agent.py

"""
Automation Agent - Main Entry Point
-----------------------------------
Orchestrates the workflow: Spreadsheet -> Agent -> Web Search -> Result

This agent reads rows from an Excel spreadsheet, processes each row,
performs a web search for the name/company, and prints the results.
"""

import sys
from pathlib import Path

from spreadsheet_reader import get_rows_as_dicts, build_search_query
from web_search import search, format_search_result


def run_agent(spreadsheet_path: str = "data.xlsx") -> None:
    """
    Run the automation agent workflow.

    Workflow:
    1. Read rows from the spreadsheet
    2. For each row, build a search query (name + company)
    3. Perform web search for each query
    4. Print results to console

    Args:
        spreadsheet_path: Path to the Excel spreadsheet
    """
    print("=" * 60)
    print("  PYTHON AUTOMATION AGENT - Data Collection Workflow")
    print("=" * 60)
    print()

    # Step 1: Read spreadsheet
    print("[1] Reading spreadsheet:", spreadsheet_path)
    try:
        rows = get_rows_as_dicts(spreadsheet_path)
    except FileNotFoundError as e:
        print(f"  ERROR: {e}")
        print("  Please ensure data.xlsx exists in the project directory.")
        sys.exit(1)

    print(f"  Found {len(rows)} rows to process.")
    print()

    # Step 2 & 3 & 4: Process each row
    print("[2] Processing rows (spreadsheet -> agent -> web search -> result)")
    print("-" * 60)

    for i, row in enumerate(rows, start=1):
        row_id = row.get("id", i)
        name = row.get("name", "")
        company = row.get("company", "")

        # Build search query from name and company
        query = build_search_query(row)

        print(f"\nRow {i} (id={row_id}): {name} @ {company}")
        print(f"  Search query: \"{query}\"")

        if not query:
            print("  [SKIP] No search query (missing name and company)")
            continue

        # Perform web search
        result = search(query)

        # Print result
        print(format_search_result(result))

    print()
    print("-" * 60)
    print("[3] Workflow complete. All rows processed.")
    print("=" * 60)


if __name__ == "__main__":
    # Default to data.xlsx in the script's directory
    script_dir = Path(__file__).parent
    data_file = script_dir / "data.xlsx"

    run_agent(str(data_file))