I need to convert some old student laboratory manual from archaic latex to some structured format like html or epub. From some research into the Web I found that the most advanced tool for that is tex4ht which is by default included in the MikTeX distribution. After some effort (quite a lot in fact) I managed to find a way to run it on my tex file and was relatively satisfied with the result.
Still, the tex4ht is presented as a highly configurable system, but I still can not find any reasonably complete documentation on its configuration. I am coming to a conclusion that there is no complete documentation at all for its features. Does anybody know where to find some help on this matter?
In fact I need to configure following things:
1) Since i am converting formulas to .gif images, I've got quite a lot of them (in the same directory as my tex source and html result). I would like to redirect them to some subdirectory.
2) My source encoding is (sadly) cp1251, and it would be good to have result in unicode, not in iso-8859-5 with a lot of &#x....-s in the page.
3) I understand that tex4ht supports sectioning, index and separate htmls for sections, bu I did not found how to configure this feature.
4) While the gif-s for my formulas are fine in the resulting html, the pictures in the tex are also transformed to gif with some obscure names (which I normally do not need) and are blank (white boxes). I would like to configure the pictures separately form formulas
Does anybody have suggestions on any of these topics?
I have not found a good documentation but I have read somewhere that it is described in the book LaTeX Web Companion.
Can you make a PDF from the LaTeX code? Then the PDF2HTML package can help. It can be found here: http://coolwanglu.github.io/pdf2htmlEX/.
Otherwise have a look at this document which contains references for other documentations: http://www.tug.org/applications/tex4ht/mn.html
I would not go this way. There is no structure in PDF. I have an experience with commercial PDF to EPUB conversion.. Its only advantage is that it makes images that are almost usable, some of them had to be corrected in gimp. Sectioning is lost, I have to go to the HTML files and fix it manually. Computer Modern fonts have quite a lot of kerning pairs which often confuse such tools so that interword spaces are often wrong. In fact, I decided to write my own convertor in perl but it requires highly structured input and does not support all LaTeX features, it can only be configured by writing a few new subroutines and is almost undocumented.
PDF to html conversion only works fine in a rather limited number of cases, especially when the mathematical formulas are not concerned. In my case it does not solve anything at all. Writing one's own script is not a brilliant solution in principle (I'm trying to maintain the position that each thing should be done solidly and by professionals), but in this case it does not work at all. The TeX is very Archaic and uses a lot of custom styles and commands.
I found the book "LaTeX Web Companion" on Google books and it indeed has a 40 pages chapter on tex4ht. Zdenek, thank you very much.
@Alexander: Have a look at the converter I linked. The results are astonishing.
Christian I tried to do as you suggest for curiosity sake. And was a bit surprises. The run the program with default parameters and it just converted each page to a picture and embedded it in html file. It is definitely good looking, but it is not html.
As far as I can understand pdf structure, there could no be good conversion from pdf to html in principle since the data representation principle in these formats is rather different. Therefore I am a bit skeptic about any such converter.
I use TeX4ht with a visually impaired student whose screen reader can speak mathml. I agree that the documentation is rather light, but I've been successful finding information using stackexchange.
http://tex.stackexchange.com/search?q=tex4ht
There is also this project:
https://gist.github.com/stefanozanella/8892211
Yes, simple PDF just dscribes the visual appearance of the page. In principle you can extract the texts and font information, feed it to filter that will analyze the font type and font size and possibly the format of a line (type of numbering) in order to decide whether it is a title and if so, at which level. It will require some nontrivial work most probably for each document unless the LaTeX class was not customized and the default fonts were used. It should be also sent through a spellcheck in order to find hyphenated words and remove interword spaces caused by positive kernings. However, equations containing complex fractions and matrices will be lost. Of course, the PDF files may have metadata, Ross Moore shows how to add them but in order to have such metadata you would have to go to the source LaTeX files and edit them manually.
Since there is still some interest to this question, I would like to point that you probably should forget about tex4ht since it is cumbersome and very badly maintained and instead focus on https://dlmf.nist.gov/LaTeXML/, which uses different approach but have a reasonable documentation and good community on the github.
I have been trying text4ht for a while. There is a lot of issues and it does not include a lot of latex libraries (the most common ones). I had to keep track of all the `.sty` files which become difficult. I am going to try LateXML and see how it goes.
- Badges
- Science topic