|
Jeremy Cline |
e62ce0 |
# -*- coding: utf-8 -*-
|
|
Jeremy Cline |
e62ce0 |
"""
|
|
Jeremy Cline |
e62ce0 |
Tests for :module:`pagure.lib.encoding_utils`.
|
|
Jeremy Cline |
e62ce0 |
"""
|
|
Jeremy Cline |
e62ce0 |
|
|
Pierre-Yves Chibon |
67d1cc |
from __future__ import unicode_literals, absolute_import
|
|
Aurélien Bompard |
626417 |
|
|
Jeremy Cline |
e62ce0 |
import chardet
|
|
Pierre-Yves Chibon |
52bc45 |
import os
|
|
Jeremy Cline |
e62ce0 |
import unittest
|
|
Pierre-Yves Chibon |
52bc45 |
import sys
|
|
Pierre-Yves Chibon |
52bc45 |
|
|
Pierre-Yves Chibon |
73d120 |
sys.path.insert(
|
|
Pierre-Yves Chibon |
73d120 |
0, os.path.join(os.path.dirname(os.path.abspath(__file__)), "..")
|
|
Pierre-Yves Chibon |
73d120 |
)
|
|
Jeremy Cline |
e62ce0 |
|
|
Jeremy Cline |
e62ce0 |
from pagure.lib import encoding_utils
|
|
Jeremy Cline |
e62ce0 |
|
|
Jeremy Cline |
e62ce0 |
|
|
Jeremy Cline |
e62ce0 |
class TestGuessEncoding(unittest.TestCase):
|
|
Jeremy Cline |
e62ce0 |
def test_guess_encoding_ascii(self):
|
|
Jeremy Cline |
e62ce0 |
"""
|
|
Jeremy Cline |
e62ce0 |
Assert when ascii-only data is provided ascii is the guessed encoding.
|
|
Jeremy Cline |
e62ce0 |
"""
|
|
Pierre-Yves Chibon |
73d120 |
data = "Twas bryllyg, and the slythy toves did gyre and gymble"
|
|
Pierre-Yves Chibon |
73d120 |
result = encoding_utils.guess_encoding(data.encode("ascii"))
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(result, "ascii")
|
|
Jeremy Cline |
e62ce0 |
|
|
Jeremy Cline |
e62ce0 |
def test_guess_encoding_favor_utf_8(self):
|
|
Jeremy Cline |
e62ce0 |
"""
|
|
Pierre-Yves Chibon |
8ac14c |
Test that strings that could be UTF-8 or ISO-8859-* result in UTF-8.
|
|
Pierre-Yves Chibon |
8ac14c |
|
|
Pierre-Yves Chibon |
8ac14c |
python-chardet-3.0.4-2.fc27.noarch detects it as ISO-8859-9
|
|
Pierre-Yves Chibon |
8ac14c |
python-chardet-2.2.1-1.el7_1.noarch detects it as ISO-8859-2
|
|
Jeremy Cline |
e62ce0 |
"""
|
|
Pierre-Yves Chibon |
73d120 |
data = "Šabata".encode("utf-8")
|
|
Jeremy Cline |
e62ce0 |
result = encoding_utils.guess_encoding(data)
|
|
Jeremy Cline |
e62ce0 |
chardet_result = chardet.detect(data)
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(result, "utf-8")
|
|
Pierre-Yves Chibon |
73d120 |
if chardet.__version__[0] == "3":
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(chardet_result["encoding"], "ISO-8859-9")
|
|
Pierre-Yves Chibon |
8ac14c |
else:
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(chardet_result["encoding"], "ISO-8859-2")
|
|
Jeremy Cline |
e62ce0 |
|
|
Jeremy Cline |
4bcf65 |
def test_guess_encoding_no_data(self):
|
|
Pierre-Yves Chibon |
826adb |
""" Test encoding_utils.guess_encoding() with an empty string """
|
|
Pierre-Yves Chibon |
73d120 |
result = encoding_utils.guess_encoding("".encode("utf-8"))
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(result, "ascii")
|
|
Jeremy Cline |
4bcf65 |
|
|
Jeremy Cline |
4bcf65 |
|
|
Pierre-Yves Chibon |
29890e |
class TestGuessEncodings(unittest.TestCase):
|
|
Pierre-Yves Chibon |
29890e |
def test_guess_encodings(self):
|
|
Pierre-Yves Chibon |
29890e |
""" Test the encoding_utils.guess_encodings() method. """
|
|
Pierre-Yves Chibon |
73d120 |
data = "Šabata".encode("utf-8")
|
|
Pierre-Yves Chibon |
29890e |
result = encoding_utils.guess_encodings(data)
|
|
Pierre-Yves Chibon |
29890e |
chardet_result = chardet.detect(data)
|
|
Pierre-Yves Chibon |
73d120 |
if chardet.__version__[0] == "3":
|
|
Aurélien Bompard |
f61bb3 |
# The first three have different confidence values
|
|
Aurélien Bompard |
f61bb3 |
self.assertListEqual(
|
|
Aurélien Bompard |
f61bb3 |
[encoding.encoding for encoding in result][:3],
|
|
Pierre-Yves Chibon |
73d120 |
["utf-8", "ISO-8859-9", "ISO-8859-1"],
|
|
Aurélien Bompard |
f61bb3 |
)
|
|
Aurélien Bompard |
f61bb3 |
# This is the one with the least confidence
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(result[-1].encoding, "windows-1255")
|
|
Aurélien Bompard |
f61bb3 |
# The values in the middle of the list all have the same confidence
|
|
Aurélien Bompard |
f61bb3 |
# value and can't be sorted reliably: use sets.
|
|
Pierre-Yves Chibon |
8ac14c |
self.assertEqual(
|
|
Aurélien Bompard |
f61bb3 |
set([encoding.encoding for encoding in result]),
|
|
Pierre-Yves Chibon |
73d120 |
set(
|
|
Pierre-Yves Chibon |
73d120 |
[
|
|
Pierre-Yves Chibon |
73d120 |
"utf-8",
|
|
Pierre-Yves Chibon |
73d120 |
"ISO-8859-9",
|
|
Pierre-Yves Chibon |
73d120 |
"ISO-8859-1",
|
|
Pierre-Yves Chibon |
73d120 |
"MacCyrillic",
|
|
Pierre-Yves Chibon |
73d120 |
"IBM866",
|
|
Pierre-Yves Chibon |
73d120 |
"TIS-620",
|
|
Pierre-Yves Chibon |
73d120 |
"EUC-JP",
|
|
Pierre-Yves Chibon |
73d120 |
"EUC-KR",
|
|
Pierre-Yves Chibon |
73d120 |
"GB2312",
|
|
Pierre-Yves Chibon |
73d120 |
"KOI8-R",
|
|
Pierre-Yves Chibon |
73d120 |
"Big5",
|
|
Pierre-Yves Chibon |
73d120 |
"IBM855",
|
|
Pierre-Yves Chibon |
73d120 |
"ISO-8859-7",
|
|
Pierre-Yves Chibon |
73d120 |
"SHIFT_JIS",
|
|
Pierre-Yves Chibon |
73d120 |
"windows-1253",
|
|
Pierre-Yves Chibon |
73d120 |
"CP949",
|
|
Pierre-Yves Chibon |
73d120 |
"EUC-TW",
|
|
Pierre-Yves Chibon |
73d120 |
"ISO-8859-5",
|
|
Pierre-Yves Chibon |
73d120 |
"windows-1251",
|
|
Pierre-Yves Chibon |
73d120 |
"windows-1255",
|
|
Pierre-Yves Chibon |
73d120 |
]
|
|
Pierre-Yves Chibon |
73d120 |
),
|
|
Aurélien Bompard |
f61bb3 |
)
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(chardet_result["encoding"], "ISO-8859-9")
|
|
Pierre-Yves Chibon |
8ac14c |
else:
|
|
Aurélien Bompard |
f61bb3 |
self.assertListEqual(
|
|
Pierre-Yves Chibon |
8ac14c |
[encoding.encoding for encoding in result],
|
|
Pierre-Yves Chibon |
73d120 |
["utf-8", "ISO-8859-2", "windows-1252"],
|
|
Pierre-Yves Chibon |
73d120 |
)
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(chardet_result["encoding"], "ISO-8859-2")
|
|
Pierre-Yves Chibon |
29890e |
|
|
Pierre-Yves Chibon |
29890e |
def test_guess_encodings_no_data(self):
|
|
Pierre-Yves Chibon |
29890e |
""" Test encoding_utils.guess_encodings() with an emtpy string """
|
|
Pierre-Yves Chibon |
73d120 |
result = encoding_utils.guess_encodings("".encode("utf-8"))
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual([encoding.encoding for encoding in result], ["ascii"])
|
|
Pierre-Yves Chibon |
29890e |
|
|
Jeremy Cline |
4bcf65 |
|
|
Pierre-Yves Chibon |
73d120 |
class TestDecode(unittest.TestCase):
|
|
Jeremy Cline |
e62ce0 |
def test_decode(self):
|
|
Pierre-Yves Chibon |
52bc45 |
""" Test encoding_utils.decode() """
|
|
Pierre-Yves Chibon |
73d120 |
data = "Šabata"
|
|
Pierre-Yves Chibon |
73d120 |
self.assertEqual(data, encoding_utils.decode(data.encode("utf-8")))
|
|
Jeremy Cline |
e62ce0 |
|
|
Jeremy Cline |
e62ce0 |
|
|
Pierre-Yves Chibon |
73d120 |
if __name__ == "__main__":
|
|
Pierre-Yves Chibon |
52bc45 |
unittest.main(verbosity=2)
|