add get command to cli

Signed-off-by: Thomas Oettli <spacefreak@noop.ch>

MANIFEST.in

@@ -1,3 +1,3 @@
 include LICENSE README.md
-recursive-include docs *
-recursive-include misc *
+recursive-include pyquarantine/docs *
+recursive-include pyquarantine/misc *

README.md

@@ -8,7 +8,7 @@ Each quarantine can be configured with a quarantine type, notification type, whi
 
 Addionally, pyquarantine-milter provides a sanitized, harmless version of the text parts of e-mails, which can be embedded in e-mail notifications. This makes it easier for users to decide, if a match is a false-positive or not. If a matching quarantine provides a quarantine ID of the original e-mail, it is also available as a template variable. This is useful if you want to add links to a webservice to notification e-mails, to give your users the ability to release e-mails or whitelist the from-address for example. The webservice then releases the e-mail from the quarantine.
 
-The project is currently in alpha status, but will soon be used in a productive enterprise environment and possibly existing bugs will be fixed.
+The project is currently in beta status, but it is already used in a productive enterprise environment which processes about a million e-mails per month.
 
 ## Requirements
 * pymilter <https://pythonhosted.org/pymilter/>
@@ -36,8 +36,8 @@ The following configuration options are mandatory in the global section:
 The following configuration options are mandatory in each quarantine section:
 * **regex**  
   Case insensitive regular expression to filter e-mail headers.
-* **quarantine_type**  
-  One of the quarantine-types described below.
+* **storage_type**  
+  One of the storage types described below.
 * **action**  
   One of the actions described below.
 * **notification_type**  
@@ -45,18 +45,18 @@ The following configuration options are mandatory in each quarantine section:
 * **whitelist_type**  
   One of the whitelist types described below.
 * **smtp_host**  
-  SMTP host to inject original e-mails. This is needed if not all recipients of an e-mail are whitelisted
+  SMTP host used to release original e-mails from the quarantine.
 * **smtp_port**  
   SMTP port
 
 The following configuration options are optional in each quarantine section:
-* **ignore_hosts**  
+* **host_whitelist**  
   Comma-separated list of host and network addresses to be ignored by this quarantine.
 * **reject_reason**  
   Reason to return to the client if action is set to reject.
 
 
-### Quarantine types
+### Storage types
 * **NONE**  
   Original e-mails scrapped, sent to nirvana, black-holed or however you want to call it.
 
@@ -64,7 +64,7 @@ The following configuration options are optional in each quarantine section:
   Original e-mails are stored on the filesystem with a unique filename. The filename is available as a
   template variable used in notifiaction templates.  
   The following configuration options are mandatory for this quarantine type:
-  * **quarantine_directory**  
+  * **storage_directory**  
     The directory in which quarantined e-mails are stored.
 
 
@@ -77,11 +77,17 @@ The following configuration options are optional in each quarantine section:
 
   The following template variables are available:
   * **{EMAIL_ENVELOPE_FROM}**  
-    E-mail from-address received by the milter.
+    E-mail sender address received by the milter.
+  * **{EMAIL_ENVELOPE_FROM_URL}**  
+    Like EMAIL_ENVELOPE_FROM, but URL encoded
   * **{EMAIL_FROM}**  
-    Value of the from header of the original e-mail.
-  * **{EMAIL_TO}**  
+    Value of the FROM header of the original e-mail.
+  * **{EMAIL_ENVELOPE_TO}**  
     E-mail recipient address of this notification.
+  * **{EMAIL_ENVELOPE_TO_URL}**  
+    Like EMAIL_ENVELOPE_TO, but URL encoded
+  * **{EMAIL_TO}**  
+    Value of the TO header of the original e-mail.
   * **{EMAIL_SUBJECT}**  
     Configured e-mail subject.
   * **{EMAIL_QUARANTINE_ID}**  
@@ -97,6 +103,10 @@ The following configuration options are optional in each quarantine section:
     Content of a named subgroup, 'subgroup_name' will be replaced by its name.
 
   The following configuration options are mandatory for this notification type:
+  * **notification_email_smtp_host**  
+    SMTP host used to send notification e-mails.
+  * **notification_email_smtp_port**  
+    SMTP port.
   * **notification_email_envelope_from**  
     Notification e-mail envelope from-address.
   * **notification_email_from**  

pyquarantine/cli.py

@@ -20,26 +20,43 @@ import logging.handlers
 import sys
 import time
 
-import pyquarantine
+from pyquarantine import QuarantineMilter, setup_milter
+from pyquarantine import __version__ as version
 
-from pyquarantine.version import __version__ as version
 
-def _get_quarantine_obj(config, quarantine):
+def _get_quarantine(quarantines, name):
     try:
-        quarantine_obj = next((q["quarantine_obj"]
-                               for q in config if q["name"] == quarantine))
+        quarantine = next((q for q in quarantines if q.name == name))
     except StopIteration:
-        raise RuntimeError("invalid quarantine '{}'".format(quarantine))
-    return quarantine_obj
+        raise RuntimeError("invalid quarantine 'name'")
+    return quarantine
 
 
-def _get_whitelist_obj(config, quarantine):
-    try:
-        whitelist_obj = next((q["whitelist_obj"]
-                              for q in config if q["name"] == quarantine))
-    except StopIteration:
-        raise RuntimeError("invalid quarantine '{}'".format(quarantine))
-    return whitelist_obj
+def _get_storage(quarantines, name):
+    quarantine = _get_quarantine(quarantines, name)
+    storage = quarantine.get_storage()
+    if not storage:
+        raise RuntimeError(
+                "storage type is set to NONE")
+    return storage
+
+
+def _get_notification(quarantines, name):
+    quarantine = _get_quarantine(quarantines, name)
+    notification = quarantine.get_notification()
+    if not notification:
+        raise RuntimeError(
+                "notification type is set to NONE")
+    return notification
+
+
+def _get_whitelist(quarantines, name):
+    quarantine = _get_quarantine(quarantines, name)
+    whitelist = quarantine.get_whitelist()
+    if not whitelist:
+        raise RuntimeError(
+                "whitelist type is set to NONE")
+    return whitelist
 
 
 def print_table(columns, rows):
@@ -56,10 +73,10 @@ def print_table(columns, rows):
         # get the length of the longest value
         lengths.append(
             len(str(max(rows, key=lambda x: len(str(x[key])))[key])))
-        # use the the longer one
+        # use the longer one
         length = max(lengths)
         column_lengths.append(length)
-        column_formats.append("{{:<{}}}".format(length))
+        column_formats.append(f"{{:<{length}}}")
 
     # define row format
     row_format = " | ".join(column_formats)
@@ -83,47 +100,70 @@ def print_table(columns, rows):
         print(row_format.format(*row))
 
 
-def list_quarantines(config, args):
+def list_quarantines(quarantines, args):
     if args.batch:
-        print("\n".join([quarantine["name"] for quarantine in config]))
+        print("\n".join([q.name for q in quarantines]))
     else:
+        qlist = []
+        for q in quarantines:
+            storage = q.get_storage()
+            if storage:
+                storage_type = q.get_storage().storage_type
+            else:
+                storage_type = "NONE"
+
+            notification = q.get_notification()
+            if notification:
+                notification_type = q.get_notification().notification_type
+            else:
+                notification_type = "NONE"
+
+            whitelist = q.get_whitelist()
+            if whitelist:
+                whitelist_type = q.get_whitelist().whitelist_type
+            else:
+                whitelist_type = "NONE"
+
+            qlist.append({
+                "name": q.name,
+                "storage": storage_type,
+                "notification": notification_type,
+                "whitelist": whitelist_type,
+                "action": q.action})
         print_table(
-            [("Name", "name"), ("Quarantine", "quarantine_type"),
-             ("Notification", "notification_type"), ("Action", "action")],
-            config
+            [("Name", "name"),
+             ("Storage", "storage"),
+             ("Notification", "notification"),
+             ("Whitelist", "whitelist"),
+             ("Action", "action")],
+            qlist
         )
 
 
-def list_quarantine_emails(config, args):
+def list_quarantine_emails(quarantines, args):
     logger = logging.getLogger(__name__)
-
-    # get quarantine object
-    quarantine = _get_quarantine_obj(config, args.quarantine)
-    if quarantine is None:
-        raise RuntimeError(
-            "quarantine type is set to None, unable to list emails")
-
+    storage = _get_storage(quarantines, args.quarantine)
     # find emails and transform some metadata values to strings
     rows = []
-    emails = quarantine.find(
-        mailfrom=args.mailfrom,
-        recipients=args.recipients,
-        older_than=args.older_than)
-    for quarantine_id, metadata in emails.items():
-        row = emails[quarantine_id]
-        row["quarantine_id"] = quarantine_id
+    emails = storage.find(
+        args.mailfrom, args.recipients, args.older_than)
+    for storage_id, metadata in emails.items():
+        row = emails[storage_id]
+        row["storage_id"] = storage_id
         row["date"] = time.strftime(
             '%Y-%m-%d %H:%M:%S',
             time.localtime(
                 metadata["date"]))
         row["mailfrom"] = metadata["mailfrom"]
         row["recipient"] = metadata["recipients"].pop(0)
-        row["subject"] = emails[quarantine_id]["headers"]["subject"][:60]
+        if "subject" not in emails[storage_id]["headers"].keys():
+            emails[storage_id]["headers"]["subject"] = ""
+        row["subject"] = emails[storage_id]["headers"]["subject"][:60].strip()
         rows.append(row)
 
         if metadata["recipients"]:
             row = {
-                "quarantine_id": "",
+                "storage_id": "",
                 "date": "",
                 "mailfrom": "",
                 "recipient": metadata["recipients"].pop(0),
@@ -137,23 +177,18 @@ def list_quarantine_emails(config, args):
         return
 
     if not emails:
-        logger.info("quarantine '{}' is empty".format(args.quarantine))
+        logger.info(f"quarantine '{args.quarantine}' is empty")
     print_table(
-        [("Quarantine-ID", "quarantine_id"), ("Date", "date"),
+        [("Quarantine-ID", "storage_id"), ("Date", "date"),
          ("From", "mailfrom"), ("Recipient(s)", "recipient"),
          ("Subject", "subject")],
         rows
     )
 
 
-def list_whitelist(config, args):
+def list_whitelist(quarantines, args):
     logger = logging.getLogger(__name__)
-
-    # get whitelist object
-    whitelist = _get_whitelist_obj(config, args.quarantine)
-    if whitelist is None:
-        raise RuntimeError(
-            "whitelist type is set to None, unable to list entries")
+    whitelist = _get_whitelist(quarantines, args.quarantine)
 
     # find whitelist entries
     entries = whitelist.find(
@@ -162,8 +197,7 @@ def list_whitelist(config, args):
         older_than=args.older_than)
     if not entries:
         logger.info(
-            "whitelist of quarantine '{}' is empty".format(
-                args.quarantine))
+            f"whitelist of quarantine '{args.quarantine}' is empty")
         return
 
     # transform some values to strings
@@ -184,14 +218,9 @@ def list_whitelist(config, args):
     )
 
 
-def add_whitelist_entry(config, args):
+def add_whitelist_entry(quarantines, args):
     logger = logging.getLogger(__name__)
-
-    # get whitelist object
-    whitelist = _get_whitelist_obj(config, args.quarantine)
-    if whitelist is None:
-        raise RuntimeError(
-            "whitelist type is set to None, unable to add entries")
+    whitelist = _get_whitelist(quarantines, args.quarantine)
 
     # check existing entries
     entries = whitelist.check(args.mailfrom, args.recipient)
@@ -229,53 +258,41 @@ def add_whitelist_entry(config, args):
     logger.info("whitelist entry added successfully")
 
 
-def delete_whitelist_entry(config, args):
+def delete_whitelist_entry(quarantines, args):
     logger = logging.getLogger(__name__)
-
-    whitelist = _get_whitelist_obj(config, args.quarantine)
-    if whitelist is None:
-        raise RuntimeError(
-            "whitelist type is set to None, unable to delete entries")
-
+    whitelist = _get_whitelist(quarantines, args.quarantine)
     whitelist.delete(args.whitelist_id)
     logger.info("whitelist entry deleted successfully")
 
 
-def notify_email(config, args):
+def notify(quarantines, args):
     logger = logging.getLogger(__name__)
-
-    quarantine = _get_quarantine_obj(config, args.quarantine)
-    if quarantine is None:
-        raise RuntimeError(
-            "quarantine type is set to None, unable to send notification")
+    quarantine = _get_quarantine(quarantines, args.quarantine)
     quarantine.notify(args.quarantine_id, args.recipient)
-    logger.info("sent notification successfully")
+    logger.info("notification sent successfully")
 
 
-def release_email(config, args):
+def release(quarantines, args):
     logger = logging.getLogger(__name__)
-
-    quarantine = _get_quarantine_obj(config, args.quarantine)
-    if quarantine is None:
-        raise RuntimeError(
-            "quarantine type is set to None, unable to release email")
-
+    quarantine = _get_quarantine(quarantines, args.quarantine)
     quarantine.release(args.quarantine_id, args.recipient)
     logger.info("quarantined email released successfully")
 
 
-def delete_email(config, args):
+def delete(quarantines, args):
     logger = logging.getLogger(__name__)
-
-    quarantine = _get_quarantine_obj(config, args.quarantine)
-    if quarantine is None:
-        raise RuntimeError(
-            "quarantine type is set to None, unable to delete email")
-
-    quarantine.delete(args.quarantine_id, args.recipient)
+    storage = _get_storage(quarantines, args.quarantine)
+    storage.delete(args.quarantine_id, args.recipient)
     logger.info("quarantined email deleted successfully")
 
 
+def get(quarantines, args):
+    storage = _get_storage(quarantines, args.quarantine)
+    fp, _ = storage.get_mail(args.quarantine_id)
+    print(fp.read().decode())
+    fp.close()
+
+
 class StdErrFilter(logging.Filter):
     def filter(self, rec):
         return rec.levelno in (logging.ERROR, logging.WARNING)
@@ -298,7 +315,7 @@ def main():
         "-c", "--config",
         help="Config files to read.",
         nargs="+", metavar="CFG",
-        default=pyquarantine.QuarantineMilter.get_configfiles())
+        default=QuarantineMilter.get_cfg_files())
     parser.add_argument(
         "-d", "--debug",
         help="Log debugging messages.",
@@ -307,7 +324,7 @@ def main():
         "-v", "--version",
         help="Print version.",
         action="version",
-        version="%(prog)s ({})".format(version))
+        version=f"%(prog)s ({version})")
     parser.set_defaults(syslog=False)
     subparsers = parser.add_subparsers(
         dest="command",
@@ -388,7 +405,7 @@ def main():
         "-a", "--all",
         help="Release email for all recipients.",
         action="store_true")
-    quarantine_notify_parser.set_defaults(func=notify_email)
+    quarantine_notify_parser.set_defaults(func=notify)
     # quarantine release command
     quarantine_release_parser = quarantine_subparsers.add_parser(
         "release",
@@ -415,7 +432,7 @@ def main():
         "-a", "--all",
         help="Release email for all recipients.",
         action="store_true")
-    quarantine_release_parser.set_defaults(func=release_email)
+    quarantine_release_parser.set_defaults(func=release)
     # quarantine delete command
     quarantine_delete_parser = quarantine_subparsers.add_parser(
         "delete",
@@ -441,7 +458,18 @@ def main():
         "-a", "--all",
         help="Delete email for all recipients.",
         action="store_true")
-    quarantine_delete_parser.set_defaults(func=delete_email)
+    quarantine_delete_parser.set_defaults(func=delete)
+    # quarantine get command
+    quarantine_get_parser = quarantine_subparsers.add_parser(
+        "get",
+        description="Get email from quarantine.",
+        help="Get email from quarantine",
+        formatter_class=formatter_class)
+    quarantine_get_parser.add_argument(
+        "quarantine_id",
+        metavar="ID",
+        help="Quarantine ID.")
+    quarantine_get_parser.set_defaults(func=get)
 
     # whitelist command group
     whitelist_parser = subparsers.add_parser(
@@ -552,8 +580,8 @@ def main():
 
     # try to generate milter configs
     try:
-        global_config, config = pyquarantine.generate_milter_config(
-            config_files=args.config, configtest=True)
+        setup_milter(
+            cfg_files=args.config, test=True)
     except RuntimeError as e:
         logger.error(e)
         sys.exit(255)
@@ -574,7 +602,7 @@ def main():
 
     # call the commands function
     try:
-        args.func(config, args)
+        args.func(QuarantineMilter.quarantines, args)
     except RuntimeError as e:
         logger.error(e)
         sys.exit(1)

pyquarantine/docs/pyquarantine.conf.example

@@ -30,12 +30,12 @@ preferred_quarantine_action = last
 
 
 [spam]
-# Option: ignore_hosts
+# Option: host_whitelist
 # Notes:  Set a list of host and network addresses to be ignored by this quarantine.
 #         All the common host/network notations are supported, including IPv6.
 # Value:  [ HOST ]
 #
-ignore_hosts = 10.0.0.0/8,172.16.0.0/12,192.168.0.0/16
+host_whitelist = 10.0.0.0/8,172.16.0.0/12,192.168.0.0/16
 
 # Option: regex
 # Notes:  Set the case insensitive regular expression to match against email headers.
@@ -57,18 +57,18 @@ smtp_host = 127.0.0.1
 #
 smtp_port = 25
 
-# Option: quarantine_type
-# Notes:  Set the quarantine type.
+# Option: storage_type
+# Notes:  Set the storage type.
 # Values: [ file | none ]
 #
-quarantine_type = file
+storage_type = file
 
-# Option: quarantine_directory
+# Option: storage_directory
 # Notes:  Set the directory to store quarantined emails.
 #         This option is needed by quarantine type 'file'.
 # Values: [ DIRECTORY ]
 #
-quarantine_directory = /var/lib/pyquarantine/spam
+storage_directory = /var/lib/pyquarantine/spam
 
 # Option: action
 # Notes:  Set the milter action to perform if email is processed by this quarantine.
@@ -88,6 +88,18 @@ reject_reason = Message rejected
 #
 notification_type = email
 
+# Option: notification_email_smtp_host
+# Notes:  Set the SMTP host. It will be used to send notification e-mails.
+# Values: [ HOSTNAME | IP_ADDRESS ]
+#
+notification_email_smtp_host = 127.0.0.1
+
+# Option: notification_email_smtp_port
+# Notes:  Set the SMTP port.
+# Values: [ PORT ]
+#
+notification_email_smtp_port = 25
+
 # Option: notification_email_envelope_from
 # Notes:  Set the envelope-from address used when sending notification emails.
 #         This option is needed by notification type 'email'.

pyquarantine/docs/templates/notification.template

@@ -10,6 +10,10 @@
     <td><b>From:</b></td>
     <td>{EMAIL_FROM}</td>
    </tr>
+   <tr>
+    <td><b>Envelope-To:</b></td>
+    <td>{EMAIL_ENVELOPE_TO}</td>
+   </tr>
    <tr>
     <td><b>To:</b></td>
     <td>{EMAIL_TO}</td>

pyquarantine/mailer.py

@@ -14,7 +14,6 @@
 
 import logging
 import smtplib
-import sys
 
 from multiprocessing import Process, Queue
 
@@ -26,6 +25,10 @@ process = None
 
 def smtp_send(smtp_host, smtp_port, mailfrom, recipient, mail):
     s = smtplib.SMTP(host=smtp_host, port=smtp_port)
+    s.ehlo()
+    if s.has_extn("STARTTLS"):
+        s.starttls()
+        s.ehlo()
     s.sendmail(mailfrom, [recipient], mail)
     s.quit()
 
@@ -41,23 +44,22 @@ def mailprocess():
             if not m:
                 break
 
-            smtp_host, smtp_port, queueid, mailfrom, recipient, mail, emailtype = m
+            smtp_host, smtp_port, qid, mailfrom, recipient, mail, emailtype = m
             try:
                 smtp_send(smtp_host, smtp_port, mailfrom, recipient, mail)
             except Exception as e:
                 logger.error(
-                    "{}: error while sending {} to '{}': {}".format(
-                        queueid, emailtype, recipient, e))
+                    f"{qid}: error while sending {emailtype} "
+                    f"to '{recipient}': {e}")
             else:
                 logger.info(
-                    "{}: successfully sent {} to: {}".format(
-                        queueid, emailtype, recipient))
+                    f"{qid}: successfully sent {emailtype} to: {recipient}")
     except KeyboardInterrupt:
         pass
     logger.debug("mailer process terminated")
 
 
-def sendmail(smtp_host, smtp_port, queueid, mailfrom, recipients, mail,
+def sendmail(smtp_host, smtp_port, qid, mailfrom, recipients, mail,
              emailtype="email"):
     "Send an email."
     global logger
@@ -77,8 +79,8 @@ def sendmail(smtp_host, smtp_port, queueid, mailfrom, recipients, mail,
     for recipient in recipients:
         try:
             queue.put(
-                (smtp_host, smtp_port, queueid, mailfrom, recipient, mail,
+                (smtp_host, smtp_port, qid, mailfrom, recipient, mail,
                  emailtype),
                 timeout=30)
-        except Queue.Full as e:
+        except Queue.Full:
             raise RuntimeError("email queue is full")

pyquarantine/notifications.py

@@ -17,26 +17,27 @@ import logging
 import re
 
 from bs4 import BeautifulSoup
-from cgi import escape
 from collections import defaultdict
+from email.policy import default as default_policy
 from email.mime.multipart import MIMEMultipart
 from email.mime.text import MIMEText
 from email.mime.image import MIMEImage
+from html import escape
 from os.path import basename
+from urllib.parse import quote
 
 from pyquarantine import mailer
 
 
 class BaseNotification(object):
     "Notification base class"
+    notification_type = "base"
 
-    def __init__(self, global_config, config, configtest=False):
-        self.quarantine_name = config["name"]
-        self.global_config = global_config
-        self.config = config
+    def __init__(self, name, global_cfg, cfg, test=False):
+        self.name = name
         self.logger = logging.getLogger(__name__)
 
-    def notify(self, queueid, quarantine_id, mailfrom, recipients, headers,
+    def notify(self, qid, storage_id, mailfrom, recipients, headers,
                fp, subgroups=None, named_subgroups=None, synchronous=False):
         fp.seek(0)
         pass
@@ -44,8 +45,7 @@ class BaseNotification(object):
 
 class EMailNotification(BaseNotification):
     "Notification class to send notifications via mail."
-    _html_text = "text/html"
-    _plain_text = "text/plain"
+    notification_type = "email"
     _bad_tags = [
         "applet",
         "embed",
@@ -109,44 +109,40 @@ class EMailNotification(BaseNotification):
         "width"
     ]
 
-    def __init__(self, global_config, config, configtest=False):
+    def __init__(self, name, global_cfg, cfg, test=False):
         super(EMailNotification, self).__init__(
-            global_config, config, configtest)
+            name, global_cfg, cfg, test)
 
-        # check if mandatory options are present in config
-        for option in [
-            "smtp_host",
-            "smtp_port",
-            "notification_email_envelope_from",
-            "notification_email_from",
-            "notification_email_subject",
-            "notification_email_template",
-                "notification_email_embedded_imgs"]:
-            if option not in self.config.keys() and option in self.global_config.keys():
-                self.config[option] = self.global_config[option]
-            if option not in self.config.keys():
-                raise RuntimeError(
-                    "mandatory option '{}' not present in config section '{}' or 'global'".format(
-                        option, self.quarantine_name))
-
-        # check if optional config options are present in config
         defaults = {
             "notification_email_replacement_img": "",
             "notification_email_strip_images": "false",
             "notification_email_parser_lib": "lxml"
         }
-        for option in defaults.keys():
-            if option not in config.keys() and \
-                    option in global_config.keys():
-                config[option] = global_config[option]
-            if option not in config.keys():
-                config[option] = defaults[option]
+        # check config
+        for opt in [
+            "notification_email_smtp_host",
+            "notification_email_smtp_port",
+            "notification_email_envelope_from",
+            "notification_email_from",
+            "notification_email_subject",
+            "notification_email_template",
+                "notification_email_embedded_imgs"] + list(defaults.keys()):
+            if opt in cfg:
+                continue
+            if opt in global_cfg:
+                cfg[opt] = global_cfg[opt]
+            elif opt in defaults:
+                cfg[opt] = defaults[opt]
+            else:
+                raise RuntimeError(
+                    f"mandatory option '{opt}' not present in config "
+                    f"section '{self.name}' or 'global'")
 
-        self.smtp_host = self.config["smtp_host"]
-        self.smtp_port = self.config["smtp_port"]
-        self.mailfrom = self.config["notification_email_envelope_from"]
-        self.from_header = self.config["notification_email_from"]
-        self.subject = self.config["notification_email_subject"]
+        self.smtp_host = cfg["notification_email_smtp_host"]
+        self.smtp_port = cfg["notification_email_smtp_port"]
+        self.mailfrom = cfg["notification_email_envelope_from"]
+        self.from_header = cfg["notification_email_from"]
+        self.subject = cfg["notification_email_subject"]
 
         testvars = defaultdict(str, test="TEST")
 
@@ -155,46 +151,48 @@ class EMailNotification(BaseNotification):
             self.from_header.format_map(testvars)
         except ValueError as e:
             raise RuntimeError(
-                "error parsing notification_email_from: {}".format(e))
+                f"error parsing notification_email_from: {e}")
 
         # test-parse subject
         try:
             self.subject.format_map(testvars)
         except ValueError as e:
             raise RuntimeError(
-                "error parsing notification_email_subject: {}".format(e))
+                f"error parsing notification_email_subject: {e}")
 
         # read and parse email notification template
         try:
             self.template = open(
-                self.config["notification_email_template"], "r").read()
+                cfg["notification_email_template"], "r").read()
             self.template.format_map(testvars)
         except IOError as e:
-            raise RuntimeError("error reading template: {}".format(e))
+            raise RuntimeError(f"error reading template: {e}")
         except ValueError as e:
-            raise RuntimeError("error parsing template: {}".format(e))
+            raise RuntimeError(f"error parsing template: {e}")
 
-        strip_images = self.config["notification_email_strip_images"].strip().upper()
+        strip_images = cfg["notification_email_strip_images"].strip().upper()
         if strip_images in ["TRUE", "ON", "YES"]:
             self.strip_images = True
         elif strip_images in ["FALSE", "OFF", "NO"]:
             self.strip_images = False
         else:
-            raise RuntimeError("error parsing notification_email_strip_images: unknown value")
+            raise RuntimeError(
+                "error parsing notification_email_strip_images: unknown value")
 
-        self.parser_lib = self.config["notification_email_parser_lib"].strip()
+        self.parser_lib = cfg["notification_email_parser_lib"].strip()
         if self.parser_lib not in ["lxml", "html.parser"]:
-            raise RuntimeError("error parsing notification_email_parser_lib: unknown value")
+            raise RuntimeError(
+                "error parsing notification_email_parser_lib: unknown value")
 
         # read email replacement image if specified
-        replacement_img = self.config["notification_email_replacement_img"].strip()
+        replacement_img = cfg["notification_email_replacement_img"].strip()
         if not self.strip_images and replacement_img:
             try:
                 self.replacement_img = MIMEImage(
                     open(replacement_img, "rb").read())
             except IOError as e:
                 raise RuntimeError(
-                    "error reading replacement image: {}".format(e))
+                    f"error reading replacement image: {e}")
             else:
                 self.replacement_img.add_header(
                     "Content-ID", "<removed_for_security_reasons>")
@@ -203,78 +201,87 @@ class EMailNotification(BaseNotification):
 
         # read images to embed if specified
         embedded_img_paths = [
-            p.strip() for p in self.config["notification_email_embedded_imgs"].split(",") if p]
+            p.strip() for p in cfg["notification_email_embedded_imgs"].split(
+                ",") if p]
         self.embedded_imgs = []
         for img_path in embedded_img_paths:
             # read image
             try:
                 img = MIMEImage(open(img_path, "rb").read())
             except IOError as e:
-                raise RuntimeError("error reading image: {}".format(e))
+                raise RuntimeError(f"error reading image: {e}")
             else:
-                img.add_header("Content-ID", "<{}>".format(basename(img_path)))
+                filename = basename(img_path)
+                img.add_header("Content-ID", f"<{filename}>")
                 self.embedded_imgs.append(img)
 
-    def get_text(self, queueid, part):
-        "Get the mail text in html form from email part."
-        mimetype = part.get_content_type()
+    def get_email_body_soup(self, qid, msg):
+        "Extract and decode email body and return it as BeautifulSoup object."
+        # try to find the body part
+        self.logger.debug(f"{qid}: trying to find email body")
+        try:
+            body = msg.get_body(preferencelist=("html", "plain"))
+        except Exception as e:
+            self.logger.error(
+                f"{qid}: an error occured in "
+                f"email.message.EmailMessage.get_body: {e}")
+            body = None
 
+        if body:
+            charset = body.get_content_charset() or "utf-8"
+            content = body.get_payload(decode=True)
+            try:
+                content = content.decode(encoding=charset, errors="replace")
+            except LookupError:
+                self.logger.info(
+                    f"{qid}: unknown encoding '{charset}', "
+                    f"falling back to UTF-8")
+                content = content.decode("utf-8", errors="replace")
+            content_type = body.get_content_type()
+            if content_type == "text/plain":
+                # convert text/plain to text/html
                 self.logger.debug(
-            "{}: extracting content of email text part".format(queueid))
-        text = part.get_payload(decode=True)
-
-        if mimetype == EMailNotification._plain_text:
-            self.logger.debug(
-                "{}: content mimetype is {}, converting to {}".format(
-                    queueid, mimetype, self._html_text))
-            text = re.sub(r"^(.*)$", r"\1<br/>",
-                          escape(text.decode()), flags=re.MULTILINE)
+                    f"{qid}: content type is {content_type}, "
+                    f"converting to text/html")
+                content = re.sub(r"^(.*)$", r"\1<br/>",
+                                 escape(content, quote=False),
+                                 flags=re.MULTILINE)
             else:
                 self.logger.debug(
-                "{}: content mimetype is {}".format(
-                    queueid, mimetype))
+                    f"{qid}: content type is {content_type}")
+        else:
+            self.logger.error(
+                f"{qid}: unable to find email body")
+            content = "ERROR: unable to find email body"
+
+        # create BeautifulSoup object
+        length = len(content)
         self.logger.debug(
-            "{}: trying to create BeatufilSoup object with parser lib {}, "
-            "text length is {} bytes".format(
-                queueid, self.parser_lib, len(text)))
-        soup = BeautifulSoup(text, self.parser_lib)
+            f"{qid}: trying to create BeatufilSoup object with "
+            f"parser lib {self.parser_lib}, "
+            f"text length is {length} bytes")
+        soup = BeautifulSoup(content, self.parser_lib)
         self.logger.debug(
-            "{}: sucessfully created BeautifulSoup object".format(queueid))
+            f"{qid}: sucessfully created BeautifulSoup object")
+
         return soup
 
-    def get_text_multipart(self, queueid, msg, preferred=_html_text):
-        "Get the mail text of a multipart email in html form."
-        soup = None
-
-        for part in msg.get_payload():
-            mimetype = part.get_content_type()
-            if mimetype in [EMailNotification._plain_text,
-                            EMailNotification._html_text]:
-                soup = self.get_text(queueid, part)
-            elif mimetype.startswith("multipart"):
-                soup = self.get_text_multipart(queueid, part, preferred)
-
-            if soup is not None and mimetype == preferred:
-                break
-        return soup
-
-    def sanitize(self, queueid, soup):
+    def sanitize(self, qid, soup):
         "Sanitize mail html text."
-        self.logger.debug("{}: sanitizing email text".format(queueid))
+        self.logger.debug(f"{qid}: sanitizing email text")
 
         # completly remove bad elements
         for element in soup(EMailNotification._bad_tags):
             self.logger.debug(
-                "{}: removing dangerous tag '{}' and its content".format(
-                    queueid, element.name))
+                f"{qid}: removing dangerous tag '{element.name}' "
+                f"and its content")
             element.extract()
 
         # remove not whitelisted elements, but keep their content
         for element in soup.find_all(True):
             if element.name not in EMailNotification._good_tags:
                 self.logger.debug(
-                    "{}: removing tag '{}', keep its content".format(
-                        queueid, element.name))
+                    f"{qid}: removing tag '{element.name}', keep its content")
                 element.replaceWithChildren()
 
         # remove not whitelisted attributes
@@ -283,45 +290,24 @@ class EMailNotification(BaseNotification):
                 if attribute not in EMailNotification._good_attributes:
                     if element.name == "a" and attribute == "href":
                         self.logger.debug(
-                            "{}: setting attribute href to '#' on tag '{}'".format(
-                                queueid, element.name))
+                            f"{qid}: setting attribute href to '#' "
+                            f"on tag '{element.name}'")
                         element["href"] = "#"
                     else:
                         self.logger.debug(
-                            "{}: removing attribute '{}' from tag '{}'".format(
-                                queueid, attribute, element.name))
+                            f"{qid}: removing attribute '{attribute}' "
+                            f"from tag '{element.name}'")
                         del(element.attrs[attribute])
         return soup
 
-    def get_html_text_part(self, queueid, msg):
-        "Get the mail text of an email in html form."
-        soup = None
-        mimetype = msg.get_content_type()
-
-        self.logger.debug(
-            "{}: trying to find text part of email".format(queueid))
-        if mimetype in [EMailNotification._plain_text,
-                        EMailNotification._html_text]:
-            soup = self.get_text(queueid, msg)
-        elif mimetype.startswith("multipart"):
-            soup = self.get_text_multipart(queueid, msg)
-
-        if soup is None:
-            self.logger.error(
-                "{}: unable to extract text part of email".format(queueid))
-            text = "ERROR: unable to extract text from email body"
-            soup = BeautifulSoup(text, "lxml", "UTF-8")
-
-        return soup
-
-    def notify(self, queueid, quarantine_id, mailfrom, recipients, headers, fp,
+    def notify(self, qid, storage_id, mailfrom, recipients, headers, fp,
                subgroups=None, named_subgroups=None, synchronous=False):
         "Notify recipients via email."
         super(
             EMailNotification,
             self).notify(
-            queueid,
-            quarantine_id,
+            qid,
+            storage_id,
             mailfrom,
             recipients,
             headers,
@@ -330,61 +316,62 @@ class EMailNotification(BaseNotification):
             named_subgroups,
             synchronous)
 
-        # extract html text from email
-        self.logger.debug(
-            "{}: extraction email text from original email".format(queueid))
-        soup = self.get_html_text_part(
-            queueid, email.message_from_binary_file(fp))
+        # extract body from email
+        soup = self.get_email_body_soup(
+            qid, email.message_from_binary_file(fp, policy=default_policy))
 
         # replace picture sources
         image_replaced = False
         if self.strip_images:
             self.logger.debug(
-                "{}: looking for images to strip".format(queueid))
+                f"{qid}: looking for images to strip")
             for element in soup("img"):
                 if "src" in element.attrs.keys():
                     self.logger.debug(
-                        "{}: strip image: {}".format(
-                            queueid, element["src"]))
+                        f"{qid}: strip image: {element['src']}")
                 element.extract()
         elif self.replacement_img:
             self.logger.debug(
-                "{}: looking for images to replace".format(queueid))
+                f"{qid}: looking for images to replace")
             for element in soup("img"):
                 if "src" in element.attrs.keys():
                     self.logger.debug(
-                        "{}: replacing image: {}".format(
-                            queueid, element["src"]))
+                        f"{qid}: replacing image: {element['src']}")
                 element["src"] = "cid:removed_for_security_reasons"
                 image_replaced = True
 
         # sanitizing email text of original email
-        sanitized_text = self.sanitize(queueid, soup)
+        sanitized_text = self.sanitize(qid, soup)
         del soup
 
         # sending email notifications
         for recipient in recipients:
             self.logger.debug(
-                "{}: generating notification email for '{}'".format(
-                    queueid, recipient))
-            self.logger.debug("{}: parsing email template".format(queueid))
+                f"{qid}: generating notification email for '{recipient}'")
+            self.logger.debug(f"{qid}: parsing email template")
 
             # generate dict containing all template variables
-            variables = defaultdict(str,
+            variables = defaultdict(
+                    str,
                     EMAIL_HTML_TEXT=sanitized_text,
-                                    EMAIL_FROM=escape(headers["from"]),
-                                    EMAIL_ENVELOPE_FROM=escape(mailfrom),
-                                    EMAIL_TO=escape(recipient),
-                                    EMAIL_SUBJECT=escape(headers["subject"]),
-                                    EMAIL_QUARANTINE_ID=quarantine_id)
+                    EMAIL_FROM=escape(headers["from"], quote=False),
+                    EMAIL_ENVELOPE_FROM=escape(mailfrom, quote=False),
+                    EMAIL_ENVELOPE_FROM_URL=escape(quote(mailfrom),
+                                                   quote=False),
+                    EMAIL_TO=escape(headers["to"], quote=False),
+                    EMAIL_ENVELOPE_TO=escape(recipient, quote=False),
+                    EMAIL_ENVELOPE_TO_URL=escape(quote(recipient)),
+                    EMAIL_SUBJECT=escape(headers["subject"], quote=False),
+                    EMAIL_QUARANTINE_ID=storage_id)
 
             if subgroups:
                 number = 0
                 for subgroup in subgroups:
-                    variables["SUBGROUP_{}".format(number)] = escape(subgroup)
+                    variables[f"SUBGROUP_{number}"] = escape(subgroup,
+                                                             quote=False)
             if named_subgroups:
                 for key, value in named_subgroups.items():
-                    named_subgroups[key] = escape(value)
+                    named_subgroups[key] = escape(value, quote=False)
                 variables.update(named_subgroups)
 
             # parse template
@@ -401,26 +388,24 @@ class EMailNotification(BaseNotification):
 
             if image_replaced:
                 self.logger.debug(
-                    "{}: attaching notification_replacement_img".format(queueid))
+                    f"{qid}: attaching notification_replacement_img")
                 msg.attach(self.replacement_img)
 
             for img in self.embedded_imgs:
-                self.logger.debug("{}: attaching imgage".format(queueid))
+                self.logger.debug(f"{qid}: attaching imgage")
                 msg.attach(img)
 
             self.logger.debug(
-                "{}: sending notification email to: {}".format(
-                    queueid, recipient))
+                f"{qid}: sending notification email to: {recipient}")
             if synchronous:
                 try:
                     mailer.smtp_send(self.smtp_host, self.smtp_port,
                                      self.mailfrom, recipient, msg.as_string())
                 except Exception as e:
                     raise RuntimeError(
-                        "error while sending email to '{}': {}".format(
-                            recipient, e))
+                        f"error while sending email to '{recipient}': {e}")
             else:
-                mailer.sendmail(self.smtp_host, self.smtp_port, queueid,
+                mailer.sendmail(self.smtp_host, self.smtp_port, qid,
                                 self.mailfrom, recipient, msg.as_string(),
                                 "notification email")
 

pyquarantine/quarantines.py

@@ -1,309 +0,0 @@
-# PyQuarantine-Milter is free software: you can redistribute it and/or modify
-# it under the terms of the GNU General Public License as published by
-# the Free Software Foundation, either version 3 of the License, or
-# (at your option) any later version.
-#
-# PyQuarantine-Milter is distributed in the hope that it will be useful,
-# but WITHOUT ANY WARRANTY; without even the implied warranty of
-# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-# GNU General Public License for more details.
-#
-# You should have received a copy of the GNU General Public License
-# along with PyQuarantineMilter.  If not, see <http://www.gnu.org/licenses/>.
-#
-
-import json
-import logging
-import os
-
-from calendar import timegm
-from datetime import datetime
-from glob import glob
-from shutil import copyfileobj
-from time import gmtime
-
-from pyquarantine import mailer
-
-
-class BaseQuarantine(object):
-    "Quarantine base class"
-
-    def __init__(self, global_config, config, configtest=False):
-        self.name = config["name"]
-        self.global_config = global_config
-        self.config = config
-        self.logger = logging.getLogger(__name__)
-
-    def add(self, queueid, mailfrom, recipients, headers,
-            fp, subgroups=None, named_subgroups=None):
-        "Add email to quarantine."
-        fp.seek(0)
-        return ""
-
-    def find(self, mailfrom=None, recipients=None, older_than=None):
-        "Find emails in quarantine."
-        return
-
-    def get_metadata(self, quarantine_id):
-        "Return metadata of quarantined email."
-        return
-
-    def delete(self, quarantine_id, recipient=None):
-        "Delete email from quarantine."
-        return
-
-    def notify(self, quarantine_id, recipient=None):
-        "Notify recipient about email in quarantine."
-        if not self.config["notification_obj"]:
-            raise RuntimeError(
-                "notification type is set to None, unable to send notifications")
-        return
-
-    def release(self, quarantine_id, recipient=None):
-        "Release email from quarantine."
-        return
-
-
-class FileQuarantine(BaseQuarantine):
-    "Quarantine class to store mails on filesystem."
-
-    def __init__(self, global_config, config, configtest=False):
-        super(FileQuarantine, self).__init__(global_config, config, configtest)
-
-        # check if mandatory options are present in config
-        for option in ["quarantine_directory"]:
-            if option not in self.config.keys() and option in self.global_config.keys():
-                self.config[option] = self.global_config[option]
-            if option not in self.config.keys():
-                raise RuntimeError(
-                    "mandatory option '{}' not present in config section '{}' or 'global'".format(
-                        option, self.name))
-        self.directory = self.config["quarantine_directory"]
-
-        # check if quarantine directory exists and is writable
-        if not os.path.isdir(self.directory) or not os.access(
-                self.directory, os.W_OK):
-            raise RuntimeError(
-                "file quarantine directory '{}' does not exist or is not writable".format(
-                    self.directory))
-        self._metadata_suffix = ".metadata"
-
-    def _save_datafile(self, quarantine_id, fp):
-        datafile = os.path.join(self.directory, quarantine_id)
-        try:
-            with open(datafile, "wb") as f:
-                copyfileobj(fp, f)
-        except IOError as e:
-            raise RuntimeError("unable save data file: {}".format(e))
-
-    def _save_metafile(self, quarantine_id, metadata):
-        metafile = os.path.join(
-            self.directory, "{}{}".format(
-                quarantine_id, self._metadata_suffix))
-        try:
-            with open(metafile, "w") as f:
-                json.dump(metadata, f, indent=2)
-        except IOError as e:
-            raise RuntimeError("unable to save metadata file: {}".format(e))
-
-    def _remove(self, quarantine_id):
-        datafile = os.path.join(self.directory, quarantine_id)
-        metafile = "{}{}".format(datafile, self._metadata_suffix)
-
-        try:
-            os.remove(metafile)
-        except IOError as e:
-            raise RuntimeError("unable to remove metadata file: {}".format(e))
-
-        try:
-            os.remove(datafile)
-        except IOError as e:
-            raise RuntimeError("unable to remove data file: {}".format(e))
-
-    def add(self, queueid, mailfrom, recipients, headers,
-            fp, subgroups=None, named_subgroups=None):
-        "Add email to file quarantine and return quarantine-id."
-        super(
-            FileQuarantine,
-            self).add(
-            queueid,
-            mailfrom,
-            recipients,
-            headers,
-            fp,
-            subgroups,
-            named_subgroups)
-        quarantine_id = "{}_{}".format(
-            datetime.now().strftime("%Y%m%d%H%M%S"), queueid)
-
-        # save mail
-        self._save_datafile(quarantine_id, fp)
-
-        # save metadata
-        metadata = {
-            "mailfrom": mailfrom,
-            "recipients": recipients,
-            "headers": headers,
-            "date": timegm(gmtime()),
-            "queue_id": queueid,
-            "subgroups": subgroups,
-            "named_subgroups": named_subgroups
-        }
-        try:
-            self._save_metafile(quarantine_id, metadata)
-        except RuntimeError as e:
-            datafile = os.path.join(self.directory, quarantine_id)
-            os.remove(datafile)
-            raise e
-
-        return quarantine_id
-
-    def get_metadata(self, quarantine_id):
-        "Return metadata of quarantined email."
-        super(FileQuarantine, self).get_metadata(quarantine_id)
-
-        metafile = os.path.join(
-            self.directory, "{}{}".format(
-                quarantine_id, self._metadata_suffix))
-        if not os.path.isfile(metafile):
-            raise RuntimeError(
-                "invalid quarantine id '{}'".format(quarantine_id))
-
-        try:
-            with open(metafile, "r") as f:
-                metadata = json.load(f)
-        except IOError as e:
-            raise RuntimeError("unable to read metadata file: {}".format(e))
-        except json.JSONDecodeError as e:
-            raise RuntimeError(
-                "invalid meta file '{}': {}".format(
-                    metafile, e))
-
-        return metadata
-
-    def find(self, mailfrom=None, recipients=None, older_than=None):
-        "Find emails in quarantine."
-        super(FileQuarantine, self).find(mailfrom, recipients, older_than)
-        if isinstance(mailfrom, str):
-            mailfrom = [mailfrom]
-        if isinstance(recipients, str):
-            recipients = [recipients]
-
-        emails = {}
-        metafiles = glob(os.path.join(
-            self.directory, "*{}".format(self._metadata_suffix)))
-        for metafile in metafiles:
-            if not os.path.isfile(metafile):
-                continue
-
-            quarantine_id = os.path.basename(
-                metafile[:-len(self._metadata_suffix)])
-            metadata = self.get_metadata(quarantine_id)
-            if older_than is not None:
-                if timegm(gmtime()) - metadata["date"] < (older_than * 86400):
-                    continue
-
-            if mailfrom is not None:
-                if metadata["mailfrom"] not in mailfrom:
-                    continue
-
-            if recipients is not None:
-                if len(recipients) == 1 and \
-                        recipients[0] not in metadata["recipients"]:
-                    continue
-                elif len(set(recipients + metadata["recipients"])) == len(recipients + metadata["recipients"]):
-                    continue
-
-            emails[quarantine_id] = metadata
-
-        return emails
-
-    def delete(self, quarantine_id, recipient=None):
-        "Delete email in quarantine."
-        super(FileQuarantine, self).delete(quarantine_id, recipient)
-
-        try:
-            metadata = self.get_metadata(quarantine_id)
-        except RuntimeError as e:
-            raise RuntimeError("unable to delete email: {}".format(e))
-
-        if recipient is None:
-            self._remove(quarantine_id)
-        else:
-            if recipient not in metadata["recipients"]:
-                raise RuntimeError("invalid recipient '{}'".format(recipient))
-
-            metadata["recipients"].remove(recipient)
-            if not metadata["recipients"]:
-                self._remove(quarantine_id)
-            else:
-                self._save_metafile(quarantine_id, metadata)
-
-    def notify(self, quarantine_id, recipient=None):
-        "Notify recipient about email in quarantine."
-        super(FileQuarantine, self).notify(quarantine_id, recipient)
-
-        try:
-            metadata = self.get_metadata(quarantine_id)
-        except RuntimeError as e:
-            raise RuntimeError("unable to release email: {}".format(e))
-
-        if recipient is not None:
-            if recipient not in metadata["recipients"]:
-                raise RuntimeError("invalid recipient '{}'".format(recipient))
-            recipients = [recipient]
-        else:
-            recipients = metadata["recipients"]
-
-        datafile = os.path.join(self.directory, quarantine_id)
-        try:
-            with open(datafile, "rb") as fp:
-                self.config["notification_obj"].notify(
-                    metadata["queue_id"], quarantine_id, metadata["mailfrom"],
-                    recipients, metadata["headers"], fp,
-                    metadata["subgroups"], metadata["named_subgroups"],
-                    synchronous=True)
-        except IOError as e:
-            raise RuntimeError
-
-    def release(self, quarantine_id, recipient=None):
-        "Release email from quarantine."
-        super(FileQuarantine, self).release(quarantine_id, recipient)
-
-        try:
-            metadata = self.get_metadata(quarantine_id)
-        except RuntimeError as e:
-            raise RuntimeError("unable to release email: {}".format(e))
-
-        if recipient is not None:
-            if recipient not in metadata["recipients"]:
-                raise RuntimeError("invalid recipient '{}'".format(recipient))
-            recipients = [recipient]
-        else:
-            recipients = metadata["recipients"]
-
-        datafile = os.path.join(self.directory, quarantine_id)
-        try:
-            with open(datafile, "rb") as f:
-                mail = f.read()
-        except IOError as e:
-            raise RuntimeError("unable to read data file: {}".format(e))
-
-        for recipient in recipients:
-            try:
-                mailer.smtp_send(
-                    self.config["smtp_host"],
-                    self.config["smtp_port"],
-                    metadata["mailfrom"],
-                    recipient,
-                    mail)
-            except Exception as e:
-                raise RuntimeError(
-                    "error while sending email to '{}': {}".format(
-                        recipient, e))
-
-            self.delete(quarantine_id, recipient)
-
-
-# list of quarantine types and their related quarantine classes
-TYPES = {"file": FileQuarantine}

pyquarantine/run.py

@@ -22,7 +22,8 @@ import sys
 
 import pyquarantine
 
-from pyquarantine.version import __version__ as version
+from pyquarantine import __version__ as version
+
 
 def main():
     "Run PyQuarantine-Milter."
@@ -35,11 +36,11 @@ def main():
         "-c", "--config",
         help="List of config files to read.",
         nargs="+",
-        default=pyquarantine.QuarantineMilter.get_configfiles())
+        default=pyquarantine.QuarantineMilter.get_cfg_files())
     parser.add_argument(
         "-s", "--socket",
         help="Socket used to communicate with the MTA.",
-        required=True)
+        default="inet:8899@127.0.0.1")
     parser.add_argument(
         "-d", "--debug",
         help="Log debugging messages.",
@@ -52,7 +53,7 @@ def main():
         "-v", "--version",
         help="Print version.",
         action="version",
-        version="%(prog)s ({})".format(version))
+        version=f"%(prog)s ({version})")
     args = parser.parse_args()
 
     # setup logging
@@ -61,24 +62,24 @@ def main():
     syslog_name = logname
     if args.debug:
         loglevel = logging.DEBUG
-        logname = "{}[%(name)s]".format(logname)
-        syslog_name = "{}: [%(name)s] %(levelname)s".format(syslog_name)
+        logname = f"{logname}[%(name)s]"
+        syslog_name = f"{syslog_name}: [%(name)s] %(levelname)s"
 
     # set config files for milter class
-    pyquarantine.QuarantineMilter.set_configfiles(args.config)
+    pyquarantine.QuarantineMilter.set_cfg_files(args.config)
     root_logger = logging.getLogger()
     root_logger.setLevel(loglevel)
 
     # setup console log
     stdouthandler = logging.StreamHandler(sys.stdout)
     stdouthandler.setLevel(logging.DEBUG)
-    formatter = logging.Formatter("%(message)s".format(logname))
+    formatter = logging.Formatter("%(message)s")
     stdouthandler.setFormatter(formatter)
     root_logger.addHandler(stdouthandler)
     logger = logging.getLogger(__name__)
     if args.test:
         try:
-            pyquarantine.generate_milter_config(args.test)
+            pyquarantine.setup_milter(test=args.test)
             print("Configuration ok")
         except RuntimeError as e:
             logger.error(e)
@@ -86,7 +87,7 @@ def main():
         else:
             sys.exit(0)
     formatter = logging.Formatter(
-        "%(asctime)s {}: [%(levelname)s] %(message)s".format(logname),
+        f"%(asctime)s {logname}: [%(levelname)s] %(message)s",
         datefmt="%Y-%m-%d %H:%M:%S")
     stdouthandler.setFormatter(formatter)
 
@@ -94,25 +95,21 @@ def main():
     sysloghandler = logging.handlers.SysLogHandler(
         address="/dev/log", facility=logging.handlers.SysLogHandler.LOG_MAIL)
     sysloghandler.setLevel(loglevel)
-    formatter = logging.Formatter("{}: %(message)s".format(syslog_name))
+    formatter = logging.Formatter(f"{syslog_name}: %(message)s")
     sysloghandler.setFormatter(formatter)
     root_logger.addHandler(sysloghandler)
 
     logger.info("PyQuarantine-Milter starting")
     try:
         # generate milter config
-        global_config, config = pyquarantine.generate_milter_config()
+        pyquarantine.setup_milter()
     except RuntimeError as e:
         logger.error(e)
         sys.exit(255)
 
-    pyquarantine.QuarantineMilter.global_config = global_config
-    pyquarantine.QuarantineMilter.config = config
-
     # register to have the Milter factory create instances of your class:
     Milter.factory = pyquarantine.QuarantineMilter
     Milter.set_exception_policy(Milter.TEMPFAIL)
-    # Milter.set_flags(0)       # tell sendmail which features we use
 
     # run milter
     rc = 0
@@ -122,6 +119,7 @@ def main():
     except Milter.milter.error as e:
         logger.error(e)
         rc = 255
+
     pyquarantine.mailer.queue.put(None)
     logger.info("PyQuarantine-Milter terminated")
     sys.exit(rc)

pyquarantine/storages.py

@@ -0,0 +1,252 @@
+# PyQuarantine-Milter is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# PyQuarantine-Milter is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with PyQuarantineMilter.  If not, see <http://www.gnu.org/licenses/>.
+#
+
+import json
+import logging
+import os
+
+from calendar import timegm
+from datetime import datetime
+from glob import glob
+from shutil import copyfileobj
+from time import gmtime
+
+
+class BaseMailStorage(object):
+    "Mail storage base class"
+    storage_type = "base"
+
+    def __init__(self, name, global_cfg, cfg, test=False):
+        self.name = name
+        self.logger = logging.getLogger(__name__)
+
+    def add(self, qid, mailfrom, recipients, headers,
+            fp, subgroups=None, named_subgroups=None):
+        "Add email to storage."
+        fp.seek(0)
+        return ""
+
+    def find(self, mailfrom=None, recipients=None, older_than=None):
+        "Find emails in storage."
+        return
+
+    def get_metadata(self, storage_id):
+        "Return metadata of email in storage."
+        return
+
+    def delete(self, storage_id, recipients=None):
+        "Delete email from storage."
+        return
+
+    def get_mail(self, storage_id):
+        "Return a file pointer to the email and metadata."
+        return
+
+
+class FileMailStorage(BaseMailStorage):
+    "Storage class to store mails on filesystem."
+    storage_type = "file"
+
+    def __init__(self, name, global_cfg, cfg, test=False):
+        super(FileMailStorage, self).__init__(name, global_cfg, cfg, test)
+
+        defaults = {}
+        # check config
+
+        for opt in ["storage_directory"] + list(defaults.keys()):
+            if opt in cfg:
+                continue
+            if opt in global_cfg:
+                cfg[opt] = global_cfg[opt]
+            elif opt in defaults:
+                cfg[opt] = defaults[opt]
+            else:
+                raise RuntimeError(
+                    f"mandatory option '{opt}' not present in config "
+                    f"section '{self.name}' or 'global'")
+        self.directory = cfg["storage_directory"]
+
+        # check if quarantine directory exists and is writable
+        if not os.path.isdir(self.directory) or not os.access(
+                self.directory, os.W_OK):
+            raise RuntimeError(
+                f"file quarantine directory '{self.directory}' does "
+                f"not exist or is not writable")
+        self._metadata_suffix = ".metadata"
+
+    def _save_datafile(self, storage_id, fp):
+        datafile = os.path.join(self.directory, storage_id)
+        try:
+            with open(datafile, "wb") as f:
+                copyfileobj(fp, f)
+        except IOError as e:
+            raise RuntimeError(f"unable save data file: {e}")
+
+    def _save_metafile(self, storage_id, metadata):
+        metafile = os.path.join(
+            self.directory, f"{storage_id}{self._metadata_suffix}")
+        try:
+            with open(metafile, "w") as f:
+                json.dump(metadata, f, indent=2)
+        except IOError as e:
+            raise RuntimeError(f"unable to save metadata file: {e}")
+
+    def _remove(self, storage_id):
+        datafile = os.path.join(self.directory, storage_id)
+        metafile = f"{datafile}{self._metadata_suffix}"
+
+        try:
+            os.remove(metafile)
+        except IOError as e:
+            raise RuntimeError(f"unable to remove metadata file: {e}")
+
+        try:
+            os.remove(datafile)
+        except IOError as e:
+            raise RuntimeError(f"unable to remove data file: {e}")
+
+    def add(self, qid, mailfrom, recipients, headers,
+            fp, subgroups=None, named_subgroups=None):
+        "Add email to file storage and return storage id."
+        super(
+            FileMailStorage,
+            self).add(
+            qid,
+            mailfrom,
+            recipients,
+            headers,
+            fp,
+            subgroups,
+            named_subgroups)
+        timestamp = datetime.now().strftime("%Y%m%d%H%M%S")
+        storage_id = f"{timestamp}_{qid}"
+
+        # save mail
+        self._save_datafile(storage_id, fp)
+
+        # save metadata
+        metadata = {
+            "mailfrom": mailfrom,
+            "recipients": recipients,
+            "headers": headers,
+            "date": timegm(gmtime()),
+            "queue_id": qid,
+            "subgroups": subgroups,
+            "named_subgroups": named_subgroups
+        }
+        try:
+            self._save_metafile(storage_id, metadata)
+        except RuntimeError as e:
+            datafile = os.path.join(self.directory, storage_id)
+            os.remove(datafile)
+            raise e
+
+        return storage_id
+
+    def get_metadata(self, storage_id):
+        "Return metadata of email in storage."
+        super(FileMailStorage, self).get_metadata(storage_id)
+
+        metafile = os.path.join(
+            self.directory, f"{storage_id}{self._metadata_suffix}")
+        if not os.path.isfile(metafile):
+            raise RuntimeError(
+                f"invalid storage id '{storage_id}'")
+
+        try:
+            with open(metafile, "r") as f:
+                metadata = json.load(f)
+        except IOError as e:
+            raise RuntimeError(f"unable to read metadata file: {e}")
+        except json.JSONDecodeError as e:
+            raise RuntimeError(
+                f"invalid metafile '{metafile}': {e}")
+
+        return metadata
+
+    def find(self, mailfrom=None, recipients=None, older_than=None):
+        "Find emails in storage."
+        super(FileMailStorage, self).find(mailfrom, recipients, older_than)
+        if isinstance(mailfrom, str):
+            mailfrom = [mailfrom]
+        if isinstance(recipients, str):
+            recipients = [recipients]
+
+        emails = {}
+        metafiles = glob(os.path.join(
+            self.directory, f"*{self._metadata_suffix}"))
+        for metafile in metafiles:
+            if not os.path.isfile(metafile):
+                continue
+
+            storage_id = os.path.basename(
+                metafile[:-len(self._metadata_suffix)])
+            metadata = self.get_metadata(storage_id)
+            if older_than is not None:
+                if timegm(gmtime()) - metadata["date"] < (older_than * 86400):
+                    continue
+
+            if mailfrom is not None:
+                if metadata["mailfrom"] not in mailfrom:
+                    continue
+
+            if recipients is not None:
+                if len(recipients) == 1 and \
+                        recipients[0] not in metadata["recipients"]:
+                    continue
+                elif len(set(recipients + metadata["recipients"])) == \
+                        len(recipients + metadata["recipients"]):
+                    continue
+
+            emails[storage_id] = metadata
+
+        return emails
+
+    def delete(self, storage_id, recipients=None):
+        "Delete email from storage."
+        super(FileMailStorage, self).delete(storage_id, recipients)
+
+        try:
+            metadata = self.get_metadata(storage_id)
+        except RuntimeError as e:
+            raise RuntimeError(f"unable to delete email: {e}")
+
+        if not recipients:
+            self._remove(storage_id)
+        else:
+            if type(recipients) == str:
+                recipients = [recipients]
+            for recipient in recipients:
+                if recipient not in metadata["recipients"]:
+                    raise RuntimeError(f"invalid recipient '{recipient}'")
+                metadata["recipients"].remove(recipient)
+                if not metadata["recipients"]:
+                    self._remove(storage_id)
+                else:
+                    self._save_metafile(storage_id, metadata)
+
+    def get_mail(self, storage_id):
+        super(FileMailStorage, self).get_mail(storage_id)
+
+        metadata = self.get_metadata(storage_id)
+        datafile = os.path.join(self.directory, storage_id)
+        try:
+            fp = open(datafile, "rb")
+        except IOError as e:
+            raise RuntimeError(f"unable to open email data file: {e}")
+        return (fp, metadata)
+
+
+# list of storage types and their related storage classes
+TYPES = {"file": FileMailStorage}

pyquarantine/version.py

@@ -1 +0,0 @@
-__version__ = "0.0.2"

pyquarantine/whitelists.py

@@ -15,7 +15,6 @@
 import logging
 import peewee
 import re
-import sys
 
 from datetime import datetime
 from playhouse.db_url import connect
@@ -24,14 +23,14 @@ from playhouse.db_url import connect
 class WhitelistBase(object):
     "Whitelist base class"
 
-    def __init__(self, global_config, config, configtest=False):
-        self.global_config = global_config
-        self.config = config
-        self.configtest = configtest
-        self.name = config["name"]
+    whitelist_type = "base"
+
+    def __init__(self, name, global_cfg, cfg, test=False):
+        self.name = name
+        self.test = test
         self.logger = logging.getLogger(__name__)
         self.valid_entry_regex = re.compile(
-            r"^[a-zA-Z0-9_.+-]*?(@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+)?$")
+            r"^[a-zA-Z0-9_.=+-]*?(@[a-zA-Z0-9-]+\.[a-zA-Z0-9-.]+)?$")
 
     def check(self, mailfrom, recipient):
         "Check if mailfrom/recipient combination is whitelisted."
@@ -73,44 +72,52 @@ class Meta(object):
 
 class DatabaseWhitelist(WhitelistBase):
     "Whitelist class to store whitelist in a database"
+    whitelist_type = "db"
     _db_connections = {}
     _db_tables = {}
 
-    def __init__(self, global_config, config, configtest=False):
+    def __init__(self, name, global_cfg, cfg, test=False):
         super(
             DatabaseWhitelist,
             self).__init__(
-            global_config,
-            config,
-            configtest)
+            global_cfg,
+            cfg,
+            test)
 
-        # check if mandatory options are present in config
-        for option in ["whitelist_db_connection", "whitelist_db_table"]:
-            if option not in self.config.keys() and option in self.global_config.keys():
-                self.config[option] = self.global_config[option]
-            if option not in self.config.keys():
+        defaults = {}
+
+        # check config
+        for opt in ["whitelist_db_connection",
+                    "whitelist_db_table"] + list(defaults.keys()):
+            if opt in cfg:
+                continue
+            if opt in global_cfg:
+                cfg[opt] = global_cfg[opt]
+            elif opt in defaults:
+                cfg[opt] = defaults[opt]
+            else:
                 raise RuntimeError(
-                    "mandatory option '{}' not present in config section '{}' or 'global'".format(
-                        option, self.name))
+                    f"mandatory option '{opt}' not present in config "
+                    f"section '{self.name}' or 'global'")
 
-        tablename = self.config["whitelist_db_table"]
-        connection_string = self.config["whitelist_db_connection"]
+        tablename = cfg["whitelist_db_table"]
+        connection_string = cfg["whitelist_db_connection"]
 
         if connection_string in DatabaseWhitelist._db_connections.keys():
             db = DatabaseWhitelist._db_connections[connection_string]
         else:
             try:
                 # connect to database
-                self.logger.debug(
-                    "connecting to database '{}'".format(
-                        re.sub(
+                conn = re.sub(
                     r"(.*?://.*?):.*?(@.*)",
                     r"\1:<PASSWORD>\2",
-                            connection_string)))
+                    connection_string)
+                self.logger.debug(
+                    f"connecting to database '{conn}'")
                 db = connect(connection_string)
             except Exception as e:
                 raise RuntimeError(
-                    "unable to connect to database: {}".format(e))
+                    f"unable to connect to database: {e}")
 
             DatabaseWhitelist._db_connections[connection_string] = db
 
@@ -118,7 +125,7 @@ class DatabaseWhitelist(WhitelistBase):
         self.meta = Meta
         self.meta.database = db
         self.meta.table_name = tablename
-        self.model = type("WhitelistModel_{}".format(self.name), (WhitelistModel,), {
+        self.model = type(f"WhitelistModel_{self.name}", (WhitelistModel,), {
             "Meta": self.meta
         })
 
@@ -127,13 +134,12 @@ class DatabaseWhitelist(WhitelistBase):
 
         if tablename not in DatabaseWhitelist._db_tables[connection_string]:
             DatabaseWhitelist._db_tables[connection_string].append(tablename)
-            if not self.configtest:
+            if not self.test:
                 try:
                     db.create_tables([self.model])
                 except Exception as e:
                     raise RuntimeError(
-                        "unable to initialize table '{}': {}".format(
-                            tablename, e))
+                        f"unable to initialize table '{tablename}': {e}")
 
     def _entry_to_dict(self, entry):
         result = {}
@@ -163,17 +169,19 @@ class DatabaseWhitelist(WhitelistBase):
 
         # generate list of possible mailfroms
         self.logger.debug(
-            "query database for whitelist entries from <{}> to <{}>".format(
-                mailfrom, recipient))
+            f"query database for whitelist entries from <{mailfrom}> "
+            f"to <{recipient}>")
         mailfroms = [""]
         if "@" in mailfrom and not mailfrom.startswith("@"):
-            mailfroms.append("@{}".format(mailfrom.split("@")[1]))
+            domain = mailfrom.split("@")[1]
+            mailfroms.append(f"@{domain}")
         mailfroms.append(mailfrom)
 
         # generate list of possible recipients
         recipients = [""]
         if "@" in recipient and not recipient.startswith("@"):
-            recipients.append("@{}".format(recipient.split("@")[1]))
+            domain = recipient.split("@")[1]
+            recipients.append(f"@{domain}")
         recipients.append(recipient)
 
         # query the database
@@ -183,7 +191,7 @@ class DatabaseWhitelist(WhitelistBase):
                     self.model.mailfrom.in_(mailfroms),
                     self.model.recipient.in_(recipients)))
         except Exception as e:
-            raise RuntimeError("unable to query database: {}".format(e))
+            raise RuntimeError(f"unable to query database: {e}")
 
         if not entries:
             # no whitelist entry found
@@ -215,7 +223,8 @@ class DatabaseWhitelist(WhitelistBase):
         try:
             for entry in list(self.model.select()):
                 if older_than is not None:
-                    if (datetime.now() - entry.last_used).total_seconds() < (older_than * 86400):
+                    delta = (datetime.now() - entry.last_used).total_seconds()
+                    if delta < (older_than * 86400):
                         continue
 
                 if mailfrom is not None:
@@ -228,7 +237,7 @@ class DatabaseWhitelist(WhitelistBase):
 
                 entries.update(self._entry_to_dict(entry))
         except Exception as e:
-            raise RuntimeError("unable to query database: {}".format(e))
+            raise RuntimeError(f"unable to query database: {e}")
 
         return entries
 
@@ -249,7 +258,7 @@ class DatabaseWhitelist(WhitelistBase):
                 comment=comment,
                 permanent=permanent)
         except Exception as e:
-            raise RuntimeError("unable to add entry to database: {}".format(e))
+            raise RuntimeError(f"unable to add entry to database: {e}")
 
     def delete(self, whitelist_id):
         "Delete entry from whitelist."
@@ -260,7 +269,7 @@ class DatabaseWhitelist(WhitelistBase):
             deleted = query.execute()
         except Exception as e:
             raise RuntimeError(
-                "unable to delete entry from database: {}".format(e))
+                f"unable to delete entry from database: {e}")
 
         if deleted == 0:
             raise RuntimeError("invalid whitelist id")
@@ -284,10 +293,11 @@ class WhitelistCache(object):
                 mailfrom, recipient)
         return self.cache[whitelist][recipient]
 
-    def get_whitelisted_recipients(self, whitelist, mailfrom, recipients):
+    def get_recipients(self, whitelist, mailfrom, recipients):
         self.load(whitelist, mailfrom, recipients)
-        return list(
-            filter(lambda x: self.cache[whitelist][x], self.cache[whitelist].keys()))
+        return list(filter(
+            lambda x: self.cache[whitelist][x],
+            self.cache[whitelist].keys()))
 
 
 # list of whitelist types and their related whitelist classes

setup.cfg

@@ -0,0 +1,2 @@
+[metadata]
+version = attr: pyquarantine.__version__

setup.py

@@ -4,11 +4,8 @@ def read_file(fname):
     with open(fname, 'r') as f:
         return f.read()
 
-version = {}
-exec(read_file("pyquarantine/version.py"), version)
 
 setup(name = "pyquarantine",
-    version = version["__version__"],
     author = "Thomas Oettli",
     author_email = "spacefreak@noop.ch",
     description = "A pymilter based sendmail/postfix pre-queue filter.",
@@ -22,7 +19,7 @@ setup(name = "pyquarantine",
         #   3 - Alpha
         #   4 - Beta
         #   5 - Production/Stable
-        "Development Status :: 3 - Alpha",
+        "Development Status :: 4 - Beta",
         "License :: OSI Approved :: GNU General Public License v3 or later (GPLv3+)",
         "Operating System :: OS Independent",
         "Programming Language :: Python",
@@ -36,5 +33,5 @@ setup(name = "pyquarantine",
         ]
     },
     install_requires = ["pymilter", "netaddr", "beautifulsoup4[lxml]", "peewee"],
-    python_requires = ">=3"
+    python_requires = ">=3.6"
 )