Add parser/reader for .mbox files

[minosvasilias]

• Adds parser for .mbox files (MboxParser)
  • This is useful to index dumps of mail inboxes, such as provided by Gmail (takeout.google.com)
  • Can be used as part of SimpleDirectoryReader
  • Supports max_count arg, determining maximum amount of messages to parse per mbox file. Useful for test-runs without using too many tokens.
  • Supports message_format arg, overriding default string formatting for messages. See MboxParser class for default value.
  • Dependencies: beautifulsoup4 (to strip messages of HTML syntax)
• Adds reader for .mbox files (MboxReader)
  • Iterates through a directory reading exclusively .mbox files

Linting

⚠️ One linting error suppressed: mbox_parser.py:66

error: Argument "factory" to "mbox" has incompatible type "Callable[[BinaryIO, bool], Message]"; expected "Optional[Callable[[IO[Any]], mboxMessage]]"  [arg-type]

Would appreciate advice on whether or not this can be worked around differently.

Testing

• Tested in a FastAPI environment using a Gmail .mbox dump.
• MboxReaderDemo notebook added

Future concerns:

• May benefit from testing with dumps from other e-mail providers
• May benefit from further params such as filtering messages by date range

[minosvasilias]

Thanks for the review! Will address comments and provide example screenshots once i'm off my day job. 👍

[minosvasilias]

For an example of this working, here are some requests prompting an index of a single GMail .mbox file with 1000 parsed messages. Some minor info redacted.

[jerryjliu]

thanks @minosvasilias! do you have a screenshot using the reader itself in python code / jupyter notebook? was looking for something like that to give users an example of the usage (i can also get from the jupyter notebook)

[jerryjliu]

i'm cutting a release soon, may just land this now

[minosvasilias]

Awesome, thank you! 🚀

Here the requested notebook screenshot: