The feedsdb.py script without any arguments operates as a CGI application. The provided feeds.css provides an example simple styling of the HTML output. Feeds can be added/removed using the form at the bottom of the generated webpage, or via the CLI.

Feeds and the items pulled from them are stored in an sqlite database, by default feeds.db in the current directory.

Dependencies

• Requires the feedparser python library.
• Assumes the user under which the CGI script runs has read/write access to the sqlite database file.
• Assumes that the feeds.css CSS file is served from the same folder as the script.

CLI

There are add, del, and list commands available (see feedsdb.py --help) for managing feeds from the command line.

PDF generation

The CLI command pdf creates a PDF of (a subset of) the articles from a given amount of time before now. It uses playwright and pikepdf. Unless you specify --non-interactive an editor (change by setting the EDITOR env var) will pop up and you can delete any articles you don't want in the PDF.

To get python-playwright you can do:

    pip install playwright        # install playwright-python
    playwright install chromium   # download chromium browser binary

Each article is downloaded using playwright to a temporary file and then they are all stitched together using pikepdf. A ToC/Outline is added to the final PDF with an entry for each article to make it easy to jump to articles.

By default (disable with --no-append) if the output PDF file already exists the selected articles will be appended to it rather than overwriting the whole thing.

The PDF generation calls into link_processor.py to get one or more dict of settings for each article. I've included my link processor as an example to work from, you can do pretty much anything here. feedsdb.py expects the dict to contain 'url' and 'desc' keys (desc is what's shown in the editor interface), and 'toc_label' if you want the article to appear in the PDF ToC.