FindCaseClash.py #1

#!/bin/env python3
# FindCaseClash.py
# Given an input file with a list of files from case senstive p4d
# Output the dir and file clashses.

# The input file should be the output of:
# sort -f filelist | uniq -Di > filedups.txt

# So all files are in case insensitive sort order


# Python 2.7/3.3 compatibility.
from __future__ import print_function

import sys
from os.path import basename, splitext
from collections import defaultdict

class ClashFinder:
    "Finds new directories in paths - seperate class for ease of testing"

    def __init__(self):
        self.dupCount = defaultdict(int)
        self.lpaths = defaultdict(list)

    def findCommonPartInd(self, parts1, parts2):
        # Returns index for common part
        i = 1
        while i <= len(parts1):
            p1 = "/".join(parts1[:i])
            p2 = "/".join(parts2[:i])
            if p1 != p2:
                return i-1
            i += 1
        return 0

    def checkCommonDirs(self, parts1, parts2):
        # We know parts aren't the same - just confirm equal lengths
        if len(parts1) != len(parts2):
            print("Problem: '%s' '%s'" % ("/".join(parts1), "/".join(parts2)))
            return

        ind = self.findCommonPartInd(parts1, parts2)
        if ind == 0 or ind == len(parts1):
            print("/".join(parts1))
            print("/".join(parts2))
            return
        # Want one more than common part
        p1 = "/".join(parts1[:ind+1])
        p2 = "/".join(parts2[:ind+1])
        self.dupCount[p1] += 1
        self.dupCount[p2] += 1
        self.lpaths[p1.lower()].append(p1)
        self.lpaths[p1.lower()].append(p2)

    # This method returns None if no new directory found, or the directory name
    def checkClash(self, paths):
        if len(paths) != 2:
            print("Error Not2: %s" % ",".join(paths))
            return
        self.checkCommonDirs(paths[0][2:].split("/"), paths[1][2:].split("/"))

    def printClashes(self):
        seen = {}
        for k in self.lpaths.keys():
            for j in self.lpaths[k]:
                if self.dupCount[j] == 1 and j not in seen:
                    print("%s" % (j))
                    seen[j] = 1
                elif j not in seen:
                    print("%d %s/" % (self.dupCount[j], j))
                    seen[j] = 1


if __name__ == '__main__':
    cf = ClashFinder()
    with open(sys.argv[1]) as f:
        lastline = None
        paths = []
        for line in f:
            line = line.rstrip()
            if line.lower() == lastline:
                paths.append(line)
            else:
                if paths:
                    cf.checkClash(paths)
                lastline = line.lower()
                paths = [line]
    cf.printClashes()