wip

django_app/redbox_app/redbox_core/views/document_views.py

@@ -1,12 +1,13 @@
 import logging
 import uuid
 from collections.abc import MutableSequence, Sequence
+from io import BytesIO
 from pathlib import Path
 
 from django.contrib.auth import get_user_model
 from django.contrib.auth.decorators import login_required
 from django.core.exceptions import FieldError, SuspiciousFileOperation, ValidationError
-from django.core.files.uploadedfile import UploadedFile
+from django.core.files.uploadedfile import InMemoryUploadedFile, UploadedFile
 from django.http import HttpRequest, HttpResponse, JsonResponse
 from django.shortcuts import get_object_or_404, redirect, render
 from django.urls import reverse
@@ -82,8 +83,10 @@ def post(self, request: HttpRequest) -> HttpResponse:
         if not uploaded_files:
             errors.append("No document selected")
 
-        for uploaded_file in uploaded_files:
+        for index, uploaded_file in enumerate(uploaded_files):
             errors += self.validate_uploaded_file(uploaded_file)
+            if not self.is_utf8_compatible(uploaded_file):
+                uploaded_files[index] = self.convert_to_utf8(uploaded_file)
 
         if not errors:
             for uploaded_file in uploaded_files:
@@ -108,22 +111,56 @@ def build_response(request: HttpRequest, errors: Sequence[str] | None = None) ->
     @staticmethod
     def validate_uploaded_file(uploaded_file: UploadedFile) -> Sequence[str]:
         errors: MutableSequence[str] = []
-
         if not uploaded_file.name:
             errors.append("File has no name")
         else:
             file_extension = Path(uploaded_file.name).suffix
             if file_extension.lower() not in APPROVED_FILE_EXTENSIONS:
                 errors.append(f"Error with {uploaded_file.name}: File type {file_extension} not supported")
-
         if not uploaded_file.content_type:
             errors.append(f"Error with {uploaded_file.name}: File has no content-type")
-
         if uploaded_file.size > MAX_FILE_SIZE:
             errors.append(f"Error with {uploaded_file.name}: File is larger than 200MB")
-
         return errors
 
+    @staticmethod
+    def is_utf8_compatible(uploaded_file: UploadedFile) -> bool:
+        try:
+            uploaded_file.open()
+            uploaded_file.read().decode("utf-8")
+            uploaded_file.seek(0)
+        except UnicodeDecodeError:
+            logger.info("File is incompatible. Converting...")
+            return False
+        else:
+            logger.info("File is compatible with utf-8 - ready for processing")
+            return True
+
+    @staticmethod
+    def convert_to_utf8(uploaded_file: UploadedFile) -> UploadedFile:
+        try:
+            uploaded_file.open()
+            content = uploaded_file.read().decode("ISO-8859-1")
+
+            # Detect and replace non-UTF-8 characters
+            new_bytes = content.encode("utf-8")
+
+            # Creating a new InMemoryUploadedFile object with the converted content
+            new_uploaded_file = InMemoryUploadedFile(
+                file=BytesIO(new_bytes),
+                field_name=uploaded_file.field_name,
+                name=uploaded_file.name,
+                content_type="application/octet-stream",
+                size=len(new_bytes),
+                charset="utf-8",
+            )
+        except Exception as e:
+            logger.exception("Error converting file %s to UTF-8.", uploaded_file, exc_info=e)
+            return uploaded_file
+        else:
+            logger.info("Conversion to UTF-8 successful")
+            return new_uploaded_file
+
     @staticmethod
     def ingest_file(uploaded_file: UploadedFile, user: User) -> Sequence[str]:
         try: