Skip to content

Latest commit

 

History

History
126 lines (109 loc) · 4.22 KB

skills.md

File metadata and controls

126 lines (109 loc) · 4.22 KB

drpy-node MCP Skills & Prompts

Skill 1: Develop DS Source (Create/Debug)

Description: Create, debug, and validate drpy JS spiders. Handles new sources, fixes, and advanced logic (encryption/lazy).

Prompt Template:

Task: [Create/Debug/Analyze] DS Source for [URL/File]
Steps:
1. **Analyze/Fetch**: Use `fetch_spider_url` to inspect HTML/Headers or `read_file` to load existing code (auto-decrypted).
2. **Develop/Refine**:
    - Use `get_spider_template` for new files.
    - Implement `rule` object (see KB below).
    - For dynamic content, use `async` parsing or `lazy` loading.
    - For novels, `lazy` returns `novel://` + JSON.
3. **Validate**:
    - Save via `write_file` to `spider/js/[Name].js`.
    - Check syntax (`check_syntax`) and structure (`validate_spider`).
    - Test rules (`debug_spider_rule`) against real content.

Skill 2: System Maintenance

Description: Monitor health, logs, DB, and config.

Prompt Template:

Task: System Health & Config Check
Steps:
1. **Diagnose**: `read_logs` for errors, `get_routes_info` for APIs, `sql_query` for DB stats.
2. **Configure**: `manage_config` (get/set) to adjust settings (e.g., timeout).
3. **Apply**: `restart_service` if config changed.

Knowledge Base (KB)

1. Source Structure (rule Object)

var rule = {
    title: 'Site', host: 'https://site.com', url: '/cat/fyclass/p/fypage',
    searchUrl: '/s?k=**&p=fypage', searchable: 2, quickSearch: 0,
    headers: { 'User-Agent': 'MOBILE_UA' },
    class_name: 'Mov&TV', class_url: 'mov&tv', // or async class_parse
    play_parse: true,
    lazy: async function() { return { parse: 1, url: '...' } }, // or return 'url'
    一级: '.list li;a&&title;img&&src;.desc&&Text;a&&href', // List: selector;attr...
    二级: '*', // Detail: '*' or { title: '...', ... }
    搜索: '*', // Search: '*' or async function
}

2. Selectors (Cheerio-based)

Format: selector;attr or selector;attr1;attr2

Func Returns Description
pdfa Array Parse List.
pdfh String Parse Node.
pd String Parse URL (auto-resolve).

Special Attrs: Text, Html, href, src, style, data-*. Syntax: && (nested), || (backup), :eq(n) (index), * (all).

3. Advanced Patterns

Async Parsing (一级/二级/搜索):

一级: async function() {
    let { input, pdfa, pdfh, pd } = this;
    return pdfa(input, '.list li').map(it => ({
        title: pdfh(it, 'a&&title'), url: pd(it, 'a&&href')
    }));
}

Batch Requests (batchFetch):

let res = await batchFetch([{url:'...', options:{}}, ...]);

Pre-processing (预处理):

预处理: async function() { rule.classes = (await request('...')).classes; }

Novel Content:

lazy: async function() { return { parse: 0, url: 'novel://' + JSON.stringify({title:'', content:''}) }; }

4. Global Helpers

Helper Description
request(url, opt) / post Async HTTP GET/POST.
log(msg) Console log (view in read_logs).
setItem(k, v) / getItem(k) Persistent storage.
urljoin(base, path) URL resolution.

5. Content Type Specifics

  • Novels (类型: '小说'):

    • lazy should return novel:// followed by a JSON string of { title, content }.
    lazy: async function() {
    let content = '...'; // Decrypted text
    let ret = JSON.stringify({ title: 'Chapter 1', content: content });
    return { parse: 0, url: 'novel://' + ret };
}

  • Comics/Images (类型: '漫画'):

    • lazy should return pics:// followed by image URLs joined by &&.
    lazy: async function() {
    let pics = ['http://img1.jpg', 'http://img2.jpg'];
    return { parse: 0, url: 'pics://' + pics.join('&&') };
}

Tool Reference

Tool Usage
fetch_spider_url Test connectivity/headers.
debug_spider_rule Test CSS/Regex selectors.
validate_spider Check rule structure.
check_syntax Validate JS syntax.
read_file Read & Auto-Decrypt.
get_spider_template Standard template.
manage_config Edit env.json.