cpc-direct-stations/CLAUDE.md

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

Overview

This skill contains a single standalone script for scraping Taiwan CPC (中油) official direct-operated gas station information. The script fetches data from https://vipmbr.cpc.com.tw/mbwebs/service_search.aspx using form POST requests and parses the HTML response.

Architecture

The script uses only Python stdlib (no external dependencies):

• HTMLParser - Custom StationTableParser class extracts data from the #MyGridView1 table
• urllib - Handles HTTP requests with proper User-Agent and form data encoding
• __VIEWSTATE/__EVENTVALIDATION - ASP.NET hidden fields are extracted and submitted in POST requests

The query flow:

1. GET initial page to extract ASP.NET hidden fields
2. POST with TypeGroup=rbGroup2 to select "direct-operated stations" (直營站)
3. Extract hidden fields again
4. POST with btnQuery=查詢 to submit the query
5. Parse HTML response table

Common Commands

Get all direct-operated stations (default text output)

python3 skills/c/cpc_direct_stations.py

Query by city

python3 skills/c/cpc_direct_stations.py --city 台北市

Filter by keyword (station name or address)

python3 skills/c/cpc_direct_stations.py --city 新北市 --keyword 中山

Export as JSON

python3 skills/c/cpc_direct_stations.py --city 台北市 --format json

Export as CSV (with BOM for Excel compatibility)

python3 skills/c/cpc_direct_stations.py --format csv --out cpc_stations.csv

City Codes

The CITY_MAP dict maps Chinese city names to CPC internal codes (e.g., "台北市": "A "). Supported cities include all Taiwan counties and cities.

Output Schema

Each station record contains:

• 縣市 - County/City
• 鄉鎮區 - District
• 類別 - Category (e.g., "直營站")
• 站名 - Station name
• 站代號 - Station code
• 地址 - Address
• 電話 - Phone number
• 營業時間 - Business hours