home

Installation of Latex2html

Last updated May 31, 2010.
By Nasser M. Abbasi

This whole page can be also found in this PDF file

1 Installation on Linux

I used fedora 13. Installation of Latex2html is very easy. Use yum package manager (GUI interface) and select latex2html package. It will install everything needed.

Once latex2html is installed, simply create a file called $HOME/.latex2html-init

This is my copy of my file, which fixes some issues mentioned below.

In addition to the above, I had to edit the file /usr/share/latex2html/styles/more_amsmath.perl for the rendering bug described below.

That is all! now latex2html is ready to be used on linux.

2 Installation on windows with MikTex 2.8

After struggling with installation of Latex2html (l2h) on windows for many days, I have decided to write down the final layout showing graphically what the tree looks like. I think a graphical illustration to describe the installation is easier for me, and I suspect for many others, to see where things are.

The diagram below shows the final installed tree. This tree layout can be used as a guide for anyone who wants to install l2h on their windows system.

click to enlarge

2.1 Installation steps

The following are the main points to watch out for in order to minimize installation problems

Install the software components needed on the same drive, where your data is and where you will be working, because if you install the software on the C: drive, and your data is on another driver, then you will find that things might not work (this was in my case).
Start by downloading all the software components and install them one after the other. These are the components needed and where I downloaded them from
- MikTex: Version 2.8 http://miktex.org/
- Perl: Active Perl 5.1 build 1006. http://www.activestate.com/activeperl/
- netpbm for windows: Version 10.27. http://gnuwin32.sourceforge.net/packages/netpbm.htm
- gsview: version 4.9. http://pages.cs.wisc.edu/~ghost/gsview/get49.htm
- ghostcript. I used AFLP ghostcript. I had problems with other versions for some reasons. May be there was something else going on, but this version works for me, so I am not going to change anything now. http://pages.cs.wisc.edu/~ghost/doc/AFPL/index.htm
- latex2html version 2008 http://www.ctan.org/tex-archive/support/latex2html/
Now install MikTex, perl, netpbm, ghostcript, gsview into the directories as shown in the tree.
untar latex2html into folder latex2html-initial/

Edit latex2hml-initial/prefs.m and add/change the corresponding lines as the following

$prefs{'EXTRAPATH'} = 'C:\\texmf\\gsAFLP\\gs8.54\\bin;C:\\texmf\\GnuWin32\\bin';
$prefs{'PREFIX'} = 'C:\\texmf\\latex2html';
$prefs{'TEXPATH'} = 'C:\\texmf\\MiKTeX2.8\\tex';
$prefs{'TMPSPACE'} = 'C:\\texmf\\tmp';

Edit latex2html-initial/config.bat and add/change the corresponding lines as the following
```
set PERL=c:\texmf\perl\bin\perl.exe
```

Add the following to the PATH environment variable on windows (Using Computer->Properties->Advanced)

C:\texmf\latex2html\bin; 
c:\texmf\perl\bin; 
C:\texmf\GnuWin32\bin; 
C:\texmf\MiKTeX2.8\miktex\bin

Add a new environment variable GS_LIB and give it the following value
```
.;C:\texmf\gsAFLP\gs8.54\lib;C:\texmf\gsAFLP\fonts
```
Notice that the above is sufficient to make latex2html happy as far as GS_LIB is concerned, and hence there is no need to edit the file cfgcache.pm and set this value there as well, which if you check the file now, most likely this variable will be empty
```
$cfg{'GS_LIB'} = q'';
```
but this was OK on my system, since it is now defined in an environment variable. But it should do no harm to also replace the above with
```
$cfg{'GS_LIB'} = q'.;C:\\texmf\\gsAFLP\\gs8.54\\lib;C:\\texmf\\gsAFLP\\fonts';
```
Even though I did not have to do it on my system as I mentioned above.
Add a new environment variable RGBDEF and give it the following value
```
C:\texmf\latex2html\styles\rgb.txt
```
There is no need to define an environment variable TEXINPUTS since install.bat when run will copy the latex2html style files and put them in location within MikTex folder where they can be located by MikTex latex.exe without the need to define such an environment variable.
Start a new DOS terminal to make sure the effect of setting the above environment variable is now in effect.

There is one bug that shows up if you are using the package amsmath which causes the generated HTML to contain extra text showing up inside the images. To fix this ¹, cd to the folder

G:\texmf\latex2html-initial\styles

and edit the file called more_amsmath.perl as follows

----- From here -----
*** styles/more_amsmath.perl.ORG        Sat Dec  2 15:15:01 2000
--- styles/more_amsmath.perl    Fri Oct  1 08:42:49 2004
***************
*** 95,100 ****
--- 95,101 ----
        }
      } else { $tag = '   ' }
      $*=0;
+     $scan =~ s/($comment_mark\d+) /$1\n/g;
      if ($labels) {
        $labels =~ s/$anchor_mark/$tag/o;
        ($labels , $scan);
----- To here -----

cd back to latex2html-initial/, run config.bat. If no errors then go to the next step.
run install.bat. Now latex2html will be installed to
```
C:\texmf\latex2html
```
run latex2html-initial/test.bat and look for any errors.
If all is well, that is all. Now you are ready to use latex2html. Go to next step

Perform post installation configuration. Edit the file

c:\texmf\latex2html\l2hconf.pm

and make any changes needed. The following are the changes I made to my l2hconf.pm to make the generate HTML and images look better for me.

$FONT_SIZE = "12pt";
$WHITE_BACKGROUND = 1;
$LOCAL_ICONS = 1;
$MAX_SPLIT_DEPTH = 4;
$SHORTEXTN = 1;
$ANTI_ALIAS = 1;
$ANTI_ALIAS_TEXT = 1;
$HTML_VERSION = '4.0';
$MATH_SCALE_FACTOR = 1.8;
$DISP_SCALE_FACTOR = 1.0;
$FIGURE_SCALE_FACTOR = 1.0;
$TRANSPARENT_FIGURES = 1;
$NO_SUBDIR = 1;
$DISCARD_PS = 0;

2.2 Some errors and possible solution

If you get this error when running test.bat that looks like something as the following
```
pstoimg.bat: Error: Ghostscript returned error status 1
pstoimg.bat: Error: Couldn't find pnm output of G:\texmf\tmp\l2h1716\image002.ps
```
Then check that you have defined GS_LIB environment variable correctly to point to the ghostscript lib/ and font/ directly as shown above.
if you get an error from test.bat that looks like something as the following
```
pstoimg.bat: Error: pnmtopng.exe -interlace -trans gray85 < p3704.pnm > img1.png" failed: No such file or directory
```
Then make sure that you have defined the env. variable RGBDEF as described above.
some of the bitmap images produced for the mathematics in the document have a solid dark bar, usually at the bottom or the side of the bitmap image. The reason for this is unknown. But I found by trial and error that setting the following values in my l2hconf.pm eliminated most if not all of these
```
$MATH_SCALE_FACTOR = 1.8;
$DISP_SCALE_FACTOR = 1.0;
```
On my system, any value less than 1.8 for the above, with everything else is fixed, produced the side solid edges again. This problem needs to be fixed.
Some mathematics equations are still underlined, even after doing the above. I found this was the case when I was using linux. If you still see the underlines, then try the following: edit l2hconf.pm and set the value of DVIPSOPT as follows
```
$DVIPSOPT = '-E' ;
```
Some mathematics formula have gray color in background. I noticed this on some equations having gray background. In this case, edit l2hconf.pm and set the following:
```
$LATEX_COLOR = "";
```
The above should remove the gray background.
Watch out for the -no_resuse option to latex2html. If one is also using -no_subdir, then Latex2html will ask the user if they want the images in the current folder deleted:
```
latex2html -no_reuse -no_subdir foo.tex

This is LaTeX2HTML Version 2008 (1.71)
....
Cannot create directory .\: File exists
(r) Reuse the images in the old directory OR
(d) *** DELETE *** the images in .\  OR
(q) Quit ?
```
And if you select option (d) then it will delete all the images in the current folder. This can include any images that were not related to latex2html earlier runs at all, and it could be your own images that you did not want deleted. This happened to me, but I had a backup copy. So, if you intend on using -no_subdir then it is safer to not use -no_reuse.

2.3 Making your latex code latex2html friendly

These are few things that I found that helped in using latex2html

Some images that I use were too large for inclusion in PDF without being scaled, even though the size looked OK for html.
So I wanted a way to use conditional logic so that when the latex file is run by pdflatex, it will scale the image, but when it is run by latex2html, the image is not scaled. I settled at this solution.
```
%  \documentclass[]{article}%
%  \usepackage{html}
%
%  \begin{document}
%
%  \begin{htmlonly}
%  \includegraphics[]{image.png}
%  \end{htmlonly}
%
%  \begin{latexonly}
%  \includegraphics[scale=0.75]{image.png}
%  \end{latexonly}
%
%  \end{document}
```

Wanted a way to modify the <HEAD> META tags in the HTML file generated by latex2html. I settled on this method for now. May be there is a better way? Thanks for old posts I found on the net by Thomas Anders and Ross Moore which helped in finding this solution.

created a file called l2h_init.pl in the same folder where my latex file is. This file looks like this

$MY_KEYWORDS = "Latex2html, HTML, MikTex";
$MY_DESCRIPTION = "This describes Latex2html";
sub meta_information {
    local($_) = @_;
    if (not defined $MY_KEYWORDS) { $MY_KEYWORDS   = "$FILE"; }
    if (not defined $MY_DESCRIPTION) { $MY_DESCRIPTION = "$_"; }
    do { s/<[^>]*>//g;
         "<!-- Do not edit here - edit TeX file $FILE.tex instead -->\n" .
         "<META NAME=\"description\" CONTENT=\"$MY_DESCRIPTION\">\n" .
         "<META NAME=\"keywords\" CONTENT=\"$MY_KEYWORDS\">\n" .
         "<META NAME=\"resource-type\" CONTENT=\"document\">\n" .
         "<META NAME=\"distribution\" CONTENT=\"global\">\n" .
         "$MY_META" } if $_;
}
1;      # This must be the last line

Now run latex2html to convert my latex file to html. Supposed my latex file is called foo.tex, then the command I used is
```
         latex2html -init_file l2h_init.pl foo.tex
```

now I looked at the generated HTML file foo.htm, and I see the header as follows

<HEAD>
<TITLE>index</TITLE>
<!-- Do not edit here - edit TeX file index.tex instead -->
<META NAME="description" CONTENT="This describes Latex2html">
<META NAME="keywords" CONTENT="Latex2html, HTML, MikTex">
<META NAME="resource-type" CONTENT="document">
<META NAME="distribution" CONTENT="global">

<META NAME="Generator" CONTENT="LaTeX2HTML v2008">
<META HTTP-EQUIV="Content-Style-Type" CONTENT="text/css">

</HEAD>

do not use \href{}{} to make links. use \htmladdnormallink{name}{URL} instead. l2h does not support \href. Remember to include the html package \usepackage{html}.
do not enclose the image in a frame using \fbox. This will cause the image not to show up correctly. i.e. do not do something as the following
```
\fbox{\includegraphics[]{image}}
```
But instead just write
```
\includegraphics[]{image}
```
I had the following problem when using l2h when it creates a subdirectory to put all the resulting HTML and image into. This is because the hyperlinks inside the HTML would now all be off by one level.
Assuming you are at folder called A/ and you run l2h on a latex file called B.tex which contains hyperlinks to other places in your site, and these links are added so that they are relative hyperlinks. i.e. the hyperlinks are relative the folder A. When l2h process the file B.tex, it would by default create a sub folder and put the result into it. This folder is called A/B/.
However, from B/ the hyperlinks are now not correct as the case would be if the file was still in A/. Therefore, I now add the following option to my l2h command -no_subdir. This means the current folder A/ will contain all the output from the l2h run, which can make the folder too messy. However, this for me was a better solution than having to change all the hyperlinks in the latex file or to make them all absolute links (which I do not think is a good idea).
I think the correct solution is to make l2h behave the same way as when Microsoft WORD saves a document as a web page, where it would leave the html file at the same level as the word document, but would then create a special folder called document_files which contains all the images and other html files generated.
Use a css file to customize the look of the resulting HTML instead of adding HTML code in the latex file using l2h special latex commands. Once you create a css file, you can tell l2h to use it with the option
```
   latex2html -style="myfile.css" myfile.tex
```
When running latex2html on a file called say foo.tex, it will create also a file index.htm in addition to foo.htm and link to foo.htm from index.htm. This can cause a problem if one allready has an index.htm in the same folder and hence it will be overwritten. To make latex2html create only foo.htm the option -no_auto_link can be used which is what I currently do.

2.4 log files

In here I have links to the output generated from running config.bat, install.bat and test.bat. I also have link to my l2hconf.pm. Notice that on my PC, I have installed everything on the G:\ drive under a folder I called LATEX and not under C:\texmf as I showed in the diagram, but as long as you have installed everything on the same drive, it does not have to be the C: drive and can be anything else.

3 installation on cygwin

These are the steps I did to install latex2html under cygwin.

I downloaded cygwin from http://www.cygwin.com/ and installed it on windows. I installed EVERYTHING. Used Unix mode. (not DOS mode).

made sure the following is installed after the above is completed

       > which latex
       /usr/bin/latex

       > latex -v
       pdfeTeX 3.141592-1.21a-2.2 (Web2C 7.5.4)
       kpathsea version 3.5.4

       > which pdflatex
       /usr/bin/pdflatex

       > pdflatex -v
       pdfeTeX 3.141592-1.21a-2.2 (Web2C 7.5.4)
       kpathsea version 3.5.4

       > cygcheck -c | grep -i tex
       tetex-base                      3.0.0-3                 OK

       > which perl
       /usr/bin/perl

       > perl -v
       This is perl, v5.10.1 (*) built for i686-cygwin-thread-multi-64int
       (with 12 registered patches, see perl -V for more detail)

        > cygcheck -c | grep -i perl
        perl                            5.10.1-3                OK

        > which gs
        /usr/bin/gs

       > gs -v
      GPL Ghostscript 8.63 (2008-08-01)
       Copyright (C) 2008 Artifex Software, Inc.  All rights reserved.

Looked to make sure netpbm package allready installed from above installation of cygwin, to check, I lookeded for one of its programs

       > which pngtopnm
       /usr/bin/pngtopnm

       > cygcheck -c | grep -i netpbm
         libnetpbm-devel                 10.49.2-1               OK
         libnetpbm10                     10.49.2-1               OK
         netpbm                          10.49.2-1               OK

downloaded latex2html version 2008 from http://www.ctan.org/tex-archive/support/latex2html/ and untar it in some directoy.

cd to latexhtml and edit the file L2hos.pm, replace the line (near the end)

       @ISA = load('L2hos', $^O);

with

       @ISA = load('L2hos', 'unix');

edited prefs.pm and set location of tmporary disk space. This step is not needed, but I like to be more sure where temporary disk space for latex2html is.
```
       $prefs{'TMPSPACE'} = '/cygdrive/G/LATEX/TMP';
```
run configure
```
      ./configure
```
Noticed that it did not use TMP defined above in prefs.pm, which is not I want TMP to point to. So I edited the file generated by running configure above, which is called cfgcache.pm located in the same folder, and forced it to use my TMPSPACE by changing the line as follows:
```
      $cfg{'TMPSPACE'} = q'/cygdrive/G/LATEX/TMP';
```
run the command make
run the command
```
       make install
```

run the command

       make test

cd to the test directory and check the output to make sure it is OK.
As of today (June 6, 2010), there is a problem with cygwin installation which is not yet resolved. Cygwin generates an error coming from one of the perl file (L2h/Unix.pm). I send a bug report to cygwin on this. Until this is fixed, I am not able to run Latex2html under cygwin

I installed cygwin 1.7.5 on windows 7 (64 bit OS), and when I run some  
command which uses perl, I get the following error:

0 [main] perl 2528 C:\cygwin\bin\perl.exe: *** fatal error - Internal  
error: TP_NUM_W_BUFS too small.


DETAILS:

I installed cygwin, all of it on windows 7, 64 bit os.
All went ok.

Then I installed latex2html, and I am trying to use latex2html under  
cygwin, which uses perl. (latex2hml is a PERL script)

When I run latex2html command on some latex file, I get many of the  
above errors each time perl is called:

$ uname -a
CYGWIN_NT-6.1-WOW64 me-PC 1.7.5(0.225/5/3) 2010-04-12 19:07 i686 Cygwin

$ cygcheck -c | grep perl
perl                            5.10.1-3                OK
perl-Error                      0.17016-1               OK
perl-ExtUtils-Depends           0.302-1                 OK
perl-ExtUtils-PkgConfig         1.12-1                  OK
perl-Graphics-Magick            1.3.7-2                 OK
perl-Image-Magick               6.4.0.6-2               OK
perl-libwin32                   0.28-3                  OK
perl-Locale-gettext             1.05-11                 OK
perl-ming                       0.4.3-1                 Incomplete
perl-SGMLSpm                    1.03ii-2                OK
perl-Tk                         804.028-3               OK
perl-Win32-GUI                  1.06-3                  OK
perl-XML-Simple                 2.18-10                 OK
perl_manpages                   5.10.1-3                OK
postgresql-plperl               8.2.11-1                OK
subversion-perl                 1.6.11-1                OK
$


$make test
.....
Converting image #2
       1 [main] perl 3104 C:\cygwin\bin\perl.exe: *** fatal error -  
Internal error: TP_NUM_W_BUFS too small.

Error while converting image

This is the file where this error comes from Unix.pm This is a more detailed error message:

> 262: my ($self,$cmd,$in,$out,$err) = @_;
> 263: carp qq{Debug (syswait): Running "$cmd"\n} if($Verbose);
> 265: my $status;
> 266: my $child_pid;
> 267: if ($child_pid = fork) {
> 268: $status = waitpid($child_pid, 0);
> 274: unless(exec($cmd)) {
> 0 [main] perl 4524 C:\cygwin\bin\perl.exe: *** fatal error - Internal error: TP_NUM_W_BUFS too small.
> 269: carp "Debug (syswait): Finished child process: #$child_pid\n"
> 270: if($Verbose);
> 271: $child_pid = 0;
> 272: return($?);
> exited L2hos::Unix::syswait

3.1 log files

In here I have links to the output generated from running above commands:

This is output from cygwin configure
This is output from make install

4 References

About this document ...

This document was generated using the LaTeX2HTML translator Version 2008 (1.71)

The command line arguments were:
latex2html -init_file l2h_init.pl -info 1 -split 0 -reuse 1 -show_section_numbers -no_subdir -local_icons -style nma.css index.tex

The translation was initiated by me on 2010-07-03

Footnotes

... this ¹: Thanks to Shigeharu TAKENO for finding the solution to this bug and posting it to the latex2htlm mailing list on 11/3/2009

me 2010-07-03