LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows
Python ai-agentsdata-engineeringdocument-aigenerative-aiidpjson-extractionllmmcp-serverocrpdf-extractionprompt-engineeringstructured-output
from rest_framework.pagination import PageNumberPagination
from utils.constants import Pagination
class CustomPagination(PageNumberPagination):
page_size = Pagination.PAGE_SIZE
page_size_query_param = Pagination.PAGE_SIZE_QUERY_PARAM
max_page_size = Pagination.MAX_PAGE_SIZE
class OptionalPagination(CustomPagination):
"""Paginate only when the caller opts in via ?page / ?page_size.
These list endpoints are shared: besides the listing page they feed
dropdowns/selectors that expect a bare array. Returning None keeps that
response untouched for callers that don't ask for a page, while the
listing page opts in and gets the {count, next, previous, results} envelope.
"""
def paginate_queryset(self, queryset, request, view=None):
params = request.query_params
# Blank values (?page=) count as not opting in, so shared callers keep
# their bare-array response instead of getting an enveloped page.
if not params.get(self.page_query_param) and not params.get(
self.page_size_query_param
):
return None
return super().paginate_queryset(queryset, request, view=view)