Skip to content

Automate browser based workflows with AI

Use suffix instead of prefix when creating temp file (#1622)

Shuchang Zheng31885113a4e1
diff --git a/skyvern/forge/sdk/api/files.py b/skyvern/forge/sdk/api/files.py
index 36e4a90d7..dfd4e34cf 100644
--- a/skyvern/forge/sdk/api/files.py
+++ b/skyvern/forge/sdk/api/files.py
@@ -24,6 +24,7 @@ async def download_from_s3(client: AsyncAWSClient, s3_uri: str) -> str:
downloaded_bytes = await client.download_file(uri=s3_uri)
filename = s3_uri.split("/")[-1] # Extract filename from the end of S3 URI
file_path = create_named_temporary_file(delete=False, file_name=filename)
+ LOG.info(f"Downloaded file to {file_path.name}")
file_path.write(downloaded_bytes)
return file_path.name
@@ -48,7 +49,7 @@ def get_file_extension_from_headers(headers: CIMultiDictProxy[str]) -> str:
async def download_file(url: str, max_size_mb: int | None = None) -> str:
try:
# Check if URL is an S3 URI
- if url.startswith("s3://skyvern-uploads/local/o_"):
+ if url.startswith(f"s3://{settings.AWS_S3_BUCKET_UPLOADS}/{settings.ENV}/o_"):
LOG.info("Downloading Skyvern file from S3", url=url)
client = AsyncAWSClient()
return await download_from_s3(client, url)
@@ -193,7 +194,17 @@ def make_temp_directory(
def create_named_temporary_file(delete: bool = True, file_name: str | None = None) -> tempfile._TemporaryFileWrapper:
temp_dir = settings.TEMP_PATH
create_folder_if_not_exist(temp_dir)
- return tempfile.NamedTemporaryFile(dir=temp_dir, delete=delete, prefix=file_name)
+
+ if file_name:
+ # Sanitize the filename to remove any dangerous characters
+ safe_file_name = sanitize_filename(file_name)
+ # Create file with exact name (without random characters)
+ file_path = os.path.join(temp_dir, safe_file_name)
+ # Open in binary mode and return a NamedTemporaryFile-like object
+ file = open(file_path, "wb")
+ return tempfile._TemporaryFileWrapper(file, file_path, delete=delete)
+
+ return tempfile.NamedTemporaryFile(dir=temp_dir, delete=delete)
def clean_up_dir(dir: str) -> None:
diff --git a/skyvern/forge/sdk/routes/agent_protocol.py b/skyvern/forge/sdk/routes/agent_protocol.py
index 352995273..5a0fab341 100644
--- a/skyvern/forge/sdk/routes/agent_protocol.py
+++ b/skyvern/forge/sdk/routes/agent_protocol.py
@@ -1,5 +1,6 @@
import datetime
import hashlib
+import os
import uuid
from enum import Enum
from typing import Annotated, Any
@@ -1094,12 +1095,25 @@ async def upload_file(
) -> Response:
bucket = app.SETTINGS_MANAGER.AWS_S3_BUCKET_UPLOADS
todays_date = datetime.datetime.now().strftime("%Y-%m-%d")
- uuid_prefixed_filename = f"{str(uuid.uuid4())}_{file.filename}"
- s3_uri = (
- f"s3://{bucket}/{app.SETTINGS_MANAGER.ENV}/{current_org.organization_id}/{todays_date}/{uuid_prefixed_filename}"
- )
- # Stream the file to S3
- uploaded_s3_uri = await aws_client.upload_file_stream(s3_uri, file.file)
+
+ # First try uploading with original filename
+ try:
+ sanitized_filename = os.path.basename(file.filename) # Remove any path components
+ s3_uri = (
+ f"s3://{bucket}/{app.SETTINGS_MANAGER.ENV}/{current_org.organization_id}/{todays_date}/{sanitized_filename}"
+ )
+ uploaded_s3_uri = await aws_client.upload_file_stream(s3_uri, file.file)
+ except Exception:
+ LOG.error("Failed to upload file to S3", exc_info=True)
+ uploaded_s3_uri = None
+
+ # If upload fails, try again with UUID prefix
+ if not uploaded_s3_uri:
+ uuid_prefixed_filename = f"{str(uuid.uuid4())}_{file.filename}"
+ s3_uri = f"s3://{bucket}/{app.SETTINGS_MANAGER.ENV}/{current_org.organization_id}/{todays_date}/{uuid_prefixed_filename}"
+ file.file.seek(0) # Reset file pointer
+ uploaded_s3_uri = await aws_client.upload_file_stream(s3_uri, file.file)
+
if not uploaded_s3_uri:
raise HTTPException(status_code=500, detail="Failed to upload file to S3.")