Strip tags from HTML, optionally from areas identified by CSS selectors

See llm, ttok and strip-tags—CLI tools for working with ChatGPT and other LLMs for more on this project.

Install this tool using pip:

pip install strip-tags

Pipe content into this tool to strip tags from it:

cat input.html | strip-tags > output.txt

Or pass a filename:

strip-tags -i input.html > output.txt

To run against just specific areas identified by CSS selectors:

strip-tags '.content' -i input.html > output.txt

This can be called with multiple selectors:

cat input.html | strip-tags '.content' '.sidebar' > output.txt

To return just the first element on the page that matches one of the selectors, use --first:

cat input.html | strip-tags .content --first > output.txt

To remove content contained by specific selectors - e.g. the <nav> section of a page, use -r or --remove:

cat input.html | strip-tags -r nav > output.txt

To minify whitespace - reducing multiple space and tab characters to a single space, removing any remaining blank lines - add -m or --minify:

cat input.html | strip-tags -m > output.txt

You can also run this command using python -m like this:

python -m strip_tags --help

When passing content to a language model, it can sometimes be useful to leave in a subset of HTML tags - <h1>This is the heading</h1> for example - to provide extra hints to the model.

The -t/--keep-tag option can be passed multiple times to specify tags that should be kept.

This example looks at the <header> section of https://datasette.io/ and keeps the tags around the list items and <h1> elements:

curl -s https://datasette.io/ | strip-tags header -t h1 -t li

<li>Uses</li>
<li>Documentation Docs</li>
<li>Tutorials</li>
<li>Examples</li>
<li>Plugins</li>
<li>Tools</li>
<li>News</li>
<h1>
    Datasette
</h1>
Find stories in data

All attributes will be removed from the tags, except for the id= and class= attribute since those may provide further useful hints to the language model.

The href attribute on links, the alt attribute on images and the name and value attributes on meta tags are kept as well.

You can also specify a bundle of tags. For example, strip-tags -t hs will keep the tag markup for all levels of headings.

The following bundles can be used:

• -t hs: <h1>, <h2>, <h3>, <h4>, <h5>, <h6>
• -t metadata: <title>, <meta>
• -t structure: <header>, <nav>, <main>, <article>, <section>, <aside>, <footer>
• -t tables: <table>, <tr>, <td>, <th>, <thead>, <tbody>, <tfoot>, <caption>, <colgroup>, <col>
• -t lists: <ul>, <ol>, <li>, <dl>, <dd>, <dt>

You can use strip-tags from Python code too. The function signature looks like this:

def strip_tags(
    input: str,
    selectors: Optional[Iterable[str]]=None,
    *,
    removes: Optional[Iterable[str]]=None,
    minify: bool=False,
    remove_blank_lines: bool=False,
    first: bool=False,
    keep_tags: Optional[Iterable[str]]=None,
    all_attrs: bool=False
) -> str:

Here's an example:

from strip_tags import strip_tags

html = """
<div>
<h1>This has tags</h1>

<p>And whitespace too</p>
</div>
Ignore this bit.
"""
stripped = strip_tags(html, ["div"], minify=True, keep_tags=["h1"])
print(stripped)

Output:

<h1>This has tags</h1>

And whitespace too

Use remove_blank_lines=True to remove any remaining blank lines from the output.

Usage: strip-tags [OPTIONS] [SELECTORS]...

  Strip tags from HTML, optionally from areas identified by CSS selectors

  Example usage:

      cat input.html | strip-tags > output.txt

  To run against just specific areas identified by CSS selectors:

      cat input.html | strip-tags .entry .footer > output.txt

Options:
  --version             Show the version and exit.
  -r, --remove TEXT     Remove content in these selectors
  -i, --input FILENAME  Input file
  -m, --minify          Minify whitespace
  -t, --keep-tag TEXT   Keep these <tags>
  --all-attrs           Include all attributes on kept tags
  --first               First element matching the selectors
  --help                Show this message and exit.

To contribute to this tool, first checkout the code. Then create a new virtual environment:

cd strip-tags
python -m venv venv
source venv/bin/activate

Now install the dependencies and test dependencies:

pip install -e '.[test]'

To run the tests:

pytest