equate 0.0.2

Guessing the pip install name from a pip package name, and other related analyses

Equate

This is a package with tools for matching things.

Dirty things, like language, files in your file system, socks and whistles.

Install

pip install equate

Moving on...

In search of a import-to-package name matcher

The problem

Ever got an import error and wondered what the pip install package name was.

Say...

ImportError: No module named skimage

But it ain't pip install skimage is it (well, it USED to not to, but you get the point...). What you actually need to do to install (with pip) is:

pip install scikit-image

I would have guessed that!

So no, it's annoying. It shouldn't be allowed. And since it is, there should be an index out there to help out, right?

pip install --just-find-it-for-me skimage

Instead of just complaining, I thought I'd throw some code at it. (I'll still complain though.)

Here's a solution: Ask the world (of semantic clouds -- otherwise known as "Google") about it...

A (fun) solution

import requests
import re
from collections import Counter

search_re = re.compile('(?<=pip install\W)[-\w]+')

def pkg_name_options(query):
    r = requests.get('https://www.google.com/search', params={'q': f'python "pip install" {query}'})
    if r.status_code == 200:
        return Counter(filter(lambda x: x != query, p.findall(r.content.decode('latin-1')))).most_common()

def best_guess(query):
    t = pkg_name_options(query)
    if t:
        return t[0][0]

>>> pkg_name_options('skimage')
[('scikit-image', 5),
 ('-e', 2),
 ('virtualenv', 1),
 ('scikit', 1),
 ('scikit-', 1),
 ('pillow', 1)]

>>> best_guess('skimage')
'scikit-image'

Yay, it works! With a sample of one! Let's try two...

>>> pkg_name_options('sklearn')
[('numpy', 3), ('scikit-learn', 2), ('-U', 2), ('scikit-', 1), ('scipy', 1)]

Okay, so it already fails.

Sure, I could parse more carefully. I could dig into the webpages and get more scope.

That'd be fun.

But that's not very nice too Google (and probably is illegal, if anyone cares).

What you'll find next is an attempt to look at the man in the mirror instead. Looking locally, where the packages actually are: In the site-packages folders...

Extract, analyze and compare site-packages info names

import pandas as pd
import numpy as np

from equate.site_names import (
    DFLT_SITE_PKG_DIR,    
    site_packages_info_df,
    print_n_null_elements_in_each_column_containing_at_least_one,
    Lidx,
)

>>> DFLT_SITE_PKG_DIR
'/Users/twhalen/.virtualenvs/382/lib/python3.8/site-packages'

>>> data = site_packages_info_df()
>>> print(f"{data.shape}")
(303, 8)
>>> data

<style scoped> .dataframe tbody tr th:only-of-type { vertical-align: middle; }

.dataframe tbody tr th {
    vertical-align: top;
}

.dataframe thead th {
    text-align: right;
}

</style>

	dist_info_dirname	info_kind	dist_name	most_frequent_record_dirname	first_line_of_top_level_txt	installer	metadata_name	pypi_url_name
0	xlrd-1.2.0.dist-info	dist-info	xlrd	xlrd	xlrd	pip	xlrd	None
1	boltons-20.2.0.dist-info	dist-info	boltons	boltons	boltons	pip	boltons	None
2	appdirs-1.4.3.dist-info	dist-info	appdirs	appdirs	appdirs	pip	appdirs	None
3	yapf-0.29.0.dist-info	dist-info	yapf	yapftests	yapf	pip	yapf	None
4	cmudict-0.4.4.dist-info	dist-info	cmudict	cmudict	cmudict	pip	cmudict	None
...	...	...	...	...	...	...	...	...
298	simplegeneric-0.8.1.dist-info	dist-info	simplegeneric	simplegeneric	simplegeneric	pip	simplegeneric	None
299	plotly-4.6.0.dist-info	dist-info	plotly	plotly	_plotly_future_	pip	plotly	None
300	rsa-3.4.2.dist-info	dist-info	rsa	rsa	rsa	pip	rsa	None
301	backcall-0.1.0.dist-info	dist-info	backcall	backcall	backcall	pip	backcall	None
302	cantools-33.1.1.dist-info	dist-info	cantools	cantools	cantools	pip	cantools	None

303 rows × 8 columns

>>> print_n_null_elements_in_each_column_containing_at_least_one(data)
most_frequent_record_dirname:	1 null values
first_line_of_top_level_txt:	6 null values
installer:	32 null values
metadata_name:	1 null values
pypi_url_name:	255 null values

>>> lidx = Lidx(data)
>>> df = data[lidx.no_nans]
>>> print(f"no nan df: {len(df)=}")
no_nans: 302
equal: 187
dash_underscore_eq: 220
('equal', 'dash_underscore_eq'): 186

>>> lidx = Lidx(df)
>>> lidx.print_diagnosis()
no_nans: 302
equal: 187
dash_underscore_eq: 220
('equal', 'dash_underscore_eq'): 186

>>> lidx = Lidx(df, 'first_line_of_top_level_txt')
>>> lidx.print_diagnosis()
no_nans: 297
equal: 182
dash_underscore_eq: 214
('equal', 'dash_underscore_eq'): 181

>>> t = Lidx(df, 'most_frequent_record_dirname')
>>> tt = Lidx(df, 'first_line_of_top_level_txt')
>>> sum(t.equal | tt.equal)
199

>> sum(t.dash_underscore_eq | tt.dash_underscore_eq)
233