annotate Resources/EncodingTests.py @ 1090:e494ceb8d763

support more encodings
author Sebastien Jodogne <s.jodogne@gmail.com>
date Tue, 05 Aug 2014 12:04:23 +0200
parents 6fd4434c1bcf
children a66224eec125
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1088
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
1 #!/usr/bin/python
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
2 # -*- coding: utf-8 -*-
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
3
1090
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
4 source = u'éäöòДΘĝדصķћ๛ネİ'
1088
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
5
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
6 encodings = {
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
7 'UTF-8' : 'Utf8',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
8 'ASCII' : 'Ascii',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
9 'ISO-8859-1' : 'Latin1',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
10 'ISO-8859-2' : 'Latin2',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
11 'ISO-8859-3' : 'Latin3',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
12 'ISO-8859-4' : 'Latin4',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
13 'ISO-8859-9' : 'Latin5',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
14 'ISO-8859-5' : 'Cyrillic',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
15 'ISO-8859-6' : 'Arabic',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
16 'ISO-8859-7' : 'Greek',
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
17 'ISO-8859-8' : 'Hebrew',
1090
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
18 'TIS-620' : 'Thai',
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
19 'SHIFT-JIS' : 'Japanese',
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
20 #'GB18030' : 'Chinese',
1088
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
21 }
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
22
1090
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
23 #from encodings.aliases import aliases
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
24 #for a, b in aliases.iteritems():
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
25 # print '%s : %s' % (a, b)
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
26
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
27
1088
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
28 # "63" corresponds to "?"
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
29 l = []
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
30 encoded = []
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
31 expected = []
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
32
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
33 def ToArray(source):
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
34 result = ''
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
35 for byte in bytearray(source):
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
36 result += '\\x%02x' % byte
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
37 return '"%s"' % result
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
38
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
39
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
40 for encoding, orthancEnumeration in encodings.iteritems():
1090
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
41 l.append('::Orthanc::Encoding_%s' % orthancEnumeration)
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
42 s = source.encode(encoding, 'ignore')
1088
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
43 encoded.append(ToArray(s))
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
44 expected.append(ToArray(s.decode(encoding).encode('utf-8')))
6fd4434c1bcf unit tests for encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents:
diff changeset
45
1090
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
46
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
47 # https://en.wikipedia.org/wiki/GB_18030#Technical_details
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
48 l.append('::Orthanc::Encoding_Chinese')
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
49 expected.append(ToArray('Þßàáâã'))
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
50 encoded.append('"\\x81\\x30\\x89\\x37\\x81\\x30\\x89\\x38\\xA8\\xA4\\xA8\\xA2\\x81\\x30\\x89\\x39\\x81\\x30\\x8A\\x30"')
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
51
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
52
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
53 if True:
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
54 print 'static const unsigned int testEncodingsCount = %d;' % len(l)
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
55 print 'static const ::Orthanc::Encoding testEncodings[] = {\n %s\n};' % (',\n '.join(l))
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
56 print 'static const char *testEncodingsEncoded[%d] = {\n %s\n};' % (len(l), ',\n '.join(encoded))
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
57 print 'static const char *testEncodingsExpected[%d] = {\n %s\n};' % (len(l), ',\n '.join(expected))
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
58 else:
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
59 for i in range(len(expected)):
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
60 print expected[i]
e494ceb8d763 support more encodings
Sebastien Jodogne <s.jodogne@gmail.com>
parents: 1088
diff changeset
61 #print '%s: %s' % (expected[i], l[i])