office-thinner.py

#! /usr/bin/python3
# -*- coding: utf-8 -*-

import os
import sys
import filecmp


def print_usage():
    print("Usage:")
    print("  cd /Applications")
    print("  sudo ./office-thinner.py 'Microsoft Word.app' 'Microsoft Excel.app' 'Microsoft PowerPoint.app' ... ")
    print("")
    print("Or generally, link same files with same paths in other_dir to main_dir")
    print("  ./office-thinner.py main_dir other_dir")


if len(sys.argv) < 3:
    print_usage()
    exit(0)

main_dir = sys.argv[1]
other_dirs = sys.argv[2:]

if not os.path.isdir(main_dir):
    print("%s does not exist." % main_dir)
    print_usage()
    exit(1)


def get_ino(path):
    return os.stat(path).st_ino


def get_common_files_helper(base_dir, dcmp):
    if os.path.islink(os.path.join(dcmp.right)):
        # print "skip dir link %s" % os.path.join(dcmp.left, base_dir)
        return []
    else:
        rs = []
        rs0 = filecmp.cmpfiles(dcmp.left, dcmp.right, dcmp.same_files, False)
        for name in rs0[0]:
            rs.append(os.path.join(base_dir, name))
        # for name in rs0[1]:
        #     print "%s is false same" % os.path.join(base_dir, name)

        for name, sub_dcmp in dcmp.subdirs.items():
            path = os.path.join(base_dir, name)
            rs += get_common_files_helper(path, sub_dcmp)

        return rs


def get_common_files(dir1, dir2):
    return get_common_files_helper("", filecmp.dircmp(dir1, dir2))

def makelink(dir1, dir2):
    common_files = []
    try:
        print("calculating common files...")
        common_files += get_common_files(dir1, dir2)

    except OSError as ex:
        print("Error: %s" % ex.strerror)
        print("Maybe you have installed Office in a non-default location.")

    n_actual_linked = 0
    n_common_files = len(common_files)
    print("n_common_files: %s" % n_common_files)
    i = 0
    for filename in common_files:
        path1 = os.path.join(dir1, filename)
        path2 = os.path.join(dir2, filename)

        if os.path.islink(path1) or os.path.islink(path2):
            print("  ignore symlink %s" % filename)
            pass
        elif get_ino(path1) == get_ino(path2):
            print("  same inode for %s" % filename)
            pass
        else:
            # print "diff inode for %s" % filename

            # os.rename won't create parent dir for me
            # os.renames will prune empty dir from src
            # Let's be lazy to skip backuping...
            os.remove(path2)
            os.link(path1, path2)
            n_actual_linked += 1
        i += 1
        if i % (n_common_files / 10) == 0:
            print('%s files linked...' % i)

    print("actually linked: ", n_actual_linked)
    print("done")


def link_if_exist(dir_app):
    if os.path.isdir(dir_app):
        print("make link for %s" % dir_app)
        makelink(main_dir, dir_app)
    else:
        print("%s not found. skipped" % dir_app)


for the_dir in other_dirs:
    link_if_exist(the_dir)

Will you extend it to support all Office products at the same time ?