpageattr{/Group <</S /Transparency /I true /CS /DeviceRGB>>}\fi
  in the preamble.

Ein Alpha-Kanal wurde in eine PDF-Ausgabe übertragen.
  Um PDFs mit Alpha-Kanal mit pdfLaTeX zu verwenden, sollten Sie
  \ifpdf\pdfpageattr{/Group <>}\fi
  in der Präambel der LaTeX-Quelle verwenden.

bmeps -c -A

Alternatively you can create preferences file (Unix/Linux) or registry (Windows) entries manually:

• On Unix/Linux systems edit $HOME/.defaults/all and add a section

  [fig2vect]
  /pdf/suppress-alpha-info=true
  

• On Windows open the registry key HKCU/Software/DkApp/fig2vect (create the key if it does not exist) and add an entry

  REGEDIT4
  
  [HKEY_CURRENT_USER\Software\DkApp\fig2vect]
  "/pdf/suppress-alpha-info"="true"
  

• Compression
  Get the latest multivalent...jar file from the MultiValent project, avaible on SourceForge at http://sourceforge.net/projects/multivalent.
  Additionally you need a Java Run-Time Environment, I suggest to use the one from Sun available at http://java.sun.com.
  To compress a PDF file, run:

  java -Xms64m -Xmx1024m -cp ...path.../Multivalent20060102.jar tool.pdf.Compress -compatible file.pdf
  

  This creates file-o.pdf.
   
  
• Fast web view
  Use GhostScript (available at http://pages.cs.wisc.edu/~ghost/) to create a fast web view:

  pdfopt input.pdf output.pdf
  

This driver supports other drivers which can not handle special text theirselves (i.e. the EPS or PDF driver).

It depends how you want to handle normal (non-special) text. There are two choices:

• Normal text is handled by the EPS driver and ignored by the TeX driver. The configuration entries should look like this:

  [eps.tex]
    normal text = handling:none,font:fig,size:fig
    skip all texts = yes
  [tex]
    normal text = handling:none
  

• Normal text is ignored by the EPS driver and handled by the TeX driver. The configuration entries should look like this:

  [eps.tex]
    normal text    = handling:tex,font:similar,size:fig,mbox
    skip all texts = yes
  [tex]
    normal text    = handling:tex,font:similar,size:fig,mbox
  

Run

fig2vect -leps.tex myfile.fig myfile.eps
fig2vect -ltex myfile.fig myfile.tex

\begin{figure}
{\centering
\input{myfile.tex}
\caption{My image caption}
\label{fig:my-image-label}
}
\end{figure}

When using this driver you can use included images of any type in the Fig file as the driver does not need to convert included images, it simply writes a reference. But you must make sure your SVG viewer can render the included image.
Special comments can be used to insert ECMA-Script code into the SVG output and to attach event handlers and links to objects.
Using the Adobe SVG viewer you can embed SVG images into other applications (i.e. Office applications) on Windows systems.
Special text can not be used with this driver.

• Install Adobe SVG viewer.
  You can obtain it from http://www.adobe.com/support/downloads/main.html.
  
• In your application choose to insert an object of type "SVG Document".
• In the context menu change the properties of the object. Change the value of the "SRC" property to the URL of the SVG file, see the screenshot below.
  
• Resize the object if necessary.

You should try to set the environment variable LC_NUMERIC like this

LC_NUMERIC=C
export LC_NUMERIC

Use the

svg version = 1.0

The driver can produce both standalone SVG images and SVG fragments for embedding in XML documents.
To produce standalone images use

embedded svg fragment = no

The SVG tag contains width/height information and a viewport specification. The width and height attributes specify the physical size of the image, the viewport attribute specifies the logical coordinates assigned to that area.
The

wh specification = inches

There are three ways to specify an SVG object's style:

1. Multiple simple object property attributes.
   Each style property is defined via a separated XML attribute, i.e.:

   <rect fill="yellow" stroke="black" stroke-width="0.9" .../>
   

2. A single style attribute.
   A style attribute is used:

   <rect style="fill: yellow; stroke: black; stroke-width: 0.9;" .../>
   

3. A class attribute.
   The class attribute in an object definition refers to a style in the CSS style section.

   <style type="text/css"><![CDATA[
   .c1 {
     fill: yellow;
     stroke: black;
     stroke-width: 0.9;
   }
   ]]></style>
   <rect class="c1" .../>
   

• it finds instructions to use JavaScript (ECMA-script) files (see below),
• event handlers are attached to objects (see below) or
• the

  prepare for modifications = yes
  

  configuration setting is used.

use css = yes

To add ECMA-script code into a <script> tag, write it into a file, i.e. "mycode.js". Use either the configuration setting

js library = mycode.js

## svg: js library = mycode.js

#FIG 3.2
Portrait
Center
Metric
A6      
100.00
Single
-2
1200 2

#FIG 3.2
Portrait
Center
Metric
A6      
100.00
Single
-2
## svg: js library = mycode.js
1200 2

Comments (and special comments) related to an object must be placed immediately before the line starting the object. Event handlers can be defined in special comments, i.e. by changing

#FIG 3.2
Portrait
Center
Metric
A6      
100.00
Single
-2
## svg: js library = mycode.js
1200 2
1 1 0 1 0 7 50 0 -1 0.000 1 0.0000 990 900 450 180 990 900 1440 720

#FIG 3.2
Portrait
Center
Metric
A6      
100.00
Single
-2
## svg: js library = mycode.js
1200 2
## svg: onmouseover = fill_yellow(evt)
## svg: onmouseout = fill_white(evt)
## svg: id = e1
1 1 0 1 0 7 50 0 -1 0.000 1 0.0000 990 900 450 180 990 900 1440 720

<ellipse transform="translate(49.2 22.1)"
rx="27" ry="10.8"
fill="none" stroke="#000000" stroke-width="0.9"
/>

<ellipse transform="translate(49.2 22.1)"
rx="27" ry="10.8"
fill="none" stroke="#000000" stroke-width="0.9"
id="e1"
onmouseover="fill_yellow(evt)"
onmouseout="fill_white(evt)"
/>

Event handlers for the entire drawing can be specified in special comments in the document level, i.e. change

#FIG 3.2
Portrait
Center
Metric
A6      
100.00
Single
-2
## svg: js library = mycode.js
1200 2

#FIG 3.2
Portrait
Center
Metric
A6      
100.00
Single
-2
## svg: js library = mycode.js
## svg: onload = handler_onload(evt)
1200 2

<svg width="255px" height="298px" viewBox="0 0 255 298"
xmlns="http://www.w3.org/2000/svg"
xmlns:xlink="http://www.w3.org/1999/xlink"
>

<svg width="255px" height="298px" viewBox="0 0 255 298"
xmlns="http://www.w3.org/2000/svg"
xmlns:xlink="http://www.w3.org/1999/xlink"
onload="handler_onload(evt)"
>

Obtain the Ghostscript fonts. Convert them to SVG fonts using the FontForge software (see below). Use the

gs svg-font directory = ...url...

Obtain the Ghostscript fonts and the sources for these fonts. Unpack both archives into one temporary directory. You should have 35 *.pfb files and 33 *.sfd files there.
Open each of the *.pfb files in FontForge. Save the font using the same name but suffix ".svg".
Optionally: Open each of the *.sfd files in FontForge. Save the fonts as *.svg:

NimbusRomanNo9L-Regu.sfd       n021003l.svg
NimbusRomNo9L-ReguItal.sfd     n021023l.svg
NimbusRomNo9L-Medi.sfd         n021004l.svg
NimbusRomNo9L-MediItal.sfd     n021024l.svg
URWGothicL-Book.sfd            a010013l.svg
URWGothicL-BookObli.sfd        a010033l.svg
URWGothicL-Demi.sfd            a010015l.svg
URWGothicL-DemiObli.sfd        a010035l.svg
URWBookmanL-Ligh.sfd           b018012l.svg
URWBookmanL-LighItal.sfd       b018032l.svg
URWBookmanL-DemiBold.sfd       b018015l.svg
URWBookmanL-DemiBoldItal.sfd   b018035l.svg
NimbusMonL-Regu.sfd            n022003l.svg
NimbusMonL-ReguObli.sfd        n022023l.svg
NimbusMonL-Bold.sfd            n022004l.svg
NimbusMonL-BoldObli.sfd        n022024l.svg
NimbusSansL-Regu.sfd           n019003l.svg
NimbusSanL-ReguItal.sfd        n019023l.svg
NimbusSansL-Bold.sfd           n019004l.svg
NimbusSanL-BoldItal.sfd        n019024l.svg
NimbusSanL-ReguCond.sfd        n019043l.svg
NimbusSanL-ReguCondItal.sfd    n019063l.svg
NimbusSanL-BoldCond.sfd        n019044l.svg
NimbusSanL-BoldCondItal.sfd    n019064l.svg
CenturySchL-Roma.sfd           c059013l.svg
CenturySchL-Ital.sfd           c059033l.svg
CenturySchL-Bold.sfd           c059016l.svg
CenturySchL-BoldItal.sfd       c059036l.svg
URWPalladioL-Roma.sfd          p052003l.svg
URWPalladioL-Ital.sfd          p052023l.svg
URWPalladioL-Bold.sfd          p052004l.svg
URWPalladioL-BoldItal.sfd      p052024l.svg
s050000l.pfb                   s050000l.svg
URWChanceryL-MediItal.sfd      z003034l.svg
d050000l.pfb                   d050000l.svg

[svg.gs]
gs svg-font directory = file:///usr/local/share/gs-fonts/svg

[svg.gs-web]
gs svg-font directory = http://www.my-org.com/gs-fonts/svg

Note: The Ghostscript fonts are licensed under the terms of the GNU General Public License. Please read the file COPYING from the Ghostscript fonts distribution for details.
If you make GPL-licensed software (i.e. the *.svg font files) accessable to other users - i.e. by placing them in a shared directory in a LAN or on a multi-user computer system or on web/ftp servers - you are responsible for making all the sources for that software available too. Access to the sources should not be more complicated than access to the software.
To fullfill the requirements of the GPL I suggest to place the archives containing Ghostscript fonts and the Ghostscript fonts sources used to build the *.svg font files in the same directories as the *.svg font files.

UNICODE is a character set using 32 bits. Documents using this character set can use characters/glyphs from different languages.
UTF-8 is an encoding for UNICODE characters using 1 byte for the glyphs used most often and up to 6 bytes (48 bits) for the not-so-popular glyphs.

The same glyph is encoded differently when using UTF-8 and traditional encodings. An application showing text must know whether or not the text is UTF-8 encoded or not.

The Fig file format does not allow to provide information whether or not UTF-8 is used.
So there are two ways to tell applications processing Fig files which encoding is used:

• command line options and
• special comments in the Fig files. These comments are processed by programs like fig2vect and ignored by other programs.

Check the LANG environment variable. If it ends on ``.UTF-8'' (not case-sensitive) -- like ``de_DE.UTF-8'' your system uses UTF-8 encoding.
If it does not end on ``.UTF-8'' -- i.e. ``de_DE'' -- your system does not use UTF-8 encoding.

Handling of native characters (i.e. german umlauts) depends on the systems LANG setting.

• If UTF-8 is turned off and you type the german umlaut ä on the keyboard, one byte -- the ANSI code for ä -- is sent from the X11 system to the application.
• If UTF-8 is turned on and you type ä two bytes -- the UTF-8 encoded ä -- is sent to the application.
  An UTF-8 aware application will inspect the two bytes, find the ä and save the ä correctly in output files (using the encoding required by the file format specification). To render texts on screen the ä is sent to X11 UTF-8 encoded.
  If the application is not UTF-8 aware it simply stores the two bytes as two characters. In most cases text is shown fine on screen (if X11 is set up to use UTF-8 it will create keyboard events containging UTF-8 encoded data, the routines showing text expect UTF-8 encoded text). When saving data to files, data is stored UTF-8 encoded no matter whether or not the file format specification allows this.

No! All desktop applications must use the same UTF-8 setting. You may change LANG from ``de_DE'' to ``en_US'' or from ``de_DE.UTF-8'' to ``en_US.UTF-8'', but not from ``de_DE'' to ``de_DE.UTF-8''.
If you want to change the UTF-8 usage, log in as root (or Administrator), change the setting globally -- i.e. in /etc/sysconfig/i18n -- and reboot the system.
If the desktop delivers not-UTF-8 encoded data and an application expects UTF-8 encoded data (or vice versa) there is a wide variety of results: unknown keyboard bindings, application crashes...

• jFig -- or one of the libraries jFig uses -- seems to handle UTF-8 encoding perfectly (from my point of view). If LANG is set to ``de_DE.UTF-8'' german umlauts are saved to Fig files using 1 byte. This indicates that the texts were recoded internally before saving to file.
• In XFig handling differs depending on where german umlauts are inserted. When typing text in the drawing area, german umlauts are ignored. In the "Edit" dialog box german umlauts are converted into two malformed characters. Text is saved to output files UTF-8 encoded, two bytes are used for ä.
  Note: The statements above are related to XFig 3.2.4 on Fedora Core 5 using ``de_DE.UTF-8''. The behaviour can differ for other XFig versions, other OS and other localization settings.

Fig2vect cares about UTF-8 encoded text in non-special text only.
In special text you can use LaTeX code directly to produce the glyphs you need.
Handling UTF-8 encoded strings depends on the driver:

• EPS and PDF: The driver can hande UNICODE characters in the range 0x00000000...0x000000FF. Texts are recoded from UTF-8 to ANSI and saved in WinAnsiEncoding in PDF files and in ISOLatin1Encoding in EPS files.
  Characters above 0x000000FF can not be handled.
• TeX and MetaPost: The texts are recoded to UNICODE, the driver uses the dkle_...() functions to find an appropriate LaTeX encoding.
  Note: The encoding tables in dklibs are very incomplete, at this time they contain support for some european languages only.
  If you have created a table for codes not yet in the list, please share it and use the bug track/feature request mechanisms on SourceForge.

The dklibs library set used by fig2vect uses a preference to find a shared directory (shared between all dklibs-based applications). This directory is /usr/local/share on Unix/Linux and c:\programme\krause\share on Windows by default. The ``c:\programme'' part can differ, it depends on your Windows version and language.
This directory contains a sub-directory ``uc2lat-t'' providing tables for translation from UNICODE to LaTeX. There is a nested directory structure containing multiple files. Files are loaded on demand.
The encoding for UNICODE 0x01234567 is found by searching the file 01/23/45.dat file for an 0x67 entry.
There are three types of entries in the file:

• * hex LaTeX
  The LaTeX code can be used both in normal and in math mode.
• t hex LaTeX
  The LaTeX code can be used in normal (non-math) mode only.
• m hex LaTeX
  The LaTeX code can be used in math mode only.

# Raute
*	0x23	\#

# Asterisk
t	0x2A	\textasteriskcentered{}
m	0x2A	*

# Smaller-than
t	0x3C	\textless{}
m	0x3C	<

It depends on your drawing application.

• jFig
  Just change LANG to ``de_DE.UTF-8'', ``en_US.UTF-8'' or ``whatever.UTF-8'' and reboot.
  Modify $sysconfdir/fig2vect/fig2vect.cfg¹, in section 1 in the [*] subsection add

  utf-8 = no
  

  as the default setting for processing your own files.
  To process files made by other people provide the

  -o utf-8=yes
  

  command line option to fig2vect if the files contain UTF-8 encoded texts.
• XFig
  Change LANG to ``de_DE.UTF-8'', ``en_US.UTF-8'' or ``whatever.UTF-8'' and reboot.
  Modify $sysconfdir/fig2vect/fig2vect.cfg¹, in section 1 in the [*] subsection add

  utf-8 = auto
  

  as the default setting (inspect the LANG variable). To process files made by other people or on other computers use the fig2vect command line option

  -o utf-8=no
  

  or

  -o utf-8=yes
  

There are different ways to change the default setting:

• On Linux/Unix systems for all users:
  As root, add a section

  [*/fig2vect]
  /utf-8	=	no
  

  or

  [fig2vect]
  /utf-8	=	yes
  

  to the $sysconfdir/appdefaults¹ file.
• On Linux/Unix systems for the current user:
  Add a section

  [fig2vect]
  /utf-8	=	no
  

  or

  [fig2vect]
  /utf-8	=	yes
  

  to the $HOME/.defaults/all file.
• On Windows systems for all users:
  In the registry key HKLM\Software\DkApp\fig2vect (create the key if necessary) add a string entry, the name is ``all/all:/utf-8'', the value is either ``yes'' or ``no''.
• On windows systems for the current user:
  In the registry key HKCU\Software\DkApp\fig2vect add a string entry, the name is ``all:/utf-8'', the value is either ``yes'' or ``no''.

Specify the command line options

-o utf-8=yes

-o utf-8=no

To process Fig files created by another user or on another computer, open the Fig file in a text editor.
In the document comment section -- between line 8 (transparent color) and line 9 (resolution and coordinates origin) -- place a special comment

## *: utf-8 = yes

## *: utf-8 = no

• Boot you computer or start a desktop session using UTF-8 encoding.
  I.e. set

  LANG="de_DE.UTF-8"
  

  in /etc/sysconfig/i18n and reboot.
• Start xfig, create a new file test.fig (in the utf-8-example subdirectory).
• Add a text. If you attempt to type german umlauts, they are ignored.
• Push the Edit button, in the Edit dialog box complete the text and type the umlauts äöü.
  For each of the umlauts you see two glyphs in the dialogbox.
  After leaving the dialog box the updated text is shown in the drawing.
  Save and close the drawing.
• Open the test.fig file in a text editor. Between line 8 (transparent color) and 9 (resolution, coordinates origin) add a special comment

  ## *: utf-8 = yes
  

  to indicate that texts in this drawing are UTF-8 encoded.
  This allows fig2vect run by other users or on other computers to correctly decode the texts if language settings differ.

  #FIG 3.2
  Landscape
  Center
  Inches
  Letter  
  100.00
  Single
  -2
  ## *: utf-8 = yes
  1200 2
  2 2 0 1 31 31 51 -1 20 0.000 0 0 -1 0 0 5
  	 1200 1200 3375 1200 3375 1725 1200 1725 1200 1200
  4 0 0 50 -1 0 12 0.0000 4 225 1425 1350 1500 Test mit \303\244\303\266\303\274\001
  

• Now run a conversion, i.e.

  fig2vect -lpdf test.fig test.pdf
  

• The result can be viewed, i.e. in xpdf. The umlauts are shown correctly.
  
• If you use jFig instead of XFig, insert

  ## *: utf-8 = no
  

  because jFig already decoded the UTF-8 encoded data from keyboard and saved the text in traditional encoding.

Apply changes to the *.ctr files only.
The *.c files are generated from the *.ctr files automatically by running the tracecc preprocessor. See http://dklibs.sourceforge.net for details, there is a manual tracecc.pdf available.