cyrillic_convert.py Python3 refactor

contrib/cyrillic_convert.py

@@ -1,92 +1,93 @@
-#!/usr/bin/python
+#!/usr/bin/python3
 
 """Convert gibberish back into Cyrillic"""
 
 import fileinput
-import os
+import argparse
 import sys
 
-usage = """
-If you have signs that should be Cyrillic, but are instead gibberish,
-this script will convert it back to proper Cyrillic.
-
-usage: python %(script)s <markers.js>
-ex. python %(script)s C:\\Inetpub\\www\\map\\markers.js
- or %(script)s /srv/http/map/markers.js
-""" % {'script': os.path.basename(sys.argv[0])}
-
-if len(sys.argv) < 2:
-    sys.exit(usage)
-
 gibberish_to_cyrillic = {
-    r"\u00c0": r"\u0410",
+    'À': 'А',
-    r"\u00c1": r"\u0411",
+    'Á': 'Б',
-    r"\u00c2": r"\u0412",
+    'Â': 'В',
-    r"\u00c3": r"\u0413",
+    'Ã': 'Г',
-    r"\u00c4": r"\u0414",
+    'Ä': 'Д',
-    r"\u00c5": r"\u0415",
+    'Å': 'Е',
-    r"\u00c6": r"\u0416",
+    'Æ': 'Ж',
-    r"\u00c7": r"\u0417",
+    'Ç': 'З',
-    r"\u00c8": r"\u0418",
+    'È': 'И',
-    r"\u00c9": r"\u0419",
+    'É': 'Й',
-    r"\u00ca": r"\u041a",
+    'Ê': 'К',
-    r"\u00cb": r"\u041b",
+    'Ë': 'Л',
-    r"\u00cc": r"\u041c",
+    'Ì': 'М',
-    r"\u00cd": r"\u041d",
+    'Í': 'Н',
-    r"\u00ce": r"\u041e",
+    'Î': 'О',
-    r"\u00cf": r"\u041f",
+    'Ï': 'П',
-    r"\u00d0": r"\u0420",
+    'Ð': 'Р',
-    r"\u00d1": r"\u0421",
+    'Ñ': 'С',
-    r"\u00d2": r"\u0422",
+    'Ò': 'Т',
-    r"\u00d3": r"\u0423",
+    'Ó': 'У',
-    r"\u00d4": r"\u0424",
+    'Ô': 'Ф',
-    r"\u00d5": r"\u0425",
+    'Õ': 'Х',
-    r"\u00d6": r"\u0426",
+    'Ö': 'Ц',
-    r"\u00d7": r"\u0427",
+    '×': 'Ч',
-    r"\u00d8": r"\u0428",
+    'Ø': 'Ш',
-    r"\u00d9": r"\u0429",
+    'Ù': 'Щ',
-    r"\u00da": r"\u042a",
+    'Ú': 'Ъ',
-    r"\u00db": r"\u042b",
+    'Û': 'Ы',
-    r"\u00dc": r"\u042c",
+    'Ü': 'Ь',
-    r"\u00dd": r"\u042d",
+    'Ý': 'Э',
-    r"\u00de": r"\u042e",
+    'Þ': 'Ю',
-    r"\u00df": r"\u042f",
+    'ß': 'Я',
-    r"\u00e0": r"\u0430",
+    'à': 'а',
-    r"\u00e1": r"\u0431",
+    'á': 'б',
-    r"\u00e2": r"\u0432",
+    'â': 'в',
-    r"\u00e3": r"\u0433",
+    'ã': 'г',
-    r"\u00e4": r"\u0434",
+    'ä': 'д',
-    r"\u00e5": r"\u0435",
+    'å': 'е',
-    r"\u00e6": r"\u0436",
+    'æ': 'ж',
-    r"\u00e7": r"\u0437",
+    'ç': 'з',
-    r"\u00e8": r"\u0438",
+    'è': 'и',
-    r"\u00e9": r"\u0439",
+    'é': 'й',
-    r"\u00ea": r"\u043a",
+    'ê': 'к',
-    r"\u00eb": r"\u043b",
+    'ë': 'л',
-    r"\u00ec": r"\u043c",
+    'ì': 'м',
-    r"\u00ed": r"\u043d",
+    'í': 'н',
-    r"\u00ee": r"\u043e",
+    'î': 'о',
-    r"\u00ef": r"\u043f",
+    'ï': 'п',
-    r"\u00f0": r"\u0440",
+    'ð': 'р',
-    r"\u00f1": r"\u0441",
+    'ñ': 'с',
-    r"\u00f2": r"\u0442",
+    'ò': 'т',
-    r"\u00f3": r"\u0443",
+    'ó': 'у',
-    r"\u00f4": r"\u0444",
+    'ô': 'ф',
-    r"\u00f5": r"\u0445",
+    'õ': 'х',
-    r"\u00f6": r"\u0446",
+    'ö': 'ц',
-    r"\u00f7": r"\u0447",
+    '÷': 'ч',
-    r"\u00f8": r"\u0448",
+    'ø': 'ш',
-    r"\u00f9": r"\u0449",
+    'ù': 'щ',
-    r"\u00fa": r"\u044a",
+    'ú': 'ъ',
-    r"\u00fb": r"\u044b",
+    'û': 'ы',
-    r"\u00fc": r"\u044c",
+    'ü': 'ь',
-    r"\u00fd": r"\u044d",
+    'ý': 'э',
-    r"\u00fe": r"\u044e",
+    'þ': 'ю',
-    r"\u00ff": r"\u044f"
+    'ÿ': 'я'
 }
+trans_table = {ord(k): v for k, v in gibberish_to_cyrillic.items()}
 
-for line in fileinput.FileInput(inplace=1):
-    for i, j in gibberish_to_cyrillic.iteritems():
-        line = line.replace(i, j)
-    sys.stdout.write(line)
 
+def convert(s):
+    return s.translate(trans_table)
+
+
+if __name__ == '__main__':
+    description = """
+    If you have signs that should be Cyrillic, but are instead gibberish,
+    this script will convert it back to proper Cyrillic.
+    """
+    parser = argparse.ArgumentParser(description=description)
+    parser.add_argument('file', metavar='markers.js')
+
+    args = parser.parse_args()
+    convert(args.markers_file)
+    for line in fileinput.input(files=markers_file, inplace=1):
+        print(convert(s), end='')

test/test_all.py

@@ -16,6 +16,7 @@ from test_settings import SettingsTest
 from test_tileset import TilesetTest
 from test_cache import TestLRU
 from test_contributors import TestContributors
+from test_cyrillic_convert import TestCyrillicConvert
 
 # DISABLE THIS BLOCK TO GET LOG OUTPUT FROM TILESET FOR DEBUGGING
 if 0:

test/test_cyrillic_convert.py

@@ -0,0 +1,11 @@
+import unittest
+import tempfile
+
+from contrib.cyrillic_convert import convert
+
+
+class TestCyrillicConvert(unittest.TestCase):
+    def test_convert(self):
+        gibberish = '{chunk: [-2, 0],y: 65,msg: "ðåëèãèè",x: -20,z: 4,type: "sign"}'
+        cyrillic = '{chunk: [-2, 0],y: 65,msg: "религии",x: -20,z: 4,type: "sign"}'
+        self.assertEqual(convert(gibberish), cyrillic)