Refactor email indexing logic: improve check for already indexed emails and enhance logging
Browse files- app/controllers/mail.py +4 -5
app/controllers/mail.py
CHANGED
@@ -78,11 +78,10 @@ def list_emails(service, messages):
|
|
78 |
metadata = {}
|
79 |
logger.info("vectorstore.index_to_docstore_id: %s", list(vectorstore.index_to_docstore_id.values()))
|
80 |
logger.info("type: %s", type(vectorstore.index_to_docstore_id.values()))
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
continue
|
86 |
for header in msg["payload"]["headers"]:
|
87 |
if header["name"] == "From":
|
88 |
metadata["from"] = header["value"]
|
|
|
78 |
metadata = {}
|
79 |
logger.info("vectorstore.index_to_docstore_id: %s", list(vectorstore.index_to_docstore_id.values()))
|
80 |
logger.info("type: %s", type(vectorstore.index_to_docstore_id.values()))
|
81 |
+
for docstore_id in list(vectorstore.index_to_docstore_id.values()):
|
82 |
+
if docstore_id.startswith(message["id"]):
|
83 |
+
logger.info("Already indexed: %s", message["id"])
|
84 |
+
continue
|
|
|
85 |
for header in msg["payload"]["headers"]:
|
86 |
if header["name"] == "From":
|
87 |
metadata["from"] = header["value"]
|