[pre-commit.ci] pre-commit autoupdate

.pre-commit-config.yaml

@@ -9,7 +9,7 @@ ci:
     - biome-check # Uses local npm/npx, not available in pre-commit CI
 repos:
   - repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v5.0.0
+    rev: v6.0.0
     hooks:
       - id: trailing-whitespace
         exclude_types:
@@ -37,7 +37,7 @@ repos:
       - id: no-commit-to-branch
 
   - repo: https://github.com/astral-sh/ruff-pre-commit
-    rev: v0.3.4
+    rev: v0.15.8
     hooks:
       - id: ruff
         args: [--fix]
@@ -56,7 +56,7 @@ repos:
   #       language: python
 
   - repo: https://github.com/asottile/pyupgrade
-    rev: v3.17.0
+    rev: v3.21.2
     hooks:
       - id: pyupgrade
         entry: pyupgrade --py39-plus --keep-runtime-typing
@@ -87,15 +87,15 @@ repos:
         types_or: [javascript, jsx, ts, tsx, json]
 
   - repo: https://github.com/igorshubovych/markdownlint-cli
-    rev: v0.42.0
+    rev: v0.48.0
     hooks:
       - id: markdownlint
         args: [--disable, MD013]
       - id: markdownlint-fix
         args: [--disable, MD013]
 
   - repo: https://github.com/gitleaks/gitleaks
-    rev: v8.18.2
+    rev: v8.30.0
     hooks:
       - id: gitleaks
 
@@ -105,7 +105,7 @@ repos:
       - id: htmlhint
 
   - repo: https://github.com/hadolint/hadolint
-    rev: v2.12.1-beta
+    rev: v2.14.0
     hooks:
       - id: hadolint-docker
         args:

backend/account_v2/custom_auth_middleware.py

@@ -1,6 +1,8 @@
 import logging
 import uuid
 
+from backend.constants import RequestHeader, RequestMethod
+from backend.internal_api_constants import INTERNAL_API_PREFIX
 from django.conf import settings
 from django.http import HttpRequest, HttpResponse, JsonResponse
 from utils.constants import Account
@@ -10,8 +12,6 @@
 from account_v2.authentication_plugin_registry import AuthenticationPluginRegistry
 from account_v2.authentication_service import AuthenticationService
 from account_v2.constants import Common
-from backend.constants import RequestHeader, RequestMethod
-from backend.internal_api_constants import INTERNAL_API_PREFIX
 
 logger = logging.getLogger(__name__)
 

backend/account_v2/models.py

@@ -1,10 +1,9 @@
 import uuid
 
+from backend.constants import FieldLengthConstants as FieldLength
 from django.contrib.auth.models import AbstractUser, Group, Permission
 from django.db import models
 
-from backend.constants import FieldLengthConstants as FieldLength
-
 NAME_SIZE = 64
 KEY_SIZE = 64
 

backend/adapter_processor_v2/adapter_processor.py

@@ -8,6 +8,13 @@
 from django.core.exceptions import ObjectDoesNotExist
 from platform_settings_v2.platform_auth_service import PlatformAuthenticationService
 from tenant_account_v2.organization_member_service import OrganizationMemberService
+from unstract.sdk1.adapters.adapterkit import Adapterkit
+from unstract.sdk1.adapters.base import Adapter
+from unstract.sdk1.adapters.x2text.constants import X2TextConstants
+from unstract.sdk1.constants import AdapterTypes
+from unstract.sdk1.embedding import EmbeddingCompat
+from unstract.sdk1.exceptions import SdkError
+from unstract.sdk1.llm import LLM
 
 from adapter_processor_v2.constants import AdapterKeys, AllowedDomains
 from adapter_processor_v2.exceptions import (
@@ -16,13 +23,6 @@
     InValidAdapterId,
     TestAdapterError,
 )
-from unstract.sdk1.adapters.adapterkit import Adapterkit
-from unstract.sdk1.adapters.base import Adapter
-from unstract.sdk1.adapters.x2text.constants import X2TextConstants
-from unstract.sdk1.constants import AdapterTypes
-from unstract.sdk1.embedding import EmbeddingCompat
-from unstract.sdk1.exceptions import SdkError
-from unstract.sdk1.llm import LLM
 
 from .models import AdapterInstance, UserDefaultAdapter
 

backend/adapter_processor_v2/exceptions.py

@@ -1,7 +1,7 @@
 from rest_framework.exceptions import APIException
+from unstract.sdk1.exceptions import SdkError
 
 from adapter_processor_v2.constants import AdapterKeys
-from unstract.sdk1.exceptions import SdkError
 
 
 class IdIsMandatory(APIException):

backend/adapter_processor_v2/management/commands/manage_deprecated_adapters.py

@@ -88,7 +88,7 @@ def _list_deprecated(self) -> None:
         )
 
         for adapter in deprecated_adapters:
-            self.stdout.write(f"\n{'='*80}")
+            self.stdout.write(f"\n{'=' * 80}")
             self.stdout.write(f"UUID: {adapter.id}")
             self.stdout.write(f"Name: {adapter.adapter_name}")
             self.stdout.write(f"Type: {adapter.adapter_type}")
@@ -258,7 +258,7 @@ def _generate_report(self) -> None:
         self.stdout.write(f"Total adapters: {total_adapters}")
         self.stdout.write(f"Deprecated adapters: {deprecated_count}")
         self.stdout.write(
-            f"Deprecation rate: {(deprecated_count/total_adapters*100):.1f}%\n"
+            f"Deprecation rate: {(deprecated_count / total_adapters * 100):.1f}%\n"
             if total_adapters > 0
             else "N/A\n"
         )

backend/adapter_processor_v2/models.py

@@ -9,17 +9,16 @@
 from django.db import models
 from django.db.models import QuerySet
 from tenant_account_v2.models import OrganizationMember
+from unstract.sdk1.constants import AdapterTypes
+from unstract.sdk1.exceptions import SdkError
+from unstract.sdk1.llm import LLM
 from utils.exceptions import InvalidEncryptionKey
 from utils.models.base_model import BaseModel
 from utils.models.organization_mixin import (
     DefaultOrganizationManagerMixin,
     DefaultOrganizationMixin,
 )
 
-from unstract.sdk1.constants import AdapterTypes
-from unstract.sdk1.exceptions import SdkError
-from unstract.sdk1.llm import LLM
-
 logger = logging.getLogger(__name__)
 
 ADAPTER_NAME_SIZE = 128

backend/adapter_processor_v2/serializers.py

@@ -2,18 +2,18 @@
 from typing import Any
 
 from account_v2.serializer import UserSerializer
+from backend.constants import FieldLengthConstants as FLC
+from backend.serializers import AuditSerializer
 from cryptography.fernet import Fernet
 from django.conf import settings
 from rest_framework import serializers
 from rest_framework.serializers import ModelSerializer
+from unstract.sdk1.constants import AdapterTypes
+from unstract.sdk1.constants import Common as common
 from utils.input_sanitizer import validate_name_field, validate_no_html_tags
 
 from adapter_processor_v2.adapter_processor import AdapterProcessor
 from adapter_processor_v2.constants import AdapterKeys
-from backend.constants import FieldLengthConstants as FLC
-from backend.serializers import AuditSerializer
-from unstract.sdk1.constants import AdapterTypes
-from unstract.sdk1.constants import Common as common
 
 from .models import AdapterInstance, UserDefaultAdapter
 

backend/api_v2/management/commands/delete_org_rate_limit.py

@@ -86,7 +86,7 @@ def handle(self, *args, **options):
         try:
             usage = APIDeploymentRateLimiter.get_current_usage(organization)
             self.stdout.write(
-                f'\nCurrent usage: {usage["org_count"]}/{default_limit} concurrent requests'
+                f"\nCurrent usage: {usage['org_count']}/{default_limit} concurrent requests"
             )
 
             if usage["org_count"] >= default_limit:

backend/api_v2/management/commands/get_org_rate_limit.py

@@ -65,10 +65,10 @@ def handle(self, *args, **options):
 
             self.stdout.write("\n--- Current Usage ---")
             self.stdout.write(
-                f'Organization: {usage["org_count"]}/{usage["org_limit"]} concurrent requests'
+                f"Organization: {usage['org_count']}/{usage['org_limit']} concurrent requests"
             )
             self.stdout.write(
-                f'Global System: {usage["global_count"]}/{usage["global_limit"]} concurrent requests'
+                f"Global System: {usage['global_count']}/{usage['global_limit']} concurrent requests"
             )
 
             # Usage percentage

backend/api_v2/management/commands/list_org_rate_limits.py

@@ -37,7 +37,7 @@ def handle(self, *args, **options):
                     usage = APIDeploymentRateLimiter.get_current_usage(org)
                     pct = (usage["org_count"] / limit * 100) if limit > 0 else 0
                     self.stdout.write(
-                        f'  Usage: {usage["org_count"]}/{limit} ({pct:.1f}%)'
+                        f"  Usage: {usage['org_count']}/{limit} ({pct:.1f}%)"
                     )
                 except Exception as e:
                     self.stdout.write(f"  Usage: Error - {e}")

backend/api_v2/management/commands/set_org_rate_limit.py

@@ -55,7 +55,7 @@ def handle(self, *args, **options):
             usage = APIDeploymentRateLimiter.get_current_usage(organization)
             self.stdout.write(
                 self.style.WARNING(
-                    f'Current usage: {usage["org_count"]}/{limit} concurrent requests'
+                    f"Current usage: {usage['org_count']}/{limit} concurrent requests"
                 )
             )
 

backend/api_v2/serializers.py

@@ -4,6 +4,7 @@
 from typing import Any
 from urllib.parse import urlparse
 
+from backend.serializers import AuditSerializer
 from django.apps import apps
 from django.core.validators import RegexValidator
 from pipeline_v2.models import Pipeline
@@ -30,7 +31,6 @@
 
 from api_v2.constants import ApiExecution
 from api_v2.models import APIDeployment, APIKey
-from backend.serializers import AuditSerializer
 
 
 class APIDeploymentSerializer(IntegrityErrorMixin, AuditSerializer):

backend/backend/exceptions.py

@@ -3,7 +3,6 @@
 from rest_framework.exceptions import APIException
 from rest_framework.response import Response
 from rest_framework.views import exception_handler
-
 from unstract.connectors.exceptions import ConnectorBaseException, ConnectorError
 
 

backend/configuration/config_registry.py

@@ -165,9 +165,9 @@ def cast_value(cls, key_name: str, raw_value: Any) -> Any:
 
             converters = {
                 ConfigType.INT: int,
-                ConfigType.BOOL: lambda v: v.lower() in ("true", "1")
-                if isinstance(v, str)
-                else bool(v),
+                ConfigType.BOOL: lambda v: (
+                    v.lower() in ("true", "1") if isinstance(v, str) else bool(v)
+                ),
                 ConfigType.JSON: lambda v: __import__("json").loads(v),
                 ConfigType.STRING: str,
             }

backend/configuration/enums.py

@@ -64,9 +64,9 @@ class ConfigKey(Enum):
     def cast_value(self, raw_value: Any):
         converters = {
             ConfigType.INT: int,
-            ConfigType.BOOL: lambda v: v.lower() in ("true", "1")
-            if isinstance(v, str)
-            else bool(v),
+            ConfigType.BOOL: lambda v: (
+                v.lower() in ("true", "1") if isinstance(v, str) else bool(v)
+            ),
             ConfigType.JSON: json.loads,
             ConfigType.STRING: str,
         }

backend/connector_auth_v2/pipeline/google.py

@@ -3,7 +3,6 @@
 from connector_auth_v2.constants import SocialAuthConstants as AuthConstants
 from connector_auth_v2.exceptions import EnrichConnectorMetadataException
 from connector_processor.constants import ConnectorKeys
-
 from unstract.connectors.filesystems.google_drive.constants import GDriveConstants
 
 

backend/connector_processor/connector_processor.py

@@ -3,23 +3,23 @@
 import logging
 from typing import Any
 
+from backend.exceptions import UnstractFSException
 from connector_auth_v2.constants import ConnectorAuthKey
 from connector_auth_v2.pipeline.common import ConnectorAuthHelper
 from connector_v2.constants import ConnectorInstanceKey as CIKey
+from unstract.connectors.base import UnstractConnector
+from unstract.connectors.connectorkit import Connectorkit
+from unstract.connectors.enums import ConnectorMode
+from unstract.connectors.exceptions import ConnectorError, FSAccessDeniedError
+from unstract.connectors.filesystems.ucs import UnstractCloudStorage
 
-from backend.exceptions import UnstractFSException
 from connector_processor.constants import ConnectorKeys
 from connector_processor.exceptions import (
     InvalidConnectorID,
     InvalidConnectorMode,
     OAuthTimeOut,
     TestConnectorInputError,
 )
-from unstract.connectors.base import UnstractConnector
-from unstract.connectors.connectorkit import Connectorkit
-from unstract.connectors.enums import ConnectorMode
-from unstract.connectors.exceptions import ConnectorError, FSAccessDeniedError
-from unstract.connectors.filesystems.ucs import UnstractCloudStorage
 
 logger = logging.getLogger(__name__)
 
@@ -145,8 +145,7 @@ def test_connectors(connector_id: str, credentials: dict[str, Any]) -> bool:
                     )
                 except Exception as exc:
                     logger.error(
-                        "Error while testing file based OAuth supported "
-                        "connectors: %s",
+                        "Error while testing file based OAuth supported connectors: %s",
                         exc,
                     )
                     raise OAuthTimeOut()

backend/connector_processor/exceptions.py

@@ -1,6 +1,5 @@
-from rest_framework.exceptions import APIException
-
 from backend.exceptions import UnstractBaseException
+from rest_framework.exceptions import APIException
 from unstract.connectors.exceptions import ConnectorError
 
 

backend/connector_processor/serializers.py

@@ -1,6 +1,6 @@
+from backend.constants import FieldLengthConstants as FLC
 from rest_framework import serializers
 
-from backend.constants import FieldLengthConstants as FLC
 from connector_processor.constants import ConnectorKeys
 
 

backend/connector_v2/migrations/0003_migrate_to_centralized_connectors.py

@@ -221,7 +221,7 @@ def _fix_remaining_duplicate_names(connector_instance_model: Any) -> int:
     for i in range(0, len(duplicates), batch_size):
         batch = duplicates[i : i + batch_size]
         logger.info(
-            f"Processing batch {i//batch_size + 1}/{(len(duplicates)-1)//batch_size + 1}"
+            f"Processing batch {i // batch_size + 1}/{(len(duplicates) - 1) // batch_size + 1}"
         )
 
         for dup_info in batch:

backend/connector_v2/models.py

@@ -3,6 +3,7 @@
 from typing import Any
 
 from account_v2.models import User
+from backend.constants import FieldLengthConstants as FLC
 from connector_auth_v2.models import ConnectorAuth
 from connector_processor.connector_processor import ConnectorProcessor
 from connector_processor.constants import ConnectorKeys
@@ -14,8 +15,6 @@
     DefaultOrganizationMixin,
 )
 
-from backend.constants import FieldLengthConstants as FLC
-
 CONNECTOR_NAME_SIZE = 128
 VERSION_NAME_SIZE = 64
 

backend/connector_v2/serializers.py

@@ -2,18 +2,18 @@
 from collections import OrderedDict
 from typing import Any
 
+from backend.serializers import AuditSerializer
 from connector_auth_v2.models import ConnectorAuth
 from connector_auth_v2.pipeline.common import ConnectorAuthHelper
 from connector_processor.connector_processor import ConnectorProcessor
 from connector_processor.constants import ConnectorKeys
 from connector_processor.exceptions import OAuthTimeOut
 from rest_framework.serializers import CharField, SerializerMethodField
+from unstract.connectors.filesystems.ucs import UnstractCloudStorage
 from utils.fields import EncryptedBinaryFieldSerializer
 from utils.input_sanitizer import validate_name_field
 
-from backend.serializers import AuditSerializer
 from connector_v2.constants import ConnectorInstanceKey as CIKey
-from unstract.connectors.filesystems.ucs import UnstractCloudStorage
 
 from .models import ConnectorInstance