Added input report maildir connector, issue #82 (#555)

Co-authored-by: Paolo Schiro <[email protected]>

docs/source/usage.md

@@ -355,6 +355,10 @@ The full set of configuration options are:
   - `port` - int: The port to use
   - `mode` - str: The GELF transport type to use. Valid modes: `tcp`, `udp`, `tls`
 
+- `maildir`
+  - `reports_folder` - str: Full path for mailbox maidir location (Default: `INBOX`)
+  - `maildir_create` - bool: Create maildir if not present (Default: False)
+
 :::{warning}
 It is **strongly recommended** to **not** use the `nameservers`
 setting. By default, `parsedmarc` uses

parsedmarc/cli.py

@@ -20,7 +20,8 @@
     parse_report_file, get_dmarc_reports_from_mbox, elastic, opensearch, \
     kafkaclient, splunk, save_output, email_results, ParserError, \
     __version__, InvalidDMARCReport, s3, syslog, loganalytics, gelf
-from parsedmarc.mail import IMAPConnection, MSGraphConnection, GmailConnection
+from parsedmarc.mail import IMAPConnection, MSGraphConnection, GmailConnection, \
+    MaildirConnection
 from parsedmarc.mail.graph import AuthMethod
 
 from parsedmarc.log import logger
@@ -504,6 +505,8 @@ def process_reports(reports_):
                      gmail_api_paginate_messages=True,
                      gmail_api_scopes=[],
                      gmail_api_oauth2_port=8080,
+                     maildir_path = None,
+                     maildir_create = False,
                      log_file=args.log_file,
                      n_procs=1,
                      ip_db_path=None,
@@ -1008,6 +1011,13 @@ def process_reports(reports_):
                 opts.gmail_api_oauth2_port = \
                     gmail_api_config.get("oauth2_port", 8080)
 
+        if "maildir" in config.sections():
+            maildir_api_config = config["maildir"]
+            opts.maildir_path = \
+                maildir_api_config.get("maildir_path")
+            opts.maildir_create = \
+                maildir_api_config.get("maildir_create")           
+
         if "log_analytics" in config.sections():
             log_analytics_config = config["log_analytics"]
             opts.la_client_id = \
@@ -1072,6 +1082,7 @@ def process_reports(reports_):
     if opts.imap_host is None \
             and opts.graph_client_id is None \
             and opts.gmail_api_credentials_file is None \
+            and opts.maildir_path is None \
             and len(opts.file_path) == 0:
         logger.error("You must supply input files or a mailbox connection")
         exit(1)
@@ -1380,6 +1391,16 @@ def process_reports(reports_):
             logger.exception("Gmail API Error")
             exit(1)
 
+    if opts.maildir_path:
+        try:
+            mailbox_connection = MaildirConnection(
+                maildir_path=opts.maildir_path,
+                maildir_create=opts.maildir_create,
+            )
+        except Exception:
+            logger.exception("Maildir Error")
+            exit(1)
+
     if mailbox_connection:
         try:
             reports = get_dmarc_reports_from_mailbox(

parsedmarc/mail/maildir.py

@@ -0,0 +1,60 @@
+from time import sleep
+
+#from imapclient.exceptions import IMAPClientError
+#from mailsuite.imap import IMAPClient
+from socket import timeout
+
+from parsedmarc.log import logger
+from parsedmarc.mail.mailbox_connection import MailboxConnection
+#import email
+import mailbox
+import os
+
+class MaildirConnection(MailboxConnection):
+    def __init__(self,
+                 maildir_path=None,
+                 maildir_create=False,
+                 ):
+        self._maildir_path = maildir_path
+        self._maildir_create = maildir_create
+        maildir_owner=os.stat(maildir_path).st_uid
+        if os.getuid() != maildir_owner:
+            if os.getuid() == 0:
+                logger.warning("Switching uid to {} to access Maildir".format(maildir_owner))
+                os.setuid(maildir_owner)
+            else:
+                raise Exception('runtime uid {} differ from maildir {} owner {}'.format(os.getuid().maildir_path,maildir_owner)) 
+        self._client = mailbox.Maildir(maildir_path, create=maildir_create)
+        self._subfolder_client={}
+
+    def create_folder(self, folder_name: str):
+        self._subfolder_client[folder_name]=self._client.add_folder(folder_name)
+        self._client.add_folder(folder_name)
+
+    def fetch_messages(self, reports_folder: str, **kwargs):
+        return self._client.keys()
+
+    def fetch_message(self, message_id):
+        return self._client.get(message_id).as_string()
+
+    def delete_message(self, message_id: str):
+        self._client.remove(message_id)
+
+    def move_message(self, message_id: str, folder_name: str):
+        message_data=self._client.get(message_id)
+        if folder_name not in self._subfolder_client.keys():
+            self._subfolder_client = mailbox.Maildir(os.join(maildir_path,folder_name), create=maildir_create)
+        self._subfolder_client[folder_name].add(message_data)
+        self._client.remove(message_id)
+
+    def keepalive(self):
+        return
+
+    def watch(self, check_callback, check_timeout):
+        while True:
+            try:
+                check_callback(self)
+            except Exception as e:
+                logger.warning("Maildir init error. {0}".format(e))
+            sleep(check_timeout)
+