Python script to find po file translations that differ by just a few characters

find_almost_duplicate_translations.py

#!/usr/bin/env python

import polib
import Levenshtein as L

pofile = polib.pofile('po/base.pot')

entries = {}

for entry in pofile:
    entry_len = len((entry.msgctxt or "") + entry.msgid)
    if entry_len not in entries:
        entries[entry_len] = []
    entries[entry_len].append(entry)

max_delta = 2

for entries_len in range(20, 4000):
    if entries_len in entries:
        for i in range(len(entries[entries_len]) - 1):
            entry = entries[entries_len][i]
            for entries_len_2 in range(entries_len, entries_len + max_delta + 1):
                if entries_len_2 in entries:
                    for entry2 in entries[entries_len_2][entries_len == entries_len_2 and i+1 or 0:]:
                        s1 = (entry.msgctxt and (entry.msgctxt + " : ") or "") + entry.msgid
                        s2 = (entry2.msgctxt and (entry2.msgctxt + " : ") or "") + entry2.msgid
                        dist = L.distance(s1=entry.msgid.lower(), s2=entry2.msgid.lower(), score_cutoff=2)
                        if dist > 0 and dist <= max_delta:
                            print("'", s1, "'  |  '", s2, "'", sep="")