main.js

const {BskyAgent, RichText, AppBskyFeedPost} = require("@atproto/api")

let Parser = require('rss-parser')
let parser = new Parser()
const entities = require("entities");

const sleep = (waitTimeInMs) => new Promise(resolve => setTimeout(resolve, waitTimeInMs));

const settings = [
  {
    account: "arxiv-stat-ml.bsky.social",
    password: "xxxx-xxxx-xxxx-xxxx",
    url: "https://export.arxiv.org/rss/stat.ML",
    endsWith: "[stat.ML])"
  },
  {
    account: "arxiv-stat-me.bsky.social",
    password: "xxxx-xxxx-xxxx-xxxx",
    url: "https://export.arxiv.org/rss/stat.ME",
    endsWith: "[stat.ME])"
  },
  {
    account: "arxiv-math-ac.bsky.social",
    password: "xxxx-xxxx-xxxx-xxxx",
    url: "https://export.arxiv.org/rss/math.AC",
    endsWith: "[math.AC])"
  },
  // ... omit
]

async function get_feeds(url, endsWith) {
  const feed = await parser.parseURL(url);
  let output = [];
  for (const item of feed.items) {
    if (item.title.endsWith(endsWith)) {
      const raw_authors = item.creator.replace(/<\/?("[^"]*"|'[^']*'|[^>])*(>|$)/g, "")
      const authors = entities.decodeHTML(raw_authors.trim())
      // const author_list = decode(raw_authors).trim().split(",")
      // if (author_list.length >= 8) {
      //   authors = author_list[0] + " et.al."
      // } else {
      //   authors = author_list.join(",")
      // }
      output.push({
        authors: authors,
        title: item.title,
        link: item.link
      });
    }
  }
  return output;
}
async function post(agent, item, execute=true) {
  const content_length = item.link.length + item.title.length;
  let authors = item.authors;
  console.log(authors.length);
  console.log(content_length);
  if (authors.length + content_length > 300) {
    authors = authors.slice(0, 300 - content_length - 5) + "..."
    console.log("[INFO] truncate");
  }
  rt = new RichText({text: authors + "\n" + item.title + "\n" + item.link});
  await rt.detectFacets(agent)
  const post = {
    $type: 'app.bsky.feed.post',
    text: rt.text,
    facets: rt.facets,
    createdAt: new Date().toISOString()
  }
  const res = AppBskyFeedPost.validateRecord(post)
  if (res.success) {
    console.log(post)
    if (execute) {
      await agent.post(post);
      await sleep(10000);
    }
  } else {
    console.log(res.error)
  }
}

async function main(setting, execute=true) {
  console.log(setting.account)
  const agent = new BskyAgent({ service: "https://bsky.social" });
  await agent.login({
    identifier: setting.account,
    password: setting.password,
  });

  let processed = new Set();
  let cursor = "";
  for (let i = 0; i < 3; ++i) {
    const response = await agent.getAuthorFeed({
      actor: setting.account,
      limit: 100,
      cursor: cursor,
    });
    cursor = response.cursor;
    for (const feed of response.data.feed) {
      const lines = feed.post.record.text.split("\n");
      processed.add(lines[1]);
      processed.add(lines[2]);
    }
  }
  for (const feed of await get_feeds(setting.url, setting.endsWith)) {
    console.log(feed);
    if (!processed.has(feed.title) && !processed.has(feed.link)) {
      console.log("post " + feed.title);
      await post(agent, feed, execute);
    } else {
      console.log("skipped " + feed.title);
    }
  }
}
/*
async function entrypoint() {
  for (const setting of settings) {
    await main(setting, false);
  }
}
entrypoint()
*/

const functions = require('@google-cloud/functions-framework');
functions.http('entrypoint', async (req, res) => {
  for (const setting of settings) {
    console.log("process " + setting.url);
    await main(setting);
  }
});