Final commit after Git identity setup

agent_app.py

@@ -0,0 +1,105 @@
+# agent_app.py
+from smolagents import CodeAgent, TransformersModel
+from smolagents import tool
+from sheet_tool import (
+    fetch_sheet_as_df,
+    create_pivot,
+    summary_stats,
+    plot_dataframe,
+    df_to_csv_bytes,
+)
+import base64
+
+# Initialize model
+model = TransformersModel(model_id="HuggingFaceTB/SmolLM-135M-Instruct")
+
+# -------------------------------
+# ✅ TOOL DEFINITIONS
+# -------------------------------
+
+@tool
+def load_sheet() -> dict:
+    """Load Google Sheet into a dataframe and return a short summary (not the full sheet)."""
+    df = fetch_sheet_as_df()
+    if df.empty:
+        return {"error": "Sheet is empty or not found."}
+    return {
+        "rows": len(df),
+        "columns": list(df.columns),
+        "head": df.head(5).to_dict(orient="records"),
+    }
+
+
+@tool
+def pivot(index_cols: str, column_cols: str, value_cols: str, aggfunc: str = "sum") -> dict:
+    """
+    Create a pivot table from the Google Sheet.
+
+    Args:
+        index_cols (str): Comma-separated list of columns to use as the pivot table index.
+        column_cols (str): Comma-separated list of columns to use as pivot table columns.
+        value_cols (str): Comma-separated list of columns to aggregate.
+        aggfunc (str, optional): Aggregation function to apply (e.g., 'sum', 'mean', 'count'). Defaults to 'sum'.
+    """
+    df = fetch_sheet_as_df()
+    if df.empty:
+        return {"error": "Sheet empty"}
+    index = [c.strip() for c in index_cols.split(",")] if index_cols else []
+    columns = [c.strip() for c in column_cols.split(",")] if column_cols else []
+    values = [c.strip() for c in value_cols.split(",")] if value_cols else []
+    pivot_df = create_pivot(df, index=index, columns=columns, values=values, aggfunc=aggfunc)
+    csv_bytes = df_to_csv_bytes(pivot_df)
+    return {
+        "pivot_preview": pivot_df.head(10).to_dict(orient="records"),
+        "csv_b64": base64.b64encode(csv_bytes).decode("utf-8")
+    }
+
+
+
+@tool
+def stats() -> dict:
+    """Generate summary statistics of the sheet."""
+    df = fetch_sheet_as_df()
+    if df.empty:
+        return {"error": "Sheet empty"}
+    s = summary_stats(df)
+    return {"summary": s.to_dict()}
+
+
+@tool
+def plot(kind: str = "bar", x: str = None, y: str = None, title: str = None) -> dict:
+    """
+    Create a plot from the Google Sheet data.
+
+    Args:
+        kind (str): Type of chart to create. Example values: 'bar', 'line', 'pie', 'scatter'.
+        x (str, optional): Column name to use for the X-axis. Example: 'Date'.
+        y (str, optional): Comma-separated column names to use for Y-axis. Example: 'Sales,Profit'.
+        title (str, optional): Chart title to display at the top.
+
+    Returns:
+        dict: A dictionary containing the base64-encoded plot image or an error message if the sheet is empty.
+    """
+    df = fetch_sheet_as_df()
+    if df.empty:
+        return {"error": "Sheet empty"}
+
+    y_list = [c.strip() for c in y.split(",")] if y else None
+    img_data_uri = plot_dataframe(df, kind=kind, x=x, y=y_list, title=title)
+    return {"image": img_data_uri}
+
+
+# -------------------------------
+# ✅ AGENT CREATION
+# -------------------------------
+agent = CodeAgent(model=model, tools=[load_sheet, pivot, stats, plot], add_base_tools=True)
+
+
+def ask_agent(nl_query: str) -> dict:
+    """Send a natural-language query to the agent and return structured response."""
+    try:
+        resp = agent.run(nl_query)
+        return {"text": str(resp)}
+    except Exception as e:
+        return {"error": str(e)}
+

requirements.txt

@@ -0,0 +1,8 @@
+gradio
+pandas
+google-auth
+google-auth-oauthlib
+google-auth-httplib2
+gspread
+smolagents
+requests

sheet_tool.py

@@ -0,0 +1,87 @@
+# sheet_tool.py
+from google.oauth2.service_account import Credentials
+import gspread
+import pandas as pd
+import matplotlib.pyplot as plt
+import io
+import base64
+from typing import Tuple
+
+# CONFIG - change these
+CREDENTIALS_FILE = "credentials.json"   # path to your service account JSON
+SHEET_ID = "1nOekWGmPsjoHj9T-MFjlNGSFyKPRogVbEjFNRxGgxuM"  # replace with your sheet id
+WORKSHEET_INDEX = 0  # first sheet
+
+SCOPES = [
+    "https://www.googleapis.com/auth/spreadsheets",
+    "https://www.googleapis.com/auth/drive",
+]
+
+def authorize_gs():
+    creds = Credentials.from_service_account_file(CREDENTIALS_FILE, scopes=SCOPES)
+    client = gspread.authorize(creds)
+    return client
+
+def fetch_sheet_as_df(sheet_id: str = SHEET_ID, worksheet_index: int = WORKSHEET_INDEX) -> pd.DataFrame:
+    client = authorize_gs()
+    sh = client.open_by_key(sheet_id)
+    worksheet = sh.get_worksheet(worksheet_index)
+    data = worksheet.get_all_values()
+    if not data:
+        return pd.DataFrame()
+    df = pd.DataFrame(data[1:], columns=data[0])
+    # try to convert numeric columns where possible
+    for col in df.columns:
+        df[col] = pd.to_numeric(df[col], errors="ignore")
+    return df
+
+def create_pivot(df: pd.DataFrame, index: list, columns: list, values: list, aggfunc: str = "sum") -> pd.DataFrame:
+    if df.empty:
+        return pd.DataFrame()
+    # pandas supports aggfunc as string or function
+    agg = getattr(pd.core.groupby.SeriesGroupBy, aggfunc, None)
+    pivot = pd.pivot_table(df, index=index, columns=columns, values=values, aggfunc=aggfunc, fill_value=0)
+    # convert MultiIndex columns to string for easier display
+    pivot = pivot.reset_index()
+    pivot.columns = [(" " .join(map(str, c)) if isinstance(c, tuple) else c).strip() for c in pivot.columns]
+    return pivot
+
+def summary_stats(df: pd.DataFrame, numeric_only: bool = True) -> pd.DataFrame:
+    if df.empty:
+        return pd.DataFrame()
+    return df.describe(include="all") if not numeric_only else df.describe()
+
+def plot_dataframe(df: pd.DataFrame, kind: str = "bar", x: str = None, y: list = None, title: str = None, figsize=(8,5)) -> str:
+    """
+    Creates a matplotlib plot and returns a base64 PNG data URI.
+    """
+    if df.empty:
+        raise ValueError("DataFrame is empty")
+
+    plt.close('all')
+    fig, ax = plt.subplots(figsize=figsize)
+
+    if kind == "bar":
+        if x is None or y is None:
+            df.plot(kind="bar", ax=ax)
+        else:
+            df.plot(kind="bar", x=x, y=y, ax=ax)
+    elif kind == "line":
+        df.plot(kind="line", x=x, y=y, ax=ax)
+    elif kind == "pie":
+        df.set_index(x)[y].plot(kind="pie", ax=ax, autopct='%1.1f%%')
+    else:
+        df.plot(kind=kind, x=x, y=y, ax=ax)
+
+    if title:
+        ax.set_title(title)
+    ax.grid(True)
+    buf = io.BytesIO()
+    fig.tight_layout()
+    fig.savefig(buf, format="png")
+    buf.seek(0)
+    b64 = base64.b64encode(buf.read()).decode("utf-8")
+    return "data:image/png;base64," + b64
+
+def df_to_csv_bytes(df: pd.DataFrame) -> bytes:
+    return df.to_csv(index=False).encode("utf-8")

web_app.py

@@ -0,0 +1,15 @@
+# web_app.py
+from flask import Flask, request, jsonify
+from agent_app import ask_agent
+
+app = Flask(__name__)
+
+@app.route("/ask", methods=["POST"])
+def ask():
+    data = request.json
+    prompt = data.get("prompt")
+    result = ask_agent(prompt)
+    return jsonify({"result": result})
+
+if __name__ == "__main__":
+    app.run(port=8000, debug=True)