blob: 07e760b8e3142cfca298cc04438bb034d67d8160 [file] [log] [blame]
Florent Xicluna8fbddf12010-03-17 20:29:51 +00001from test.support import run_unittest, check_warnings
Jeremy Hyltond9827c42000-08-03 22:11:43 +00002import cgi
3import os
4import sys
Thomas Wouters00ee7ba2006-08-21 19:07:27 +00005import tempfile
Thomas Wouters89f507f2006-12-13 04:49:30 +00006import unittest
Senthil Kumaran6b102f22013-01-23 02:50:15 -08007from collections import namedtuple
Victor Stinner5c23b8e2011-01-14 13:05:21 +00008from io import StringIO, BytesIO
Jeremy Hyltond9827c42000-08-03 22:11:43 +00009
10class HackedSysModule:
11 # The regression test will have real values in sys.argv, which
Fred Drake004d5e62000-10-23 17:22:08 +000012 # will completely confuse the test of the cgi module
Jeremy Hyltond9827c42000-08-03 22:11:43 +000013 argv = []
14 stdin = sys.stdin
15
16cgi.sys = HackedSysModule()
17
Jeremy Hyltond9827c42000-08-03 22:11:43 +000018class ComparableException:
19 def __init__(self, err):
20 self.err = err
21
22 def __str__(self):
23 return str(self.err)
24
Guido van Rossum47b9ff62006-08-24 00:41:19 +000025 def __eq__(self, anExc):
Jeremy Hyltond9827c42000-08-03 22:11:43 +000026 if not isinstance(anExc, Exception):
Guido van Rossum47b9ff62006-08-24 00:41:19 +000027 return NotImplemented
28 return (self.err.__class__ == anExc.__class__ and
29 self.err.args == anExc.args)
Jeremy Hyltond9827c42000-08-03 22:11:43 +000030
31 def __getattr__(self, attr):
Guido van Rossum846d6db2001-01-17 15:08:37 +000032 return getattr(self.err, attr)
Jeremy Hyltond9827c42000-08-03 22:11:43 +000033
34def do_test(buf, method):
35 env = {}
36 if method == "GET":
37 fp = None
38 env['REQUEST_METHOD'] = 'GET'
39 env['QUERY_STRING'] = buf
40 elif method == "POST":
Victor Stinner5c23b8e2011-01-14 13:05:21 +000041 fp = BytesIO(buf.encode('latin-1')) # FieldStorage expects bytes
Jeremy Hyltond9827c42000-08-03 22:11:43 +000042 env['REQUEST_METHOD'] = 'POST'
43 env['CONTENT_TYPE'] = 'application/x-www-form-urlencoded'
44 env['CONTENT_LENGTH'] = str(len(buf))
45 else:
Collin Winter3add4d72007-08-29 23:37:32 +000046 raise ValueError("unknown method: %s" % method)
Jeremy Hyltond9827c42000-08-03 22:11:43 +000047 try:
48 return cgi.parse(fp, env, strict_parsing=1)
Guido van Rossumcd16bf62007-06-13 18:07:49 +000049 except Exception as err:
Jeremy Hyltond9827c42000-08-03 22:11:43 +000050 return ComparableException(err)
51
Neil Schemenauer66edb622004-07-19 15:38:11 +000052parse_strict_test_cases = [
Jeremy Hyltond9827c42000-08-03 22:11:43 +000053 ("", ValueError("bad query field: ''")),
54 ("&", ValueError("bad query field: ''")),
55 ("&&", ValueError("bad query field: ''")),
Jeremy Hyltonafde7e22000-09-15 20:06:57 +000056 (";", ValueError("bad query field: ''")),
57 (";&;", ValueError("bad query field: ''")),
Jeremy Hyltond9827c42000-08-03 22:11:43 +000058 # Should the next few really be valid?
59 ("=", {}),
60 ("=&=", {}),
Jeremy Hyltonafde7e22000-09-15 20:06:57 +000061 ("=;=", {}),
Jeremy Hyltond9827c42000-08-03 22:11:43 +000062 # This rest seem to make sense
63 ("=a", {'': ['a']}),
64 ("&=a", ValueError("bad query field: ''")),
65 ("=a&", ValueError("bad query field: ''")),
66 ("=&a", ValueError("bad query field: 'a'")),
67 ("b=a", {'b': ['a']}),
68 ("b+=a", {'b ': ['a']}),
69 ("a=b=a", {'a': ['b=a']}),
70 ("a=+b=a", {'a': [' b=a']}),
71 ("&b=a", ValueError("bad query field: ''")),
72 ("b&=a", ValueError("bad query field: 'b'")),
73 ("a=a+b&b=b+c", {'a': ['a b'], 'b': ['b c']}),
74 ("a=a+b&a=b+a", {'a': ['a b', 'b a']}),
75 ("x=1&y=2.0&z=2-3.%2b0", {'x': ['1'], 'y': ['2.0'], 'z': ['2-3.+0']}),
Jeremy Hyltonafde7e22000-09-15 20:06:57 +000076 ("x=1;y=2.0&z=2-3.%2b0", {'x': ['1'], 'y': ['2.0'], 'z': ['2-3.+0']}),
77 ("x=1;y=2.0;z=2-3.%2b0", {'x': ['1'], 'y': ['2.0'], 'z': ['2-3.+0']}),
Jeremy Hyltond9827c42000-08-03 22:11:43 +000078 ("Hbc5161168c542333633315dee1182227:key_store_seqid=400006&cuyer=r&view=bustomer&order_id=0bb2e248638833d48cb7fed300000f1b&expire=964546263&lobale=en-US&kid=130003.300038&ss=env",
79 {'Hbc5161168c542333633315dee1182227:key_store_seqid': ['400006'],
80 'cuyer': ['r'],
81 'expire': ['964546263'],
82 'kid': ['130003.300038'],
83 'lobale': ['en-US'],
84 'order_id': ['0bb2e248638833d48cb7fed300000f1b'],
85 'ss': ['env'],
86 'view': ['bustomer'],
87 }),
Fred Drake004d5e62000-10-23 17:22:08 +000088
Jeremy Hyltond9827c42000-08-03 22:11:43 +000089 ("group_id=5470&set=custom&_assigned_to=31392&_status=1&_category=100&SUBMIT=Browse",
90 {'SUBMIT': ['Browse'],
91 '_assigned_to': ['31392'],
92 '_category': ['100'],
93 '_status': ['1'],
94 'group_id': ['5470'],
95 'set': ['custom'],
96 })
97 ]
98
Guido van Rossum47b9ff62006-08-24 00:41:19 +000099def norm(seq):
Guido van Rossumcc2b0162007-02-11 06:12:03 +0000100 return sorted(seq, key=repr)
Jeremy Hyltond9827c42000-08-03 22:11:43 +0000101
102def first_elts(list):
Guido van Rossumc1f779c2007-07-03 08:25:58 +0000103 return [p[0] for p in list]
Jeremy Hyltond9827c42000-08-03 22:11:43 +0000104
105def first_second_elts(list):
Guido van Rossumc1f779c2007-07-03 08:25:58 +0000106 return [(p[0], p[1][0]) for p in list]
107
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000108def gen_result(data, environ):
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000109 encoding = 'latin-1'
110 fake_stdin = BytesIO(data.encode(encoding))
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000111 fake_stdin.seek(0)
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000112 form = cgi.FieldStorage(fp=fake_stdin, environ=environ, encoding=encoding)
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000113
114 result = {}
115 for k, v in dict(form).items():
Florent Xicluna8fbddf12010-03-17 20:29:51 +0000116 result[k] = isinstance(v, list) and form.getlist(k) or v.value
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000117
118 return result
Jeremy Hyltond9827c42000-08-03 22:11:43 +0000119
Thomas Wouters89f507f2006-12-13 04:49:30 +0000120class CgiTests(unittest.TestCase):
Neil Schemenauer66edb622004-07-19 15:38:11 +0000121
Senthil Kumaran6b102f22013-01-23 02:50:15 -0800122 def test_parse_multipart(self):
123 fp = BytesIO(POSTDATA.encode('latin1'))
124 env = {'boundary': BOUNDARY.encode('latin1'),
125 'CONTENT-LENGTH': '558'}
126 result = cgi.parse_multipart(fp, env)
127 expected = {'submit': [b' Add '], 'id': [b'1234'],
128 'file': [b'Testing 123.\n'], 'title': [b'']}
129 self.assertEqual(result, expected)
130
131 def test_fieldstorage_properties(self):
132 fs = cgi.FieldStorage()
133 self.assertFalse(fs)
134 self.assertIn("FieldStorage", repr(fs))
135 self.assertEqual(list(fs), list(fs.keys()))
136 fs.list.append(namedtuple('MockFieldStorage', 'name')('fieldvalue'))
137 self.assertTrue(fs)
138
Senthil Kumaran47b5ddb2012-03-13 01:50:27 -0700139 def test_escape(self):
140 self.assertEqual("test & string", cgi.escape("test & string"))
141 self.assertEqual("&lt;test string&gt;", cgi.escape("<test string>"))
142 self.assertEqual("&quot;test string&quot;", cgi.escape('"test string"', True))
143
Thomas Wouters89f507f2006-12-13 04:49:30 +0000144 def test_strict(self):
145 for orig, expect in parse_strict_test_cases:
146 # Test basic parsing
147 d = do_test(orig, "GET")
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000148 self.assertEqual(d, expect, "Error parsing %s method GET" % repr(orig))
Thomas Wouters89f507f2006-12-13 04:49:30 +0000149 d = do_test(orig, "POST")
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000150 self.assertEqual(d, expect, "Error parsing %s method POST" % repr(orig))
Thomas Wouters89f507f2006-12-13 04:49:30 +0000151
152 env = {'QUERY_STRING': orig}
Thomas Wouters89f507f2006-12-13 04:49:30 +0000153 fs = cgi.FieldStorage(environ=env)
Florent Xicluna8fbddf12010-03-17 20:29:51 +0000154 if isinstance(expect, dict):
Thomas Wouters89f507f2006-12-13 04:49:30 +0000155 # test dict interface
Georg Brandl49d1b4f2008-05-11 21:42:51 +0000156 self.assertEqual(len(expect), len(fs))
Ezio Melotti263cbdf2010-11-29 02:02:10 +0000157 self.assertCountEqual(expect.keys(), fs.keys())
Georg Brandl49d1b4f2008-05-11 21:42:51 +0000158 ##self.assertEqual(norm(expect.values()), norm(fs.values()))
159 ##self.assertEqual(norm(expect.items()), norm(fs.items()))
Thomas Wouters89f507f2006-12-13 04:49:30 +0000160 self.assertEqual(fs.getvalue("nonexistent field", "default"), "default")
161 # test individual fields
162 for key in expect.keys():
163 expect_val = expect[key]
Benjamin Peterson577473f2010-01-19 00:09:57 +0000164 self.assertIn(key, fs)
Thomas Wouters89f507f2006-12-13 04:49:30 +0000165 if len(expect_val) > 1:
Thomas Wouters89f507f2006-12-13 04:49:30 +0000166 self.assertEqual(fs.getvalue(key), expect_val)
167 else:
Thomas Wouters89f507f2006-12-13 04:49:30 +0000168 self.assertEqual(fs.getvalue(key), expect_val[0])
Thomas Wouters89f507f2006-12-13 04:49:30 +0000169
170 def test_log(self):
171 cgi.log("Testing")
Senthil Kumaran6b102f22013-01-23 02:50:15 -0800172 cgi.logfile = "fail/"
173 cgi.initlog("%s", "Testing initlog")
Thomas Wouters89f507f2006-12-13 04:49:30 +0000174 cgi.logfp = StringIO()
175 cgi.initlog("%s", "Testing initlog 1")
176 cgi.log("%s", "Testing log 2")
177 self.assertEqual(cgi.logfp.getvalue(), "Testing initlog 1\nTesting log 2\n")
178 if os.path.exists("/dev/null"):
179 cgi.logfp = None
180 cgi.logfile = "/dev/null"
181 cgi.initlog("%s", "Testing log 3")
Ezio Melotti7ebb7062011-07-14 15:18:57 +0300182 def log_cleanup():
183 """Restore the global state of the log vars."""
184 cgi.logfile = ''
185 cgi.logfp.close()
186 cgi.logfp = None
187 cgi.log = cgi.initlog
188 self.addCleanup(log_cleanup)
Thomas Wouters89f507f2006-12-13 04:49:30 +0000189 cgi.log("Testing log 4")
190
191 def test_fieldstorage_readline(self):
192 # FieldStorage uses readline, which has the capacity to read all
193 # contents of the input file into memory; we use readline's size argument
194 # to prevent that for files that do not contain any newlines in
195 # non-GET/HEAD requests
196 class TestReadlineFile:
197 def __init__(self, file):
198 self.file = file
199 self.numcalls = 0
200
201 def readline(self, size=None):
202 self.numcalls += 1
203 if size:
204 return self.file.readline(size)
Jeremy Hyltond9827c42000-08-03 22:11:43 +0000205 else:
Thomas Wouters89f507f2006-12-13 04:49:30 +0000206 return self.file.readline()
Jeremy Hyltond9827c42000-08-03 22:11:43 +0000207
Thomas Wouters89f507f2006-12-13 04:49:30 +0000208 def __getattr__(self, name):
209 file = self.__dict__['file']
210 a = getattr(file, name)
211 if not isinstance(a, int):
212 setattr(self, name, a)
213 return a
Jeremy Hyltond9827c42000-08-03 22:11:43 +0000214
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000215 f = TestReadlineFile(tempfile.TemporaryFile("wb+"))
Benjamin Peterson807a5a12010-10-30 22:59:37 +0000216 self.addCleanup(f.close)
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000217 f.write(b'x' * 256 * 1024)
Thomas Wouters89f507f2006-12-13 04:49:30 +0000218 f.seek(0)
219 env = {'REQUEST_METHOD':'PUT'}
220 fs = cgi.FieldStorage(fp=f, environ=env)
Benjamin Peterson807a5a12010-10-30 22:59:37 +0000221 self.addCleanup(fs.file.close)
Thomas Wouters89f507f2006-12-13 04:49:30 +0000222 # if we're not chunking properly, readline is only called twice
223 # (by read_binary); if we are chunking properly, it will be called 5 times
224 # as long as the chunksize is 1 << 16.
Benjamin Petersonc9c0f202009-06-30 23:06:06 +0000225 self.assertTrue(f.numcalls > 2)
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000226 f.close()
Jeremy Hyltond9827c42000-08-03 22:11:43 +0000227
Thomas Wouters89f507f2006-12-13 04:49:30 +0000228 def test_fieldstorage_multipart(self):
229 #Test basic FieldStorage multipart parsing
Victor Stinnerf0e293c2011-03-01 23:09:57 +0000230 env = {
231 'REQUEST_METHOD': 'POST',
232 'CONTENT_TYPE': 'multipart/form-data; boundary={}'.format(BOUNDARY),
233 'CONTENT_LENGTH': '558'}
234 fp = BytesIO(POSTDATA.encode('latin-1'))
235 fs = cgi.FieldStorage(fp, environ=env, encoding="latin-1")
Ezio Melottib3aedd42010-11-20 19:04:17 +0000236 self.assertEqual(len(fs.list), 4)
Thomas Wouters89f507f2006-12-13 04:49:30 +0000237 expect = [{'name':'id', 'filename':None, 'value':'1234'},
238 {'name':'title', 'filename':None, 'value':''},
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000239 {'name':'file', 'filename':'test.txt', 'value':b'Testing 123.\n'},
Thomas Wouters89f507f2006-12-13 04:49:30 +0000240 {'name':'submit', 'filename':None, 'value':' Add '}]
241 for x in range(len(fs.list)):
242 for k, exp in expect[x].items():
243 got = getattr(fs.list[x], k)
Ezio Melottib3aedd42010-11-20 19:04:17 +0000244 self.assertEqual(got, exp)
Thomas Wouters00ee7ba2006-08-21 19:07:27 +0000245
Victor Stinnerf0e293c2011-03-01 23:09:57 +0000246 def test_fieldstorage_multipart_non_ascii(self):
247 #Test basic FieldStorage multipart parsing
248 env = {'REQUEST_METHOD':'POST',
249 'CONTENT_TYPE': 'multipart/form-data; boundary={}'.format(BOUNDARY),
250 'CONTENT_LENGTH':'558'}
251 for encoding in ['iso-8859-1','utf-8']:
252 fp = BytesIO(POSTDATA_NON_ASCII.encode(encoding))
253 fs = cgi.FieldStorage(fp, environ=env,encoding=encoding)
254 self.assertEqual(len(fs.list), 1)
255 expect = [{'name':'id', 'filename':None, 'value':'\xe7\xf1\x80'}]
256 for x in range(len(fs.list)):
257 for k, exp in expect[x].items():
258 got = getattr(fs.list[x], k)
259 self.assertEqual(got, exp)
260
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000261 _qs_result = {
262 'key1': 'value1',
263 'key2': ['value2x', 'value2y'],
264 'key3': 'value3',
265 'key4': 'value4'
266 }
267 def testQSAndUrlEncode(self):
268 data = "key2=value2x&key3=value3&key4=value4"
269 environ = {
270 'CONTENT_LENGTH': str(len(data)),
271 'CONTENT_TYPE': 'application/x-www-form-urlencoded',
272 'QUERY_STRING': 'key1=value1&key2=value2y',
273 'REQUEST_METHOD': 'POST',
274 }
275 v = gen_result(data, environ)
276 self.assertEqual(self._qs_result, v)
277
278 def testQSAndFormData(self):
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000279 data = """---123
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000280Content-Disposition: form-data; name="key2"
281
282value2y
283---123
284Content-Disposition: form-data; name="key3"
285
286value3
287---123
288Content-Disposition: form-data; name="key4"
289
290value4
291---123--
292"""
293 environ = {
294 'CONTENT_LENGTH': str(len(data)),
295 'CONTENT_TYPE': 'multipart/form-data; boundary=-123',
296 'QUERY_STRING': 'key1=value1&key2=value2x',
297 'REQUEST_METHOD': 'POST',
298 }
299 v = gen_result(data, environ)
300 self.assertEqual(self._qs_result, v)
301
302 def testQSAndFormDataFile(self):
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000303 data = """---123
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000304Content-Disposition: form-data; name="key2"
305
306value2y
307---123
308Content-Disposition: form-data; name="key3"
309
310value3
311---123
312Content-Disposition: form-data; name="key4"
313
314value4
315---123
316Content-Disposition: form-data; name="upload"; filename="fake.txt"
317Content-Type: text/plain
318
319this is the content of the fake file
320
321---123--
322"""
323 environ = {
324 'CONTENT_LENGTH': str(len(data)),
325 'CONTENT_TYPE': 'multipart/form-data; boundary=-123',
326 'QUERY_STRING': 'key1=value1&key2=value2x',
327 'REQUEST_METHOD': 'POST',
328 }
329 result = self._qs_result.copy()
330 result.update({
Victor Stinner5c23b8e2011-01-14 13:05:21 +0000331 'upload': b'this is the content of the fake file\n'
Benjamin Petersondcf97b92008-07-02 17:30:14 +0000332 })
333 v = gen_result(data, environ)
334 self.assertEqual(result, v)
335
Facundo Batistabd18fd62008-09-08 00:23:29 +0000336 def test_deprecated_parse_qs(self):
Florent Xicluna8fbddf12010-03-17 20:29:51 +0000337 # this func is moved to urllib.parse, this is just a sanity check
338 with check_warnings(('cgi.parse_qs is deprecated, use urllib.parse.'
339 'parse_qs instead', DeprecationWarning)):
Facundo Batistaa27244b2008-09-09 02:43:19 +0000340 self.assertEqual({'a': ['A1'], 'B': ['B3'], 'b': ['B2']},
341 cgi.parse_qs('a=A1&b=B2&B=B3'))
Facundo Batistabd18fd62008-09-08 00:23:29 +0000342
343 def test_deprecated_parse_qsl(self):
Florent Xicluna8fbddf12010-03-17 20:29:51 +0000344 # this func is moved to urllib.parse, this is just a sanity check
345 with check_warnings(('cgi.parse_qsl is deprecated, use urllib.parse.'
346 'parse_qsl instead', DeprecationWarning)):
Facundo Batistaa27244b2008-09-09 02:43:19 +0000347 self.assertEqual([('a', 'A1'), ('b', 'B2'), ('B', 'B3')],
348 cgi.parse_qsl('a=A1&b=B2&B=B3'))
Facundo Batistabd18fd62008-09-08 00:23:29 +0000349
Fred Drake9a0a65b2008-12-04 19:24:50 +0000350 def test_parse_header(self):
351 self.assertEqual(
352 cgi.parse_header("text/plain"),
353 ("text/plain", {}))
354 self.assertEqual(
355 cgi.parse_header("text/vnd.just.made.this.up ; "),
356 ("text/vnd.just.made.this.up", {}))
357 self.assertEqual(
358 cgi.parse_header("text/plain;charset=us-ascii"),
359 ("text/plain", {"charset": "us-ascii"}))
360 self.assertEqual(
361 cgi.parse_header('text/plain ; charset="us-ascii"'),
362 ("text/plain", {"charset": "us-ascii"}))
363 self.assertEqual(
364 cgi.parse_header('text/plain ; charset="us-ascii"; another=opt'),
365 ("text/plain", {"charset": "us-ascii", "another": "opt"}))
366 self.assertEqual(
367 cgi.parse_header('attachment; filename="silly.txt"'),
368 ("attachment", {"filename": "silly.txt"}))
369 self.assertEqual(
370 cgi.parse_header('attachment; filename="strange;name"'),
371 ("attachment", {"filename": "strange;name"}))
372 self.assertEqual(
373 cgi.parse_header('attachment; filename="strange;name";size=123;'),
374 ("attachment", {"filename": "strange;name", "size": "123"}))
Senthil Kumaran1ef0c032011-10-20 01:05:44 +0800375 self.assertEqual(
376 cgi.parse_header('form-data; name="files"; filename="fo\\"o;bar"'),
377 ("form-data", {"name": "files", "filename": 'fo"o;bar'}))
378
Fred Drake9a0a65b2008-12-04 19:24:50 +0000379
Victor Stinnerf0e293c2011-03-01 23:09:57 +0000380BOUNDARY = "---------------------------721837373350705526688164684"
381
382POSTDATA = """-----------------------------721837373350705526688164684
383Content-Disposition: form-data; name="id"
384
3851234
386-----------------------------721837373350705526688164684
387Content-Disposition: form-data; name="title"
388
389
390-----------------------------721837373350705526688164684
391Content-Disposition: form-data; name="file"; filename="test.txt"
392Content-Type: text/plain
393
394Testing 123.
395
396-----------------------------721837373350705526688164684
397Content-Disposition: form-data; name="submit"
398
399 Add\x20
400-----------------------------721837373350705526688164684--
401"""
402
403POSTDATA_NON_ASCII = """-----------------------------721837373350705526688164684
404Content-Disposition: form-data; name="id"
405
406\xe7\xf1\x80
407-----------------------------721837373350705526688164684
408"""
409
Fred Drake9a0a65b2008-12-04 19:24:50 +0000410
Thomas Wouters89f507f2006-12-13 04:49:30 +0000411def test_main():
412 run_unittest(CgiTests)
413
414if __name__ == '__main__':
415 test_main()