update

app.py

@@ -10,7 +10,7 @@ from src.leaderboard.student_results import (
 )
 import time
 from src.submission.student_queue import queue_student_submission
-from src.envs import EVAL_RESULTS_PATH, RESULTS_REPO, TOKEN
+from src.envs import EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, QUEUE_REPO, RESULTS_REPO, TOKEN
 
 
 demo = gr.Blocks(css=custom_css)
@@ -69,6 +69,7 @@ with demo:
                         return f"Timed out waiting for results. Please check back later. Submission ID: {ts}"
                     
                     try:
+                        # Refresh both results and requests so status/error and metrics stay in sync
                         snapshot_download(
                             repo_id=RESULTS_REPO,
                             local_dir=EVAL_RESULTS_PATH,
@@ -77,6 +78,14 @@ with demo:
                             etag_timeout=30,
                             token=TOKEN,
                         )
+                        snapshot_download(
+                            repo_id=QUEUE_REPO,
+                            local_dir=EVAL_REQUESTS_PATH,
+                            repo_type="dataset",
+                            tqdm_class=None,
+                            etag_timeout=30,
+                            token=TOKEN,
+                        )
                     except Exception:
                         pass # Ignore network errors during polling
                     
@@ -183,7 +192,7 @@ with demo:
                 if not group_id:
                     return "Please enter a valid Group ID.", pd.DataFrame()
 
-                # Ensure we have the latest evaluation results locally
+                # Ensure we have the latest evaluation results and request metadata locally
                 try:
                     snapshot_download(
                         repo_id=RESULTS_REPO,
@@ -193,6 +202,14 @@ with demo:
                         etag_timeout=30,
                         token=TOKEN,
                     )
+                    snapshot_download(
+                        repo_id=QUEUE_REPO,
+                        local_dir=EVAL_REQUESTS_PATH,
+                        repo_type="dataset",
+                        tqdm_class=None,
+                        etag_timeout=30,
+                        token=TOKEN,
+                    )
                 except Exception:
                     # Network/cache errors shouldn't break the UI; we'll just use whatever is cached locally.
                     pass

src/leaderboard/student_results.py

@@ -5,7 +5,7 @@ from typing import Dict, Optional, Tuple
 import numpy as np
 import pandas as pd
 
-from src.envs import EVAL_RESULTS_PATH, PROJ_DIR
+from src.envs import EVAL_REQUESTS_PATH, EVAL_RESULTS_PATH, PROJ_DIR
 
 
 def _load_all_attempts() -> pd.DataFrame:
@@ -100,22 +100,23 @@ def get_student_leaderboard_df(dataset_name: Optional[str] = None) -> Tuple[pd.D
 
 
 def get_student_status(group_id: str) -> str:
-    attempts_dir = os.path.join(EVAL_RESULTS_PATH, PROJ_DIR, "student_attempts")
+    attempts_dir = os.path.join(EVAL_REQUESTS_PATH, PROJ_DIR)
     if not os.path.isdir(attempts_dir):
         return "No submissions found."
 
     rows = []
-    for fname in os.listdir(attempts_dir):
-        if not fname.endswith(".json"):
-            continue
-        try:
-            with open(os.path.join(attempts_dir, fname), "r") as f:
-                data = json.load(f)
-                # Ensure group_id comparison is robust (string vs int)
-                if str(data.get("group_id", "")).strip() == str(group_id).strip():
-                    rows.append(data)
-        except Exception:
-            continue
+    for root, _, files in os.walk(attempts_dir):
+        for fname in files:
+            if not fname.endswith(".json"):
+                continue
+            try:
+                with open(os.path.join(root, fname), "r") as f:
+                    data = json.load(f)
+                    # Ensure group_id comparison is robust (string vs int)
+                    if str(data.get("group_id", "")).strip() == str(group_id).strip():
+                        rows.append(data)
+            except Exception:
+                continue
 
     if not rows:
         return "No submissions found."
@@ -137,22 +138,24 @@ def get_student_status(group_id: str) -> str:
 
 
 def get_latest_submission(group_id: str) -> Optional[Dict]:
-    attempts_dir = os.path.join(EVAL_RESULTS_PATH, PROJ_DIR, "student_attempts")
-    if not os.path.isdir(attempts_dir):
+    """Return the latest request.json entry for a given group_id (used for status + error)."""
+    requests_dir = os.path.join(EVAL_REQUESTS_PATH, PROJ_DIR)
+    if not os.path.isdir(requests_dir):
         return None
 
     rows = []
-    for fname in os.listdir(attempts_dir):
-        if not fname.endswith(".json"):
-            continue
-        try:
-            with open(os.path.join(attempts_dir, fname), "r") as f:
-                data = json.load(f)
-                # Ensure group_id comparison is robust (string vs int)
-                if str(data.get("group_id", "")).strip() == str(group_id).strip():
-                    rows.append(data)
-        except Exception:
-            continue
+    for root, _, files in os.walk(requests_dir):
+        for fname in files:
+            if not fname.endswith(".json"):
+                continue
+            try:
+                with open(os.path.join(root, fname), "r") as f:
+                    data = json.load(f)
+                    # Ensure group_id comparison is robust (string vs int)
+                    if str(data.get("group_id", "")).strip() == str(group_id).strip():
+                        rows.append(data)
+            except Exception:
+                continue
 
     if not rows:
         return None
@@ -163,28 +166,64 @@ def get_latest_submission(group_id: str) -> Optional[Dict]:
 
 
 def get_group_submission_history(group_id: str) -> pd.DataFrame:
-    """Return all submissions for a given group_id, sorted by timestamp (latest first)."""
-    attempts_dir = os.path.join(EVAL_RESULTS_PATH, PROJ_DIR, "student_attempts")
-    if not os.path.isdir(attempts_dir):
-        return pd.DataFrame()
+    """
+    Return all submissions for a given group_id, sorted by timestamp (latest first).
 
-    rows = []
-    for fname in os.listdir(attempts_dir):
-        if not fname.endswith(".json"):
-            continue
-        try:
-            with open(os.path.join(attempts_dir, fname), "r") as f:
-                data = json.load(f)
-                # Ensure group_id comparison is robust (string vs int)
-                if str(data.get("group_id", "")).strip() == str(group_id).strip():
-                    rows.append(data)
-        except Exception:
-            continue
+    Status and error are sourced from request.json entries (request dataset),
+    while performance metrics are merged in from the results student_attempts
+    table when available.
+    """
+    requests_dir = os.path.join(EVAL_REQUESTS_PATH, PROJ_DIR)
+    if not os.path.isdir(requests_dir):
+        return pd.DataFrame()
 
-    if not rows:
+    request_rows = []
+    for root, _, files in os.walk(requests_dir):
+        for fname in files:
+            if not fname.endswith(".json"):
+                continue
+            try:
+                with open(os.path.join(root, fname), "r") as f:
+                    data = json.load(f)
+                    if str(data.get("group_id", "")).strip() == str(group_id).strip():
+                        # Normalize fields we care about
+                        datasets = data.get("datasets") or []
+                        dataset = datasets[0] if isinstance(datasets, list) and datasets else ""
+                        request_rows.append(
+                            {
+                                "group_id": str(data.get("group_id", "")).strip(),
+                                "alias": data.get("alias"),
+                                "timestamp": data.get("timestamp", ""),
+                                "status": data.get("status", "UNKNOWN"),
+                                "error": data.get("error", ""),
+                                "dataset": dataset,
+                            }
+                        )
+            except Exception:
+                continue
+
+    if not request_rows:
         return pd.DataFrame()
 
-    df = pd.DataFrame(rows)
-    if "timestamp" in df.columns:
-        df = df.sort_values("timestamp", ascending=False)
-    return df
+    requests_df = pd.DataFrame(request_rows)
+
+    # Merge in performance metrics from results student_attempts when possible.
+    attempts_df = _load_all_attempts()
+    if not attempts_df.empty:
+        attempts_df = attempts_df.copy()
+        attempts_df["group_id"] = attempts_df["group_id"].astype(str).str.strip()
+        requests_df["group_id"] = requests_df["group_id"].astype(str).str.strip()
+
+        merged = requests_df.merge(
+            attempts_df,
+            on=["group_id", "timestamp"],
+            how="left",
+            suffixes=("", "_results"),
+        )
+    else:
+        merged = requests_df
+
+    if "timestamp" in merged.columns:
+        merged = merged.sort_values("timestamp", ascending=False)
+
+    return merged