artsite/service/cron-jobs/fetch-meta.js

const models = require('../../models');
const fetch = require('node-fetch');

let apiKey = require('process').env.ARK_API_KEY;
const url = "https://ark.cn-beijing.volces.com/api/v3/chat/completions";
let headers = {
  'Authorization': `Bearer ${apiKey}`,
  'Content-Type': 'application/json'
}

/**
 * 豆包视觉API
 * 从豆包获取图片的标题和文案描述
 * @param {*} imageUrl 
 */
async function fetchMetaByImageFromDoubao(imageUrl) {
  let data = {
    "model": "ep-20250206115552-7qg5c", // Doubao-1.5...ion-pro-32k 当前最新，贵，响应慢，效果好
    // "model": "ep-20250204231910-4phb8", // Doubao-vision-lite-32k  便宜点，相应速度快
    "messages": [
      {
        "role": "user", "content": [
          {
            "type": "text",
            "text": "这是一张填色页图片，根据图片生成简短标题(title，无需包含填色页关键字)、SEO标题(seotitle, 用于网站SEO 60个字符以内，包含填色页关键字)、简要描述（description, 用于网站SEO的meta description，160个字符以内，包含填色页关键字）、文案描述（copy, 200字左右）。以json格式输出，支持语言中文(zh)、英语(en)、西班牙语(es)、葡萄牙语(pt)、日语(ja)，形如： { title: {zh:'', en:'', es: '', pt: '', ja: ''}, description: {zh:'', en:'', es: '', pt: '', ja: ''}, copy:{zh:'', en:'', es: '', pt: '', ja: ''} }"
          },
          {
            "type": "image_url",
            "image_url": {
              "url": `${imageUrl}`
            }
          }
        ]
      }
    ]
  }

  console.log(data);

  const jsonData = JSON.stringify(data);

  const response = await fetch(url, { method: 'POST', headers, body: jsonData });

  let responseJson = await response.json();

  console.log(responseJson);

  if (responseJson.choices) return responseJson.choices[0].message.content;
  else return null;
}

/**
 * 豆包对话API
 * @param {*} text 
 */
async function fetchMetaByTxtFromDoubao(text) {
  let data = {
    "model": "ep-20250206115552-7qg5c", // Doubao-1.5...ion-pro-32k 当前最新，贵，响应慢，效果好
    // "model": "ep-20250204231910-4phb8", // Doubao-vision-lite-32k  便宜点，相应速度快
    "messages": [
      { "role": "user", "content": text }
    ]
  }

  console.log(data);

  const jsonData = JSON.stringify(data);

  const response = await fetch(url, { method: 'POST', headers, body: jsonData });

  let responseJson = await response.json();

  console.log(responseJson);

  return responseJson.choices[0].message.content;
}


async function runAlbumMeta() {
  let done = 0;
  let duration = 0;
  let hour, minute, second;
  let start = Date.now();

  // 筛选出所有album
  let query = { pid: 'art', enabled: true, $or: [{ seoTitle: { $exists: false } }, { seoTitle: null }] }
  let docs = await models.ArtAlbum
    .find(query)
    .sort({ order: 'asc' })
    .populate('title')
    .select('tag title slogon');


  let total = docs.length;
  console.log('total:', total);
  if (total <= 0) return;

  for (let doc of docs) {
    console.log(`process album ${doc._id}`);

    console.time(doc._id);

    try {
      let text = `帮我生成填色页专辑《${doc.title.zh}》的SEO title(60个字符以内) 和 description(160个字符以内), 以json格式输出，支持语言中文(zh)、英语(en)、西班牙语(es)、葡萄牙语(pt)、日语(ja)，形如： { title: {zh:'', en:'', es: '', pt: '', ja: ''}, description: {zh:'', en:'', es: '', pt: '', ja: ''}`;
      let metaInfo = await fetchMetaByTxtFromDoubao(text);
      console.log(metaInfo);
      let metaInfoJson = JSON.parse(metaInfo);
      let titleJson = metaInfoJson.title;
      let descJson = metaInfoJson.description;
      let title = JSON.stringify(titleJson);
      let desc = JSON.stringify(descJson);

      doc.seoTitle = title;
      doc.seoDescription = desc;
      await doc.save();

    } catch (e) {
      console.error(e.message);
    }

    console.timeEnd(doc._id);

    done++;
    duration = (Date.now() - start) / 1000;
    hour = (Math.floor(duration / 60 / 60)).toString().padStart(2, '0');
    minute = (Math.floor(duration / 60) % 60).toString().padStart(2, '0');
    second = (Math.floor(duration) % 60).toString().padStart(2, '0');

    console.log('progress: ' + Math.floor((100 * done / total)) + '% used time: ' + hour + ':' + minute + ':' + second);

  }
}


async function runVideoStoryMeta() {
  let done = 0;
  let duration = 0;
  let hour, minute, second;
  let start = Date.now();

  // 筛选出所有videostory
  let query = { enabled: true, $or: [{ seoTitle: { $exists: false } }, { seoTitle: null }] }
  let docs = await models.ArtVideoStory
    .find(query)
    .sort({ order: 'asc' })
    .select('name');


  let total = docs.length;
  console.log('total:', total);
  if (total <= 0) return;

  for (let doc of docs) {
    console.log(`process video-story ${doc._id}`);

    console.time(doc._id);

    try {
      let text = `帮我生成填色页视频故事《${doc.name}》的SEO title(60个字符以内) 和 description(160个字符以内), 以json格式输出，支持语言中文(zh)、英语(en)、西班牙语(es)、葡萄牙语(pt)、日语(ja)，形如： { title: {zh:'', en:'', es: '', pt: '', ja: ''}, description: {zh:'', en:'', es: '', pt: '', ja: ''}`;
      let metaInfo = await fetchMetaByTxtFromDoubao(text);
      console.log(metaInfo);
      let metaInfoJson = JSON.parse(metaInfo);
      let titleJson = metaInfoJson.title;
      let descJson = metaInfoJson.description;
      let title = JSON.stringify(titleJson);
      let desc = JSON.stringify(descJson);

      doc.seoTitle = title;
      doc.seoDescription = desc;
      await doc.save();

    } catch (e) {
      console.error(e.message);
    }

    console.timeEnd(doc._id);

    done++;
    duration = (Date.now() - start) / 1000;
    hour = (Math.floor(duration / 60 / 60)).toString().padStart(2, '0');
    minute = (Math.floor(duration / 60) % 60).toString().padStart(2, '0');
    second = (Math.floor(duration) % 60).toString().padStart(2, '0');

    console.log('progress: ' + Math.floor((100 * done / total)) + '% used time: ' + hour + ':' + minute + ':' + second);

  }
}

async function runArtMeta() {
  let done = 0;
  let duration = 0;
  let hour, minute, second;
  let start = Date.now();


  // 筛选出所有已经ready并且还没有title的图
  let query = { status: { $gte: 7000 }, $or: [{ seoTitle: { $exists: false } }, { seoTitle: null }] };
  let docs = await models.Art.find(query).limit(1000).sort({ publishTime: 'desc' });  // 内存有限，每次跑1000个


  let total = docs.length;
  console.log('total:', total);
  if (total <= 0) return;

  for (let doc of docs) {
    console.log(`process ${doc._id}`);

    let thumbUrl = `http://color.jccytech.cn/thumbs/v2/work/320/${doc._id}.png`;
    if (doc.hasSpecial) {
      thumbUrl = `http://color.jccytech.cn/thumbs/v2/special/320/${doc._id}.png`;
    }

    console.time(doc._id);

    try {
      let metaInfo = await fetchMetaByImageFromDoubao(thumbUrl);
      if (!metaInfo) {
        console.error("something wrong, quit!");
        process.exit();
      }
      console.log(metaInfo);
      let metaInfoJson = JSON.parse(metaInfo);
      let titleJson = metaInfoJson.title;
      let seotitleJson = metaInfoJson.seotitle;
      let descJson = metaInfoJson.description;
      let copyJson = metaInfoJson.copy;
      let title = JSON.stringify(titleJson);
      let seotitle = JSON.stringify(seotitleJson);
      let desc = JSON.stringify(descJson);
      let copy = JSON.stringify(copyJson);

      doc.seoTitle = seotitle;
      doc.seoDescription = desc;
      doc.title = title;
      doc.desc = copy;
      await doc.save();

    } catch (e) {
      console.error(e.message);
    }

    console.timeEnd(doc._id);

    done++;
    duration = (Date.now() - start) / 1000;
    hour = (Math.floor(duration / 60 / 60)).toString().padStart(2, '0');
    minute = (Math.floor(duration / 60) % 60).toString().padStart(2, '0');
    second = (Math.floor(duration) % 60).toString().padStart(2, '0');

    console.log('progress: ' + Math.floor((100 * done / total)) + '% used time: ' + hour + ':' + minute + ':' + second);

  }

  runArtMeta();

}

async function run() {
  await runAlbumMeta();
  await runVideoStoryMeta();
  await runArtMeta();
}


async function test() {
  let metaInfo = await fetchMetaByImageFromDoubao("https://color.jccytech.cn/thumbs/v2/work/640/67a254ec4f9d65537938e5c5.png");
  console.log(metaInfo);
  let metaInfoJson = JSON.parse(metaInfo);
  let titleJson = metaInfoJson.title;
  let descJson = metaInfoJson.copy;
  let title = JSON.stringify(titleJson);
  let desc = JSON.stringify(descJson);
  console.log(title);
  console.log(desc);
}


module.exports = { run }


if (require.main == module) {
  run();
}