|
To: A. Problems resolved in LaTeX2HTML
|
 |
LATEX is a high-quality document preparation system that is particularly powerful if you have a lot of math or cross-links in your document. LATEX2HTML is an add-on program to convert these documents into web pages instead of printed hard copies.
Of course, before you can use LATEX2HTML, you need to know how to create documents in LATEX format. There are books like that of Leslie Lamport (see the bibliography of the original online manual). A powerful free resource is the web page of the TeX Users Group. In my opinion, the current dominant free implementation of LATEX for Windows is MiKTeX. It has a useful manual, and there is an easy install script to add LATEX2HTML-FU to it.
But let me give a really really basic intro first if you are really
clueless about it all.
On MS Windows, download and install the emTeX/LATEX2HTML-FU bundle using
the easy install script;
on Unix, install TeXLive from the TeX Users Group and add LATEX2HTML-FU
according to instructions. Also add the ghostview and perl programs.
Create a folder for your first document; avoid spaces
and points in the folder name, they are just trouble. On Windows, you
might create a new folder myfirsttry in the top C: disk.
Open Windows Notepad, or your favorite Unix editor (pico, nedit, ...).
In it, type in the following text, case-sensitive, avoid typos, wrong brackets:
\documentclass{article}
\begin{document}
Hello world.
\end{document}
Now save this to the folder myfirsttry that you created, naming
the saved file index.tex. Windows users beware! You have to
specify the name between double quotes as "index.tex", or
Notepad will save it as index.tex.txt instead. It’s Microsoft.
Now on Windows double-click the Windows_XP (or 98-ME icon for older
Windows) in the “l2hsup” folder in the top C:
drive, or on Unix open an xterm or telnet command window. In the
created window, cd to the folder with the index.tex file you just
created (for example, in Windows use cd \myfirsttry). Enter
the following commands:
latex index makepdf
Then examine the created
pdf file index.pdf in the folder myfirstry;
It should show the text “Hello World”, as shown
here.
To make web pages instead of a pdf file, use:
latex index l2hand examine the created web page
index.html.
It should look like
this.
Like HTML, LATEX is a “mark-up language”, which means that you type in formatting information instead of pick it from menus. For example, a more realistic document would look like:
\documentclass{article}
\begin{document}
\title{My Second Document}
\author{J. Doe}
\maketitle
\section{Quadratic equations}
\label{sec:quad}
The general quadratic equation
\begin{equation}
a x^2 + b x + c = 0 \label{eq:quad}
\end{equation}
has two solutions given by
\begin{equation}
x_{1,2} = \frac{-b \pm \sqrt{b^2 - 4ac}}{2a} \label{eq:quadsol}
\end{equation}
\section{Numerical solution}
\label{sec:numsol}
In section \ref{sec:quad}, we gave the solution of the quadratic
equation. This solution can be put in a computer program as is, but
is then likely to suffer from round-off error problems.
The better way to do it is to find only one solution, call it $x_1$,
from the given formulae (\ref{eq:quadsol}), selecting $x_1$ as {\em
the solution for which the two term in the top of the ratio have the
same sign.} Then compute the other solution from
\begin{equation}
x_2 = \frac{c}{a x_1} \label{eq:altsol}
\end{equation}
\end{document}
Note that fractions and square root signs are obtained by typing them in. If you do a lot of math, you will find that this is actually much quicker and more reliable than picking items from menus. Also note that I use mnemonics instead of actual section and equation numbers. That avoids having to do do extensive renumberings in large documents if you insert a new section or equation number somewhere, and it cuts down tremendously on incorrect numbers. LATEX2HTML likes the equation labels to be following the math. There seems to be a lot of other typing involved, but a LATEX-aware editor like Emacs will put it in for you.
Process as:
latex index latex index makepdf l2hThe pdf file so produced is here and the web pages are here.
See the sources mentioned above to get more in depth info. The FAQ on tug.org is especially useful.
If you get into trouble processing documents with latextohtml-FU, the first thing you have to make sure of is that your index.tex file processes correctly under latex (and dvips). If not, everything stops right there.
Browsers are buggy too. Try hitting “reload” or “refresh”.
If the document is correct, the next thing to ask is whether it is fairly standard LATEX. LATEX2HTML has extensive translation capability, but there are very definite limits. If necessary, create a very standard form index.tex to convert into web pages.
If you made elective changes from my installation procedure, however seemingly minor, repeat installation and don't.
Check the log file C:\Temp\l2h.log (in the Temp directory of
the C: disk) for clues. (For Unix users, the log file is in either
$L2HTOP/temp or $HOME/temp.)
If you specify “ -debug” behind “l2h”,
you will get much more error information on the screen and in
C:\Temp\l2h.log. It will also leave the ghostscript files in
a subdirectory of “Temp”, and their conversions. You
can look at the .ps files using Ghostview, and at the pnm files using
an image viewer like CompuPic. Alternatively, you can convert them to
.bmp using the Netpbm utilities you installed and look at them with
Paint.
To slow down the screen a bit, in the DOS window use:
set L2H_SLEEP=1(or 2) before running l2h. Do not try output redirection.
If there were problems processing images.tex, see section 9.20.1 for more.
If you are familiar with Perl, try setting -d option behind perl.exe and $PERL in latex2html.bat and stepping with n, s, and r.
Unix users should make sure that the L2HTOP name is set and that its l2hsup directory is in their PATH. If not, use
setenv L2HTOP "$HOME/l2h" (or where the LaTeX2HTML files have been installed) setenv PATH "$L2HTOP/l2hsup:$PATH" rehashMake sure that
L2HTOP/l2hsup is before other parts of the path
that may have netpbm files of the wrong version number, or strange things
may happen.
If that does not solve it, the next subsections are some solutions to various potential problems that I am aware of, sorted by type.
The official online manual also has a list of known problems and solutions in Chapter 6 that you also want to read through.
You can also have a look at the sample documents in section 6 for examples.
Good luck.
The official online manual also has a list of known problems and solutions in Chapter 6 that you should read through.
The manual list some sources of support for the official version of LATEX2HTML. I do not know how active they are. There is a TEXusergroup, which includes a very useful FAQ and much software.
E-mail me problems, section 3, and I will add them to the list. If they are due to my stuff, or if they are important enough to me, I will fix them or devise a work-around.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h.
All I know is in the respective installation section. When people send me corrections or additions, I add it there. Or weird stuff maybe here.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h.
See item 4 of the installation instructions.
There are some potential problems that cause symptoms that are just weird. If you are baffled, try each of them and see whether it solves the problem. None of them can hurt:
command /e:3072Now try again. This will give you 3072 bytes of environment space, for this DOS window only. (This goes at the expense of other DOS memory. You also lose additional memory from now having two copies of command.com running, so if you have to do this frequently, consider increasing the corresponding value in file CONFIG.SYS.)
setenv L2HTOP "$HOME/l2h" (or where the LaTeX2HTML files have been installed) setenv PATH "$L2HTOP/l2hsup:$PATH" rehashMake sure that
L2HTOP/l2hsup is before other parts of
the path that may have netpbm files of the wrong version number, or
strange things may happen.
Browsers are buggy too. Try hitting “reload” or “refresh”.
For your comparison, I have full copies of my own Windows 98 CONFIG.SYS and AUTOEXEC.SYS files. Do not use such files on Windows XP.
The official online manual includes some other potential problems.
If the screen freezes, and there are notices on the screen that .tex and .log files are being moved to the Temp folder, a problem occurred in latexl2h.bat. Examine the log files moved into the Temp folder for the problem. Also have a look at rb.err. Then hit Return. L2h should continue.
If the DOS window freezes solidly, see section 9.6 on weird problems.
See section 9.2 on how to use the debug option and the log file.
See section 9.6 for random DOS problems.
Review the Installation section for potential installation problems.
No, it does not. Reboot time.
If it still occurs, make sure the FILES value in CONFIG.SYS is big enough. L2hcrop opens three files.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h.
LATEX2HTML works normally, (or rather, as it should.) Run as
C:\l2h\bin\latex2html or
$L2HTOP/bin/latex2html, using
options as in the manual in the C:\l2hsup\l2hdocs folder.
The orginal .latex2html-init file that comes with the
unmodified LATEX2HTML can be found in the l2h
directory under name dot.latex2html-init. Copy the file to
the directory with your index.tex file and rename it to
.latex2html-init; in other words, remove dot.
(This will make the file invisible on Unix.) LATEX2HTML has its
own style sheet index.css hard-coded.
However, do make sure that html4_0.pl is loaded. Various
fixes have only been impemented for that version (and not for
the default html3_2.pl.
The critical html package (file
html.sty) should have been installed with
your installation. If not, try putting a copy of it in with your
document.
You need to have the epsf.sty and
hyperref packages installed in your version of TEX. If
you do not yet have them, TUG has them
in downloads/CTAN.
However, as of 2005 or so, hyperref is no longer backward compatible
with the htmlimage command as described in the LATEX2HTML manual.
The solution is to follow every htmlimage command with an empty pair
of braces, {}.
However, as of 2006 or so, hyperref is no longer backward comptable
with html.sty. The solution is to load hyperref before
html.sty:
\usepackage[draft]{hyperref}
\usepackage{html}
However, as of 2007 or so, hyperref needs package kvoptions,
and TeXlive 2007 does not seem to include kvoptions. You can
find the kvoptions on TUG in
downloads/CTAN. (Or you can just put
kvoptions.sty in with your file.)
However, as of 2008 or so, ...
The procedure that LATEX2HTML uses to determine the correct section, figure, and table numbers can fail. The results are unsightly, with missing or wrong numbers. I reduced its likelyhood of this occurring very much, but nothing is perfect.
In particular, if you put HTML-only sort of stuff in a figure or table caption or in a sectional title, (like a latexhtml, htmlcode or htmlonly command), it will prevent its number to be found. Latex2html will then guess at the number. To fix this, provide the optional shorter caption or title, and leave out the html stuff from it. (Do not forget that there may no line break between ] and {.)
I believe the above is nowadays enough to fix the problem. But a few older suggestions might still be useful.
Avoid discretionary hyphens in captions and section names. Embedded math or accents may also cause problems; try to rewrite it without.
Remove duplicate caption or section titles that give problems. For example, if you have two sections called “Introduction”, rename one “Introduction to ...”
Examine the log file C:\Temp\l2h.log; it often shows why
LATEX2HTML did not find the latex caption matching the html one.
Try rephrasing the part that does not match.
Or, if you are not afraid of a bit of Perl, edit file
C:\l2h\bin\latex2html.bat, and find the string
“leon: use brute force now”. The last-resort number
search is below it. For example, the statement
$cap_key_s =~ s/\\'/é/g; makes the latex
\' match the html e-acute symbol.
See the next item.
This will produce erroneous processing if done for the shorter version of a caption or title or similar. Break inside the brackets instead.
If definitions do not work in image environments or similar, they may be missing in images.tex. Try putting a copy of the definition in an “imagesonly” environment.
Do not put a \input or \include command inside a
\[re]newcommand or \def. It does not work properly.
(Yes, this is a bit of a pain. But it is probably unavoidable.)
Yes. For example, if you try to redefine a command like
-, there
will be an erroneous space produced. This author does not know how to
fix this. However,
0 through
9 work correctly. So if you need a
one-character, and/or nonalphabetic command, use these.
The manual says that you need the linefeed behind the
\begin{rawhtml} or before the \end{rawhtml}. But
that is not always needed.
See also subsection 9.20.17.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, look at settings.pl, not .latex2html-init.
\HTMLcode only accepts tags from the corresponding official
HTML version. For example, a NOBR tag, which prevents line breaking,
is not official, so refused. To fix this, go into the
versions subdirectory of the top LATEX2HTML directory.
Edit file html4_0.pl or whatever is the version of html that
you have set in your .latex2html-init file. Add the tag to
either closed_tags_list or unclosed_tags_list,
depending whether there is a closing tag. For example, NOBR has a
closing tag /NOBR.
Instead you could try to use the rawhtml environment.
Unfortunately that adds unwanted space before and after the tags.
Problems with images (figures and mathematical formulae) are varied. Images are made through the following steps:
Obviously, any of the above steps can cause problems. Your best bet is then to specify the debug option and examine the log file and images as in section 9.2.
If there were problems processing images.tex, you can create a temporary subfolder “tmp” in the same folder as index.tex, move images.tex to that subfolder, and then process it from the DOS window like:
C: cd \My Documents\mydoc\tmp C:\l2hsup\latexl2h imagesThis should give you a better view of LATEX error messages and problems.
Unix users should make sure that L2HTOP/l2hsup is in their
PATH before other parts that may have netpbm files of the wrong
version number, or strange things may happen. Their log file is
in either $L2HTOP/temp or $HOME/temp.
Note that some web graphics used through the graphics or graphicx
packages might conceivably not be converted correctly since ppmquant
no longer exists. Solution if this would actually happen: convert the
figures to eps, or edit l2h/styles/graphics-support.perl and replace
ppmquant by an equivalent current utility.
See the next item.
If you have [re]newcommands in which the command name is not within curly brackets, latex may eat it, but images.tex may not have a documentclass line. Of course, everything stops there.
Do not put labels of equations on otherwise empty lines. It will
cause LATEX to shout a lot when it processes images.tex.
The reason is that LATEX2HTML removes the label, leaving a blank line in
the formula.
It is recommended that you put the label immediately following the math and hide the end-of-line by following it with a percent sign. If not, you will get errors in images.log.
See the next item.
This will produce erroneous processing if done for the shorter version of a caption or title or similar. Break inside the brackets instead.
This sounds like an installation problem. Review the installation procedure.
Review the general notes on images in subsection 9.20.1
See section 9.2 on how to use the debug option and the log file.
See section 9.6 for random DOS problems.
When processing the images, LATEX2HTML uses its own page size.
So whitespace may be stretched differently. To fix this, immediately
behind the \begin{document} line, put
\latexhtml{}
{\setlength{\textheight}{606.5pt}\setlength{\textwidth}{405pt}
\setlength{\vsize}{606.5pt}\setlength{\hsize}{405pt}}
Change the numbers according to your document text size. Here 1pt
is 1/72 inch.
If you use emTeX, eps file names must be 8 characters or less (excluding the .eps), or the bounding box will not be read. Rename or add the bounding box explicitly to the epsffile command.
LATEX2HTML does not know that frameboxes in pictures are different.
The following is cheating, but it works for me. In the preamble. define:
\def\picfrbox(#1,#2){\framebox(#1,#2)}
\begin{imagesonly}
\def\picfrbox(#1,#2){\framebox(#1,#2)}
\end{imagesonly}
then use \picfrbox instead of \framebox in your pictures.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, you should edit settings.pl, not .latex2html-init.
If you misspell “transparent” as “transparant”
inside \htmlimage, the figure is going to fail. Take it from me.
Do not use \hsize or \textwidth in pictures,
images.tex may not conform to your document. Specify actual
dimensions in pt or in.
If a picture has too much empty space around it, it may cause a
problem if \resizebox is used. Try reducing the declared the
picture size to closer conform to the actual size.
Make sure .eps files have proper bounding boxes. Use
ps2eps -l -B. Don’t think commercial software puts out
correct postscript files; most do not. If the bounding box is wrong,
the picture may only show up half or not at all.
In picture environments, a circle at the far left may show up as half
a circle, as only the center is considered as area that is written to.
To fix it up, put a tiny \rule{.2pt}{.2pt} on the leftmost
perimeter point of the circle.
Plain TEX \def definitions in the preamble may not always be
correctly transferred to the images.tex file, causing figures using
them to fail. One fix to this bug is to add the line
$TEXDEFS=1;to your
.latex2html-init file. Alternately, the
-notex_defs command line option could be used with (make)l2h.
(The .latex2html-init file is hidden on Unix.)
More simply, add the definitions explicitly to images.tex using
the imagesonly environment.
Examine the images.log file and check that the figures can be found.
Note that image processing takes place in a subdirectory, so, for example
\epsffile{myfigs/myfig.eps} would fail to find the
figure. Use a complete path, or simply put the figures in with
index.tex and get rid of myfigs/.
If they are figures using \special commands or other complex
graphics procedures, and your dvips supports the -E option, delete the bad
figures, if they were made at all, and try the following in the DOS
window:
set L2H_NOEPS=s l2h (or makel2h) set L2H_NOEPS=This uses the non-eps version of latex2html, which may more robust. (Unix users use setenv and unsetenv, with no equals signs.) However, I find this trick no longer works on at least some new versions of TEX; it will make things worse. It works fine with the emTeX version, though. Sorry, I do not know why, the headers of the postscript versions seem equivalent except for the dvips version number and I am using the same gswin32c for both. So, do not set L2H_NOEPS as the default!
Alternatively, and better, convert the figure into a proper eps file.
Copy index.tex to folder Temp, in the preamble set
\pagestyle{empty}, and remove the caption(s) and any other
text and headers on the page(s) of the bad figure(s). In the DOS
window:
c: cd Temp dvips -Ppdf -E -p XXX -n 1 -o index.ps indexwhere XXX is the page number with a bad figure. Load the generated index.ps file into Ghostview, turning OFF "EPS Clip" under options. From the “File” menu, select “PS to EPS”. Use automatic bounding box, and save the file under name figpXXX.eps. Open that file using Ghostview and check that it now has a proper bounding box. Move the file back to the original index.tex document folder, and replace the specials garbage in index.tex there with
\epsffile{figpXXX.eps} (if you use the epsf package.) Repeat
with the other bad figures.
Some additional items for the sophisticated: There are some perl scripts in folder l2hsup that are supposed to do much the same thing without ghostview. If in DOS you
set L2H_PUTPS=1the postscript files generated by latex2html will be moved to the temp folder for messing around with. If you
set L2H_GETPNM=1pstoimg will import matching pnm files from the Temp folder, bypassing normal conversions.
MiKTeX 2.5, and probably any other LaTeX loading the Aug 2006 hyperref
package, produces random cryptic errors in figure environments with an
htmlimage command in it, typically complaints about 'extra }'
or 'Missing \endcsname inserted.' Some problems go beyond
the figure environment.
The problem appears to be incompatibility of the html package with the new hyperref package loaded by MiKTeX 2.5. To fix the problem, before loading the html package, turn off hyperref using the line:
\usepackage[draft]{hyperref}
If this is not possible for some reason, put an empty pair of brackets behind the htmlimage command, for example:
\htmlimage{extrascale=3}{}
It is in fact recommended that you use such an empty pair of braces
with recent versions of LATEX whether you have problems or not.
Put \htmlimage{notransparent}{} in the figure environment. Do
not misspell this!
Gif files have only 256 different colors, unlike jpeg ones which have 16 million. You can create jpeg versions of the figures. Unfortunately, jpeg files might have a bit of visible “ghosting”. Little is ideal is this world.
Jpeg files are typically smaller than gif ones. Try using
them, with a L2H_JPGQ of 75.
There are also utilities on the web that compress images to the maximum, letting you see the effects while doing it.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, the l2h menu should take care of this. And edit settings.pl, not .latex2html-init.
If you change the background color from white, this is likely to
happen. You need to get ghostscript to use a different background
color in its anti-aliasing, by changing the pagecolor in the
preamble of index.tex or in .latex2html-init (set WHITE_BACKGROUND
to 0 and the LATEX_COLOR to appropriate RGB fractions.)
(The .latex2html-init file is hidden on Unix.)
Delete the bad images and remake.
If you have inline math with inline math inside it (like in a mbox), latex2html puts a comment in front of it that introduces an additional space.
To fix that, go into the latex2html script. Inside sub
replace_verbatim_marks, find
s!$verbatim_mark(\w*[vV]erbatim\*?|tex2html_code)(\d+)#\n?!$tmp=$verbatim{$2};
$tmp.(($tmp =~/\n\s*$/s)? '':"\n")!eg;
Get rid of the \n inside the double quotes.
At the image state, there is no way for LATEX2HTML to know about whitespace that extends beyond the area with ink on it. Ignore it, rewrite it not to have leading or trailing whitespace, or if it is a real problem, put a tiny dot (ruler) in front or behind the whitespace to extend the area with ink on it.
A browser problem. Press refresh.
This is really a Ghostscript problem. Ghostscript does not anti-alias
embedded bit maps in the postscript files it processes. Delete the
bad gif file. Inside the figure environment of the bad figure, put
the command \htmlimage{extrascale=3}{}. (Make sure that you are
using the html package.) Rerun l2h or makel2h.
Unfortunately, this only works for figure environments. If you are
using displaymath to force a figure in a given place (What is this
with LATEX changing [h] into [ht]?? If a user with a brain says
[h], some stupid program with none should not second-guess that), you
will need to use figures and strategically placed \newpage
commands instead. Or use the float package and specify the figure
location as “[H]”.
The first step is to press “Refresh” or “Reload” on your browser, to check that you are really see the latest version and not a cache one. Also, in Mozilla, use “Edit” / “Preferences” / click on the plus sign in front of “Advanced”, select “Cache”, and select “Every time I view the page”.
The following problem no longer exists for versions of LATEX2HTML-FU downloaded after Sep. 20, 2007. For older version only: Unfortunately, LATEX2HTML is not very observant about changes in figures and math. Just delete the image (use “Properties” or “View Image” in your browser to find its name). It will be remade according to the new latex code next application of l2h or makel2h.
The first step is to press “Refresh” or “Reload” on your browser, to check that you are really see the latest version and not a cache one. Also, in Mozilla, use “Edit” / “Preferences” / click on the plus sign in front of “Advanced”, select “Cache”, and select “Every time I view the page”.
Sometimes, latex2html gets all mixed up about what image goes where.
The only thing to do then is clrl2h all
(keep the user-modifiable files!) and start from scratch.
The following problem no longer exists for versions of LATEX2HTML-FU
downloaded after Sep. 20, 2007. For older version only:
Two formula or figures starting with the same long LATEX string may be
considered equal even though the LATEX is eventually different.
Put a \strut at the start of one of the formulae to make
them different.
Pestilential critters.
If a sidebar occurs for a figure or formulae in which you do tricky
things like lowering and raising items, or where you write outside of
bounds in a picture, putting a suitably sized
\rule[..]{..}{0pt} strut at the start may help. The idea is
to make the formatter recognize what the size of the image really is.
I had a sidebar in a big makeimage that could be fixed by
moving a renew of arraystretch outside, in front, of it.
Otherwise, make sure you have the latest and greatest version of the alignment bar fixes, posted Sep 8, 2006 or later.
If so, to have me fix this, e-mail me the corresponding:
index.tex file, cut down to just a single example of the
math that creates an image with an extra bar.
.latex2html-init file (hidden on unix)
index.css (the one in the folder with the web pages if using
makel2h)
To try to fix it yourself, edit C:\l2hsup\doc\l2hcrop.f and
try increasing the barthb and barths values a unit.
Recompile l2hcrop with g77 according to instructions in the
file.
Make sure, in l2h.log, that GraphicsAlphaBits is not used by Ghostscript in processing the images.
See section 9.12
Do not use two captions in the same figure environment. Use two figure environments. Or combine the two captions into one.
If you use an optional argument to the caption, make sure there is
no linebreak between the ] and the {. Or the caption will
become part of the image, probably with an incorrect figure number.
Try to keep math out, and try to avoid captions with the same text
save for the math. If the caption involves (even implicitly)
\html or \latexhtml commands, the correct figure
number may not be found.
It will. To avoid this, put the text in a caption, and use
\addtocounter{figure}{-1} in front of the figure
environment.
Or live with the text being graphics.
You may be able to play around with the makeimage command described in the LATEX2HTML manual.
This issue is obsolete. Using the “e” browser, Internet Explorer changes the vertical positioning of inline images if there are other images in the same line. Since that depends on the browser width setting, there is nothing LATEX2HTML can do.
However, nowadays L2H_BROWSER is by default “a”, which has little alignment error. Switching to “x” may improve alignment even more.
The best way to include figures is as .eps files, typically using the epsf package. Unfortunately, much graphics software does not, or not correctly, produce .eps files. Use your netpbm files to convert from the existing picture format to .pnm, and then to eps.
If your figure is a Windows .bmp, as produced by MS Paint, you can use
C:\l2hsup\bmptoeps. (It will tell you how to use it when called.)
If latex processes the figure correctly, consider using
dvips to create an eps file of it, by putting the figure by
itself on a page with an empty page style. Use the -p and
-n options of dvips to select the page.
The LATEX2HTML version of \includegraphics
is old and has problems:
\DefineGraphicsExtensions command.
You can force the \includegraphics command to be processed
by LATEX instead of LATEX2HTML by including the following
text in your settings.pl or .latex2html-init
file:
&::process_commands_in_tex (<<_RAW_ARG_CMDS_);
fbox # {}
framebox # [] # [] # {}
includegraphics # [] # [] # {}
_RAW_ARG_CMDS_
Inside the figure environment, use a command like
\htmlimage{align=left}{} (normal image, left aligned)
\htmlimage{thumbnail=0.5,align=right}{} (thumbnail of the image, at the right)
Be sure to use lower case.
Put the figure at a place where there is some following text to flow around it, not right at the end of the web page.
Use the “rawhtml” environment. Quantum Mechanics for Engineers has examples of animations, activated by a roll-over, in the sections on wave packets and following (search for “animation”).
(The animation themselves were made using a Fortran program to generate the curve data, a plotting program to convert them into postscript curves, the netpbm libraries to convert them into pnm, l2hcrop.exe to crop them to the frame, l2hcomp.exe to compress them in size, ppmquant to reduce the colors to 256, netpbm to convert them into gifs, and “gifsicle” to combine them into a moving gif. But all that has nothing to do with LATEX2HTML.)
The navigation panels are not readily modified to include rollovers. You would have to go into the perl, I think.
If you try to customize lists such as itemize, enumerate, and list, maybe decrease the font size of the entire list or so, you will find it does not work. While you are encapsulating the list in your font size or whatever, LATEX2HTML unencapsulates the list just as quickly. This is obviously deliberately, though the reason is unknown.
The only solution I have is to put in the html yourself. To create quoted small, bold, italic lists in the preamble define:
\def\listsheader{
\latexhtml{\begin{quote}\small\bf\em}{
\begin{rawhtml}
<blockquote><small class="small"><b><em>
\end{rawhtml}
}}
\def\listsfooter{
\latexhtml{\end{quote}}{
\begin{rawhtml}
</em></b></small></blockquote>
\end{rawhtml}
}}
Feel free to leave out some of the blockquote, small, b, or em
<...> tags if you like your tables less garish. But do
not change indentation or line breaks, they must be as shown.
Then in your document, use them as:
\listsheader
\begin{itemize}
\item Item a.
\item Item b.
\end{itemize}
\listsfooter
Also LATEX2HTML may not reset counters to zero if you define a list and use it a second time. The following works:
% questions list environment
\newenvironment{questions}[1]
{\setcounter{enumiv}{0}\listbeg\centerline{\bf #1 Review Questions}
\begin{list}{{\bf \arabic{enumiv}}}{\usecounter{enumiv}
\settowidth{\labelwidth}{{\bf 99}}
\setlength{\leftmargin}{\labelwidth}\addtolength{\leftmargin}{\labelsep}}}
{\end{list}\listend}
Note the explicitly added \setcounter{enumiv}{0} before the list.
If you define new counters, LATEX2HTML puts their definition twice
in the images.tex file, leading to nasty error messages. It will not
hurt anything, but if it drives you crazy like it does me, don't define
new counters but simply use enumiv.
In description environments, the index key is put in bold case just
like in LATEX, but on a line by itself. If you do not like that,
you can enter the following HTML style information using the
rawhtml environment:
<STYLE TYPE='text/css'>
DT {float: left; margin: 0px 0px 0px 0px}
DD {margin: 0px 0px 0px 45px}
</STYLE>
If you put this in your index.css file instead, leave out the
two STYLE lines. Do note that this will mess up your index
unless you use the short_index option. The margins are in
order top, right, bottom, left.
The item keys in the bibliography and index are by default put in
separate lines. To fix, use the short_index option and
redefine the style of the HTML DT element as described in the previous
subsubsection.
To automatically add blank lines 25 px high in the HTML enumerate
environments, you can enter the following HTML style information using
the rawhtml environment:
<STYLE TYPE='text/css'>
OL LI {margin: 0px 0px 25px 0px}
</STYLE>
If you put this in your index.css file instead, leave out the
two STYLE lines. The margins are in order top, right,
bottom, left.
It may be noted that LATEX2HTML seems to respect blank lines between entries put in the list environment, but not the enumerate one. In any case, try that first.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, use the second l2h menu to change the icons. Create a theme to make the icons generally available. Do not edit .latex2html-init, edit settings.pl.
Use C:\l2hsup\giftobmp next to turn the “next”
icon next.gif into bitmap next.bmp. Load that in MS Paint and edit
away. (If you change the icon size using attributes, you must also
edit the .latex2html-init file to put in the new size;
the .latex2html-init file is hidden on Unix.)
After completion, use C:\l2hsup\bmptogif next FFFFFF to
convert back, making the color white
(FFFFFF) transparent.
It is not a good idea to use “New” in MS Paint instead
of C:\l2hsup\giftobmp, since Paint will mess up your colors
on saving.
To make changed icons available to all future documents, use the DOS commands
c: cd \l2hsup xcopy iconsby myiconsand answer “D” when prompted. Now edit the icons in subfolder “myicons” of folder “l2hsup” at will. They can be included in future documents by editing .latex2html-init in subfolder “files” of folder “l2hsup”. (The
.latex2html-init file is hidden on Unix.)
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, you must edit settings.pl, not .latex2html-init. And you can insert your image directly from index.tex.
The easiest is to do it with the wiz. Also use it to change the bar colors to match your logo. But if you need special customizations, here is the manual process.
Edit file .latex2html-init, (hidden on Unix), in the
files subdirectory of the l2hsup directory and
locate the line
"<TD><TABLE WIDTH=\"100\%" BORDER=0 CELLPADDING=2 CELLSPACING=0><TR>\n" .This line exists at two locations; the first is for the header, the second for the footer. Immediately behind the line, add a line
"<TD COLSPAN=3 ALIGN=\"aaa\"><IMG SRC=\"uuu\"></TD></TR><TR>\n" .substituting for “aaa” the desired alignment,
left, middle, or right,
and substituting for “uuu” the url of the image, e.g,
mylogo.jpg. Don't forget the trailing point.
To make the logo clickable, use instead
"<TD COLSPAN=3 ALIGN=\"aaa\"><A HREF=\"ccc\"><IMG\n" . " SRC=\"uuu\" BORDER=0></A></TD></TR><TR>" .substituting for “ccc” the desired target url, e.g.
http://www.mycompany.com. Don't forget the trailing points.
You may need to also play around with the TD.runner colors in
file index.css to get it to match your logo's text and
background colors.
To make the changes retroactive for existing documents, make the same
changes to .latex2html-init in the directory with the
index.tex file and index.css in the directory with
the web pages. Putting the image itself in the directory with the
index.tex file is strongly recommended, in case somebody
decides to move a centrally located image file.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, you must edit settings.pl, not .latex2html-init. And try the second l2h menu first.
Most changes are best made with the wiz, but here is some more.
To add a border around the bars, copy files template.html
and select.html in the files subfolder of
l2hsup over to the folder with your index.tex
file, renaming select.html into select0.html.
Edit these files and find the
BORDER="0" CELLPADDING="0"
TABLE tag pairs (two occurences per file). Change the border
value from zero to the desired border size in pixels, say 5.
Run the wiz and select “change headers and footers”.
No changes need to be made, they just must be rebuild. Rerun (make)l2h.
Other changes to the HTML code can be made similarly, within reason.
If the changes you make are too big for the wiz to implement, the
actual code that is used is in file .latex2html-init.
If the style changes that the wiz can do are not sufficient, edit index.css (current document) and/or index.css in subfolder “files” of folder “l2hsup” (future documents) and change the values of TD.runner(b) to change bar colors, font, etcetera. Change the a.runner(b) values to change the appearance of active links in the bars. (B stands for bottom.)
Remake the web pages with (make)l2h.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, use the second l2h menu for this.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, you must edit settings.pl, not .latex2html-init. And try the second l2h menu first.
(Without using the wiz, that is.)
Edit .latex2html-init in the folder of the current document and/or
.latex2html-init in subfolder “files” of folder
“l2hsup” (for future documents) and set
WHITE_BACKGROUND to zero and the desired color, in RGB
fractions, in LATEX_COLOR.
(The .latex2html-init files are hidden on Unix.)
Also edit index.css (current document) and/or index.css in subfolder “files” of folder “l2hsup” (future documents). Put in the desired color in the BODY style.
Editing index.css and then reloading a web page in that directory is a good way to experiment with possible background colors.
Existing images not made on the correct background color should be deleted so that they can be remade on the right color, or ghosting will result.
As far as your figures are concerned, they should be at least one of the following:
\htmlimage{notransparent}{} in the figure environment. This will
show the background color of the figures on which they were created.
Before running l2h or makel2h, in the DOS box enter:
set TRANSPARENT_COLOR=#123456where 123456 is the hex specification of your background color.
I do not know an automatic way of doing that. Given time, you could
start your document with section one, in the right background color as
described above. Process. Add section two to index.tex, and process
giving it the right background color (ignoring that web page one will
now also assume that color.) Repeat until all sections have been
added. Now edit index.css and remove the background-color line from the
BODY tag. Edit index.html and each node...html and change
<BODY > into <BODY BGCOLOR="#CCCCCC">, where CCCCCC
is the desired color of that page.
Try to ensure that images from one page are not being used on another, or they will show ghosting on the later pages. Maybe put in a small strut of varying vertical size?
To correct an error, reprocess in the color of the corrected web page and repeat the editing of the html files.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, you must edit settings.pl, not .latex2html-init. And try the second l2h menu first.
In LATEX's rgb model, (also used near the end in .latex2html-init,) you specify the amount of red, green, and blue, separated by commas. Each is on a scale from 0 to 1, so {0.23, 0.23, 0} is 23% red, 23% green, no blue: in other words, a dark yellow.
In HTML, and in index.css, you scale these numbers up by a factor 255,
so in the example above you get 59, 59, 0. Each is then converted in
a two “digit” hex number as follows: divide by 16 to
find the first digit; the division remainder is the second. For
example 59/16=3.68, so the first digit is 3. The second digit is 59 -
3
16=11. Now the hex system takes digit 10 to be
“A”, digit 11 to be “B”, etcetera, until
15 is “F”. So 59 is 3B and the full example color
specification is 3B3B00.
Remember that 000000 is black (all colors off) and FFFFFF is white (all digits to the max.) Also remember that the color in html is preceeded by #, e.g. #FFFFFF for white.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, selecting the original_frames theme will give frames. And you must edit settings.pl, not .latex2html-init.
For your dose of prison bars, follow the example of the manual. Note that you will not see much frames unless you click on the contents or index buttons.
Use clrl2h all in the folder with your
index.tex file to clean up any old web pages you may have made.
Then download the
.latex2html-init file, (this is a hidden file on Unix,) and
the index.css file and put them in with your document.
Run latex index and l2h or makel2h to process.
Frames are not compatible with the wiz. In particular,
you will need to edit .latex2html-init to change colors.
Font sizes and styles can still be changed with the wiz, though.
Rewrite index.tex to use more standard LATEX. Or use the htmlonly/latexonly environment described in the manual.
The Proceedings of the Royal Society style does not define
\paragraph, causing the html package to crash.
I used \newcommand{\paragraph}{\subsubsection}
before loading html.
No. But by default the titles of starred chapters and sections are
added to the contents web page, so try to make use of that. You might
be able to do some tricks with \htmlonly (see the manual)
too.
This may not be as bad as it sounds. Check the output.
Otherwise, try to avoid the need to explicitly change chapter numbers or whatever.
This happens if there are inline images, especially big ones, and more for Internet Explorer than Mozilla. It is due to the need for additional whitespace around the images to get them at the right height in the line.
Relatively small images with a vertical ink-region center about half-way the letter “a” are least susceptible.
Consider turning big inline mathematics into displaymath.
Use {\lq} for left quotes, {\rq} for right ones.
Some accented characters are unknown to LATEX2HTML and
represented by images. To get the character to align well,
put it in an mbox in a math environment. For example, use
$\mbox{\.p}$ instead of \.p.
Play around with index.css.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. For one, you must change settings.pl, not .latex2html-init. And try the second l2h menu first.
To change the font size of the current document, edit the index.css
file in the directory of the web pages and set the desire body font
size. Do not put a blank before “px”. You
also want to change the sizes of the headings H1 through H6, captions,
small and large styles,
table cells, and verbatim stuff correspondingly.
In addition, edit the file .latex2html-init in the directory
with the index.tex file and change the value of
$MATH_SCALE_FACTOR according to the formula given there.
(The .latex2html-init file is hidden on Unix.)
Delete all images in the directory with the web pages and rerun
l2h or
makel2h. Repeat for the web page directory of the second
browser, if any, copying the index.css file over from the first.
To make the changes apply to future documents, instead edit
index.css and .latex2html-init in the directory
C:\l2hsup\files.
Note that if you make the font size much less than 15px, subscripts may become ink splots, and at about 8px, ordinary text becomes splots.
The vertical alignment of inline images is computed based on data for
the Arial font family. If you select a body font that is very
different from Arial, you might have to play games with the
scaling factor used to compute font size in C:\l2h\bin\pstoimg.
Alternatively, you can force the font size being passed to l2hcrop directly
by setting the environment variable L2H_FONT_SIZE equal to
the desired font size in px.
The online manual notes that double quotes should not be used as terminators of verbs.
If you are using tildes in a url, you probably want to use the hthtml package instead. Documentation is in the docs directory of l2h.
The online manual notes to use \~{} to put in tildes.
LATEX2HTML is not very strong on redefinitions. Try to contain them within an environment or within curly braces. Or just redefine the variable back to the original value after the need for the change is gone.
In general, encapsulating redefinitions within math environments or other environments send to LATEX for processing seems to keep them out of the main loop.
The end of an environment may not end a font change as it does in LATEX. Simply put additional curly brackets around the font change and its text.
The online manual notes that the font change in {\LARGE $a_b$}
does not work, since the surrounding text is not included in math
processing. So, put it inside the math: {$\mbox{\LARGE $a_b$}$}.
The manual notes that you may sometimes need to enclose the argument in the square brackets also with curly brackets.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. For one, you must change settings.pl, not .latex2html-init.
In LATEX2HTML the index keys are sorted case-insensitive. Worse,
if you have say an entry with sort key “D” followed by
one with sort key “d,” the two entries end up as a
single entry “d”. Since the index coding is a mess, the
only solution I have is to index the second entry as, say,
\index{d-@d|...}.
The short_index option, or $SHORT_INDEX = 1; in
.latex2html-init, is highly recommended if you use section
numbers.
You might want a blank line in front of each starting letter, (to
visually separate different starting letters in the index). The
standard way is to use \index{D| }. But that does not work
for me, as this D is put inside an HTML DT tag even using SHORT_INDEX.
I float the DT tags (labels) in my style sheet to get them on the same
line as the DD data (descriptions) of description environments.
However, \index{D@\HTMLcode{BR}D|...} does work. But you
must of course have entries for each capital letter.
If you want an index entry refering to a sectioning title, put the index command on the same line as the sectioning command. Alternatively, end the sectioning command with a percent sign and put the index command on the next line. If you have more than one index command for the same title, make sure to put them all on the same line. Do not break this line, even using a percent sign to comment away to line break. For some reason, that does not work.
Unlike LATEX, LATEX2HTML bolds the entries in the index. That
may be a problem, for example if you use bold for vectors. I also
think it is ugly. If you do not like it, go into the
latex2html script, find sub add_idx and
remove <STRONG> and </STRONG> from the second line.
Ntabib's \citeindextrue does not work correctly in l2hfu and
versions of l2h up to and including 1.1. This will be fixed in the
next version of l2h, still to be posted at the time of writing, Sep. 1
2013.
If you want links from the bibliography entries to where they are
cited in the text, note that latextohtml does not implement
the hyperref package. So it does not implement its backref option.
However, such links will be made using \citeindextrue when
fixed.
Do not put two \input{...} commands on the same line.
LATEX2HTML will not read the second file.
Somehow, LATEX2HTML confuses \input{...} with
\begin{...}, so make sure that your input file names do not
look anything like latex environments. You will see [the start of]
all your input file names show up in the "unrecognized commands"
listing at the end. They will have been faithfully read, however.
To cut down on the error messages, try starting each input file name with
a z.
Once upon a time, on a world far, far, away from you and me, the wise men of the entire world came together to establish how web pages should understand the words of their designers. They talked much, and thought many deep thoughts, about what the designer would mean with saying that an image should be aligned in the middle of the line, vertically speaking. And their thoughts allowed them to conclude that aligning with the middle of the line clearly should be understood to mean, aligning with the bottom of the line, rather than the middle of it. And so it was done.
But a small band of malcontents thought their browsers might grab a bigger audience if it would align the middle of images with the middle of the line, instead of its bottom. And so they did it. And so they grabbed pretty much all of the audience.
Still, not everybody was happy. A very rich, but not very bright, kid noticed that everyone was having a good time with browsers, and that he did not have any piece of the action, having been unaware that something was up earlier. He thought long about how he could get in on the action, which did not help. He designed his own browser, but it was not very good, since he did not understand well where the middle of a line is supposed to be, and people did not like it. And he became very angry and decided to use his riches to give away his browser for free. He hired thugs to ensure that even people who did not want his browser got one for free or by force. And he paid for vast amounts of web pages to be created with poor formatting. This was really pretty clever, if you think of it, since while it did not improve his browser's appearance, it made other browsers look bad too. So people no longer bothered with the browsers that were not free.
And in this way the rich little kid made web pages mediocre for everyone with his browser, which was a pretty good feeling. But it got pretty boring soon, so the rich little kid went on to spread mediocrity elsewhere.
And new, more interesting browsers were then created by people who still liked their images to be centered in the middle of the line. And some poor kids got all excited about improving the new browsers. And one of them noticed the words the wise men had spoken many years ago. She saw it all. “Aligning images in the middle of the line means aligning them to the bottom of the line!”, she exclaimed to her friends, “The wise men have decreed it!”
And so it came about that half a score of years after the wise men had spoken, their words of wisdom were finally heard and adhered to. And it was established all over that world that the middle is the bottom. Just like for people.
Ahem. I now see what you mean. Since the capabilties of HTML, even its derivatives, in producing good math formulae, are abysmal, LATEX2HTML uses LATEX to create proper formulae, and then converts them to images. A great idea, but of course, those images have to be put in the right place in the web page. No big deal if they are displayed formulae, but tricky if the math appears in the middle of a sentence. Horizontally speaking, it is still not much of a problem; browsers simply put them in the place whereLATEX2HTML says to put them.
The difficulty is the vertical position of the image compared to the
surrounding line. For a math symbol such as 
this is not much
of a problem; when properly directed, the browser will align the
bottom of with the bottom of the line. Or rather, with the
“baseline”; the bottom of letters such as a, b, and c,
that don't have appendages sticking down. Definitely, bottom
alignment with the baseline is not a problem. (Except under some
conditions for Internet Explorer, which is another story.)
While is OK, where should the browser put an image of the
letter 
, descending below the baseline? The browser does not
understand the meaning of the image, and so has no clue what part of
it should be at the height of the baseline if it is not the
bottom. The original LATEX2HTML cleverly solved this conundrum by
adding enough blank space so that if the image is properly centered
vertically (instead of bottom-aligned), the image will be
at the right height in the line.
Unfortunately, browsers do not manage to properly center images vertically in the line. The way to do it was specified in international standards, but they were ignored by all browsers I have seen.
My own initial take on the matter was that of a realist; I just measured where the various leading browsers did put the image vertically, and then sized the image so that it would end up at about the right height. Which was a significant improvement.
Yet there was a difficulty. The practical standard of where to vertically center images was set by Mosaic and Netscape. Yet, when Microsoft entered the browser market with Internet Explorer, they did not just ignore the internationally agrred upon standards, like everyone else, they also ignored the reigning standard established by Mosaic and Netscape. I have the suspicion that the real reason was that the Microsoft software designers did not really know where the baselines of their lines were. It appears that they only kept track of where the overall highest and lowest points in their lines were, and that they averaged the two to get some sort of approximate “baseline”. Of course, for very large images, the highest and lowest points are the image itself, and those images are aligned with their own center, a singular problem that led Mircrosoft to a “baseline” that is the center of the image. Gee, just like the international standard stipulate! Unfortunately, this is only true for very large images. Realistically, if the image is at the right height in Internet Explorer, it is not in Mosaic, Netscape, Mozilla, or Firefox, and vice versa.
So I set things to create middle-aligned images of two different sizes, one for Explorer, and one for the rest of the world. Not ideal, but it seemed the best I could do. Then Firefox 1.5 threw a monkey wrench into that idea too, by deciding out-of-the-blue to suddenly follow the international standards. A good idea, but much, much, too late in my view. I had now ended up with the need of sizing math images three different ways.
It got me to think about style sheet alignment. Style sheets are a later development in HTML, specifying in greater detail the visual appearance of web pages. In other words, style sheets appeared when browser designers might have developed a vague clue what was going on and might have gotten some of it right. And indeed, it turns out that style sheet alignment is much more uniform among browsers than the normal version. Enough to create a single size of images and have them OK for all browsers.
In fact, the international standard for style sheet aligment is very clear: the center of the image must be aligned with the center of the lower case letter “x”. True, Internet Explorer thinks the middle of the letter x is two third of the way up, instead of where the lines cross; they probably got confused with the capital letter X. But still, the differences between browsers remain small if you average them. So, I switched back to one type of images for all, though you can still generate the other versions if you want alignment to be more precise.
Note: Much of what is in this section is outdated. For current software and its usage instructions go to either http://www.eng.fsu.edu/~dommelen/l2h or http://www.dommelen.net/l2h. In particular, only the a-style is currently still supported.
Right. So, as explained in the last section, optimal vertical alignment of math images requires different size images for different browsers. The following sizings are available:
