From Sylvain.Thenault@logilab.fr Tue Sep 5 10:25:33 2006

Return-Path: <Sylvain.Thenault@logilab.fr>
X-Original-To: adim@logilab.fr
Delivered-To: adim@logilab.fr
Received: from tucana.logilab.fr (tucana.logilab.fr [172.17.0.4])
	by orion.logilab.fr (Postfix) with ESMTP id B9BF0EA9FF
	for <adim@logilab.fr>; Tue,  5 Sep 2006 10:25:33 +0200 (CEST)
Received: from smtp1-g19.free.fr (smtp1-g19.free.fr [212.27.42.27])
	by tucana.logilab.fr (Postfix) with ESMTP id A751571403E
	for <adim@logilab.fr>; Tue,  5 Sep 2006 10:25:50 +0200 (CEST)
Received: from localhost.localdomain (ram31-2-82-228-88-123.fbx.proxad.net
	[82.228.88.123])
	by smtp1-g19.free.fr (Postfix) with ESMTP id 670B090E2E
	for <adim@logilab.fr>; Tue,  5 Sep 2006 10:25:45 +0200 (CEST)
Content-Type: multipart/mixed; boundary="===============0640966485=="
MIME-Version: 1.0
Message-Id: <patchbomb.1157444811@localhost.localdomain>
Date: Tue,  5 Sep 2006 10:26:51 +0200
To: adim@logilab.fr
X-Spambayes-Classification: ham; 0.02
Status: RO
X-Status: F
Content-Length: 160
Lines: 8
--===============0640966485==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit



--===============0640966485==--

From Sylvain.Thenault@logilab.fr  Tue Sep  5 10:25:34 2006
Return-Path: <Sylvain.Thenault@logilab.fr>
X-Original-To: adim@logilab.fr
Delivered-To: adim@logilab.fr
Received: from tucana.logilab.fr (tucana.logilab.fr [172.17.0.4])
	by orion.logilab.fr (Postfix) with ESMTP id 08C46EAA65
	for <adim@logilab.fr>; Tue,  5 Sep 2006 10:25:34 +0200 (CEST)
Received: from smtp1-g19.free.fr (smtp1-g19.free.fr [212.27.42.27])
	by tucana.logilab.fr (Postfix) with ESMTP id EBF277143EE
	for <adim@logilab.fr>; Tue,  5 Sep 2006 10:25:50 +0200 (CEST)
Received: from localhost.localdomain (ram31-2-82-228-88-123.fbx.proxad.net
	[82.228.88.123])
	by smtp1-g19.free.fr (Postfix) with ESMTP id 744699A992
	for <adim@logilab.fr>; Tue,  5 Sep 2006 10:25:46 +0200 (CEST)
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Subject: [PATCH 1 of 2] blabla
X-Mercurial-Node: c9a40d354365f82164a6c0ffcf9c80deb644e0d5
Message-Id: <c9a40d354365f82164a6.1157444812@localhost.localdomain>
In-Reply-To: <patchbomb.1157444811@localhost.localdomain>
Date: Tue,  5 Sep 2006 10:26:52 +0200
From: syt@logilab.fr
To: adim@logilab.fr
X-Spambayes-Classification: ham; 0.00
Status: RO
X-Status: F
Content-Length: 510
Lines: 17

blabla

3 files changed, 116 insertions, 2 deletions

diff --git a/ChangeLog b/ChangeLog
index e27fe72..4c1569c 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -2,9 +2,11 @@ ChangeLog for logilab.common
 ============================
 
  --
-   * new "date" otion type in optik_ext
+   * new umessage module which provides a class similar to the standard
+     email.Message class but returning unicode strings
    * new clcommands module to handle commands based command line tool 
      (based on the configuration module)
+   * new "date" otion type in optik_ext
    * new AttrObject in testlib to create objects in test with arbitrary attributes
    * add pytest to run project's tests and get rid of all runtests.py
    * add pytest option to enable design-by-contract using aspects
diff --git a/testlib.py b/testlib.py
index b69359f..dba29f6 100644
--- a/testlib.py
+++ b/testlib.py
@@ -1063,6 +1063,6 @@ def enable_dbc(*args):
     return True
 
     
-class AttrObject:
+class AttrObject: # XXX cf mock_object
     def __init__(self, **kwargs):
         self.__dict__.update(kwargs)
diff --git a/umessage.py b/umessage.py
new file mode 100644
index 0000000..a42cb40
--- /dev/null
+++ b/umessage.py
@@ -0,0 +1,112 @@
+"""unicode email support"""
+
+import email
+from email.Utils import parseaddr, parsedate
+from email.Header import decode_header
+
+from mx.DateTime import DateTime
+
+def decode_QP(string):
+    parts = []
+    for decoded, charset in decode_header(string):
+        if charset is None:
+            charset = 'iso-8859-15'
+        parts.append(unicode(decoded, charset))
+
+    return u' '.join(parts)
+
+def message_from_file(fd):
+    try:
+        return UMessage(email.message_from_file(fd))
+    except email.Errors.MessageParseError:
+        return ''
+    
+class UMessage:
+    """Encapsulates an email.Message instance and returns only unicode objects"""
+
+    def __init__(self, message):
+        self.message = message
+
+    # email.Message interface #################################################
+    
+    def get(self, header, default=None):
+        value = self.message.get(header, default)
+        if value:
+            return decode_QP(value)
+        return value
+
+    def get_all(self, header, default=None):
+        return [decode_QP(val) for val in self.message.get_all(header, default)
+                if val is not None]
+    
+    def get_payload(self, index=None, decode=False):
+        message = self.message
+        if index is None:
+            payload = message.get_payload(index, decode)
+            if isinstance(payload, list):
+                return [UMessage(msg) for msg in payload]
+            if message.get_content_maintype() != 'text':
+                return payload
+            return unicode(payload or '', message.get_charset() or 'iso-8859-15')
+        else:
+            payload = UMessage(message.get_payload(index, decode))
+        return payload
+
+    def is_multipart(self):
+        return self.message.is_multipart()
+
+    def get_boundary(self):
+        return self.message.get_boundary()
+
+    def walk(self):
+        for part in self.message.walk():
+            yield UMessage(part)
+    
+    def get_content_maintype(self):
+        return unicode(self.message.get_content_maintype())
+
+    def get_content_type(self):
+        return unicode(self.message.get_content_type())
+
+    def get_filename(self, failobj=None):
+        value = self.message.get_filename(failobj)
+        if value is failobj:
+            return value
+        try:
+            return unicode(value)
+        except UnicodeDecodeError:
+            return u'error decoding filename'
+
+    # other convenience methods ###############################################
+
+    def headers(self):
+        """return an unicode string containing all the message's headers"""
+        values = []
+        for header in self.message.keys():
+            values.append(u'%s: %s' % (header, self.get(header)))
+        return '\n'.join(values)
+
+    def multi_addrs(self, header):
+        """return a list of 2-uple (name, address) for the given address (which
+        is exepected to be an header containing address such as from, to, cc...)
+        """
+        persons = []
+        for person in self.get_all(header, ()):
+            name, mail = parseaddr(person)
+            persons.append((name, mail))
+        return persons
+    
+    def date(self):
+        """return a mx.DateTime object for the email's date or None if no date is
+        set or if it can't be parsed
+        """
+        value = self.get('date')
+        if value:
+            datetuple = parsedate(value)
+            if datetuple:
+                return DateTime(*datetuple[:6])
+        return None
+
+    
+
+