Fix parsing from address with special chars

qreu/address.py

@@ -1,6 +1,10 @@
-from __future__ import absolute_import
+# coding: utf-8
+from __future__ import absolute_import, unicode_literals
 
 from collections import namedtuple
+
+import six
+
 try:
     from collections import UserList
 except ImportError:
@@ -45,3 +49,28 @@ class AddressList(UserList):
     @property
     def addresses(self):
       return [x[1] for x in getaddresses(self.data) if x[1]]
+
+
+def normalize_display_address(addr_string):
+    """
+    Ensures that the display name in an email address is properly quoted
+    if it contains special characters. If a display name is present before
+    a '<...>' address block and is not already quoted, it will be quoted.
+
+    :param addr_string: A string containing a full email address
+                        (e.g. RAMOS ESCOLÀ, PEPITA <pepita@example.com>)
+    :return: A normalized email string with quoted display name if needed
+    """
+    if isinstance(addr_string, six.binary_type):
+        addr_string = addr_string.decode('utf-8')
+    if '<' in addr_string and '>' in addr_string:
+        name_part, addr_part = addr_string.split('<', 1)
+        name = name_part.strip()
+        email = addr_part.strip('> ').strip()
+
+        if name and not (name.startswith('"') and name.endswith('"')):
+            name = '"{}"'.format(name)
+
+        return '{} <{}>'.format(name, email)
+    return addr_string  # If no angle brackets found, return as-is
+

qreu/email.py

@@ -466,7 +466,9 @@ def from_(self):
 
         :return: `address.Address`
         """
-        return address.parse(self.header('From', ''))
+        return address.parse(address.normalize_display_address(
+            self.header('From', ''))
+        )
 
     @property
     def to(self):

spec/address_spec.py

@@ -1,5 +1,5 @@
 # coding=utf-8
-from qreu.address import parse, parse_list, AddressList, Address
+from qreu.address import parse, parse_list, AddressList, Address, normalize_display_address
 from expects import *
 
 
@@ -38,3 +38,29 @@
             expect((a1 + a2).addresses).to(contain_exactly(
                 'u@example.com', 'u2@example.com'
             ))
+
+with context('normalizing address display name'):
+    with it('must quote display name if it contains commas'):
+        addr_str = 'RAMOS ESCOLÀ, PEPITA <pepita@example.com>'
+        normalized = normalize_display_address(addr_str)
+        expect(normalized).to(equal(u'"RAMOS ESCOLÀ, PEPITA" <pepita@example.com>'))
+
+    with it('must not quote display name if already quoted'):
+        addr_str = u'"RAMOS ESCOLÀ, PEPITA" <pepita@example.com>'
+        normalized = normalize_display_address(addr_str)
+        expect(normalized).to(equal(addr_str))
+
+    with it('must not alter address without display name'):
+        addr_str = 'pepita@example.com'
+        normalized = normalize_display_address(addr_str)
+        expect(normalized).to(equal(addr_str))
+
+    with it('must quote display name with semicolon'):
+        addr_str = 'Admin; Name <admin@example.com>'
+        normalized = normalize_display_address(addr_str)
+        expect(normalized).to(equal('"Admin; Name" <admin@example.com>'))
+
+    with it('must ignore address without angle brackets'):
+        addr_str = 'Admin Name admin@example.com'
+        normalized = normalize_display_address(addr_str)
+        expect(normalized).to(equal(addr_str))

spec/fixtures/7.txt

@@ -0,0 +1,3 @@
+From: "RAMOS ESCOLÀ, PEPITA" <pepitaramos@example.com>
+To: SAC <sac@example.com>
+CC: "Bústia General" <general@example.com>

spec/fixtures/8.txt

@@ -0,0 +1 @@
+From: =?iso-8859-1?Q?RAMOS_ESCOL=C0=2C_PEPITA?= <pepita@example.com>

spec/qreu_spec.py

@@ -19,7 +19,7 @@
 with description('Parsing an Email'):
     with before.all:
         self.raw_messages = []
-        for fixture in range(0, 7):
+        for fixture in range(0, 9):
             with open('spec/fixtures/{0}.txt'.format(fixture)) as f:
                 self.raw_messages.append(f.read())
 
@@ -115,6 +115,17 @@
     with it('should parse correctly address from weird header'):
         c = Email.parse(self.raw_messages[5])
         expect(c.from_).to(have_property('address', 'monica@example.com'))
+        expect(c.from_).to(have_property('display_name', u'Mónica de Test Example'))
+
+        c = Email.parse(self.raw_messages[8])
+        expect(c.from_).to(have_property('address', 'pepita@example.com'))
+        expect(c.from_).to(have_property('display_name', u'RAMOS ESCOLÀ, PEPITA'))
+
+    with it('should parse correctly address with accents'):
+        c = Email.parse(self.raw_messages[7])
+        expect(c.from_).to(have_property('display_name', u'RAMOS ESCOLÀ, PEPITA'))
+        expect(c.from_.address).to(equal('pepitaramos@example.com'))
+        expect(c.cc.addresses).to(contain_exactly(u'general@example.com'))
 
 with description("Creating an Email"):
     with context("empty"):