Update app.py

app.py

@@ -1,17 +1,16 @@
 #!/usr/bin/env python
 import os
-os.environ["GRADIO_LANGUAGE"] = "en"   
+import json
+from typing import List, Tuple
 
+os.environ["GRADIO_LANGUAGE"] = "en"   
 
 RESULT_DIR = os.environ.get("MOECAP_RESULT_DIR")
 if not RESULT_DIR:
     raise RuntimeError(
-        "MOECAP_RESULT_DIR is not set. Please set MOECAP_RESULT_DIR before running app.py"
+        "MOECAP_RESULT_DIR is not set. Please set MOECAP_RESULT_DIR (HF Repo ID) before running app.py"
     )
 
-import json
-from typing import List, Tuple
-
 import gradio as gr
 import pandas as pd
 from datasets import load_dataset
@@ -77,8 +76,6 @@ def json_to_row(path: str, metrics: dict) -> dict:
     return row
 
 
-# uoload
-
 def build_leaderboard_from_files(files: List[gr.File], prev_rows: list | None = None):
     if prev_rows is None:
         prev_rows = []
@@ -103,7 +100,7 @@ def build_leaderboard_from_files(files: List[gr.File], prev_rows: list | None =
             else:
                 links.append(str(name))
         models_str = ", ".join(links)
-        summary_md = f"**Loaded {len(prev_rows)} result files.**  \n**Models:** {models_str}"
+        summary_md = f"**Loaded {len(prev_rows)} result files.** \n**Models:** {models_str}"
         table_html = df.to_html(escape=False, index=False, classes="metrics-table")
         return summary_md, table_html, prev_rows
 
@@ -144,20 +141,31 @@ def build_leaderboard_from_files(files: List[gr.File], prev_rows: list | None =
         else:
             links.append(str(name))
     models_str = ", ".join(links)
-    summary_md = f"**Loaded {len(all_rows)} result files.**  \n**Models:** {models_str}"
+    summary_md = f"**Loaded {len(all_rows)} result files.** \n**Models:** {models_str}"
 
     table_html = df.to_html(escape=False, index=False, classes="metrics-table")
 
     return summary_md, table_html, all_rows
 
 
-def load_from_dir(dir_path: str):
+
+def load_from_dir(dir_path: str, force_refresh: bool = False):
+
     try:
         pattern = f"hf://datasets/{dir_path}/**/*.json"
-        ds = load_dataset("json", data_files={"train": pattern}, split="train")
+        
+        dl_mode = "force_redownload" if force_refresh else None
+        
+        print(f"Fetching from {pattern} (mode={dl_mode})...")
+        ds = load_dataset(
+            "json", 
+            data_files={"train": pattern}, 
+            split="train",
+            download_mode=dl_mode
+        )
     except Exception as e:
-        empty_html = "<p>No files loaded.</p>"
-        return f"Failed to load dataset `{dir_path}`: {e}", empty_html
+        empty_html = "<p>No files loaded or Dataset not found.</p>"
+        return f"Failed to load dataset `{dir_path}`: {str(e)}", empty_html
 
     rows = []
     for i, example in enumerate(ds):
@@ -194,7 +202,7 @@ def load_from_dir(dir_path: str):
     models_str = ", ".join(links)
 
     summary_md = (
-        f"**Loaded {len(rows)} result files from dataset `{dir_path}`.**  \n"
+        f"**Loaded {len(rows)} result files from dataset `{dir_path}`.** \n"
         f"**Models:** {models_str}"
     )
 
@@ -203,6 +211,11 @@ def load_from_dir(dir_path: str):
     return summary_md, table_html
 
 
+def auto_refresh_from_dir(dir_path: str):
+    return load_from_dir(dir_path, force_refresh=True)
+
+
+
 # Gradio UI
 
 def build_app() -> gr.Blocks:
@@ -250,41 +263,23 @@ def build_app() -> gr.Blocks:
                 )
 
             with gr.Column(scale=1):
-                # manual upload 
-                # files_input = gr.Files(
-                #     label="Upload `cap_metrics_*.json` files",
-                #     file_types=[".json"],
-                #     file_count="multiple",
-                # )
-                # run_button = gr.Button("Parse Uploaded Files")
-
                 dir_path = gr.Textbox(
-                    label="Load from output directory",
+                    label="HF Dataset Repo ID", 
                     value=RESULT_DIR, 
                     lines=1,
+                    placeholder="username/dataset-name"
                 )
-                load_dir_button = gr.Button("Load from directory")
-
-        # upload_summary = gr.Markdown(label="Upload Summary")
-        # upload_table = gr.HTML(label="Upload Metrics")
+                load_dir_button = gr.Button("Load from Dataset")
 
         summary_output = gr.Markdown(label="Directory Summary")
         leaderboard_output = gr.HTML(label="Directory Metrics")
 
-        # run_button.click(
-        #     fn=build_leaderboard_from_files,
-        #     inputs=files_input,
-        #     outputs=[upload_summary, upload_table],
-        # )
-
         load_dir_button.click(
             fn=load_from_dir,
             inputs=dir_path,
             outputs=[summary_output, leaderboard_output],
         )
 
-
-
         timer = gr.Timer(5.0)
         timer.tick(
             fn=auto_refresh_from_dir,