]>
Commit | Line | Data |
---|---|---|
df91e0f9 JJ |
1 | ## @file\r |
2 | # Unit tests for AutoGen.UniClassObject\r | |
3 | #\r | |
4 | # Copyright (c) 2015, Intel Corporation. All rights reserved.<BR>\r | |
5 | #\r | |
2e351cbe | 6 | # SPDX-License-Identifier: BSD-2-Clause-Patent\r |
df91e0f9 JJ |
7 | #\r |
8 | \r | |
9 | ##\r | |
10 | # Import Modules\r | |
11 | #\r | |
12 | import os\r | |
13 | import unittest\r | |
14 | \r | |
15 | import codecs\r | |
16 | \r | |
17 | import TestTools\r | |
18 | \r | |
19 | from Common.Misc import PathClass\r | |
20 | import AutoGen.UniClassObject as BtUni\r | |
21 | \r | |
22 | from Common import EdkLogger\r | |
23 | EdkLogger.InitializeForUnitTest()\r | |
24 | \r | |
25 | class Tests(TestTools.BaseToolsTest):\r | |
26 | \r | |
27 | SampleData = u'''\r | |
28 | #langdef en-US "English"\r | |
29 | #string STR_A #language en-US "STR_A for en-US"\r | |
30 | '''\r | |
31 | \r | |
32 | def EncodeToFile(self, encoding, string=None):\r | |
33 | if string is None:\r | |
34 | string = self.SampleData\r | |
dadfab5b JJ |
35 | if encoding is not None:\r |
36 | data = codecs.encode(string, encoding)\r | |
37 | else:\r | |
38 | data = string\r | |
df91e0f9 JJ |
39 | path = 'input.uni'\r |
40 | self.WriteTmpFile(path, data)\r | |
41 | return PathClass(self.GetTmpFilePath(path))\r | |
42 | \r | |
43 | def ErrorFailure(self, error, encoding, shouldPass):\r | |
44 | msg = error + ' should '\r | |
45 | if shouldPass:\r | |
46 | msg += 'not '\r | |
47 | msg += 'be generated for '\r | |
48 | msg += '%s data in a .uni file' % encoding\r | |
49 | self.fail(msg)\r | |
50 | \r | |
51 | def UnicodeErrorFailure(self, encoding, shouldPass):\r | |
52 | self.ErrorFailure('UnicodeError', encoding, shouldPass)\r | |
53 | \r | |
54 | def EdkErrorFailure(self, encoding, shouldPass):\r | |
55 | self.ErrorFailure('EdkLogger.FatalError', encoding, shouldPass)\r | |
56 | \r | |
57 | def CheckFile(self, encoding, shouldPass, string=None):\r | |
58 | path = self.EncodeToFile(encoding, string)\r | |
59 | try:\r | |
60 | BtUni.UniFileClassObject([path])\r | |
61 | if shouldPass:\r | |
62 | return\r | |
63 | except UnicodeError:\r | |
64 | if not shouldPass:\r | |
65 | return\r | |
66 | else:\r | |
67 | self.UnicodeErrorFailure(encoding, shouldPass)\r | |
68 | except EdkLogger.FatalError:\r | |
69 | if not shouldPass:\r | |
70 | return\r | |
71 | else:\r | |
72 | self.EdkErrorFailure(encoding, shouldPass)\r | |
73 | except Exception:\r | |
74 | pass\r | |
75 | \r | |
76 | self.EdkErrorFailure(encoding, shouldPass)\r | |
77 | \r | |
78 | def testUtf16InUniFile(self):\r | |
79 | self.CheckFile('utf_16', shouldPass=True)\r | |
80 | \r | |
dadfab5b JJ |
81 | def testSupplementaryPlaneUnicodeCharInUtf16File(self):\r |
82 | #\r | |
83 | # Supplementary Plane characters can exist in UTF-16 files,\r | |
84 | # but they are not valid UCS-2 characters.\r | |
85 | #\r | |
86 | # This test makes sure that BaseTools rejects these characters\r | |
87 | # if seen in a .uni file.\r | |
88 | #\r | |
89 | data = u'''\r | |
90 | #langdef en-US "English"\r | |
91 | #string STR_A #language en-US "CodePoint (\U00010300) > 0xFFFF"\r | |
92 | '''\r | |
93 | \r | |
94 | self.CheckFile('utf_16', shouldPass=False, string=data)\r | |
95 | \r | |
96 | def testSurrogatePairUnicodeCharInUtf16File(self):\r | |
97 | #\r | |
98 | # Surrogate Pair code points are used in UTF-16 files to\r | |
99 | # encode the Supplementary Plane characters. But, a Surrogate\r | |
100 | # Pair code point which is not followed by another Surrogate\r | |
101 | # Pair code point might be interpreted as a single code point\r | |
102 | # with the Surrogate Pair code point.\r | |
103 | #\r | |
104 | # This test makes sure that BaseTools rejects these characters\r | |
105 | # if seen in a .uni file.\r | |
106 | #\r | |
fe906312 | 107 | data = codecs.BOM_UTF16_LE + b'//\x01\xd8 '\r |
dadfab5b JJ |
108 | \r |
109 | self.CheckFile(encoding=None, shouldPass=False, string=data)\r | |
110 | \r | |
156d6d65 JJ |
111 | def testValidUtf8File(self):\r |
112 | self.CheckFile(encoding='utf_8', shouldPass=True)\r | |
113 | \r | |
114 | def testValidUtf8FileWithBom(self):\r | |
115 | #\r | |
116 | # Same test as testValidUtf8File, but add the UTF-8 BOM\r | |
117 | #\r | |
118 | data = codecs.BOM_UTF8 + codecs.encode(self.SampleData, 'utf_8')\r | |
119 | \r | |
120 | self.CheckFile(encoding=None, shouldPass=True, string=data)\r | |
121 | \r | |
15c3a04c JJ |
122 | def test32bitUnicodeCharInUtf8File(self):\r |
123 | data = u'''\r | |
124 | #langdef en-US "English"\r | |
125 | #string STR_A #language en-US "CodePoint (\U00010300) > 0xFFFF"\r | |
126 | '''\r | |
127 | \r | |
128 | self.CheckFile('utf_16', shouldPass=False, string=data)\r | |
129 | \r | |
130 | def test32bitUnicodeCharInUtf8File(self):\r | |
131 | data = u'''\r | |
132 | #langdef en-US "English"\r | |
133 | #string STR_A #language en-US "CodePoint (\U00010300) > 0xFFFF"\r | |
134 | '''\r | |
135 | \r | |
136 | self.CheckFile('utf_8', shouldPass=False, string=data)\r | |
137 | \r | |
138 | def test32bitUnicodeCharInUtf8Comment(self):\r | |
139 | data = u'''\r | |
140 | // Even in comments, we reject non-UCS-2 chars: \U00010300\r | |
141 | #langdef en-US "English"\r | |
142 | #string STR_A #language en-US "A"\r | |
143 | '''\r | |
144 | \r | |
145 | self.CheckFile('utf_8', shouldPass=False, string=data)\r | |
146 | \r | |
8fb5a0ca JJ |
147 | def testSurrogatePairUnicodeCharInUtf8File(self):\r |
148 | #\r | |
149 | # Surrogate Pair code points are used in UTF-16 files to\r | |
150 | # encode the Supplementary Plane characters. In UTF-8, it is\r | |
151 | # trivial to encode these code points, but they are not valid\r | |
152 | # code points for characters, since they are reserved for the\r | |
153 | # UTF-16 Surrogate Pairs.\r | |
154 | #\r | |
155 | # This test makes sure that BaseTools rejects these characters\r | |
156 | # if seen in a .uni file.\r | |
157 | #\r | |
fe906312 | 158 | data = b'\xed\xa0\x81'\r |
8fb5a0ca JJ |
159 | \r |
160 | self.CheckFile(encoding=None, shouldPass=False, string=data)\r | |
161 | \r | |
162 | def testSurrogatePairUnicodeCharInUtf8FileWithBom(self):\r | |
163 | #\r | |
164 | # Same test as testSurrogatePairUnicodeCharInUtf8File, but add\r | |
165 | # the UTF-8 BOM\r | |
166 | #\r | |
fe906312 | 167 | data = codecs.BOM_UTF8 + b'\xed\xa0\x81'\r |
8fb5a0ca JJ |
168 | \r |
169 | self.CheckFile(encoding=None, shouldPass=False, string=data)\r | |
170 | \r | |
df91e0f9 JJ |
171 | TheTestSuite = TestTools.MakeTheTestSuite(locals())\r |
172 | \r | |
173 | if __name__ == '__main__':\r | |
174 | allTests = TheTestSuite()\r | |
175 | unittest.TextTestRunner().run(allTests)\r |