organisciak / bookwormpython

Python library for working with a Bookworm server

Hi there 👋, I'm Peter Organisciak!

• 🎓 I'm a tenured Associate Professor at the University of Denver, focusing on creativity and AI, and massive-scale text analysis.
• 🛠️ Currently on a research sabbatical until September 2024, I'm engaged in building new and interesting things and open to new partnerships or short-term collaborations.
• 💡 Ask me about my work with Large Language Models in creativity scoring, or working with relationship tagging of massive digital libraries.
• 🙋 Pronouns: identify He/Him, fine with ungendered alternatives
• 🌐 Visit my website at porg.dev for more about my work and access to my online tools.
• 📄 My online CV, and recent publications at Google Scholar.
• 🎨 Something fun: try this small color composition game: Hue Hunter (hues.red).

bookwormpython's Issues

Support multiple search_limits

Not currently supported - multiple searches in one query:

                  {"word":["tea"],"date_year":{"$gte":1760,"$lte":2010},"publication_country__id":["2"]},
                  {"word":["coffee"],"date_year":{"$gte":1760,"$lte":2010},"publication_country__id":["2"]}
]```

It can run if you turn off validation, but the parsing code in `BWResults` wouldn't know what to do with it.

rename counttype to count_type

Allow non-list counttype

Python 2 support

Should work in Python 2, but I've been neglecting testing it. Mostly a note-to-self.

Pip build

limited_field_values

Need a method that tells you all the possible fields for the current search limit.

compare_limits warps results

Remove it from the default.

See: Bookworm-project/BookwormDB#144

Use return_pandas method?

This might be better filed as an issue on BookwormAPI called "Should return_pandas method be retired?", but I just want to get your thoughts without spamming everyone, so filing it as a feature request insincerely.

You say here that JSON is "more transparent and safer."

    def _fetch(self, query, type="pandas"):
        ''' Get results from a bookworm Server
            This method calls JSON and converts to Pandas, rather than using
            Bookworm's built-in DataFrame return method, as JSON is a more
            transparent and safer format for data interchange.

Care to elucidate? If you're right, it might make sense simply to retire the return_pandas method entirely (it's never been fully documented): the two uses cases for it I can think of are

1. This one: a numpy based set of code for using the API
2. As the internal interchange format in a meta-Bookworm aggregating across servers.

If it's not safe enough to use for 1, as you imply, it might not be safe enough for number 2 either. So what are the safety issues involved?

Test suite

conda support

insecure connection error

Error:
InsecureRequestWarning: Unverified HTTPS request is being made. Adding certificate verification is strongly advised. See: https://urllib3.readthedocs.io/en/latest/advanced-usage.html#ssl-warnings InsecureRequestWarning)

plot() in BWResults

Automatically sort frame in BWResults by first group

I expect the most common use case will want results sorted by counts rather than key names.