Python code to compare the loaded packages in two R sessions

compare_sessions.py

#!/usr/bin/env python3
#coding=utf-8
"""
Save the output of R's `sessionInfo()` into a text file. 
"""
__author__ = 'Simon J. Greenhill <[email protected]>'
__copyright__ = 'Copyright (c) 2017 Simon J. Greenhill'
__license__ = 'New-style BSD'

import re
import codecs

is_counter = re.compile(r"""\[\d+\]""")
is_whitespace = re.compile(r"""\s+""")

def parse_package(package, rver=None):
    package = package.split("_")
    if len(package) == 1:
        package = tuple([package[0], rver])
    return tuple(package)


def parse_session(filename):
    sections = {
        'base': set(),
        'packages': set(),
        'namespace': set(),
    }
    rver = None
    with codecs.open(filename, 'r', encoding="utf8") as handle:
        section = None
        for line in handle.readlines():
            line = line.strip()
            if len(line) == 0:
                continue
            elif line.startswith("R version"):
                rver = is_whitespace.split(line)[2]
                continue
            elif line.strip() == 'attached base packages:':
                section = 'base'
                continue
            elif line.strip() == 'other attached packages:':
                section = 'packages'
                continue
            elif line.strip() == 'loaded via a namespace (and not attached):':
                section = 'namespace'
                continue
            elif section is None:
                continue
            else:
                line = is_counter.sub("", line)
                line = [
                    parse_package(_, rver) for _ in is_whitespace.split(line) if len(_)
                ]
                sections[section].update(line)
    return sections


def match(package, packages):
    try:
        return [p for p in packages if p[0] == package][0]
    except IndexError:
        return None

def get_version(package):
    if package is None:
        return "Not Installed"
    try:
        return package[1]
    except IndexError:
        return None


if __name__ == '__main__':
    import argparse
    parser = argparse.ArgumentParser(description='Does something.')
    parser.add_argument("filename1", help='filename')
    parser.add_argument("filename2", help='filename')
    args = parser.parse_args()

    s1 = parse_session(args.filename1)
    s2 = parse_session(args.filename2)
    
    for section in s1:
        allp = [p[0] for p in s1[section]] + [p[0] for p in s2[section]]
        for p in sorted(allp):
            m1 = match(p, s1[section])
            m2 = match(p, s2[section])
            if m1 != m2:
                print(
                    section.ljust(10), p.ljust(10),
                    get_version(m1).ljust(10),
                    get_version(m2)
                )