backup

app.py

@@ -254,6 +254,62 @@ def generate_json_files(source_image, responses, post_id):
     return post_folder
 
 
+def get_statistics():
+    """
+    Scan the data folder and return statistics about the responses
+    """
+    data_dir = Path("./data")
+    if not data_dir.exists():
+        return "No data directory found"
+
+    # Count folders with metadata.json
+    total_posts = 0
+    posts_with_responses = 0
+    total_responses = 0
+    responses_per_post = []  # List to track number of responses for each post
+
+    for metadata_file in data_dir.glob("*/metadata.json"):
+        total_posts += 1
+        try:
+            with open(metadata_file, "r") as f:
+                metadata = json.load(f)
+                num_responses = len(metadata.get("responses", []))
+                responses_per_post.append(num_responses)
+                if num_responses > 0:
+                    posts_with_responses += 1
+                    total_responses += num_responses
+        except:
+            continue
+
+    # Calculate additional statistics
+    if responses_per_post:
+        responses_per_post.sort()
+        median_responses = responses_per_post[len(responses_per_post) // 2]
+        max_responses = max(responses_per_post)
+        avg_responses = (
+            total_responses / posts_with_responses if posts_with_responses > 0 else 0
+        )
+    else:
+        median_responses = max_responses = avg_responses = 0
+
+    stats = f"""
+    📊 Collection Statistics:
+    
+    Overall Progress:
+    - Total Posts Processed: {total_posts}
+    - Posts with Responses: {posts_with_responses}
+    - Total Individual Responses: {total_responses}
+    - Completion Rate: {(posts_with_responses/len(VALID_DATASET_POST_IDS)*100):.2f}%
+    
+    Response Distribution:
+    - Median Responses per Post: {median_responses}
+    - Average Responses per Post: {avg_responses:.2f}
+    - Maximum Responses for a Post: {max_responses}
+    - Posts with No Responses: {total_posts - posts_with_responses}
+    """
+    return stats
+
+
 # Build the Gradio app
 with gr.Blocks() as demo:
     gr.Markdown("# Image Response Collector")
@@ -289,6 +345,22 @@ with gr.Blocks() as demo:
         submit_btn = gr.Button("Submit All Responses")
         clear_btn = gr.Button("Clear Form")
 
+    # Add statistics accordion
+    with gr.Accordion("Collection Statistics", open=False):
+        stats_text = gr.Markdown("Loading statistics...")
+        refresh_stats_btn = gr.Button("Refresh Statistics")
+
+        def update_stats():
+            return get_statistics()
+
+        refresh_stats_btn.click(fn=update_stats, outputs=[stats_text])
+
+        # Move the load event inside the Blocks context
+        demo.load(
+            fn=get_statistics,
+            outputs=[stats_text],
+        )
+
     def submit_responses(
         source_img, post_id, instruction, simplified_instruction, *response_data
     ):