Create app.py

app.py

@@ -0,0 +1,162 @@
+import io
+import pandas as pd
+import gradio as gr
+
+def _load_excel(file_obj):
+    """
+    file_obj is a tempfile-like object from Gradio.
+    Returns: (file_bytes, excel_file, sheet_names)
+    """
+    if file_obj is None:
+        raise ValueError("No file uploaded.")
+
+    # Gradio passes an object with a .name (path)
+    file_path = getattr(file_obj, "name", None)
+    if not file_path:
+        raise ValueError("Invalid uploaded file object.")
+
+    with open(file_path, "rb") as f:
+        file_bytes = f.read()
+
+    if not file_bytes:
+        raise ValueError("Uploaded file is empty.")
+
+    # Read workbook
+    excel = pd.ExcelFile(io.BytesIO(file_bytes), engine="openpyxl")
+    sheets = excel.sheet_names
+    if not sheets:
+        raise ValueError("No sheets found in this workbook.")
+
+    return file_bytes, excel, sheets
+
+
+def on_upload(file_obj):
+    """
+    Triggered when user uploads the Excel file.
+    """
+    try:
+        file_bytes, excel, sheets = _load_excel(file_obj)
+
+        # Auto-load first sheet
+        sheet = sheets[0]
+        df = pd.read_excel(io.BytesIO(file_bytes), sheet_name=sheet, engine="openpyxl")
+
+        if df is None or df.empty:
+            rows, cols, missing = 0, 0, 0
+        else:
+            rows, cols = df.shape
+            missing = int(df.isna().sum().sum())
+
+        return (
+            gr.Dropdown(choices=sheets, value=sheet, interactive=True),
+            df,
+            rows,
+            cols,
+            missing,
+            file_bytes,     # store bytes in State
+            sheet,          # store selected sheet in State
+        )
+
+    except Exception as e:
+        # Reset UI safely
+        return (
+            gr.Dropdown(choices=[], value=None, interactive=False),
+            pd.DataFrame(),
+            0,
+            0,
+            0,
+            None,
+            None,
+        )
+
+
+def on_sheet_change(sheet_name, file_bytes):
+    """
+    When sheet is changed, reload dataframe from bytes.
+    """
+    try:
+        if not file_bytes:
+            raise ValueError("Please upload an Excel file first.")
+        if not sheet_name:
+            raise ValueError("Please select a sheet.")
+
+        df = pd.read_excel(io.BytesIO(file_bytes), sheet_name=sheet_name, engine="openpyxl")
+
+        if df is None or df.empty:
+            rows, cols, missing = 0, 0, 0
+        else:
+            rows, cols = df.shape
+            missing = int(df.isna().sum().sum())
+
+        return df, rows, cols, missing, sheet_name
+
+    except Exception:
+        return pd.DataFrame(), 0, 0, 0, None
+
+
+def make_clean_csv(file_bytes, sheet_name, drop_empty_cols=True, drop_empty_rows=True):
+    """
+    Create a cleaned CSV (bytes) for download.
+    """
+    if not file_bytes:
+        raise ValueError("Please upload an Excel file first.")
+    if not sheet_name:
+        raise ValueError("Please select a sheet first.")
+
+    df = pd.read_excel(io.BytesIO(file_bytes), sheet_name=sheet_name, engine="openpyxl")
+
+    if drop_empty_cols:
+        df = df.dropna(axis=1, how="all")
+    if drop_empty_rows:
+        df = df.dropna(axis=0, how="all")
+
+    csv_bytes = df.to_csv(index=False).encode("utf-8")
+    filename = f"cleaned__{sheet_name}.csv"
+    return (filename, csv_bytes)
+
+
+with gr.Blocks(title="CGA Excel Analyzer") as demo:
+    gr.Markdown("## 📊 CGA Excel Analyzer (Gradio)")
+    gr.Markdown("Upload an Excel file (.xlsx). Select a sheet, preview it, see basic stats, and download cleaned CSV.")
+
+    file_state = gr.State(None)   # stores raw Excel bytes
+    sheet_state = gr.State(None)  # stores selected sheet name
+
+    with gr.Row():
+        upload = gr.File(label="Upload Excel (.xlsx)", file_types=[".xlsx"])
+        sheet_dd = gr.Dropdown(label="Sheet", choices=[], interactive=False)
+
+    with gr.Row():
+        rows_out = gr.Number(label="Rows", value=0, precision=0)
+        cols_out = gr.Number(label="Columns", value=0, precision=0)
+        missing_out = gr.Number(label="Missing cells", value=0, precision=0)
+
+    df_view = gr.Dataframe(label="Preview", interactive=False, wrap=True)
+
+    with gr.Accordion("🧹 Cleaning options (for CSV export)", open=False):
+        drop_cols = gr.Checkbox(label="Drop fully empty columns", value=True)
+        drop_rows = gr.Checkbox(label="Drop fully empty rows", value=True)
+
+    download_btn = gr.Button("⬇️ Download cleaned CSV")
+    download_file = gr.File(label="Your cleaned CSV")
+
+    # Events
+    upload.change(
+        fn=on_upload,
+        inputs=[upload],
+        outputs=[sheet_dd, df_view, rows_out, cols_out, missing_out, file_state, sheet_state],
+    )
+
+    sheet_dd.change(
+        fn=on_sheet_change,
+        inputs=[sheet_dd, file_state],
+        outputs=[df_view, rows_out, cols_out, missing_out, sheet_state],
+    )
+
+    download_btn.click(
+        fn=make_clean_csv,
+        inputs=[file_state, sheet_state, drop_cols, drop_rows],
+        outputs=[download_file],
+    )
+
+demo.launch()