finding from import statements in python

Question

Let's say I have a module with some code in it and I want to find the imported libraries for this module.

import os
import pandas as pd
import numpy as np
from os.path import basename, join

def export(df, folder):
    """some export thing (not really important)"""
    f = join(folder, "test.csv")
    df.to_csv(f)
    return f

So this is just a dummy example. What the methods do aren't really important in this context of the question

From a previous question, I was able to get the import statements that import modules, not realizing that the from foo import bar will not show.

[(x, y.__name__) for x,y in inspect.getmembers(sys.modules[__name__], inspect.ismodule)]

This will return a tuple:

[('inspect', 'inspect'), ('np', 'numpy'), ('os', 'os'), ('pd', 'pandas'), ('sys', 'sys')]

I can then take that list and if the tuple position 0 != position 1 then it's an import foo as f situation. What is missing are situations where the from os.path impor basename, join and situations where from os import path. How do I include this types of imports?

Thanks

Are you trying do this determination at runtime for some particular reason? Or are you trying to do static code analysis, and just doing it at runtime so you can utilize `inspect` and Python's runtime import information? — PaulMcG, May 29 '18 at 12:41
*why*? you want to find the dependencies of a particular module perhaps? — Chris_Rands, May 29 '18 at 13:19
@Chris_Rands to move code from local machine to a server at runtime. — code base 5000, May 29 '18 at 14:15

Mazdak · Answer 1 · 2018-05-29T13:10:09.103

You need to create a custom predicate function that accepts both modules and function. In that case you'll be able to get the name of all the function as well and then by filtering the members based on your need you can get the desired result.

def get_members(module):
    predicate = lambda x: inspect.ismodule or  inspect.function

    for x,y in inspect.getmembers(module, predicate):
        try:
            md = y.__module__
            if md != "__main__":
                yield (x, y.__name__, y.__module__)
        except:
            if inspect.ismodule(y):
                yield (x, y.__name__)

Demo:

# test.py

import os
import pandas as pd
import numpy as np
from os.path import basename, join
from itertools import product
import inspect
import sys

def export(df, folder):
    """some export thing (not really important)"""
    f = join(folder, "test.csv")
    df.to_csv(f)
    return f

def get_members(module):
    predicate = lambda x: inspect.ismodule or  inspect.function

    for x,y in inspect.getmembers(module, predicate):
        try:
            md = y.__module__
            if md != "__main__":
                yield (x, y.__name__, y.__module__)
        except:
            if inspect.ismodule(y):
                yield (x, y.__name__)    


print(list(get_members(sys.modules[__name__])))

Output:

~/Desktop$ python3 test.py 
[('__builtins__', 'builtins'), ('basename', 'basename', 'posixpath'), ('inspect', 'inspect'), ('join', 'join', 'posixpath'), ('np', 'numpy'), ('os', 'os'), ('pd', 'pandas'), ('product', 'product', 'itertools'), ('sys', 'sys')]

@codebase5000 It's exactly like your code except for imported functions there is an extra `y.__module__`. — Mazdak, May 29 '18 at 15:11

finding from import statements in python

1 Answers1