handle acl

smclab · Apr 11, 2023 · fcbeabc · fcbeabc
1 parent 89d429a
commit fcbeabc
Show file tree

Hide file tree

Showing 3 changed files with 45 additions and 12 deletions.
diff --git a/email-parser/app/imap/imap_extraction.py b/email-parser/app/imap/imap_extraction.py
@@ -37,7 +37,8 @@
 
 class AsyncEmailExtraction(threading.Thread):
 
-    def __init__(self, mail_server, port, username, password, timestamp, datasource_id, folder, schedule_id, tenant_id):
+    def __init__(self, mail_server, port, username, password, timestamp, datasource_id, folder, schedule_id, tenant_id,
+                 index_acl):
         super(AsyncEmailExtraction, self).__init__()
 
         self.mail_server = mail_server
@@ -49,6 +50,7 @@ def __init__(self, mail_server, port, username, password, timestamp, datasource_
         self.folder = folder
         self.schedule_id = schedule_id
         self.tenant_id = tenant_id
+        self.index_acl = index_acl
 
         self.status_logger = logging.getLogger('email-logger')
 
@@ -111,21 +113,23 @@ def extract(self):
 
                     payload = {
                         "datasourceId": self.datasource_id,
-                        "contentId": msg_id,
+                        "contentId": str(msg_id).replace("<", "").replace(">", ""),
                         "parsingDate": int(end_timestamp),
-                        "rawContent": raw_msg,
+                        "rawContent": "",
                         "datasourcePayload": datasource_payload,
                         "resources": {
-                            "binaries": binaries,
+                            "binaries": [],
                             "splitBinaries": True
                         },
                         "scheduleId": self.schedule_id,
                         "tenantId": self.tenant_id,
-                        "acl": {
-                            "email": acl_list
-                        }
                     }
 
+                    if self.index_acl:
+                        payload["acl"] = acl_list
+                    else:
+                        payload["acl"] = []
+
                     try:
                         post_message(ingestion_url, payload, 10)
                         # self.status_logger.info(payload)

diff --git a/email-parser/app/imap/util/utility.py b/email-parser/app/imap/util/utility.py
@@ -84,20 +84,47 @@ def parse_email(fetched_msg):
 
     msg_date = datetime.fromtimestamp(local_date).strftime("%a, %d %b %Y %H:%M:%S")
 
-    acl_list = [email.utils.parseaddr(msg_from)[1]]
-
     raw_msg = ""
     if msg_subject is not None:
         raw_msg = raw_msg + "Subject: " + msg_subject
+
+    msg_from_email = email.utils.parseaddr(msg_from)[1]
+    msg_from_user = email.utils.parseaddr(msg_from)[0]
+    if len(msg_from_user) == 0:
+        msg_from_user = None
+
+    acl_list = [msg_from_email]
+
     raw_msg = raw_msg + " Date: " + msg_date + " From: " + msg_from
+
+    msg_to_email = msg_to_user = msg_cc_email = msg_cc_user = None
+
     if msg_to is not None:
+
+        msg_to_email = [email.utils.parseaddr(to)[1] for to in str(msg_to).split(",")]
+        msg_to_user = [email.utils.parseaddr(to)[0] if len(to) > 0 else None for to in str(msg_to).split(",")]
+
         raw_msg = raw_msg + " To: " + msg_to
-        acl_list.append(msg_to)
+
+        acl_list = acl_list + msg_to_email
+
     if msg_cc is not None:
+
+        msg_cc_email = [email.utils.parseaddr(cc)[1] for cc in str(msg_cc).split(",")]
+        msg_cc_user = [email.utils.parseaddr(cc)[0] if len(cc) > 0 else None for cc in str(msg_cc).split(",")]
+
         raw_msg = raw_msg + " CC: " + msg_cc
+
+        acl_list = acl_list + msg_cc_email
+
     raw_msg = raw_msg + " " + body
 
-    struct_msg = {'date': (local_date * 1000), 'from': msg_from, 'subject': msg_subject, 'to': msg_to, 'cc': msg_cc,
+    struct_msg = {'date': (local_date * 1000),
+                  'from_user': msg_from_user, 'from_email': msg_from_email,
+                  'to_user': msg_to_user, 'to_email': msg_to_email,
+                  'cc_user': msg_cc_user, 'cc_email': msg_cc_email,
+                  'subject': msg_subject,
+                  'to': msg_to, 'cc': msg_cc,
                   'body': body, "htmlBody": raw_body}
 
     return raw_msg, struct_msg, msg_id, binaries, acl_list
diff --git a/email-parser/app/main.py b/email-parser/app/main.py
@@ -33,6 +33,7 @@ class ImapRequest(BaseModel):
     scheduleId: str
     folder: str
     tenantId: str
+    indexAcl: bool
 
 
 @app.post("/execute")
@@ -49,9 +50,10 @@ def get_data(request: ImapRequest):
     folder = request["folder"]
     schedule_id = request["scheduleId"]
     tenant_id = request["tenantId"]
+    index_acl = request["indexAcl"]
 
     email_extraction_task = AsyncEmailExtraction(mail_server, port, username, password, timestamp, datasource_id,
-                                                 folder, schedule_id, tenant_id)
+                                                 folder, schedule_id, tenant_id, index_acl)
 
     thread = threading.Thread(target=email_extraction_task.extract())
     thread.start()