class CharsetConverterTest(unittest.TestCase):
def setUp(self):
self.converter = CharsetConverter()
self.samples = path(resource_filename(__package__, "samples"))
def test_detection(self):
file = self.samples / "sample1-ISO-8859-1.txt"
self.assertEquals("ISO-8859-1",
self.converter.detect_encoding(file).charset)
file = self.samples / "sample1-UTF-8.txt"
self.assertEquals("UTF-8",
self.converter.detect_encoding(file).charset)
def test_convertion_from_iso_to_utf(self):
# setup files
iso = self.samples / "sample1-ISO-8859-1.txt"
iso_checksum = self.__checksum(iso)
utf = self.samples / "sample1-UTF-8.txt"
utf_checksum = self.__checksum(utf)
# create temp file
fd, filename = tempfile.mkstemp(prefix="baudot")
tmp = path(filename)
tmp_checksum = self.__checksum(tmp)
# validate before convertion
self.assertNotEquals(iso_checksum, utf_checksum)
self.assertNotEquals(tmp_checksum, utf_checksum)
self.assertNotEquals(iso_checksum, tmp_checksum)
# convert files
self.converter.convert_encoding(iso, tmp, "ISO-8859-1", "UTF-8")
tmp_checksum = self.__checksum(tmp)
# validate output
self.assertNotEquals(iso_checksum, tmp_checksum)
self.assertEquals(tmp_checksum, utf_checksum)
tmp.remove()
self.assertFalse(tmp.exists())
def test_get_encodings(self):
available = self.converter.get_encodings()
self.assertIn("UTF-8", available)
self.assertIn("ISO-8859-1", available)
self.assertIn("windows-1251", available)
def __checksum(self, file):
block_size = 0x10000
def upd(m, data):
m.update(data)
return m
fd = open(file, "rb")
try:
contents = iter(lambda: fd.read(block_size), "")
m = reduce(upd, contents, md5())
return m.hexdigest()
finally:
fd.close()
class CharsetConverterTest(unittest.TestCase):
def setUp(self):
self.converter = CharsetConverter()
self.samples = path(resource_filename(__package__, "samples"))
def test_detection(self):
file = self.samples / "sample1-ISO-8859-1.txt"
self.assertEquals("ISO-8859-1", self.converter.detect_encoding(file).charset)
file = self.samples / "sample1-UTF-8.txt"
self.assertEquals("UTF-8", self.converter.detect_encoding(file).charset)
def test_convertion_from_iso_to_utf(self):
# setup files
iso = self.samples / "sample1-ISO-8859-1.txt"
iso_checksum = self.__checksum(iso)
utf = self.samples / "sample1-UTF-8.txt"
utf_checksum = self.__checksum(utf)
# create temp file
fd, filename = tempfile.mkstemp(prefix="baudot")
tmp = path(filename)
tmp_checksum = self.__checksum(tmp)
# validate before convertion
self.assertNotEquals(iso_checksum, utf_checksum)
self.assertNotEquals(tmp_checksum, utf_checksum)
self.assertNotEquals(iso_checksum, tmp_checksum)
# convert files
self.converter.convert_encoding(iso, tmp, "ISO-8859-1", "UTF-8")
tmp_checksum = self.__checksum(tmp)
# validate output
self.assertNotEquals(iso_checksum, tmp_checksum)
self.assertEquals(tmp_checksum, utf_checksum)
tmp.remove()
self.assertFalse(tmp.exists())
def test_get_encodings(self):
available = self.converter.get_encodings()
self.assertIn("UTF-8", available)
self.assertIn("ISO-8859-1", available)
self.assertIn("windows-1251", available)
def __checksum(self, file):
block_size = 0x10000
def upd(m, data):
m.update(data)
return m
fd = open(file, "rb")
try:
contents = iter(lambda: fd.read(block_size), "")
m = reduce(upd, contents, md5())
return m.hexdigest()
finally:
fd.close()
def setUp(self):
self.converter = CharsetConverter()
self.samples = path(resource_filename(__package__, "samples"))
def setUp(self):
self.converter = CharsetConverter()
self.samples = path(resource_filename(__package__, "samples"))
class FileManagerTest(unittest.TestCase):
def setUp(self):
self.converter = CharsetConverter()
self.fm = FileManager()
self.samples = path(ResourceManager().resource_filename(__package__, "samples"))
def test_add_search(self):
dir1 = self.samples / "dir1"
txt = dir1 / "sample1-ISO-8859-1.txt"
image = dir1 / "locked.svg"
empty = dir1 / "empty"
self.assertEqual(0, len(self.fm))
cmd = self.fm.add(dir1)
cmd.start()
cmd.join()
self.assertEqual(1, len(self.fm))
dir_row = cmd.search(dir1)
self.assertIsNotNone(dir_row)
self.assertEqual(6, len(dir1.listdir()))
self.assertEqual(dir1, dir_row[0])
self.assertEqual(gtk.STOCK_DIRECTORY, dir_row[1])
self.assertEqual(dir1, dir_row[2])
self.assertEqual("4 items", dir_row[3])
self.assertEqual("Folder", dir_row[4])
self.assertIsNone(dir_row[5])
txt_row = cmd.search(txt)
self.assertIsNotNone(txt_row)
self.assertEqual(txt, txt_row[0])
self.assertEqual(gtk.STOCK_FILE, txt_row[1])
self.assertEqual("sample1-ISO-8859-1.txt", txt_row[2])
self.assertEqual("1.16 KB", txt_row[3])
self.assertEqual("text/plain", txt_row[4])
self.assertEqual("ISO-8859-1", txt_row[5])
self.assertTrue(image.exists())
self.assertIsNone(cmd.search(image))
self.assertTrue(empty.exists())
self.assertIsNone(cmd.search(empty))
cmd = self.fm.add(empty)
cmd.start()
cmd.join()
self.assertEqual(2, len(self.fm))
def test_add_duplicate(self):
dir1 = self.samples / "dir1"
self.assertEqual(0, len(self.fm))
cmd = self.fm.add(dir1)
cmd.start()
cmd.join()
self.assertEqual(1, len(self.fm))
mock = MockListener()
cmd = self.fm.add(dir1)
cmd.connect("command-aborted", mock.slot)
cmd.start()
cmd.join()
self.assertTrue(mock.invoked)
self.assertEqual(1, len(self.fm))
def test_convert_copy(self):
orig = self.samples / "dir1"
copy = path(tempfile.mkdtemp())
try:
cmd = self.fm.add(orig)
cmd.start()
cmd.join()
cmd = self.fm.convert("ISO-8859-1", copy)
cmd.start()
cmd.join()
converted = copy / "sample2-UTF-8.txt"
self.assertTrue(converted.exists())
self.assertEqual("ISO-8859-1", self.converter.detect_encoding(converted).charset)
finally:
copy.rmtree()
def test_convert_in_place(self):
orig = self.samples / "dir1"
tmp = path(tempfile.mkdtemp())
copy = tmp / "copy"
orig.copytree(copy)
try:
cmd = self.fm.add(copy)
cmd.start()
cmd.join()
cmd = self.fm.convert("ISO-8859-1")
cmd.start()
cmd.join()
converted = copy / "sample2-UTF-8.txt"
self.assertTrue(converted.exists())
self.assertEqual("ISO-8859-1", self.converter.detect_encoding(converted).charset)
finally:
tmp.rmtree()
self.assertFalse(tmp.exists())
class FileManagerTest(unittest.TestCase):
def setUp(self):
self.converter = CharsetConverter()
self.fm = FileManager()
self.samples = path(ResourceManager().resource_filename(__package__, "samples"))
def test_add_search(self):
dir1 = self.samples / "dir1"
txt = dir1 / "sample1-ISO-8859-1.txt"
image = dir1 / "locked.svg"
empty = dir1 / "empty"
self.assertEqual(0, len(self.fm))
cmd = self.fm.add(dir1)
cmd.start()
cmd.join()
self.assertEqual(1, len(self.fm))
dir_row = cmd.search(dir1)
self.assertIsNotNone(dir_row)
self.assertEqual(6, len(dir1.listdir()))
self.assertEqual(dir1, dir_row[0])
self.assertEqual(gtk.STOCK_DIRECTORY, dir_row[1])
self.assertEqual(dir1, dir_row[2])
self.assertEqual("4 items", dir_row[3])
self.assertEqual("Folder", dir_row[4])
self.assertIsNone(dir_row[5])
txt_row = cmd.search(txt)
self.assertIsNotNone(txt_row)
self.assertEqual(txt, txt_row[0])
self.assertEqual(gtk.STOCK_FILE, txt_row[1])
self.assertEqual("sample1-ISO-8859-1.txt", txt_row[2])
self.assertEqual("1.16 KB", txt_row[3])
self.assertEqual("text/plain", txt_row[4])
self.assertEqual("ISO-8859-1", txt_row[5])
self.assertTrue(image.exists())
self.assertIsNone(cmd.search(image))
self.assertTrue(empty.exists())
self.assertIsNone(cmd.search(empty))
cmd = self.fm.add(empty)
cmd.start()
cmd.join()
self.assertEqual(2, len(self.fm))
def test_add_duplicate(self):
dir1 = self.samples / "dir1"
self.assertEqual(0, len(self.fm))
cmd = self.fm.add(dir1)
cmd.start()
cmd.join()
self.assertEqual(1, len(self.fm))
mock = MockListener()
cmd = self.fm.add(dir1)
cmd.connect("command-aborted", mock.slot)
cmd.start()
cmd.join()
self.assertTrue(mock.invoked)
self.assertEqual(1, len(self.fm))
def test_convert_copy(self):
orig = self.samples / "dir1"
copy = path(tempfile.mkdtemp())
try:
cmd = self.fm.add(orig)
cmd.start()
cmd.join()
cmd = self.fm.convert("ISO-8859-1", copy)
cmd.start()
cmd.join()
converted = copy / "sample2-UTF-8.txt"
self.assertTrue(converted.exists())
self.assertEqual("ISO-8859-1", self.converter.detect_encoding(converted).charset)
finally:
copy.rmtree()
def test_convert_in_place(self):
orig = self.samples / "dir1"
tmp = path(tempfile.mkdtemp())
copy = tmp / "copy"
orig.copytree(copy)
try:
cmd = self.fm.add(copy)
cmd.start()
cmd.join()
cmd = self.fm.convert("ISO-8859-1")
cmd.start()
cmd.join()
converted = copy / "sample2-UTF-8.txt"
self.assertTrue(converted.exists())
self.assertEqual("ISO-8859-1", self.converter.detect_encoding(converted).charset)
finally:
tmp.rmtree()
self.assertFalse(tmp.exists())