LLM-Driven Extraction of Unstructured Data — Built for API Deployments & ETL Pipeline Workflows
Python ai-agentsdata-engineeringdocument-aigenerative-aiidpjson-extractionllmmcp-serverocrpdf-extractionprompt-engineeringstructured-output
[Fix] Multi Doc support in prompt studio (#1078)
jagadeeswaran-zipstack02a1ac221a44
diff --git a/backend/utils/FileValidator.py b/backend/utils/FileValidator.pyindex 173e41f84..7b715a22e 100644--- a/backend/utils/FileValidator.py+++ b/backend/utils/FileValidator.py@@ -8,6 +8,7 @@ from django.core.files.uploadedfile import InMemoryUploadedFilefrom django.template.defaultfilters import filesizeformatfrom django.utils.translation import gettext_lazy as _from typing_extensions import NotRequired, Unpack+from unstract.sdk.file_storage.constants import FileOperationParamsclass FileValidationParam(TypedDict):@@ -69,7 +70,10 @@ class FileValidator:raise ValidationError(message)def _check_file_mime_type(self, file: InMemoryUploadedFile) -> None:- mimetype = magic.from_buffer(file.read(2048), mime=True)+ # TODO: Need to optimise, istead of reading entire file.+ mimetype = magic.from_buffer(+ file.read(FileOperationParams.READ_ENTIRE_LENGTH), mime=True+ )file.seek(0) # Reset the file pointer to the startif self.allowed_mimetypes and mimetype not in self.allowed_mimetypes: