auth and fetch user details

README.md

@@ -4,6 +4,7 @@ emoji: 🥇
 colorFrom: green
 colorTo: indigo
 sdk: gradio
+hf_oauth: true
 app_file: app.py
 pinned: true
 license: apache-2.0

app.py

@@ -2,6 +2,7 @@ import gradio as gr
 import pandas as pd
 from apscheduler.schedulers.background import BackgroundScheduler
 from gradio_leaderboard import ColumnFilter, Leaderboard, SelectColumns
+from huggingface_hub import whoami
 
 from src.about import CITATION_BUTTON_LABEL, CITATION_BUTTON_TEXT, EVALUATION_QUEUE_TEXT, INTRODUCTION_TEXT, TITLE
 from src.datamodel.data import F1Data
@@ -11,7 +12,7 @@ from src.display.utils import AutoEvalColumn, ModelType, fields
 from src.envs import API, CODE_PROBLEMS_REPO, REPO_ID, RESULTS_REPO, SUBMISSIONS_REPO
 from src.logger import get_logger
 from src.populate import get_leaderboard_df
-from src.submission.submit import add_new_solutions
+from src.submission.submit import add_new_solutions, fetch_sub_claim
 from src.validation.validate import MAX_INPUT_LENGTH, MIN_INPUT_LENGTH, is_submission_file_valid, is_valid
 
 logger = get_logger(__name__)
@@ -77,6 +78,75 @@ def init_leaderboard(dataframe: pd.DataFrame):
     )
 
 
+def add_solution_cbk(
+    system_name: str,
+    org: str,
+    sys_type: str,
+    submission_path: str,
+    profile: gr.OAuthProfile | None,
+    token: gr.OAuthToken | None,
+):
+    logger.info("Fetching user details for submission")
+    if profile is None or token is None:
+        return styled_error("Please sign in with Hugging Face before submitting.")
+
+    # Display handle and display name (good for the UI)
+    handle = profile.username
+    logger.info(f"User handle: {handle}")
+    display_name = profile.name or handle
+    logger.info(f"Displaying name: {display_name}")
+
+    # Optional: verify handle/orgs via Hub API
+    info = whoami(token.token)  # {'name': 'user', 'orgs': [{'name': 'org1'}, ...], ...}
+    logger.info(f"User info: {info}")
+
+    # Stable account id (for dedup, enforcement, joins)
+    claims = fetch_sub_claim(token) or {}
+    logger.info(f"Claims: {claims}")
+    stable_id = claims.get("sub")
+    logger.info(f"User stable ID: {stable_id}")
+
+    try:
+        # Validating the submission file.
+        if len(submission_path) == 0:
+            return styled_error("Please upload JSONL submission file.")
+
+        if not is_submission_file_valid(
+            submission_path,
+            is_warmup_dataset=(SPLIT == "warmup"),
+        ):
+            return styled_error("Failed to read JSONL submission file. Please try again later.")
+
+        # Validating all user-supplied arguments.
+        for val, val_name in [
+            (system_name, "System name"),
+            (org, "Organisation name"),
+            (sys_type, "System type"),
+        ]:
+            if len(val) == 0:
+                return styled_error(f"Please fill in the '{val_name}' field.")
+
+            if not is_valid(val):
+                return styled_error(
+                    f"{val_name} is invalid! Must only contain characters [a-zA-Z0-9], spaces, "
+                    + "or the special characters '-' and '.', and be of length between "
+                    + f"{MIN_INPUT_LENGTH} and {MAX_INPUT_LENGTH}."
+                )
+    except Exception:
+        logger.warning("Failed to process user submission", exc_info=True)
+        return styled_error("An error occurred. Please try again later.")  # Intentionally vague.
+
+    return add_new_solutions(
+        lbdb,
+        system_name,
+        org,
+        sys_type,
+        submission_path,
+        is_warmup_dataset=(SPLIT == "warmup"),
+        ensure_all_present=ENSURE_ALL_PRESENT,
+    )
+
+
 blocks = gr.Blocks(css=custom_css)
 with blocks:
     gr.Image(
@@ -126,81 +196,43 @@ with blocks:
             with gr.Row():
                 gr.Markdown("# ✉️✨ Submit your solutions here!", elem_classes="markdown-text")
 
-            with gr.Row():
-                with gr.Column():
-                    system_name_textbox = gr.Textbox(label=AutoEvalColumn.system.name)
-                    org_textbox = gr.Textbox(label=AutoEvalColumn.organization.name)
-                    sys_type_dropdown = gr.Dropdown(
-                        choices=[t.to_str() for t in ModelType],
-                        label=AutoEvalColumn.system_type.name,
-                        multiselect=False,
-                        value=ModelType.LLM.to_str(),
-                        interactive=True,
-                    )
-
-                    submission_file = gr.File(label="JSONL solutions file", file_types=[".jsonl"])
-
-            logger.info("Submit button")
-            submit_button = gr.Button("Submit")
-            submission_result = gr.Markdown()
-
-            def add_solution_cbk(
-                system_name: str,
-                org: str,
-                sys_type: str,
-                submission_path: str,
-            ):
-
-                try:
-                    # Validating the submission file.
-                    if len(submission_path) == 0:
-                        return styled_error("Please upload JSONL submission file.")
-
-                    if not is_submission_file_valid(
-                        submission_path,
-                        is_warmup_dataset=(SPLIT == "warmup"),
-                    ):
-                        return styled_error("Failed to read JSONL submission file. Please try again later.")
-
-                    # Validating all user-supplied arguments.
-                    for val, val_name in [
-                        (system_name, "System name"),
-                        (org, "Organisation name"),
-                        (sys_type, "System type"),
-                    ]:
-                        if len(val) == 0:
-                            return styled_error(f"Please fill in the '{val_name}' field.")
-
-                        if not is_valid(val):
-                            return styled_error(
-                                f"{val_name} is invalid! Must only contain characters [a-zA-Z0-9], spaces, "
-                                + "or the special characters '-' and '.', and be of length between "
-                                + f"{MIN_INPUT_LENGTH} and {MAX_INPUT_LENGTH}."
-                            )
-                except Exception:
-                    logger.warning("Failed to process user submission", exc_info=True)
-                    return styled_error("An error occurred. Please try again later.")  # Intentionally vague.
-
-                return add_new_solutions(
-                    lbdb,
-                    system_name,
-                    org,
-                    sys_type,
-                    submission_path,
-                    is_warmup_dataset=(SPLIT == "warmup"),
-                    ensure_all_present=ENSURE_ALL_PRESENT,
-                )
+            # Shown when logged OUT
+            login_box = gr.Group(visible=True)
+            with login_box:
+                gr.Markdown("### Submit your results\nPlease sign in to continue.")
+                gr.LoginButton()
 
-            submit_button.click(
-                add_solution_cbk,
-                [
-                    system_name_textbox,
-                    org_textbox,
-                    sys_type_dropdown,
-                    submission_file,
-                ],
-                submission_result,
-            )
+            # Shown when logged IN
+            submit_panel = gr.Group(visible=False)
+            with submit_panel:
+                with gr.Row():
+                    with gr.Column():
+                        system_name_textbox = gr.Textbox(label=AutoEvalColumn.system.name)
+                        org_textbox = gr.Textbox(label=AutoEvalColumn.organization.name)
+                        sys_type_dropdown = gr.Dropdown(
+                            choices=[t.to_str() for t in ModelType],
+                            label=AutoEvalColumn.system_type.name,
+                            multiselect=False,
+                            value=ModelType.LLM.to_str(),
+                            interactive=True,
+                        )
+
+                        submission_file = gr.File(label="JSONL solutions file", file_types=[".jsonl"])
+
+                logger.info("Submit button")
+                submit_button = gr.Button("Submit")
+                submission_result = gr.Markdown()
+
+                submit_button.click(
+                    add_solution_cbk,
+                    [
+                        system_name_textbox,
+                        org_textbox,
+                        sys_type_dropdown,
+                        submission_file,
+                    ],
+                    submission_result,
+                )
 
     with gr.Row():
         logger.info("Citation")

src/submission/submit.py

@@ -1,9 +1,12 @@
 import time
 from datetime import datetime, timezone
+import os
+import requests
 
 import pandas as pd
 from datasets import Dataset
 from pandas.api.types import is_integer_dtype
+import gradio as gr
 
 from src.datamodel.data import F1Data
 from src.display.formatting import styled_error, styled_message
@@ -15,31 +18,6 @@ from src.validation.validate import is_submission_file_valid, is_valid
 logger = get_logger(__name__)
 
 
-def _validate_all_submissions_present(
-    lbdb: F1Data,
-    pd_ds: pd.DataFrame,
-):
-    logger.info(f"Validating DS size {len(pd_ds)} columns {pd_ds.columns} set {set(pd_ds.columns)}")
-    expected_cols = ["problem_id", "solution"]
-
-    if set(pd_ds.columns) != set(expected_cols):
-        return ValueError(f"Expected attributes: {expected_cols}, Got: {pd_ds.columns.tolist()}")
-
-    if not is_integer_dtype(pd_ds["problem_id"]):
-        return ValueError("problem_id must be str convertible to int")
-
-    if any(type(v) is not str for v in pd_ds["solution"]):
-        return ValueError("solution must be of type str")
-
-    submitted_ids = set(pd_ds.problem_id.astype(str))
-    if submitted_ids != lbdb.code_problem_ids:
-        missing = lbdb.code_problem_ids - submitted_ids
-        unknown = submitted_ids - lbdb.code_problem_ids
-        raise ValueError(f"Mismatched problem IDs: {len(missing)} missing, {len(unknown)} unknown")
-    if len(pd_ds) > len(lbdb.code_problem_ids):
-        return ValueError("Duplicate problem IDs exist in uploaded file")
-
-
 def add_new_solutions(
     lbdb: F1Data,
     system_name: str,
@@ -70,7 +48,7 @@ def add_new_solutions(
             "An error occurred. Please try again later."
         )  # Use same message as external error. Avoid infoleak.
 
-    submission_id = f"{system_name}_{org}_{sys_type}_{datetime.now(timezone.utc).strftime('%Y%m%d_%H%M%S')}"
+    submission_id = f"{datetime.now(timezone.utc).strftime('%Y%m%d_%H%M%S')}_{system_name}_{org}_{sys_type}"
 
     # Seems good, creating the eval.
     logger.info(f"Adding new submission: {submission_id}")
@@ -99,3 +77,61 @@ def add_new_solutions(
         "Your request has been submitted to the evaluation queue!\n"
         + "Results may take up to 24 hours to be processed and shown in the leaderboard."
     )
+
+
+def _validate_all_submissions_present(
+    lbdb: F1Data,
+    pd_ds: pd.DataFrame,
+):
+    logger.info(f"Validating DS size {len(pd_ds)} columns {pd_ds.columns} set {set(pd_ds.columns)}")
+    expected_cols = ["problem_id", "solution"]
+
+    if set(pd_ds.columns) != set(expected_cols):
+        return ValueError(f"Expected attributes: {expected_cols}, Got: {pd_ds.columns.tolist()}")
+
+    if not is_integer_dtype(pd_ds["problem_id"]):
+        return ValueError("problem_id must be str convertible to int")
+
+    if any(type(v) is not str for v in pd_ds["solution"]):
+        return ValueError("solution must be of type str")
+
+    submitted_ids = set(pd_ds.problem_id.astype(str))
+    if submitted_ids != lbdb.code_problem_ids:
+        missing = lbdb.code_problem_ids - submitted_ids
+        unknown = submitted_ids - lbdb.code_problem_ids
+        raise ValueError(f"Mismatched problem IDs: {len(missing)} missing, {len(unknown)} unknown")
+    if len(pd_ds) > len(lbdb.code_problem_ids):
+        return ValueError("Duplicate problem IDs exist in uploaded file")
+
+
+def fetch_sub_claim(oauth_token: gr.OAuthToken | None) -> dict | None:
+    if oauth_token is None:
+        return None
+    provider = os.getenv("OPENID_PROVIDER_URL")
+    if not provider:
+        return None
+    try:
+        oidc_meta = requests.get(f"{provider}/.well-known/openid-configuration", timeout=5).json()
+        userinfo_ep = oidc_meta["userinfo_endpoint"]
+        claims = requests.get(userinfo_ep, headers={"Authorization": f"Bearer {oauth_token.token}"}, timeout=5).json()
+        # Typical fields: sub (stable id), preferred_username, name, picture
+        return {
+            "sub": claims.get("sub"),
+            "preferred_username": claims.get("preferred_username"),
+            "name": claims.get("name"),
+        }
+    except Exception:
+        return None
+
+
+# --- on-load gate: show/hide submit panel based on login state ---
+def gate_submission(profile: gr.OAuthProfile | None):
+    """
+    Returns:
+      - login_box visibility (True if logged OUT)
+      - submit_panel visibility (True if logged IN)
+      - status line
+    """
+    if profile is None:
+        return gr.update(visible=True), gr.update(visible=False), "You're not signed in."
+    return gr.update(visible=False), gr.update(visible=True), f"Signed in as @{profile.username}"