lib-python.2.5.test.test_quopri.py Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of jython Show documentation
Show all versions of jython Show documentation
Jython is an implementation of the high-level, dynamic, object-oriented
language Python written in 100% Pure Java, and seamlessly integrated with
the Java platform. It thus allows you to run Python on any Java platform.
from test import test_support
import unittest
import sys, os, cStringIO, subprocess
import quopri
ENCSAMPLE = """\
Here's a bunch of special=20
=A1=A2=A3=A4=A5=A6=A7=A8=A9
=AA=AB=AC=AD=AE=AF=B0=B1=B2=B3
=B4=B5=B6=B7=B8=B9=BA=BB=BC=BD=BE
=BF=C0=C1=C2=C3=C4=C5=C6
=C7=C8=C9=CA=CB=CC=CD=CE=CF
=D0=D1=D2=D3=D4=D5=D6=D7
=D8=D9=DA=DB=DC=DD=DE=DF
=E0=E1=E2=E3=E4=E5=E6=E7
=E8=E9=EA=EB=EC=ED=EE=EF
=F0=F1=F2=F3=F4=F5=F6=F7
=F8=F9=FA=FB=FC=FD=FE=FF
characters... have fun!
"""
# First line ends with a space
DECSAMPLE = "Here's a bunch of special \n" + \
"""\
\xa1\xa2\xa3\xa4\xa5\xa6\xa7\xa8\xa9
\xaa\xab\xac\xad\xae\xaf\xb0\xb1\xb2\xb3
\xb4\xb5\xb6\xb7\xb8\xb9\xba\xbb\xbc\xbd\xbe
\xbf\xc0\xc1\xc2\xc3\xc4\xc5\xc6
\xc7\xc8\xc9\xca\xcb\xcc\xcd\xce\xcf
\xd0\xd1\xd2\xd3\xd4\xd5\xd6\xd7
\xd8\xd9\xda\xdb\xdc\xdd\xde\xdf
\xe0\xe1\xe2\xe3\xe4\xe5\xe6\xe7
\xe8\xe9\xea\xeb\xec\xed\xee\xef
\xf0\xf1\xf2\xf3\xf4\xf5\xf6\xf7
\xf8\xf9\xfa\xfb\xfc\xfd\xfe\xff
characters... have fun!
"""
def withpythonimplementation(testfunc):
def newtest(self):
# Test default implementation
testfunc(self)
# Test Python implementation
if quopri.b2a_qp is not None or quopri.a2b_qp is not None:
oldencode = quopri.b2a_qp
olddecode = quopri.a2b_qp
try:
quopri.b2a_qp = None
quopri.a2b_qp = None
testfunc(self)
finally:
quopri.b2a_qp = oldencode
quopri.a2b_qp = olddecode
newtest.__name__ = testfunc.__name__
return newtest
class QuopriTestCase(unittest.TestCase):
# Each entry is a tuple of (plaintext, encoded string). These strings are
# used in the "quotetabs=0" tests.
STRINGS = (
# Some normal strings
('hello', 'hello'),
('''hello
there
world''', '''hello
there
world'''),
('''hello
there
world
''', '''hello
there
world
'''),
('\201\202\203', '=81=82=83'),
# Add some trailing MUST QUOTE strings
('hello ', 'hello=20'),
('hello\t', 'hello=09'),
# Some long lines. First, a single line of 108 characters
('xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx\xd8\xd9\xda\xdb\xdc\xdd\xde\xdfxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx',
'''xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx=D8=D9=DA=DB=DC=DD=DE=DFx=
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx'''),
# A line of exactly 76 characters, no soft line break should be needed
('yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy',
'yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy'),
# A line of 77 characters, forcing a soft line break at position 75,
# and a second line of exactly 2 characters (because the soft line
# break `=' sign counts against the line length limit).
('zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz',
'''zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz=
zz'''),
# A line of 151 characters, forcing a soft line break at position 75,
# with a second line of exactly 76 characters and no trailing =
('zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz',
'''zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz=
zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz'''),
# A string containing a hard line break, but which the first line is
# 151 characters and the second line is exactly 76 characters. This
# should leave us with three lines, the first which has a soft line
# break, and which the second and third do not.
('''yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy
zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz''',
'''yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy=
yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy
zzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzzz'''),
# Now some really complex stuff ;)
(DECSAMPLE, ENCSAMPLE),
)
# These are used in the "quotetabs=1" tests.
ESTRINGS = (
('hello world', 'hello=20world'),
('hello\tworld', 'hello=09world'),
)
# These are used in the "header=1" tests.
HSTRINGS = (
('hello world', 'hello_world'),
('hello_world', 'hello=5Fworld'),
)
@withpythonimplementation
def test_encodestring(self):
for p, e in self.STRINGS:
self.assert_(quopri.encodestring(p) == e)
@withpythonimplementation
def test_decodestring(self):
for p, e in self.STRINGS:
self.assert_(quopri.decodestring(e) == p)
@withpythonimplementation
def test_idempotent_string(self):
for p, e in self.STRINGS:
self.assert_(quopri.decodestring(quopri.encodestring(e)) == e)
@withpythonimplementation
def test_encode(self):
for p, e in self.STRINGS:
infp = cStringIO.StringIO(p)
outfp = cStringIO.StringIO()
quopri.encode(infp, outfp, quotetabs=False)
self.assert_(outfp.getvalue() == e)
@withpythonimplementation
def test_decode(self):
for p, e in self.STRINGS:
infp = cStringIO.StringIO(e)
outfp = cStringIO.StringIO()
quopri.decode(infp, outfp)
self.assert_(outfp.getvalue() == p)
@withpythonimplementation
def test_embedded_ws(self):
for p, e in self.ESTRINGS:
self.assert_(quopri.encodestring(p, quotetabs=True) == e)
self.assert_(quopri.decodestring(e) == p)
@withpythonimplementation
def test_encode_header(self):
for p, e in self.HSTRINGS:
self.assert_(quopri.encodestring(p, header=True) == e)
@withpythonimplementation
def test_decode_header(self):
for p, e in self.HSTRINGS:
self.assert_(quopri.decodestring(e, header=True) == p)
def test_scriptencode(self):
(p, e) = self.STRINGS[-1]
process = subprocess.Popen([sys.executable, "-mquopri"],
stdin=subprocess.PIPE, stdout=subprocess.PIPE)
cout, cerr = process.communicate(p)
# On Windows, Python will output the result to stdout using
# CRLF, as the mode of stdout is text mode. To compare this
# with the expected result, we need to do a line-by-line comparison.
self.assert_(cout.splitlines() == e.splitlines())
def test_scriptdecode(self):
(p, e) = self.STRINGS[-1]
process = subprocess.Popen([sys.executable, "-mquopri", "-d"],
stdin=subprocess.PIPE, stdout=subprocess.PIPE)
cout, cerr = process.communicate(e)
self.assert_(cout.splitlines() == p.splitlines())
def test_main():
test_support.run_unittest(QuopriTestCase)
if __name__ == "__main__":
test_main()