Python script for cleaning LaTeX sources for dissemination. Helpful for producing camera-ready versions or arXiv pre-prints.
TLDR:
- Parses a LaTex source file and produces a new directory with a cleaned version
- Distills a single file with matching BibTeX entries
- Renames figures and copies directly included resources from TiKz figures
- Can crop fragments of larger images to reduce file size
- Can compile standalone figures and include them as PDFs instead
- Not thoroughly tested, and possibly buggy but saves me time
usage: tex-camera-ready.py [-h] [-o OUTPUT] [-c] [-v] [-f] [-b] filename
LaTeX source cleanup: take a working LaTeX sources and export a copy for
dissemination (with resource refactoring, bibtex items selection, etc.)
positional arguments:
filename input file (*.tex)
optional arguments:
-h, --help show this help message and exit
-o OUTPUT, --output OUTPUT
Output directory, default: ./final
-c, --crop Crop bitmaps based on LaTeX trim parameters
-v, --verbose Print analysis summary to stdout
-f, --force Force output to an existing directory
-b, --bib Cleanup Bibtex entries (leave only cited)
-t, --tikz Compile standalone TikZ figures and include resulting
PDFs
> python3 tex-camera-ready.py -b document.tex
Loaded 588 lines from document.tex
Writing to ./final_new
+ figure 01 : ./figures/Architecture.pdf
> includes/figure_01.pdf
+ figure 02 : ./figures/Pipeline.pdf
> includes/figure_02.pdf
+ figure 03 : ./figures/Results.tex
> includes/figure_03.tex
Refactoring file ./figures/Results.tex -> ./final_new/includes/figure_03.tex
../figures/Results-000.png
../figures/Results-001.png
../figures/Results-002.png
../figures/Results-003.png
Missing dependencies (you may need to handle them manually):
+ figures/boxplot_accuracy.tex
\inetdata
Found 40 citations:
[1] Chen2008
...
[40] Zhou2018
Found 3 Bibtex databases: ['../bib/surveys', '../bib/general', '../bib/software']
Parsing ../bib/surveys
Parsing ../bib/general
Parsing ../bib/software
Matched 40 entries
.
├── bib
│ └── references.bib
├── document.tex
├── includes
│ ├── figure_01.pdf
│ ├── figure_02.pdf
│ └── figure_03.tex
└── resources
└── figure_03
├── Results-000.png
├── Results-001.png
├── Results-002.png
└── Results-003.png
I extensively use the standalone package for figures. In order to ensure correct building of both the main document, and the figures (when built independently), I define new commands to handle path differences. In the figure *.tex while, this would be:
\newcommand*{\FigPath}{../figures/}
...
\includegraphics[width=2in]{\FigPath schematic/graphics.jpg}}
In the main document, this becomes:
\newcommand*{\FigPath}{./figures/}
The tool will refactor \FigPath and \DataPath commands.
LaTeX allows you to trim included graphics files, but the bitmaps embedded in the PDF are actually full size. In order to reduce file size, this tool can crop the files for you. The graphics will be opened, trimmed, and saved in a new location. The parser looks for a trimming definition ,trim=A S D F,clip, so the following line:
\subfloat[Thumbnails]{\includegraphics[width=\columnwidth,trim=10 10 10 10,clip]{figures/graphics.jpg}}
will be replaced with
\subfloat[Thumbnails]{\includegraphics[width=\columnwidth]{includes/figure_01.jpg}}
The current implementation requires BibTeX entries to have the final closing bracket in a separate line. Attributes should also be defined in lower case. There might be more limitations. Entries copied from Google Scholar are typically handled correctly.
The tool can compile standalone figures and include them as graphics files (see -t option). In the process, latexmk -pdf will be called, and the resulting PDFs will be used in the include commands (includestandalone becomes includegraphics). For this to work, make sure that latexmk can build your figures with the figure directory as the current dir.