Fix bug in header parsing

[jkamdjou]

This PR fixes a bug where a CRLF ("folding" according to the RFC) in ADDRESSES_HEADERS would cause incorrect value extraction. For example, getaddresses() fails to parse this correctly, which will lead to an incorrect value in the respective mailparser field.

For example, this will fail:

foo = '"test \r\n" <foo@bar.com>'
print(email.utils.getaddresses([foo]))
[('', 'test ')]

According to RFC822:

Unfolding is accomplished by regarding CRLF immediately followed by a LWSP-char 
as equivalent to the LWSP-char.

LWSP-char as defined by the RFC:

LWSP-char   =  SPACE / HTAB                 ; semantics = SPACE

References

See RFC on "folding": https://tools.ietf.org/html/rfc822#section-3.1.1
See "linesep": https://docs.python.org/3/library/email.header.html#email.header.Header.encode

[fedelemantuano]

Hi @jkamdjou,
can you attach an email example?

[jkamdjou]

Sure (note the incorrect from field):

test.eml.txt

mailparser -f test.eml.txt --json
{
  "body": "Test",
  "to_domains": [
    "foo.com"
  ],
  "to": [
    [
      "",
      "bar@foo.com"
    ]
  ],
  "from": [
    [
      "",
      "test"
    ]
  ],
  "subject": "Test",
  "timezone": "-4.0",
  "date": "2019-10-21T18:23:24",
  "has_defects": false
}

[fedelemantuano]

Hi @jkamdjou,
I can't merge this PR because the error is in email library.
When the header passes to your fix, it's already wrong:

h = decode_header_part(self.message.get(name_header, six.text_type()))

The problem is in self.message.get(name_header). I tested your solution but it doesn't work.