LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows
Python ai-agentsdata-engineeringdocument-aigenerative-aiidpjson-extractionllmmcp-serverocrpdf-extractionprompt-engineeringstructured-output
import json
import logging
from datetime import UTC, datetime
from django.conf import settings
from django.http import HttpRequest
from rest_framework import status, viewsets
from rest_framework.decorators import action
from rest_framework.response import Response
from utils.cache_service import CacheService
from utils.user_session import UserSessionUtils
from .log_service import LogService
from .serializers import StoreLogMessagesSerializer
logger = logging.getLogger(__name__)
class LogsHelperViewSet(viewsets.ModelViewSet):
"""Viewset to handle all Tool Studio prompt related API logics."""
@action(detail=False, methods=["get"])
def get_logs(self, request: HttpRequest) -> Response:
# Extract the session ID
session_id: str = UserSessionUtils.get_session_id(request=request)
# Construct the Redis key pattern to match keys
# associated with the session ID
redis_key = LogService.generate_redis_key(session_id=session_id)
# Retrieve keys matching the pattern
keys = CacheService.get_all_keys(f"{redis_key}*")
# Retrieve values corresponding to the keys and sort them by timestamp
logs = []
for key in keys:
log_data = CacheService.get_key(key)
logs.append(log_data)
# Sort logs based on timestamp
sorted_logs = sorted(logs, key=lambda x: float(x["timestamp"]))
return Response({"data": sorted_logs}, status=status.HTTP_200_OK)
# This API will be triggered whenever a notification message
# pops up in the UI.
@action(detail=False, methods=["post"])
def store_log(self, request: HttpRequest) -> Response:
"""Store log message in Redis."""
# Extract the session ID
logs_expiry = settings.LOGS_EXPIRATION_TIME_IN_SECOND
session_id: str = UserSessionUtils.get_session_id(request=request)
serializer = StoreLogMessagesSerializer(data=request.data)
serializer.is_valid(raise_exception=True)
# Extract the log message from the validated data
log: str = serializer.validated_data.get("log")
log_data = json.loads(log)
timestamp = datetime.now(UTC).timestamp()
redis_key = f"{LogService.generate_redis_key(session_id=session_id)}:{timestamp}"
CacheService.set_key(redis_key, log_data, logs_expiry)
return Response({"message": "Successfully stored the message in redis"})