how to turn a DVI to tex?

Question

Any idea how to take DVI files and turn them into tex?

something similar was today: http://stackoverflow.com/questions/1620002/pdf-to-latex-linux — P Shved, Oct 25 '09 at 19:53
There is a related question about this in TeX.stackexhange: http://tex.stackexchange.com/q/46779/10944 — madth3, Jan 31 '13 at 18:19

score 10 · Answer 1 · answered Oct 25 '09 at 22:45

10

This is similar to the problem of turning PDF into XML which is referred to as "trying to turn a hamburger back into a cow". Both TeX->DVI and XML->PDF lose information, both in the structure of the document and its semantics.

It requires a great deal of heuristics and a large corpus to recreate (some of) the original document. It is never usually 100%. The text strings may be possible, the vectors are harder. Bitmaps are almost impossible.

answered Oct 25 '09 at 22:45

peter.murray.rust

37,407
44
153
217

2

Oh, I really like the hamburger/cow picture! Very very descriptive. – Boldewyn Oct 26 '09 at 13:46
@Boldewyn I got it from Mike Kay (Saxon) but he got it from somewhere else I thinnk – peter.murray.rust Oct 26 '09 at 13:53

score 4 · Answer 2 · answered Oct 25 '09 at 19:55

What you are asking is not possible. I think that (same as in PostScript) even recognizing words in a DVI files may require heuristics. A DVI file is a description of where to place individual letters on a piece of paper, and nothing more.

You can get partway there by either dvi2tty, or by running dvips followed by ps2ascii, whichever gives the best results.

score 4 · Answer 3 · answered Oct 26 '09 at 20:13

4

There's also catdvi, dvitype, and dvi2tty, available from ctan.

answered Oct 26 '09 at 20:13

lhf

70,581
9
108
149

score 2 · Answer 4 · answered Oct 25 '09 at 19:53

2

I am pretty sure this is not possible. DVI contains informations about rendering the page and not which tex commands it has.

answered Oct 25 '09 at 19:53

Mouk

1,807
2
18
26

I think there should be no doubt that this is possible. The issue is can it be done well enough to be worthwhile. – Charles Stewart Dec 07 '09 at 13:49

score 2 · Answer 5 · answered Nov 12 '09 at 09:59

2

for whom ever finds this question again, or for all you who answered I found the best answer for me: what I was looking for is how indeed difficult, it's trying to figure out what could be an original tex that would compile to a given DVI (or pdf for that matter since i can turn the DVI into pdf easily). and InftyReader does it. it works prefect, i tried i a bunch of pdfs on it and then re-made them into pdfs and it was perfect!

answered Nov 12 '09 at 09:59

jarer

71
1
3

Yes, good call! OCR systems tend not be smart about linebreaks though: have you looked at how it handles multi-line equations. – Charles Stewart Dec 07 '09 at 14:06

score 0 · Answer 6 · answered Oct 26 '09 at 13:42

0

Read Description of the DVI file format and write the programm. Result of your program will not be original text but it will be suitable.

answered Oct 26 '09 at 13:42

Alexey Malistov

26,407
13
68
88

score 0 · Answer 7 · edited May 23 '17 at 12:26

Err, well, sort of.

The path of least resistance will involve, I think, a dvi->rtf convertor. I've posted a question: Q#1859373 dvi2rtf: who can convert DVI files to RTF. And there I post an untested implementation, which gives a bad solution that throws away all formatting.

With such a thing, then you could use word2007/8 and the excellent docx2tex utility to turn the rtf to tex.

The results would be unpleasant to read, but I can see some use cases for doing such.

how to turn a DVI to tex?

7 Answers7

Linked