Update app.py

app.py

@@ -1,48 +1,128 @@
 import gradio as gr
 import pandas as pd
 import os
-from translator_utils import process_csv, download_dataset, LOCAL_FILENAME
+from translator_utils import process_rows, download_dataset, LOCAL_FILENAME
 
 def load_data():
-    """Load the dataset for display."""
+    """Load the dataset safely."""
     if not os.path.exists(LOCAL_FILENAME):
         try:
             download_dataset()
         except:
-            return pd.DataFrame({"Message": ["Dataset not found and could not be downloaded."]})
-    
+            return pd.DataFrame()
     try:
         df = pd.read_csv(LOCAL_FILENAME)
+        # Ensure translation columns exist
+        if 'Judul_Inggris' not in df.columns:
+            df['Judul_Inggris'] = ""
+        if 'Isi_Inggris' not in df.columns:
+            df['Isi_Inggris'] = ""
+        df = df.fillna("")
         return df
-    except Exception as e:
-        return pd.DataFrame({"Error": [str(e)]})
+    except:
+        return pd.DataFrame()
 
-def translate_wrapper(token, progress=gr.Progress()):
-    """Wrapper for translation with Gradio progress."""
-    try:
-        df, upload_msg = process_csv(token=token, progress=progress)
-        return df, f"Translation Complete! {upload_msg}"
-    except Exception as e:
-        return load_data(), f"Error: {str(e)}"
+def get_pending_data(full_df):
+    """Filter for rows that are NOT fully translated."""
+    if full_df.empty:
+        return full_df
+    # A row is pending if either Judul_Inggris or Isi_Inggris is empty
+    mask = (full_df['Judul_Inggris'].str.strip() == "") | (full_df['Isi_Inggris'].str.strip() == "")
+    # Add a 'Select' column for checkbox, initialized to False
+    df_view = full_df[mask].copy()
+    df_view.insert(0, "Select", False)
+    return df_view
 
-with gr.Blocks() as app:
-    gr.Markdown("# Auto AI News Translator")
-    gr.Markdown("This app automatically translates 'Judul' and 'Isi' columns of the news dataset from Indonesian to English.")
-    gr.Markdown("**Persistence**: Enter your Hugging Face Token (with write access) to save changes back to the dataset repo.")
+def get_history_data(full_df, date_filter=""):
+    """Filter for rows that ARE fully translated."""
+    if full_df.empty:
+        return full_df
+    mask = (full_df['Judul_Inggris'].str.strip() != "") & (full_df['Isi_Inggris'].str.strip() != "")
+    df_view = full_df[mask].copy()
+    
+    if date_filter:
+        df_view = df_view[df_view['Tanggal'].str.contains(date_filter, case=False, na=False)]
+        
+    return df_view
+
+def on_translate_selected(df_data, token, progress=gr.Progress()):
+    """
+    Callback when user clicks 'Translate Selected'.
+    df_data is the dataframe from the component (with checkboxes).
+    """
+    # Filter rows where 'Select' is True
+    # Gradio Dataframe sends back a pandas DataFrame
+    selected_rows = df_data[df_data["Select"] == True]
     
-    with gr.Row():
-        hf_token = gr.Textbox(label="HF Token (Optional)", placeholder="hf_...", type="password")
-        translate_btn = gr.Button("Translate & Save", variant="primary")
+    if selected_rows.empty:
+        return gr.DataFrame(), gr.DataFrame(), "No rows selected!"
     
-    status_output = gr.Textbox(label="Status", value="Ready", interactive=False)
+    # We need the ORIGINAL indices. 
+    # Since we filtered the view, the index of the dataframe should still be the original index 
+    # IF we didn't reset_index. 
+    # Let's check. Gradio might reset it or treat it as data.
+    # To be safe, we rely on the fact that we passed the dataframe with original index?
+    # Actually Gradio usually converts index to 0..n.
+    # We should have preserved the original index as a column or rely on matching content?
+    # Better approach: We passed the DF. Gradio returns the modified DF.
+    # If the index is preserved, we are good.
+    # Let's try to trust the index.
+    
+    indices = selected_rows.index.tolist()
+    
+    # Call backend
+    updated_df, msg = process_rows(indices, token=token, progress=progress)
     
-    data_display = gr.DataFrame(value=load_data(), label="Dataset Preview", interactive=False)
+    # Refresh views
+    pending = get_pending_data(updated_df)
+    history = get_history_data(updated_df)
     
+    return pending, history, msg
+
+def filter_history(date_input):
+    df = load_data()
+    return get_history_data(df, date_input)
+
+with gr.Blocks(title="Auto AI News Translator") as app:
+    gr.Markdown("# Auto AI News Translator")
+    gr.Markdown("Translate news from Indonesian to English (with Chunking). Select rows to translate.")
+    
+    # Store full data state if needed, but we read from disk to be single-source-of-truth safe
+    # token input
+    hf_token = gr.Textbox(label="HF Token (Optional for Persistence)", placeholder="hf_...", type="password")
+    status_output = gr.Textbox(label="Status", value="Ready", interactive=False)
+    
+    with gr.Tabs():
+        with gr.TabItem("Translate Pending"):
+            gr.Markdown("Select rows to translate and click Submit.")
+            # We use wrap=True for long text
+            # We use interactive=True to allow checking the 'Select' box
+            pending_df_comp = gr.DataFrame(
+                value=get_pending_data(load_data()),
+                label="Pending Translations",
+                interactive=True,
+                wrap=True,
+                datatype=["bool"] + ["str"] * 6 # approximate types
+            )
+            translate_btn = gr.Button("Translate Selected Rows", variant="primary")
+            
+        with gr.TabItem("History"):
+            date_filter = gr.Textbox(label="Filter by Date (YYYY-MM-DD)")
+            history_df_comp = gr.DataFrame(
+                value=get_history_data(load_data()),
+                label="Translation History",
+                interactive=False,
+                wrap=True
+            )
+            date_filter.change(fn=filter_history, inputs=[date_filter], outputs=[history_df_comp])
+
     translate_btn.click(
-        fn=translate_wrapper,
-        inputs=[hf_token],
-        outputs=[data_display, status_output]
+        fn=on_translate_selected,
+        inputs=[pending_df_comp, hf_token],
+        outputs=[pending_df_comp, history_df_comp, status_output]
     )
+    
+    # Auto refresh on load? (Handled by initial value)
 
 if __name__ == "__main__":
     app.launch()