This is an attempt to put into a working docker the Maxtract program. The executable are downloaded from official site and are put in a working environment inside a docker.
docker build . -t maxtract
Create the directory data into which you can put some pdf files.
In this repository there is the default test.pdf file from Maxtract.
docker run -v $(pwd)/data:/data -it maxtract:latest ./process_file.sh test.pdf
This should produce a file test.pdf.tex inside the data folder.
-
I'm not the author of Maxtract, I just put together a working image because I needed it.
-
The produced LaTeX files are good but need some tuning by hand to be more usable.
-
If you want to know how the original program works, you should look at the paper or at the code in other repositories. If you want to know how I got it to work, look at the file in
maxtract/process_file.sh. No magic involved, I just read the help of all given commands and figured out a way to get it to work. -
To compile the resulting tex file you can use as preamble:
\documentclass[a4paper,11pt]{article} \usepackage{ifthen} \usepackage{amsmath,amssymb} \begin{document}Then insert the whole tex file produced and after it:
\end{document}
By default we use the latex output, but you can change this using the settings of the `process_file.sh` script.
In particular the script input is:
```
process_file.sh PDF_FILE_TO_PROCESS DRIVER LAYOUT TYPE
```
where the defaults are `DRIVER=latex2`, `LAYOUT=latex`, `TYPE=LINES`.
Be aware that many combinations do produce errors and so look carefully at the terminal output.
For convenience the relevant help of `anderson_post.opt` is reported here (do not ask me what they mean because I don't have an idea):
```
$ ./anderson_post.opt -drivers
dummy: A dummy expression driver that serves as default.
festival: A basic driver for the Festival Speech Synthesis System based
exclusively on character information.
index: A driver to provide XML annotation for Indexing.
latex: A simple LaTeX driver based exclusively on character information.
(Retained for compatibility!)
latex2: A complex LaTeX driver, using font and spacing information.
(Retained for compabitility!)
latex_com: A complex LaTeX driver, using font and spacing information.
latex_med: A medium LaTeX driver, using some font and size information.
latex_sim: A simple LaTeX driver based exclusively on character information.
mathml: A simple MathML driver based exclusively on character information.
pdf: A LaTeX driver that adds Latex and MathML annotations for math
expressions in PDF.
pdf_latex: A LaTeX driver that adds Latex annotations for math expressions in
PDF.
pdf_mathml: A LaTeX driver that adds MathML annotations for math expressions
in PDF.
word: A simple Word driver ignoring all Math.
xml: A simple XML translation with linearity and minimal argument tags
xml_bbox: A complex XML translation with linearity, minimal argument tags
and bounding box information for all sub-expressions.
xml_tree: A simple XML translation faithful to the tree structure
$ ./anderson_post.opt -layouts
dummy: A dummy layout driver that serves as default.
Implements:
festival: Layout analysis for Festival documents.
Implements: EXPR EXPRS LINE LINES AREA PAGE DOC
festival_latex: Layout analysis for Latex documents to simple text.
Implements: EXPR EXPRS CEXPRS LEXPRS LINE LINES AREA PAGE DOC
gts: Ground truth data.
Implements: EXPR EXPRS LINE LINES AREA PAGE DOC
index: Layout analysis to provide XML annotation for Indexing.
Implements: EXPR EXPRS LINE LINES AREA PAGE DOC
latex: Layout analysis for Latex documents.
Implements: EXPR EXPRS CEXPRS LEXPRS LINE LINES AREA PAGE DOC
legacy: Invoking legacy drivers.
Implements: MKM09-latex MKM09-mathml DAS10-latex EuDML-latex
EuDML-mathml ICDAR11-latex ICDAR11-gts
mathml: Layout analysis for Mathml documents.
Implements: EXPR EXPRS CEXPRS LINE LINES AREA PAGE DOC
pdf_both: Multilayer PDF documents containing LaTeX code and normal text
underneath.
Implements: EXPR EXPRS LINE LINES AREA PAGE DOC
pdf_code: Multilayer PDF documents containing LaTeX code underneath.
Implements: EXPR EXPRS LINE LINES AREA PAGE DOC
pdf_text: Multilayer PDF documents containing plain text underneath.
Implements: EXPR EXPRS LINE LINES AREA PAGE DOC
tralics: Translation to MathML using Tralics via a given Latex driver.
Implements: LINE LINES AREA PAGE DOC
word: Getting Words out of documents, etc.
Implements: LINE LINES AREA PAGE DOC
xml: Layout analysis for Latex documents.
Implements: EXPR EXPRS CEXPRS LEXPRS LINE LINES AREA PAGE DOC
$ ./anderson_post.opt -types
AREA: Parsing an area.
Implemented by drivers: festival festival_latex gts index latex mathml
pdf_both pdf_code pdf_text tralics word xml
CEXPRS: Parsing expressions for comparison with image files.
Implemented by drivers: festival_latex latex mathml xml
DOC: Parsing an entire document.
Implemented by drivers: festival festival_latex gts index latex mathml
pdf_both pdf_code pdf_text tralics word xml
EXPR: Parsing single Math expressions.
Implemented by drivers: festival festival_latex gts index latex mathml
pdf_both pdf_code pdf_text xml
EXPRS: Parsing several Math expressions.
Implemented by drivers: festival festival_latex gts index latex mathml
pdf_both pdf_code pdf_text xml
LEXPRS: Parsing expressions separately and including separate files into a
master file (rather than including the actual expressions).
Implemented by drivers: festival_latex latex xml
LINE: Parsing a single line.
Implemented by drivers: festival festival_latex gts index latex mathml
pdf_both pdf_code pdf_text tralics word xml
LINES: Parsing several lines.
Implemented by drivers: festival festival_latex gts index latex mathml
pdf_both pdf_code pdf_text tralics word xml
PAGE: Parsing a page.
Implemented by drivers: festival festival_latex gts index latex mathml
pdf_both pdf_code pdf_text tralics word xml
Some drivers might implement other, undocumented layout types.
```
Have fun with such thing.
If you are one of the author of the original project, please provide the community with working source files as well as drivers description and differences.