Add authentication

app.py

@@ -1,23 +1,21 @@
-import gradio as gr
-import json
-import time
-import traceback
-from validation import validate_json, validate_croissant, validate_records, generate_validation_report
 import requests
-import requests, os
+import os
 
+# Make sure the HF token is loaded
 HF_TOKEN = os.environ.get("HF_TOKEN")
-AUTH_HEADERS = {"Authorization": f"Bearer {HF_TOKEN}"} if HF_TOKEN else {}
 
-_original_get = requests.get
-def authenticated_get(url, *args, **kwargs):
-    if "headers" not in kwargs:
-        kwargs["headers"] = AUTH_HEADERS
-    else:
-        kwargs["headers"].update(AUTH_HEADERS)
-    return _original_get(url, *args, **kwargs)
+# Set the environment variables Croissant expects
+os.environ["CROISSANT_BASIC_AUTH_USERNAME"] = "hf_user"
+os.environ["CROISSANT_BASIC_AUTH_PASSWORD"] = HF_TOKEN or ""
+
+print("[DEBUG] HF_TOKEN is", "set" if HF_TOKEN else "missing")
+print("[DEBUG] Basic auth env set for Croissant")
 
-requests.get = authenticated_get
+import gradio as gr
+import json
+import time
+import traceback
+from validation import validate_json, validate_croissant, validate_records, generate_validation_report
 
 def process_file(file):
     results = []

validation.py

@@ -1,23 +1,20 @@
-import json
-import traceback
-import mlcroissant as mlc
-import func_timeout
-import os
 import requests
+import os
 
+# Make sure the HF token is loaded
 HF_TOKEN = os.environ.get("HF_TOKEN")
-AUTH_HEADERS = {"Authorization": f"Bearer {HF_TOKEN}"} if HF_TOKEN else {}
 
-_original_get = requests.get
+# Set the environment variables Croissant expects
+os.environ["CROISSANT_BASIC_AUTH_USERNAME"] = "hf_user"
+os.environ["CROISSANT_BASIC_AUTH_PASSWORD"] = HF_TOKEN or ""
 
-def authenticated_get(url, *args, **kwargs):
-    if "headers" not in kwargs:
-        kwargs["headers"] = AUTH_HEADERS
-    else:
-        kwargs["headers"].update(AUTH_HEADERS)
-    return _original_get(url, *args, **kwargs)
+print("[DEBUG] HF_TOKEN is", "set" if HF_TOKEN else "missing")
+print("[DEBUG] Basic auth env set for Croissant")
 
-requests.get = authenticated_get
+import mlcroissant as mlc
+import func_timeout
+import json
+import traceback
 
 WAIT_TIME = 10 * 60  # seconds