crawler-lib/src/lib/ids.ts

import { request, RequestConfig } from './request'
import { createHmac } from 'crypto'

export interface IdsConfig {
  endpoint: string
  accessKeyId: string
  accessKeySecret: string
}

export enum IdsChannel {
  Csdn = 1,
  Wechat = 2,
  Zhihu = 3,
  Juejin = 4
}

export type IdsCrawlRule = 'fulldata' | 'extdata'

export interface IdsArticle {
  author_id: string | number
  cover_url: string
  code: IdsChannel
  title: string
  author_name?: string
  sn_code: string
  original_url?: string
  source_address?: string
  is_headline?: 0 | 1
  label?: string
  published_at: Date
  body: string
  is_original?: 0 | 1
  external_read_num?: number
  external_see_num?: number
  external_like_num?: number
  external_comment_num?: number
}

export class Ids {
  config: IdsConfig

  constructor(config: IdsConfig) {
    this.config = config
  }

  private async request(config: RequestConfig) {
    const { endpoint, accessKeyId, accessKeySecret } = this.config
    config.baseURL = endpoint
    config.url = config.url || '/'

    const timestamp = Math.floor(Date.now() / 1000)
    const data = config.url.toLowerCase() + timestamp
    const signature = createHmac('sha256', accessKeySecret).update(data).digest('base64')

    config.headers = config.headers || {}
    config.headers['Authorization'] = `IDS-HMAC-SHA256 Credential=${accessKeyId}/${timestamp},Signature=${signature}`

    const res = await request(config)

    if (res.data.status != 1) {
      console.error(res.data)
      throw new Error(`[${res.data.status}] ${res.data.info}`)
    }

    return res.data
  }

  getCrawlAuthors(channel: IdsChannel) {
    return this.request({
      method: 'POST',
      url: '/api/ids/getCrawlAuthors',
      data: {
        code: channel
      }
    })
  }

  getCrawlArticleRules(params: { ids: string[]; channel: IdsChannel }) {
    const { ids, channel } = params

    return this.request({
      method: 'POST',
      url: '/api/ids/getCrawlArticleRules',
      data: {
        code: channel,
        sn_codes: ids
      }
    })
  }

  putArticle(data: IdsArticle, rule: IdsCrawlRule) {
    return this.request({
      method: 'POST',
      url: `/api/ids/putArticleData?crawl=${rule}`,
      data
    })
  }
}