Merge branch 'release/3.6.0'

.coveragerc

@@ -0,0 +1,13 @@
+[run]
+source = src/modules/
+
+[report]
+omit = mailparser/version.py
+
+exclude_lines =
+    pragma: no cover
+    except OSError
+    def __repr__
+    raise AssertionError
+    raise NotImplementedError
+    if __name__ == .__main__.:

README.md

@@ -56,6 +56,7 @@ There are other properties to get:
   - attachments
   - sender IP address
   - to domains
+  - timezone
 
 To get custom headers you should replace "-" with "\_".
 Example for header `X-MSMail-Priority`:
@@ -172,6 +173,7 @@ mail.text_plain: only text plain mail parts in a list
 mail.text_html: only text html mail parts in a list
 mail.to
 mail.to_domains
+mail.timezone: returns the timezone, offset from UTC
 ```
 
 ## Usage from command-line

README.rst

@@ -61,7 +61,7 @@ headers <https://www.iana.org/assignments/message-headers/message-headers.xhtml>
 -  to
 
 There are other properties to get: - body - body html - body plain -
-headers - attachments - sender IP address - to domains
+headers - attachments - sender IP address - to domains - timezone
 
 To get custom headers you should replace “-” with “\_”. Example for
 header ``X-MSMail-Priority``:
@@ -178,6 +178,7 @@ Then you can get all parts
    mail.text_html: only text html mail parts in a list
    mail.to
    mail.to_domains
+   mail.timezone: returns the timezone, offset from UTC
 
 Usage from command-line
 -----------------------

mailparser/const.py

@@ -84,4 +84,6 @@
     "body",
     "date",
     "received",
-    "to_domains"])
+    "timezone",
+    "to_domains",
+])

mailparser/mailparser.py

@@ -547,10 +547,23 @@ def date(self):
         conv = None
 
         try:
-            conv = convert_mail_date(date)
+            conv, _ = convert_mail_date(date)
         finally:
             return conv
 
+    @property
+    def timezone(self):
+        """
+        Return timezone. Offset from UTC.
+        """
+        date = self.message.get('date')
+        timezone = 0
+
+        try:
+            _, timezone = convert_mail_date(date)
+        finally:
+            return timezone
+
     @property
     def date_json(self):
         """

mailparser/utils.py

@@ -282,9 +282,15 @@ def receiveds_parsing(receiveds):
 
 
 def convert_mail_date(date):
+    log.debug("Date to parse: {!r}".format(date))
     d = email.utils.parsedate_tz(date)
+    log.debug("Date parsed: {!r}".format(d))
     t = email.utils.mktime_tz(d)
-    return datetime.datetime.utcfromtimestamp(t)
+    log.debug("Date parsed in timestamp: {!r}".format(t))
+    date_utc = datetime.datetime.utcfromtimestamp(t)
+    timezone = d[9] / 3600 if d[9] else 0
+    timezone = "{:+.0f}".format(timezone)
+    return date_utc, timezone
 
 
 def receiveds_not_parsed(receiveds):
@@ -342,7 +348,7 @@ def receiveds_format(receiveds):
             # "for <eboktor@romolo.com>; Tue, 7 Mar 2017 14:29:24 -0800",
             i["date"] = i["date"].split(";")[-1]
             try:
-                j["date_utc"] = convert_mail_date(i["date"])
+                j["date_utc"], _ = convert_mail_date(i["date"])
             except TypeError:
                 j["date_utc"] = None
 
@@ -418,22 +424,22 @@ def get_mail_keys(message):
     return all_parts
 
 
-def safe_print(data):
+def safe_print(data):  # pragma: no cover
     try:
         print(data)
     except UnicodeEncodeError:
         print(data.encode('utf-8'))
 
 
-def print_mail_fingerprints(data):
+def print_mail_fingerprints(data):  # pragma: no cover
     md5, sha1, sha256, sha512 = fingerprints(data)
     print("md5:\t{}".format(md5))
     print("sha1:\t{}".format(sha1))
     print("sha256:\t{}".format(sha256))
     print("sha512:\t{}".format(sha512))
 
 
-def print_attachments(attachments, flag_hash):
+def print_attachments(attachments, flag_hash):  # pragma: no cover
     if flag_hash:
         for i in attachments:
             if i.get("content_transfer_encoding") == "base64":

mailparser/version.py

@@ -17,7 +17,7 @@
 limitations under the License.
 """
 
-__version__ = "3.5.1"  # pragma: no cover
+__version__ = "3.6.0"
 
-if __name__ == "__main__":  # pragma: no cover
+if __name__ == "__main__":
     print(__version__)

tests/test_mail_parser.py

@@ -29,12 +29,15 @@
 
 import mailparser
 from mailparser.utils import (
+    convert_mail_date,
     fingerprints,
     get_header,
     get_to_domains,
     msgconvert,
     ported_open,
-    receiveds_parsing)
+    ported_string,
+    receiveds_parsing,
+)
 
 from mailparser.exceptions import MailParserEnvironmentError
 
@@ -463,6 +466,9 @@ def test_from_file_obj(self):
         result = mail.defects
         self.assertIsInstance(result, list)
 
+        result = mail.timezone
+        self.assertEquals(result, "+1")
+
     def test_get_to_domains(self):
         m = mailparser.parse_from_file(mail_test_6)
 
@@ -477,6 +483,28 @@ def test_get_to_domains(self):
 
         self.assertIsInstance(m.to_domains_json, six.text_type)
 
+    def test_convert_mail_date(self):
+        s = "Mon, 20 Mar 2017 05:12:54 +0600"
+        d, t = convert_mail_date(s)
+        self.assertEquals(t, "+6")
+        self.assertEquals(str(d), "2017-03-19 23:12:54")
+        s = "Mon, 20 Mar 2017 05:12:54 -0600"
+        d, t = convert_mail_date(s)
+        self.assertEquals(t, "-6")
+
+    def test_ported_string(self):
+        raw_data = ""
+        s = ported_string(raw_data)
+        self.assertEquals(s, six.text_type())
+
+        raw_data = "test "
+        s = ported_string(raw_data)
+        self.assertEquals(s, "test")
+
+        raw_data = u"test "
+        s = ported_string(raw_data)
+        self.assertEquals(s, "test")
+
 
 if __name__ == '__main__':
     unittest.main(verbosity=2)