Autoparts-DB/vehicle_scraper/manual_input_simple.py

"""
Manual Data Extraction Guide for RockAuto.com

Since RockAuto has strong anti-bot measures, here's a manual approach to extract vehicle data:

1. Visit https://www.rockauto.com/
2. Click on "Catalog" in the navigation menu
3. You'll see a list of vehicle manufacturers (makes)
4. For each make, manually note down the models, years, and engines

This script provides a framework to input the manually collected data into your database.
"""

import sqlite3
from typing import List, Dict


class ManualDataInput:
    def __init__(self, db_path: str = "../vehicle_database/vehicle_database.db"):
        self.db_path = db_path
    
    def add_vehicle_data(self, make: str, model: str, year: int, engine: str = "Unknown"):
        """Add a single vehicle entry to the database"""
        conn = sqlite3.connect(self.db_path)
        cursor = conn.cursor()
        
        try:
            # Insert brand
            cursor.execute(
                "INSERT OR IGNORE INTO brands (name) VALUES (?)",
                (make,)
            )
            cursor.execute("SELECT id FROM brands WHERE name = ?", (make,))
            brand_id = cursor.fetchone()[0]
            
            # Insert year
            cursor.execute(
                "INSERT OR IGNORE INTO years (year) VALUES (?)",
                (year,)
            )
            cursor.execute("SELECT id FROM years WHERE year = ?", (year,))
            year_id = cursor.fetchone()[0]
            
            # Insert engine
            cursor.execute(
                "INSERT OR IGNORE INTO engines (name) VALUES (?)",
                (engine,)
            )
            cursor.execute("SELECT id FROM engines WHERE name = ?", (engine,))
            engine_id = cursor.fetchone()[0]
            
            # Insert model
            cursor.execute(
                "INSERT OR IGNORE INTO models (brand_id, name) VALUES (?, ?)",
                (brand_id, model)
            )
            cursor.execute("SELECT id FROM models WHERE brand_id = ? AND name = ?", (brand_id, model))
            model_id = cursor.fetchone()[0]
            
            # Link model, year, and engine
            cursor.execute(
                """INSERT OR IGNORE INTO model_year_engine 
                   (model_id, year_id, engine_id) VALUES (?, ?, ?)""",
                (model_id, year_id, engine_id)
            )
            
            conn.commit()
            print(f"Added: {year} {make} {model} with {engine}")
            
        except Exception as e:
            print(f"Error adding vehicle: {e}")
        finally:
            conn.close()
    
    def add_multiple_vehicles(self, vehicles: List[Dict]):
        """Add multiple vehicles at once"""
        for vehicle in vehicles:
            self.add_vehicle_data(
                make=vehicle.get('make', ''),
                model=vehicle.get('model', ''),
                year=vehicle.get('year', 0),
                engine=vehicle.get('engine', 'Unknown')
            )
    
    def show_extraction_guide(self):
        """Show the manual extraction guide"""
        guide = """
        ================================================
        Manual RockAuto Data Extraction Guide
        ================================================
        
        1. OPEN YOUR WEB BROWSER and go to: https://www.rockauto.com
        
        2. CLICK on the "Catalog" link in the navigation menu
        
        3. YOU WILL SEE a list of vehicle manufacturers (makes) like:
           - Acura
           - Audi
           - BMW
           - Chevrolet
           - Ford
           - Honda
           - Toyota
           - And many more...
        
        4. FOR EACH MANUFACTURER:
           a) Click on the manufacturer name
           b) You'll see a page with vehicle models organized by year
           c) Note down the models and years you see
           d) Example format: 2020 Honda Civic, 2019 Ford F-150, etc.
        
        5. TO FIND ENGINE INFORMATION:
           a) Click on a specific model/year combination
           b) You'll see parts categories for that vehicle
           c) Look for "Engine" or "Engine Mechanical" category
           d) Note down the engine type/specifications
        
        6. USE THE FOLLOWING COMMANDS to add data to your database:
        
        Example Python commands:
        >>> from manual_input import ManualDataInput
        >>> input_tool = ManualDataInput()
        >>> input_tool.add_vehicle_data("Toyota", "Camry", 2020, "2.5L 4-Cylinder")
        >>> input_tool.add_vehicle_data("Honda", "Civic", 2019, "1.5L Turbo")
        
        Or add multiple at once:
        >>> vehicles = [
        ...     {"make": "Ford", "model": "F-150", "year": 2021, "engine": "3.5L V6"},
        ...     {"make": "BMW", "model": "X3", "year": 2020, "engine": "2.0L 4-Cylinder Turbo"}
        ... ]
        >>> input_tool.add_multiple_vehicles(vehicles)
        
        7. TIPS FOR EFFICIENT DATA COLLECTION:
           - Focus on popular makes/models first
           - Record data in a spreadsheet as you go
           - Take screenshots of pages for reference
           - Be systematic - go alphabetically or by make popularity
        
        ================================================
        """
        print(guide)


def main():
    print("Manual RockAuto Data Extraction Tool")
    print("=====================================")
    
    input_tool = ManualDataInput()
    
    # Show the extraction guide
    input_tool.show_extraction_guide()
    
    # Add sample vehicles to database
    print("\nAdding sample vehicles to database:")
    sample_vehicles = [
        {"make": "Toyota", "model": "Camry", "year": 2020, "engine": "2.5L 4-Cylinder"},
        {"make": "Honda", "model": "Civic", "year": 2019, "engine": "1.5L Turbo"},
        {"make": "Ford", "model": "F-150", "year": 2021, "engine": "3.5L V6"},
        {"make": "BMW", "model": "X3", "year": 2020, "engine": "2.0L 4-Cylinder Turbo"},
        {"make": "Chevrolet", "model": "Silverado", "year": 2022, "engine": "5.3L V8"}
    ]
    
    input_tool.add_multiple_vehicles(sample_vehicles)
    print("\nSample vehicles added to database!")
    
    print("\nYou can now use the ManualDataInput class to add more vehicles manually.")
    print("Import it in Python with: from manual_input import ManualDataInput")


if __name__ == "__main__":
    main()