test_rfc822.py

来自「mallet是自然语言处理、机器学习领域的一个开源项目。」· Python 代码 · 共 212 行

import rfc822
import sys
import test_support
import unittest

try:
    from cStringIO import StringIO
except ImportError:
    from StringIO import StringIO


class MessageTestCase(unittest.TestCase):
    def create_message(self, msg):
        return rfc822.Message(StringIO(msg))

    def test_get(self):
        msg = self.create_message(
            'To: "last, first" <[e-mail]>\n\ntest\n')
        self.assert_(msg.get("to") == '"last, first" <[e-mail]>')
        self.assert_(msg.get("TO") == '"last, first" <[e-mail]>')
        self.assert_(msg.get("No-Such-Header") is None)
        self.assert_(msg.get("No-Such-Header", "No-Such-Value")
                     == "No-Such-Value")

    def test_setdefault(self):
        msg = self.create_message(
            'To: "last, first" <[e-mail]>\n\ntest\n')
        self.assert_(not msg.has_key("New-Header"))
        self.assert_(msg.setdefault("New-Header", "New-Value") == "New-Value")
        self.assert_(msg.setdefault("New-Header", "Different-Value")
                     == "New-Value")
        self.assert_(msg["new-header"] == "New-Value")

        self.assert_(msg.setdefault("Another-Header") == "")
        self.assert_(msg["another-header"] == "")

    def check(self, msg, results):
        """Check addresses and the date."""
        m = self.create_message(msg)
        i = 0
        for n, a in m.getaddrlist('to') + m.getaddrlist('cc'):
            try:
                mn, ma = results[i][0], results[i][1]
            except IndexError:
                print 'extra parsed address:', repr(n), repr(a)
                continue
            i = i + 1
            if mn == n and ma == a:
                pass
            else:
                print 'not found:', repr(n), repr(a)

        out = m.getdate('date')
        if out:
            self.assertEqual(out,
                             (1999, 1, 13, 23, 57, 35, 0, 0, 0),
                             "date conversion failed")


    # Note: all test cases must have the same date (in various formats),
    # or no date!

    def test_basic(self):
        self.check(
            'Date:    Wed, 13 Jan 1999 23:57:35 -0500\n'
            'From:    Guido van Rossum <[e-mail]>\n'
            'To:      "Guido van\n'
            '\t : Rossum" <[e-mail]>\n'
            'Subject: test2\n'
            '\n'
            'test2\n',
            [('Guido van\n\t : Rossum', '[e-mail]')])

        self.check(
            'From: Barry <[e-mail]\n'
            'To: [e-mail] (Guido: the Barbarian)\n'
            'Subject: nonsense\n'
            'Date: Wednesday, January 13 1999 23:57:35 -0500\n'
            '\n'
            'test',
            [('Guido: the Barbarian', '[e-mail]')])

        self.check(
            'From: Barry <[e-mail]\n'
            'To: [e-mail] (Guido: the Barbarian)\n'
            'Cc: "Guido: the Madman" <[e-mail]>\n'
            'Date:  13-Jan-1999 23:57:35 EST\n'
            '\n'
            'test',
            [('Guido: the Barbarian', '[e-mail]'),
             ('Guido: the Madman', '[e-mail]')
             ])

        self.check(
            'To: "The monster with\n'
            '     the very long name: Guido" <[e-mail]>\n'
            'Date:    Wed, 13 Jan 1999 23:57:35 -0500\n'
            '\n'
            'test',
            [('The monster with\n     the very long name: Guido',
              '[e-mail]')])

        self.check(
            'To: "Amit J. Patel" <[e-mail]>\n'
            'CC: Mike Fletcher <[e-mail]>,\n'
            '        "\'[e-mail]\'" <[e-mail]>\n'
            'Cc: [e-mail], [e-mail]\n'
            'Cc: [e-mail]\n'
            'Date:    Wed, 13 Jan 1999 23:57:35 -0500\n'
            '\n'
            'test',
            [('Amit J. Patel', '[e-mail]'),
             ('Mike Fletcher', '[e-mail]'),
             ("'[e-mail]'", '[e-mail]'),
             ('', '[e-mail]'),
             ('', '[e-mail]'),
             ('', '[e-mail]'),
             ])

        self.check(
            'To: Some One <[e-mail]>\n'
            'From: Anudder Persin <[e-mail]>\n'
            'Date:\n'
            '\n'
            'test',
            [('Some One', '[e-mail]')])

        self.check(
            'To: [e-mail] (User J. Person)\n\n',
            [('User J. Person', '[e-mail]')])

    def test_twisted(self):
        # This one is just twisted.  I don't know what the proper
        # result should be, but it shouldn't be to infloop, which is
        # what used to happen!
        self.check(
            'To: <[smtp:[e-mail]][e-mail]>\n'
            'Date:    Wed, 13 Jan 1999 23:57:35 -0500\n'
            '\n'
            'test',
            [('', ''),
             ('', '[e-mail]'),
             ('', '[e-mail]'),
             ])

    def test_commas_in_full_name(self):
        # This exercises the old commas-in-a-full-name bug, which
        # should be doing the right thing in recent versions of the
        # module.
        self.check(
            'To: "last, first" <[e-mail]>\n'
            '\n'
            'test',
            [('last, first', '[e-mail]')])

    def test_quoted_name(self):
        self.check(
            'To: (Comment stuff) "Quoted name"@somewhere.com\n'
            '\n'
            'test',
            [('Comment stuff', '"Quoted name"@somewhere.com')])

    def test_bogus_to_header(self):
        self.check(
            'To: :\n'
            'Cc: [e-mail]\n'
            'Date:    Wed, 13 Jan 1999 23:57:35 -0500\n'
            '\n'
            'test',
            [('', '[e-mail]')])

    def test_addr_ipquad(self):
        self.check(
            'To: guido@[132.151.1.21]\n'
            '\n'
            'foo',
            [('', 'guido@[132.151.1.21]')])

    def test_rfc2822_phrases(self):
        # RFC 2822 (the update to RFC 822) specifies that dots in phrases are
        # obsolete syntax, which conforming programs MUST recognize but NEVER
        # generate (see $4.1 Miscellaneous obsolete tokens).  This is a
        # departure from RFC 822 which did not allow dots in non-quoted
        # phrases.
        self.check('To: User J. Person <[e-mail]>\n\n',
                   [('User J. Person', '[e-mail]')])

    # This takes to long to add to the test suite
##    def test_an_excrutiatingly_long_address_field(self):
##        OBSCENELY_LONG_HEADER_MULTIPLIER = 10000
##        oneaddr = ('Person' * 10) + '@' + ('.'.join(['dom']*10)) + '.com'
##        addr = ', '.join([oneaddr] * OBSCENELY_LONG_HEADER_MULTIPLIER)
##        lst = rfc822.AddrlistClass(addr).getaddrlist()
##        self.assertEqual(len(lst), OBSCENELY_LONG_HEADER_MULTIPLIER)


    def test_parseaddr(self):
        eq = self.assertEqual
        eq(rfc822.parseaddr('<>'), ('', ''))
        eq(rfc822.parseaddr('[e-mail]'), ('', '[e-mail]'))
        eq(rfc822.parseaddr('[e-mail] (Bea A. Person)'),
           ('Bea A. Person', '[e-mail]'))
        eq(rfc822.parseaddr('Cynthia Person <[e-mail]>'),
           ('Cynthia Person', '[e-mail]'))

def test_main():
    test_support.run_unittest(MessageTestCase)


if __name__ == "__main__":
    test_main()