The companion Perl script, tex_to_html.pl, can be used to produce a collection of html files, one per chapter and one per section, to present the book on the web. The script works primarily by doing pattern matching in the $\TeX$ source files and translating to html. This is not particularly robust, so that more-or-less equivalent $\TeX$ source will not always provide good html output, and perfectly reasonable book sources may not render properly if they go beyond what the script knows about.

Many $\TeX$ macros are simply ignored by the script; if you use a macro that the script doesn't know how to translate and doesn't know to ignore, it will probably show up as the literal macro name in the output. If it should be ignored, you can wrap it in a texonly environment:

\texonly
\unknownmacro...
\endtexonly

There is a similar \htmlonly…endhtmlonly environment; the contents will be ignored by $\TeX$ but processed by tex_to_html.pl. This is very useful for providing html versions of figures, which may, for example, be interactive figures. The script includes support for JSXGraph, three.js, and Sagecells in the html files it produces. A variety of Javascript and CSS files are needed for support.

The script assumes the book is produced in accord with the format described here, but in addition, certain guidelines must be followed. The chapter files must be named chapterxx.tex, where xx is the chapter number, and likewise section files are sectionxx.tex. The main $\TeX$ file must include a sequence of lines like \input chapterxx. Each chapter file must include optionally some introductory material, and then a sequence of lines like \input sectionxx. Subsections are to be contained within the section files.

If the main file contains \input preface before \input chapter01 it will be translated as if it were a chapter, and appear before chapter 1.

Other $\TeX$ files, like appendices, are ignored. These can be inserted by telling tex_to_html.pl that they are chapters, as described below.

The script is invoked as

tex_to_html.pl mainfile.tex output_directory

For my books, I have a subdirectory of the main book directory called html, which contains soft links to all the needed $\TeX$ files, at least the main file and the chapter and section files. Then I invoke the script as

tex_to_html.pl html/mainfile.tex output_directory

The script assumes that there is a subdirectory called html_figures (so in my case this would be a subdirectory of html) containing html code that will be incorporated into the output verbatim; this is primarily used to include figures.

The script also assumes that there is a file called front_page.html which will be the first page in the book. This might contain copyright and contact information, for example.

To include an appendix you must make it appear to the script to be a chapter. First, create a pointer from the html subdirectory to the appendix $\TeX$ file, but name the link chapterxx.tex, where of course xx is one larger than the last chapter of the book. Then in the main $\TeX$ file, include this where the appendix file was input:

\htmlonly
\input chapterxx
\endhtmlonly

You will also presumably need to modify appendix.tex to include something like

\texonly
\appendix{Appendix Title}
\endtexonly
\htmlonly
\chapter{Appendix Title}{}
\endhtmlonly

Then the macro \appendix{Appendix Title} will only be processed by $\TeX$ and \chapter{Appendix Title}{} will only be processed by tex_to_html.pl.

The script was originally written to process an open source calculus book, which comes in multiple flavors: single variable or multivariable, early transcendentals or late transcendentals. The script also knows about these; the default is to produce the multivariable, early transcendentals version. This can be changed at the command line:

tex_to_html.pl -s

will produce the single variable version, and

tex_to_html.pl -l

will produce the late transcendentals version. Both switches may be given. The versions are identified in the $\TeX$ source using something like this:

\iflatetranscendentals
...
\elselatetranscendentals
...
\filatetranscendentals

and

\ifmultivariable
...
\elsemultivariable
...
\fimultivariable

These macros are not included in bookmacros.tex; they are in calcmacros.tex in the source for the calculus book. These could be repurposed for processing any variants of a book, although it would make more sense to use different names for the macros, and then to alter the script to match.

The html pages will include a search box, which uses Google to do the searching. For this to work properly, you will need to provide the correct search path for the document in a file called search_path.txt in the directory containing the html input files; in my case, this is in the subdirectory called html. The file should consist of a single line giving the site url, something like www.whitman.edu/mathematics/book_macros/.