(20241203) Mail sync'ing started. You can now hit a dedicated API endpoint to trigger a mail-sync process (in the background if you want) and save the mails to your database.
This commit is contained in:
@@ -0,0 +1,542 @@
|
||||
"""
|
||||
|
||||
AUTHOR:
|
||||
|
||||
Khushal P Soonderji
|
||||
|
||||
DATE:
|
||||
|
||||
tuesday, 3rd Dec., 2024
|
||||
|
||||
OBJECTIVE:
|
||||
|
||||
From here we sync all mails between the mail client's server and TheCAOffice's database.
|
||||
|
||||
REFERENCES:
|
||||
|
||||
N/A
|
||||
|
||||
DOWNLOADS:
|
||||
|
||||
N/A
|
||||
|
||||
"""
|
||||
|
||||
# *****************************************************************************************************************
|
||||
# ***** ****
|
||||
# *** IMPORT ***
|
||||
# ***** ****
|
||||
# *****************************************************************************************************************
|
||||
|
||||
|
||||
# To make sibling directories accessible for imports:
|
||||
import sys
|
||||
sys.path.append(".")
|
||||
sys.path.append("..")
|
||||
|
||||
# For Quart:
|
||||
from quart import current_app
|
||||
|
||||
# My async utils:
|
||||
from utils_v2.string import json
|
||||
from utils_v2.date_time import date_time
|
||||
from utils_v2.database.async_mysql_v2 import AsyncMySQL
|
||||
from utils_v2.database.async_mongo_v2 import AsyncMongo, AsyncMongoStorage
|
||||
|
||||
# Mail Clients:
|
||||
from utils_v2.goog.gmail.gmail_client import AsyncGMailClient
|
||||
from utils_v2.goog.models.data.auth_tokens import GoogleAuthTokens
|
||||
|
||||
# Base model:
|
||||
from models.behaviour.base import BaseModel
|
||||
|
||||
# Data models:
|
||||
from models.data.mail.sync import MailSyncOneResult, MailSyncManyResults
|
||||
|
||||
# To work with MongoDB:
|
||||
from bson import ObjectId
|
||||
from pymongo import InsertOne, UpdateOne, ReplaceOne
|
||||
|
||||
# To work with LLMs:
|
||||
from langchain_openai import ChatOpenAI
|
||||
from langchain_core.prompts import ChatPromptTemplate
|
||||
|
||||
# To work with datatypes:
|
||||
from typing import Literal, List, Dict, Any
|
||||
|
||||
# To make deep-copies:
|
||||
import copy
|
||||
|
||||
# To work with base-64 encoding:
|
||||
import base64
|
||||
|
||||
# To work with date and time:
|
||||
import datetime
|
||||
|
||||
# For asynchronous activities:
|
||||
import asyncio
|
||||
|
||||
|
||||
# *****************************************************************************************************************
|
||||
# ***** ****
|
||||
# *** MACROS / ONE-TIME INIT ***
|
||||
# ***** ****
|
||||
# *****************************************************************************************************************
|
||||
|
||||
|
||||
# --- Nothing Yet
|
||||
|
||||
|
||||
# *****************************************************************************************************************
|
||||
# ***** ****
|
||||
# *** VARIABLES ***
|
||||
# ***** ****
|
||||
# *****************************************************************************************************************
|
||||
|
||||
|
||||
# --- Nothing Yet
|
||||
|
||||
|
||||
# *****************************************************************************************************************
|
||||
# ***** ****
|
||||
# *** FUNCTIONS ***
|
||||
# ***** ****
|
||||
# *****************************************************************************************************************
|
||||
|
||||
|
||||
# --- Nothing Yet
|
||||
|
||||
|
||||
# *****************************************************************************************************************
|
||||
# ***** ****
|
||||
# *** CLASSES ***
|
||||
# ***** ****
|
||||
# *****************************************************************************************************************
|
||||
|
||||
|
||||
class MailSyncModel(BaseModel):
|
||||
|
||||
# For MongoDB:
|
||||
AUTH_COLLECTION = "_authTokens"
|
||||
MAIL_COLLECTION = "_messages"
|
||||
|
||||
# For AI Magic through LLMs:
|
||||
prompt_template = ChatPromptTemplate.from_messages([
|
||||
(
|
||||
"system",
|
||||
"You're a mail summary expert that summarizes mails in 150 chars or less. HIDE SENSITIVE INFO (LIKE OTPs) FROM THE SUMMARY."
|
||||
),
|
||||
(
|
||||
"user",
|
||||
"Please summarize this mail: \"\"\"{mail}\"\"\""
|
||||
)
|
||||
])
|
||||
|
||||
@staticmethod
|
||||
async def __save_one_attachment(
|
||||
self,
|
||||
session_token: str,
|
||||
attachment: Dict[str, Any],
|
||||
attachment_tags: List[str],
|
||||
attachment_metadata: dict,
|
||||
retry_count: int = 1,
|
||||
retry_delay: int = 1,
|
||||
backoff_multiplier: float = 1.1
|
||||
) -> Dict[str, Any]:
|
||||
|
||||
"""
|
||||
Saves one attachment and generates a URL that can be later used to retrieve it.
|
||||
:param session_token: The session token of the uer who is trying to upload this file.
|
||||
:param attachment: The JSON that describes the attachment.
|
||||
:param attachment_tags: Any tags to put on the file for easy search later.
|
||||
:param attachment_metadata: Any metadata to put on the file for easy search later.
|
||||
:param retry_count: How many max. retries to do in case of failure.
|
||||
:param retry_delay: The interval between the delays.
|
||||
:param backoff_multiplier: By what rate the delay between 2 attempts must change.
|
||||
:return: The JSON that describes the same attachment, except that the payload's data is replaced by the id and
|
||||
url of where to find the attachment.
|
||||
"""
|
||||
|
||||
# Make a deep-copy of the attachment JSON,
|
||||
# and process the payload in advance:
|
||||
attachment_copy = copy.deepcopy(attachment)
|
||||
attachment_payload = attachment_copy.pop("payload").encode()
|
||||
if attachment_copy.pop("contentTransferEncoding", "?").strip().lower() == "base64":
|
||||
attachment_payload = base64.b64decode(attachment_payload)
|
||||
|
||||
# Start by assuming failure,
|
||||
# and retry as many times as asked:
|
||||
attachment_copy["id"] = None
|
||||
attachment_copy["url"] = None
|
||||
for _ in range(retry_count):
|
||||
|
||||
# Make the upload:
|
||||
api_response = await current_app.http_client.post(
|
||||
url = current_app.script_data["fileUpload"]["url"],
|
||||
headers = {
|
||||
"X-Session-Token": session_token,
|
||||
"X-File-Name": attachment["filename"],
|
||||
"X-File-Private": "false",
|
||||
"X-File-Tags": json.to_string(attachment_tags, no_space = True),
|
||||
"X-File-Metadata": json.to_string(attachment_metadata, no_space = True)
|
||||
},
|
||||
data = attachment_payload
|
||||
)
|
||||
|
||||
# If the upload was successful:
|
||||
if api_response.status_code in [200]:
|
||||
api_data = api_response.json()["data"]
|
||||
attachment_copy["id"] = api_data["id"]
|
||||
attachment_copy["url"] = api_data["url"]
|
||||
break
|
||||
|
||||
# Done here:
|
||||
return attachment_copy
|
||||
|
||||
async def __save_many_attachments(
|
||||
self,
|
||||
session_token: str,
|
||||
attachments: List[Dict[str, Any]],
|
||||
attachment_tags: List[str],
|
||||
attachment_metadata: dict,
|
||||
retry_count: int = 1,
|
||||
retry_delay:int = 1,
|
||||
backoff_multiplier: float = 1.1
|
||||
) -> List[Dict[str, Any]]:
|
||||
|
||||
"""
|
||||
Saves all the attachments received in the mail (whether inline or otherwise) and makes them available through
|
||||
simple download URLs.
|
||||
:param session_token: The session token of the uer who is trying to upload this file.
|
||||
:param attachments: The JSON that describes the attachments.
|
||||
:param attachment_tags: Any tags to put on the file for easy search later.
|
||||
:param attachment_metadata: Any metadata to put on the file for easy search later.
|
||||
:param retry_count: How many max. retries to do in case of failure.
|
||||
:param retry_delay: The interval between the delays.
|
||||
:param backoff_multiplier: By what rate the delay between 2 attempts must change.
|
||||
:return: The JSON that describes the same attachments, except that the payload's data is replaced by the id and
|
||||
url of where to find each attachment.
|
||||
"""
|
||||
|
||||
# Create and fire all the tasks
|
||||
# needed to save the files:
|
||||
tasks = [
|
||||
self.__save_one_attachment(
|
||||
session_token = session_token,
|
||||
attachment = attachment,
|
||||
attachment_tags = attachment_tags,
|
||||
attachment_metadata = attachment_metadata,
|
||||
retry_count = retry_count,
|
||||
retry_delay = retry_delay,
|
||||
backoff_multiplier = backoff_multiplier
|
||||
) for attachment in attachments
|
||||
]
|
||||
uploaded_attachments = await asyncio.gather(*tasks)
|
||||
|
||||
# Done here:
|
||||
return uploaded_attachments
|
||||
|
||||
async def __sync_one_gmail(
|
||||
self,
|
||||
session_token: str,
|
||||
mongo_conn: AsyncMongo,
|
||||
mail_client: AsyncGMailClient,
|
||||
tokens: GoogleAuthTokens,
|
||||
message_id: str,
|
||||
llm: ChatOpenAI = None,
|
||||
force_sync: bool = False
|
||||
) -> MailSyncOneResult:
|
||||
|
||||
"""
|
||||
Sync on mail from GMail.
|
||||
:param session_token: The session token of the uer who is trying to upload this file.
|
||||
:param mongo_conn: The instance of the connection to the database to use.
|
||||
:param mail_client: The instance of the mail client to use to perform the action.
|
||||
:param tokens: The tokens to use to fetch the mails.
|
||||
:param message_id: The id that Google uses to identify this mail. This will be received in the 'list_messages'
|
||||
method.
|
||||
:param llm: The instance of the LLM to use to summarize the mail's content.
|
||||
:param force_sync: Whether you would like to forcefully re-sync the mail even if it is already present in the
|
||||
database.
|
||||
:return:
|
||||
"""
|
||||
|
||||
# Start by assuming failure:
|
||||
sync_result = MailSyncOneResult()
|
||||
|
||||
# If we've not been forced to re-sync the mail message,
|
||||
# we first check if the mail already exists in our database:
|
||||
if not force_sync:
|
||||
mail_record = await mongo_conn.find_one(
|
||||
collection = self.MAIL_COLLECTION,
|
||||
filter = mongo_conn.dict_to_dot_notation({
|
||||
"payload": {
|
||||
"messageId": message_id
|
||||
}
|
||||
}),
|
||||
projection = {
|
||||
"_id": False,
|
||||
"readTs": "payload.readTs"
|
||||
}
|
||||
)
|
||||
if mail_record:
|
||||
sync_result.success = True
|
||||
sync_result.message = f"gmail message '{message_id}' already sync'd on '{mail_record['readTs']} (UTC)'"
|
||||
return sync_result
|
||||
|
||||
# Now that we know that we have to fetch the mail from GMail:
|
||||
client_response = await mail_client.get_message(
|
||||
tokens = tokens,
|
||||
message_id = message_id,
|
||||
return_raw = False
|
||||
)
|
||||
if not client_response.success:
|
||||
sync_result.message = f"gmail (messageId: '{message_id}'): {client_response.message}"
|
||||
|
||||
# We upload the attachments:
|
||||
client_response.data["attachments"] = await self.__save_many_attachments(
|
||||
session_token = session_token,
|
||||
attachments = client_response.data["attachments"],
|
||||
attachment_tags = [
|
||||
"email",
|
||||
"gmail",
|
||||
client_response.data["from"][0]["name"],
|
||||
client_response.data["from"][0]["email"],
|
||||
tokens.email,
|
||||
],
|
||||
attachment_metadata = {
|
||||
"project": "tcaoff",
|
||||
"serviceType": "email",
|
||||
"client": "gmail",
|
||||
"from": client_response.data["from"][0]["email"],
|
||||
"to": tokens.email
|
||||
},
|
||||
retry_count = 3
|
||||
)
|
||||
|
||||
# Give a quick indicator of whether this mail is an inbox mail or sent mail:
|
||||
all_recipients = []
|
||||
for field in ["to", "cc", "bcc"]: all_recipients += [item["email"] for item in client_response.data[field]]
|
||||
if tokens.email in all_recipients: client_response.data["isInbox"] = True
|
||||
else: client_response.data["isInbox"] = False
|
||||
|
||||
# If an LLM is given, we add an AI summary:
|
||||
llm_json = None
|
||||
if llm:
|
||||
llm_response = response = await llm.ainvoke(
|
||||
self.prompt_template.invoke({
|
||||
"mail": client_response.data["unformattedText"]
|
||||
})
|
||||
)
|
||||
llm_json = {
|
||||
"snippet": llm_response.content,
|
||||
"usage": llm_response.usage_metadata
|
||||
}
|
||||
client_response.data["aiSnippet"] = llm_json
|
||||
|
||||
# Done here:
|
||||
sync_result.success = True
|
||||
sync_result.mailMessage = client_response.data
|
||||
return sync_result
|
||||
|
||||
async def __sync_many_gmail(
|
||||
self,
|
||||
session_token: str,
|
||||
mongo_conn: AsyncMongo,
|
||||
account_identifier: ObjectId,
|
||||
mail_client: AsyncGMailClient,
|
||||
tokens: GoogleAuthTokens,
|
||||
llm: ChatOpenAI = None,
|
||||
force_sync: bool = False,
|
||||
start_date: datetime.datetime = None,
|
||||
end_date: datetime.datetime = None,
|
||||
max_count: int = 100
|
||||
) -> MailSyncManyResults:
|
||||
|
||||
"""
|
||||
Sync many mails from GMail in one shot.
|
||||
:param session_token: The session token of the uer who is trying to upload this file.
|
||||
:param mongo_conn: The instance of the connection to the database to use.
|
||||
:param account_identifier: The id of the document in the database that holds the tokens to access the account.
|
||||
Needed only for refreshing the tokens and saving them.
|
||||
:param mail_client: The instance of the mail client to use to perform the action.
|
||||
:param tokens: The tokens to use to fetch the mails.
|
||||
:param llm: The instance of the LLM to use to summarize the mail's content.
|
||||
:param force_sync: Whether you would like to forcefully re-sync the mail even if it is already present in the
|
||||
database.
|
||||
:param start_date: The starting date (inclusive) from when to sync the mails.
|
||||
:param end_date: The ending date (inclusive) from when to sync the mails.
|
||||
:param max_count: The max. no. of mails to sync.
|
||||
:return: The result of the sync'ing.
|
||||
"""
|
||||
|
||||
# Start by assuming failure:
|
||||
sync_results = MailSyncManyResults()
|
||||
|
||||
# Refresh the tokens (if needed):
|
||||
tokens_refreshed = await tokens.arefresh(
|
||||
http_client = current_app.http_client,
|
||||
client_id = mail_client.client_id,
|
||||
client_secret = mail_client.client_secret
|
||||
)
|
||||
if tokens_refreshed: await current_app.mail_oauth_model.set_token(
|
||||
db_conn = current_app.sql_writer,
|
||||
mongo_conn = mongo_conn,
|
||||
account_identifier = account_identifier,
|
||||
email_id = tokens.email,
|
||||
token = tokens,
|
||||
session_token = session_token
|
||||
)
|
||||
|
||||
# Let's build the query:
|
||||
sub_queries = []
|
||||
if start_date: sub_queries.append(start_date.strftime("after:%Y/%m/%d"))
|
||||
if end_date: sub_queries.append((end_date + datetime.timedelta(days = 1)).strftime("before:%Y/%m/%d"))
|
||||
query_string = " ".join(sub_queries)
|
||||
|
||||
# Let's enlist all the mails that fall in the date range:
|
||||
client_response = await mail_client.list_messages(
|
||||
tokens = tokens,
|
||||
max_count = max_count,
|
||||
query = query_string
|
||||
)
|
||||
if not client_response.success:
|
||||
sync_results["message"] = f"gmail: {client_response.message}"
|
||||
return sync_results
|
||||
messages_list = client_response.data["messages"]
|
||||
|
||||
# Now, for every mail in the list, we fetch the mail and note the results:
|
||||
tasks = [
|
||||
self.__sync_one_gmail(
|
||||
session_token = session_token,
|
||||
mongo_conn = mongo_conn,
|
||||
mail_client = mail_client,
|
||||
tokens = tokens,
|
||||
message_id = v["id"],
|
||||
llm = llm,
|
||||
force_sync = force_sync
|
||||
) for v in messages_list.values()
|
||||
]
|
||||
individual_sync_results = await asyncio.gather(*tasks)
|
||||
|
||||
# Now we create operations for each mail,
|
||||
# and maintain success/failure counters:
|
||||
sync_results.totalCount = len(individual_sync_results)
|
||||
mongo_operations = []
|
||||
for result in individual_sync_results:
|
||||
if result.success: sync_results.successCount += 1
|
||||
else: sync_results.failureCount += 1
|
||||
if result.mailMessage: mongo_operations.append(ReplaceOne(
|
||||
filter = {
|
||||
"serviceType": "email",
|
||||
"$or": [
|
||||
{
|
||||
"client": "gmail",
|
||||
"payload.messageId": result.mailMessage["messageId"]
|
||||
}
|
||||
]
|
||||
},
|
||||
replacement = {
|
||||
"version": "1.0.0",
|
||||
"accountId": ObjectId(account_identifier),
|
||||
"serviceType": "email",
|
||||
"client": "gmail",
|
||||
"payload": result.mailMessage
|
||||
},
|
||||
upsert = True
|
||||
))
|
||||
|
||||
# Make the bulk write:
|
||||
if mongo_operations:
|
||||
mongo_count = await mongo_conn.bulk_write(
|
||||
collection = self.MAIL_COLLECTION,
|
||||
requests = mongo_operations
|
||||
)
|
||||
|
||||
# Done here:
|
||||
sync_results.message = f"{sync_results.successCount}/{sync_results.totalCount} mail(s) sync'd from gmail"
|
||||
return sync_results
|
||||
|
||||
async def sync(
|
||||
self,
|
||||
session_token: str,
|
||||
mongo_conn: AsyncMongo,
|
||||
account_identifier: ObjectId,
|
||||
llm: ChatOpenAI = None,
|
||||
force_sync: bool = False,
|
||||
start_date: datetime.datetime = None,
|
||||
end_date: datetime.datetime = None,
|
||||
max_count: int = 100
|
||||
) -> MailSyncManyResults:
|
||||
|
||||
"""
|
||||
Sync many mails at once from many types of clients. Use this as a common entry point after which you internally
|
||||
route the request to the appropriate clients.
|
||||
:param session_token: The session token of the uer who is trying to upload this file.
|
||||
:param mongo_conn: The instance of the connection to the database to use.
|
||||
:param account_identifier: The id of the document in the database that holds the tokens to access the account.
|
||||
Needed only for refreshing the tokens and saving them.
|
||||
:param llm: The instance of the LLM to use to summarize the mail's content.
|
||||
:param force_sync: Whether you would like to forcefully re-sync the mail even if it is already present in the
|
||||
database.
|
||||
:param start_date: The starting date (inclusive) from when to sync the mails.
|
||||
:param end_date: The ending date (inclusive) from when to sync the mails.
|
||||
:param max_count: The max. no. of mails to sync.
|
||||
:return: The result of the sync'ing.
|
||||
"""
|
||||
|
||||
# Start by assuming failure:
|
||||
sync_results = MailSyncManyResults()
|
||||
|
||||
# ┏┓ ┓ ┏┳┓ ┓
|
||||
# ┣ ┏┓╋┏┣┓ ┃ ┏┓┃┏┏┓┏┓┏
|
||||
# ┻ ┗ ┗┗┛┗ ┻ ┗┛┛┗┗ ┛┗┛
|
||||
|
||||
# We first load the authorization tokens:
|
||||
auth_json = await current_app.mail_oauth_model.get_token(
|
||||
mongo_conn = mongo_conn,
|
||||
account_identifier = account_identifier,
|
||||
)
|
||||
|
||||
# If we failed to load the authorization tokens:
|
||||
if not auth_json:
|
||||
sync_results.message = f"no such account identifier '{account_identifier}'"
|
||||
return sync_results
|
||||
|
||||
# ┏┓ ┏┓┳┳┓ •┓
|
||||
# ┣ ┏┓┏┓ ┃┓┃┃┃┏┓┓┃
|
||||
# ┻ ┗┛┛ ┗┛┛ ┗┗┻┗┗
|
||||
|
||||
if auth_json["client"] == "gmail":
|
||||
return await self.__sync_many_gmail(
|
||||
session_token = session_token,
|
||||
mongo_conn = mongo_conn,
|
||||
account_identifier = account_identifier,
|
||||
mail_client = current_app.gmail_client,
|
||||
tokens = GoogleAuthTokens(**auth_json["token"]),
|
||||
llm = llm,
|
||||
force_sync = force_sync,
|
||||
start_date = start_date,
|
||||
end_date = end_date,
|
||||
max_count = max_count
|
||||
)
|
||||
|
||||
# ┳ ┓• ┓ ┏┓┓•
|
||||
# ┃┏┓┓┏┏┓┃┓┏┫ ┃ ┃┓┏┓┏┓╋
|
||||
# ┻┛┗┗┛┗┻┗┗┗┻ ┗┛┗┗┗ ┛┗┗
|
||||
|
||||
# If we haven't been able to sync mail due to not entering any 'if' condition:
|
||||
sync_results.message = f"no such mail client '{auth_json['client']}'"
|
||||
return sync_results
|
||||
|
||||
|
||||
# *****************************************************************************************************************
|
||||
# ***** ****
|
||||
# *** MAIN PROGRAM ***
|
||||
# ***** ****
|
||||
# *****************************************************************************************************************
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
|
||||
pass
|
||||
Reference in New Issue
Block a user