panTeX 0.0.1

Quickly build beautiful LaTeX documents from python/pandas

pretty-reports

Generating pretty reports using pandas and matplotlib.

Installation

• pip install pretty-reports to install the main application
• Install MiKTeX (on Windows) or TBD on Linux
• Install pandoc
• Install browser-sync (for browser mode): npm install -g browser-sync

Quickstart via Command Line

To run in browser mode: pretty-reports --browser ./my_template_file.md
To prduce a pretty LaTeX pdf report: pretty-reports ./my_template_file.md
To export context variables in a Python script: save_context({'title': 'Hello World', 'table': df.head()})
To add to context in a Python script: save_context({'title': 'Hello World', 'table': df.head()}, append=True)
To read the current context in a Python script: read_context()

Quickstart via Python

You can also use pantex directly from Python e.g., if you have a business process that pulls data from various sources and summarizes it in a LaTeX pdf document.

How it works

Inspired by traditional website rendering, a report generation process is divided into two parts:

1. A markdown template with Python-native template variables.
2. A context dictionary, which is saved at ./assets/context.pkl

For example, say you have a markdown file at ./my_template.md, containing:

# ${my_header}
${my_table}
Some text

In Python, you can write:

save_context({'my_header': 'Hello World!', 'my_table': df.head()})

pretty-reports will combine the markdown template and the Python context and produce a pretty pdf LaTeX report. Note that pandas DataFrame objects will be automatically rendered as a table. Matplotlib/Seaborn charts and LaTeX style equations are also supported. (The LaTeX equations must be contained in \begin{equation}/\end{equation} tags.)

The Development Server

pretty-reports --browser enables an in-browser version of the LaTeX report. It's not quite as pretty as the pdf version, but it offers a near real time rendering experience. Rendering pdf reports takes a bit more time.

Notes

• I had to add *.md to .prettierignore. LaTeX equations were getting inappropriately edited in vscode.
• $$ works for indicating equations on Linux. However, for Windows, I had to use \begin{equation} and \end{equation}.

Image Formats

matplotlib supports certain image formats. I tried each one and report the results here:

• eps: very nice; text can be highlighted ✔️
• jpeg: just an image; text is blurry & cannot be highlighted❌
• jpg: just an image; text is blurry & cannot be highlighted❌
• pdf: text can be highlighted; however, the marks look distorted when zoomed out❌
• pgf: returned non-zero exit status 43❌
• png: just an image; text is blurry & cannot be highlighted❌
• ps: it ran, the but the chart position is botched in the pdf file❌
• raw: hit an infinite loop❌
• rgba: hit an infinite loop❌
• svg: got an exception could not convert image❌
• svgz: got an exception could not convert image❌
• tif: just an image; text is blurry & cannot be highlighted❌
• tiff: just an image; text is blurry & cannot be highlighted❌

The Stack

• string.Template is used to render a markdown template
• pandas.DataFrame.to_markdown is used to create markdown tables
• pandoc is used to the convert markdown file output into a pdf file (markdown 🡲 LaTeX 🡲 pdf)
• browser-sync (npm module) is used to run the --browser version

Requirements

• See requirements.txt for the Python package requirements
• Also requires pandoc and MiKTeX to be installed (on Windows)
• browser-sync is required for --dev mode

Workflow

• Develop the document using a web page using browser-sync start --server --files "*.html" --index "output.html"
• Print to pdf when you're ready