search_schools.py

import json
import os
import argparse

# Define path relative to the script location
script_dir = os.path.dirname(os.path.abspath(__file__))
data_file_path = os.path.join(script_dir, 'data', 'school_data_final.json')

def load_data(file_path):
    """Loads school data from the JSON file."""
    try:
        with open(file_path, 'r', encoding='utf-8') as f:
            data = json.load(f)
            if 'schools' in data and isinstance(data['schools'], list):
                return data['schools']
            else:
                print(f"Error: 'schools' key not found or is not a list in {file_path}")
                return []
    except FileNotFoundError:
        print(f"Error: Data file not found at {file_path}")
        return []
    except json.JSONDecodeError:
        print(f"Error: Could not decode JSON from {file_path}")
        return []
    except Exception as e:
        print(f"An unexpected error occurred while loading data: {e}")
        return []

def search_schools(schools, name=None, min_rank=None, max_rank=None, suburb=None, school_type=None):
    """Searches and filters the list of schools based on provided criteria."""
    results = schools

    if name:
        results = [s for s in results if name.lower() in s.get('name', '').lower()]

    # Filter by rank range
    if min_rank is not None:
        try:
            min_r = int(min_rank)
            results = [s for s in results if s.get('rank') is not None and s.get('rank') >= min_r]
        except ValueError:
            print("Warning: Invalid minimum rank provided. Rank must be an integer.")
    if max_rank is not None:
        try:
            max_r = int(max_rank)
            results = [s for s in results if s.get('rank') is not None and s.get('rank') <= max_r]
        except ValueError:
            print("Warning: Invalid maximum rank provided. Rank must be an integer.")

    if suburb:
        results = [s for s in results if suburb.lower() in s.get('suburb_name', '').lower()]

    if school_type:
        results = [s for s in results if school_type.lower() == s.get('type', '').lower()]

    return results

def display_results(results):
    """Prints the search results in a readable format."""
    if not results:
        print("No schools found matching your criteria.")
        return

    print(f"\nFound {len(results)} school(s):")
    print("-" * 30)
    for school in results:
        print(f"  Name: {school.get('name', 'N/A')}")
        print(f"  Rank: {school.get('rank', 'N/A')}")
        print(f"  Type: {school.get('type', 'N/A')}")
        print(f"  Suburb: {school.get('suburb_name', 'N/A')} (Postcode: {school.get('suburb', 'N/A')})")
        print(f"  Score 2023: {school.get('score_2023', 'N/A')}")
        print("-" * 30)

def main():
    parser = argparse.ArgumentParser(description='Search and filter schools in the dataset.')
    parser.add_argument('-n', '--name', help='Filter by school name (partial match)')
    parser.add_argument('--min-rank', help='Filter by minimum rank (inclusive)')
    parser.add_argument('--max-rank', help='Filter by maximum rank (inclusive)')
    parser.add_argument('-s', '--suburb', help='Filter by suburb name (partial match)')
    parser.add_argument('-t', '--type', help='Filter by school type (exact match, e.g., Public, Catholic, Independent/Private)')

    args = parser.parse_args()

    # Check if at least one filter is provided
    if not any([args.name, args.min_rank, args.max_rank, args.suburb, args.type]):
        parser.print_help()
        print("\nPlease provide at least one filter criterion.")
        return

    all_schools = load_data(data_file_path)
    if not all_schools:
        return # Error message already printed by load_data

    found_schools = search_schools(all_schools, name=args.name, min_rank=args.min_rank, max_rank=args.max_rank, suburb=args.suburb, school_type=args.type)
    display_results(found_schools)

if __name__ == "__main__":
    main()