Add alternative method by selecting headers to match

pyproject.toml

@@ -7,10 +7,10 @@ Homepage = "https://susurrando.com"
 
 [project]
 name = "remove_duplicate_imap_messages"
-version = "0.0.1"
+version = "0.0.2"
 description = "Given an IMAP folder look for duplicate messages and optionally delete them"
 readme = "README.md"
-authors = [{ name = "Antonio J. Delgado", email = "" }]
+authors = [{ name = "Antonio J. Delgado", email = "ad@susurrando.com" }]
 license = { file = "LICENSE" }
 classifiers = [
     "License :: OSI Approved :: GPLv3 License",

remove_duplicate_imap_messages/remove_duplicate_imap_messages.py

@@ -37,6 +37,11 @@ class RemoveDuplicateImapMessages:
         self._init_log()
         signal(SIGINT, self._signal_handler)
         self.messages_hashes = []
+        if self.config['method'] == 'headers' and 'header' in self.config and len(self.config['header']) == 0:
+            self._log.error(
+                "Error in parameters. If you specify the 'headers' method, you must indicate at least one --header to check"
+            )
+            sys.exit(1)
         self.duplicates_count = 0
         if 'mailbox' not in self.config:
             self.config['mailbox'] = []
@@ -82,6 +87,19 @@ class RemoveDuplicateImapMessages:
         self.imap.expunge()
         return True
 
+    def _get_header(self, message, header):
+        decoded_header = email.header.decode_header(message.get(header, ""))
+        if isinstance(decoded_header[0][0], str):
+            header_value = decoded_header[0][0]
+        else:
+            header_value = decoded_header[0][0].decode()
+        self._log.debug(
+            "Field '%s' is '%s'",
+            header,
+            header_value
+        )
+        return header_value
+
     def _process_message(self, message_id, data):
         '''Process a mail message'''
         if isinstance(data[1], int):
@@ -94,16 +112,16 @@ class RemoveDuplicateImapMessages:
         part = data[1].decode('utf-8')
         message = email.message_from_string(part)
         hash_obj = hashlib.sha256()
-        hash_obj.update(message.as_bytes())
+        if self.config['method'] == 'headers':
+            for header in self.config['header']:
+                hash_obj.update(self._get_header(message, header).encode('UTF-8'))
+        else:
+            hash_obj.update(message.as_bytes())
         hash_obj.digest()
         msg_hash = hash_obj.hexdigest()
         self._log.debug("Hash '%s'", msg_hash)
         if msg_hash in self.messages_hashes:
-            decoded_subject = email.header.decode_header(message.get("Subject", ""))
+            msg_subject = self._get_header(message, 'Subject')
-            if isinstance(decoded_subject[0][0], str):
-                msg_subject = decoded_subject[0][0]
-            else:
-                msg_subject = decoded_subject[0][0].decode()
             self._log.info(
                 "Message with subject '%s' is duplicate (hash check) of another",
                 msg_subject
@@ -241,6 +259,22 @@ class RemoveDuplicateImapMessages:
     multiple=True,
     help='IMAP mailboxes to check. Will compare messages between all mailboxes. Default: INBOX'
 )
+@click.option(
+    '--method',
+    '-m',
+    default='full',
+    type=click.Choice(
+        ["full", "headers"],
+        case_sensitive=False,
+    ),
+    help="Method to decide messages are duplicated. The 'full' method will check that the whole message and headers are the same. The 'headers' method will check that selected headers (with --header) are the same."
+)
+@click.option(
+    '--header', '-f',
+    multiple=True,
+    help="Fields to compare when method is 'headers'"
+)
+
 @click_config_file.configuration_option()
 def __main__(**kwargs):
     return RemoveDuplicateImapMessages(**kwargs)

setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = remove_duplicate_imap_messages
-version = 0.0.1
+version = 0.0.2
 
 [options]
 packages = remove_duplicate_imap_messages

setup.py

@@ -13,7 +13,7 @@ setuptools.setup(
     author="Antonio J. Delgado",
     version=config['metadata']['version'],
     name=config['metadata']['name'],
-    author_email="",
+    author_email="ad@susurrando.com",
     url="https://susurrando.com",
     description="Given an IMAP folder look for duplicate messages and optionally delete them",
     long_description="README.md",