Add application file

app.py

@@ -1,17 +1,130 @@
-from data_handler import load_dataset, load_progress
-from ui_handler import create_ui
-
-# Load dataset
-dataset = load_dataset()
-if dataset is None:
-    raise RuntimeError("Failed to load dataset. Check your Hugging Face token.")
-
-# Load annotation progress
-current_index = load_progress()
-
-# Create UI and launch
-demo = create_ui(dataset, current_index)
-demo.launch()
+import gradio as gr
+from datasets import load_dataset
+from datasets import Dataset
+import os
+import csv
+
+# Configuration
+DATASET_NAME = "anna-tch/generation-results"  # Update this
+PROGRESS_FILE = "progress.csv"
+HF_TOKEN = os.getenv("HF_TOKEN")  # Set your Hugging Face token
+
+# Load dataset and prepare initial state
+def load_data():
+    dataset = load_dataset(DATASET_NAME, use_auth_token=HF_TOKEN)['train']
+    unannotated = [i for i, ex in enumerate(dataset) if ex['annotation'] is None]
+    return {
+        "dataset": dataset,
+        "unannotated_indices": unannotated,
+        "current_index": 0,
+        "annotations": {}
+    }
+
+def save_progress(annotation_data):
+    with open(PROGRESS_FILE, 'a', newline='') as f:
+        writer = csv.writer(f)
+        writer.writerow([
+            annotation_data['id'],
+            annotation_data['grammar'],
+            annotation_data['coherence']
+        ])
+
+def update_dataset(state):
+    state["dataset"].push_to_hub(DATASET_NAME, token=HF_TOKEN)
+
+def get_current_example(state):
+    idx = state["unannotated_indices"][state["current_index"]]
+    example = state["dataset"][idx]
+    return example, idx
+
+def update_display(state):
+    example, idx = get_current_example(state)
+    annotation = state["annotations"].get(idx, {"grammar": None, "coherence": None})
+    return (
+        example["generated_text"],
+        annotation["grammar"],
+        annotation["coherence"],
+        f"Example {state['current_index'] + 1} of {len(state['unannotated_indices'])}"
+    )
+
+def next_example(state):
+    if state["current_index"] < len(state["unannotated_indices"]) - 1:
+        state["current_index"] += 1
+    return update_display(state), state
+
+def prev_example(state):
+    if state["current_index"] > 0:
+        state["current_index"] -= 1
+    return update_display(state), state
+
+def submit(grammar, coherence, state):
+    example, idx = get_current_example(state)
+    
+    # Save annotation
+    annotation = {
+        "id": idx,
+        "grammar": grammar,
+        "coherence": coherence
+    }
+    state["annotations"][idx] = annotation
+    save_progress(annotation)
+    
+    # Update dataset
+    state["dataset"] = state["dataset"].map(
+        lambda ex, idx: {"annotation": {"grammar": grammar, "coherence": coherence} 
+                        if idx == annotation["id"] else ex},
+        with_indices=True
+    )
+    update_dataset(state)
+    
+    # Move to next example
+    return next_example(state)
+
+with gr.Blocks() as app:
+    state = gr.State(load_data)
+    
+    gr.Markdown("## Text Annotation Tool")
+    
+    with gr.Row():
+        counter = gr.Markdown()
+    text_display = gr.Textbox(label="Generated Text", interactive=False)
+    with gr.Row():
+        grammar = gr.Radio(choices=[1, 2, 3, 4, 5], label="Grammar Score")
+        coherence = gr.Radio(choices=[1, 2, 3, 4, 5], label="Coherence Score")
+    
+    with gr.Row():
+        prev_btn = gr.Button("Previous")
+        next_btn = gr.Button("Next")
+        submit_btn = gr.Button("Submit")
+
+    # Event handlers
+    prev_btn.click(
+        prev_example,
+        inputs=[state],
+        outputs=[text_display, grammar, coherence, counter, state]
+    )
+    
+    next_btn.click(
+        next_example,
+        inputs=[state],
+        outputs=[text_display, grammar, coherence, counter, state]
+    )
+    
+    submit_btn.click(
+        submit,
+        inputs=[grammar, coherence, state],
+        outputs=[text_display, grammar, coherence, counter, state]
+    )
+    
+    # Initial load
+    app.load(
+        update_display,
+        inputs=[state],
+        outputs=[text_display, grammar, coherence, counter]
+    )
+
+if __name__ == "__main__":
+    app.launch()
 
 # import gradio as gr
 # import datasets