store file uploads as plain text when preserve is not required

api/src/main.py

@@ -25,6 +25,7 @@
     generate_custom_source,
 )
 from .utils.custom_sources.manage_attachments import ManageAttachments
+from .utils.custom_sources.read_content import ReadContent
 from .utils.custom_sources.save_copied_source import CopiedPasteSourceRequest, save_copied_source
 from .utils.custom_sources.save_uploaded_sources import UploadedFiles
 from .utils.decorators.retry_decorator import RetryConfig, retry
@@ -142,7 +143,7 @@ async def get_signed_url_endpoint(blobname: str):
     """
 
     @retry(RetryConfig(max_retries=3, delay=5, backoff=1.5))
-    def handler():
+    def handler() -> str | None:
         return StorageManager().get_signed_url(blobname=blobname)
 
     url = handler()
@@ -226,24 +227,30 @@ async def detect_category_endpoint(request: DetectContentCategoryRequest):
 
 
 @app.post("/store-file-upload", response_model=str)
-async def store_file_upload(file: UploadFile, filename: str = Form(...)):
+async def store_file_upload(file: UploadFile, filename: str = Form(...), preserve: bool = Form(False)):
     """
     Store file uploaded from the frontend
     """
-    storage_manager = StorageManager()
-
-    file_obj = BytesIO(await file.read())
+    print(f"Storing file: {filename}. Preserve: {preserve}")
 
+    storage_manager = StorageManager()
     file_exists = storage_manager.check_blob_exists(filename)
     if file_exists:
         return storage_manager.get_gcs_url(filename)
 
+    file_content = await ReadContent()._read_file(file, preserve=preserve)
+    content_type = (
+        file.content_type or "application/octet-stream"
+        if preserve or isinstance(file_content, BytesIO)
+        else "text/plain"
+    )
+
     result = storage_manager.upload_to_gcs(
-        item=file_obj,
+        item=file_content,
         blobname=f"{BLOB_BASE_URI}/{filename}",
         params=UploadItemParams(
             cache_control="public, max-age=31536000",
-            content_type=file.content_type or "application/octet-stream",
+            content_type=content_type,
         ),
     )
 

api/src/utils/custom_sources/read_content.py

@@ -1,5 +1,6 @@
 from io import BytesIO
 
+from fastapi import UploadFile
 from pypdf import PdfReader
 
 
@@ -19,3 +20,18 @@ def _read_pdf(self, content: bytes) -> tuple[str, PdfReader]:
 
     def _read_txt(self, content: bytes) -> str:
         return content.decode()
+
+    async def _read_file(self, file: UploadFile, preserve: bool):
+        file_bytes = await file.read()
+
+        if preserve:
+            return BytesIO(file_bytes)
+
+        if file.content_type == "application/pdf":
+            text_content, _ = self._read_pdf(file_bytes)
+        elif file.content_type == "text/plain":
+            text_content = self._read_txt(file_bytes)
+        else:
+            return BytesIO(file_bytes)
+
+        return text_content

api/src/utils/decorators/retry_decorator.py

@@ -2,17 +2,17 @@
 from dataclasses import dataclass
 from functools import wraps
 from time import sleep
-from typing import Any
+from typing import Any, Optional
 
 
 @dataclass
 class RetryConfig:
     max_retries: int = 3
     delay: float = 1.0
-    backoff: float | None = None
+    backoff: Optional[float] = None
 
 
-def retry(retry_config: RetryConfig | None, default_return: Any = None) -> Any:
+def retry(retry_config: RetryConfig | None, default_return: Any = None):
     """
     Retry logic for async functions with exponential backoff.
     """