Add Configurable HTML Parser Wrappers for BeautifulSoup and Resiliparse

[silentninja]

Fixes #43

This PR introduces wrapper classes that provide a uniform API for common operations across two HTML parsers: BeautifulSoup and Resiliparse. The goal is to simplify switching between the parsers and ensure consistency in usage patterns.

Additionally, the PR adds support for configuring the HTML parser wrapper at runtime by introducing a new --html_parser option when submitting a Spark job. This allows users to specify their preferred parser dynamically during execution.

[silentninja]

We can enhance this further by adding support for third-party HTML parser wrappers.

The idea is to allow users to pass the path to a custom wrapper module as a runtime argument. This approach would enable developers to use community-contributed libraries without requiring modifications to the existing codebase.

Proposed Implementation

• Use Python's importlib to dynamically load the custom wrapper module at runtime.
• Define an abstract base class that all custom wrapper classes must extend to ensure a consistent interface.
• Add runtime validation to verify that the custom wrapper class correctly implements all required methods and adheres to the expected interface.

It would be worth considering this when there is an interest in third-party wrappers

[wumpus]

I like the idea of a wrapper that makes it easy to switch parsers.

[sebastian-nagel]

Hi @silentninja, thanks for your contribution! I'll try it out soon.

[sebastian-nagel]

Have you verified that it it possible to install and use FastWARC and Resiliparse with different versions?

[sebastian-nagel]

There's only one class which uses a HTML parser: CCIndexWordCountJob. Shouldn't the option be moved to the class were it is used?

If it's a global option, there may be confusion, such as "I thought WET files contain plain text from parsed HTML. Why I should specify a HTML parser?"