Snawoot/rutracker-topic-watch.gs

## rutracker-topic-watch.gs
const TRACKED_TOPICS = [
  6027310,
  6046402,
  5887367,
  5440705,
  5591249,
];

const CONTENT_CHARSET = 'cp1251';
const TOPIC_URL_TMPL = 'https://rutracker.org/forum/viewtopic.php?t=%d&start=%d';
const POST_ID_RE = /^post_(\d+)$/;
const POST_URL_TMPL = 'https://rutracker.org/forum/viewtopic.php?p=%d#%d';

const scriptProperties = PropertiesService.getScriptProperties();

function parseQuery(url) {
  const query = url.split("?")[1];
  if (query) {
    return query.split("&")
    .reduce(function(o, e) {
      var temp = e.split("=");
      var key = temp[0].trim();
      var value = temp[1].trim();
      value = isNaN(value) ? value : Number(value);
      o[key] = [value];
      return o;
    }, {});
  }
  return null;
}

function notifyOwner(topics) {
  const email = Session.getEffectiveUser().getEmail();
  const htmlBody = "<html><body>New posts found in following topics:<br>" +
  topics
    .map(([topicID, postID]) => Utilities.formatString('<a href="%s">%s</a> &lt;= <a href="%s">%s</a><br/>',
      makeTopicURL(topicID, 0), makeTopicURL(topicID, 0), makePostURL(postID), makePostURL(postID)))
    .join('\n')
  + '</body></html>'
  const options = {
    htmlBody: htmlBody,
  };
  const textBody = "New posts found in following topics:\n" +
  topics
    .map(([topicID, postID]) => Utilities.formatString('%s <= %s', makeTopicURL(topicID, 0), makePostURL(postID)))
    .join('\n');
  MailApp.sendEmail(email,
                    "rutracker topic watcher",
                    textBody,
                    options);
  Logger.log("User %s notified!", email);
}

function makeTopicURL(topicID, startParam) {
  return Utilities.formatString(TOPIC_URL_TMPL, topicID, startParam);
}

function makePostURL(postID) {
  return Utilities.formatString(POST_URL_TMPL, postID, postID);
}

function fetchTopicPage(topicID, startParam) {
  const url = makeTopicURL(topicID, startParam);
  const resp = UrlFetchApp.fetch(url);
  if (resp.getResponseCode() != 200) {
    throw "bad response code from URL "+url;
  }
  return resp.getContentText(CONTENT_CHARSET);
}

function parseHTML(html) {
  return Cheerio.load(html)
}

function getLatestPage(topicID) {
  const page = fetchTopicPage(topicID, 0);
  const doc = parseHTML(page);
  const latestIdx = doc('a.pg').get().map(
    (el) => {
      const href = doc(el).attr('href')
      if (href != null) {
        const start = parseQuery(href)["start"];
        if (isNaN(start)) {
          console.log("start is null", href)
          return 0;
        }
        return start;
      } else {
        return 0;
      }
    })
    .reduce((p, v) => {
      return ( p > v ? p : v );
    }, 0);
  console.log(`topic ${topicID} latestIdx=${latestIdx}`)
  if (latestIdx > 0) {
    console.log("refetch")
    return parseHTML(fetchTopicPage(topicID, latestIdx));
  }
  return doc;
}

function latestPost(doc) {
  const posts = doc('tbody')
    .filter((_, el) => {
      const idAttr = doc(el).attr('id');
      return idAttr != null && idAttr.match(POST_ID_RE);
    })
  if (!posts.length) {
    throw "posts not found";
  }
  return posts.last().attr('id').match(POST_ID_RE)[1];
}

function processTopic(topicID) {
  const doc = getLatestPage(topicID);
  const latest = latestPost(doc);
  Logger.log(latest);

  if (scriptProperties.getProperty("topic_"+topicID+"_latest_post") != latest) {
    scriptProperties.setProperty("topic_"+topicID+"_latest_post", latest);
    return latest;
  }
  return null;
}

function main() {
  const errors = {};
  const updated = [];
  Logger.log("The rutracker topic watcher starting...");

  for (const topicID of TRACKED_TOPICS) {
    try {
      const res = processTopic(topicID);
      if (res != null) {
        updated.push([topicID, res]);
      }
    } catch(err) {
      errors[topicID] = err;
    }
  }

  if (updated.length) {
    notifyOwner(updated);
  }

  Logger.log("updated=%s", updated);
  Logger.log("errors=%s", errors);
  if (errors.length) {
    throw errors;
  }
  Logger.log("rutracker topic watcher run finished.");
}

function trigger(e) {
  Logger.log("Trigger activated: %s", JSON.stringify(e));
  main();
}
	const TRACKED_TOPICS = [
	6027310,
	6046402,
	5887367,
	5440705,
	5591249,
	];

	const CONTENT_CHARSET = 'cp1251';
	const TOPIC_URL_TMPL = 'https://rutracker.org/forum/viewtopic.php?t=%d&start=%d';
	const POST_ID_RE = /^post_(\d+)$/;
	const POST_URL_TMPL = 'https://rutracker.org/forum/viewtopic.php?p=%d#%d';

	const scriptProperties = PropertiesService.getScriptProperties();

	function parseQuery(url) {
	const query = url.split("?")[1];
	if (query) {
	return query.split("&")
	.reduce(function(o, e) {
	var temp = e.split("=");
	var key = temp[0].trim();
	var value = temp[1].trim();
	value = isNaN(value) ? value : Number(value);
	o[key] = [value];
	return o;
	}, {});
	}
	return null;
	}

	function notifyOwner(topics) {
	const email = Session.getEffectiveUser().getEmail();
	const htmlBody = "<html><body>New posts found in following topics:<br>" +
	topics
	.map(([topicID, postID]) => Utilities.formatString('<a href="%s">%s</a> <= <a href="%s">%s</a><br/>',
	makeTopicURL(topicID, 0), makeTopicURL(topicID, 0), makePostURL(postID), makePostURL(postID)))
	.join('\n')
	+ '</body></html>'
	const options = {
	htmlBody: htmlBody,
	};
	const textBody = "New posts found in following topics:\n" +
	topics
	.map(([topicID, postID]) => Utilities.formatString('%s <= %s', makeTopicURL(topicID, 0), makePostURL(postID)))
	.join('\n');
	MailApp.sendEmail(email,
	"rutracker topic watcher",
	textBody,
	options);
	Logger.log("User %s notified!", email);
	}

	function makeTopicURL(topicID, startParam) {
	return Utilities.formatString(TOPIC_URL_TMPL, topicID, startParam);
	}

	function makePostURL(postID) {
	return Utilities.formatString(POST_URL_TMPL, postID, postID);
	}

	function fetchTopicPage(topicID, startParam) {
	const url = makeTopicURL(topicID, startParam);
	const resp = UrlFetchApp.fetch(url);
	if (resp.getResponseCode() != 200) {
	throw "bad response code from URL "+url;
	}
	return resp.getContentText(CONTENT_CHARSET);
	}

	function parseHTML(html) {
	return Cheerio.load(html)
	}

	function getLatestPage(topicID) {
	const page = fetchTopicPage(topicID, 0);
	const doc = parseHTML(page);
	const latestIdx = doc('a.pg').get().map(
	(el) => {
	const href = doc(el).attr('href')
	if (href != null) {
	const start = parseQuery(href)["start"];
	if (isNaN(start)) {
	console.log("start is null", href)
	return 0;
	}
	return start;
	} else {
	return 0;
	}
	})
	.reduce((p, v) => {
	return ( p > v ? p : v );
	}, 0);
	console.log(`topic ${topicID} latestIdx=${latestIdx}`)
	if (latestIdx > 0) {
	console.log("refetch")
	return parseHTML(fetchTopicPage(topicID, latestIdx));
	}
	return doc;
	}

	function latestPost(doc) {
	const posts = doc('tbody')
	.filter((_, el) => {
	const idAttr = doc(el).attr('id');
	return idAttr != null && idAttr.match(POST_ID_RE);
	})
	if (!posts.length) {
	throw "posts not found";
	}
	return posts.last().attr('id').match(POST_ID_RE)[1];
	}

	function processTopic(topicID) {
	const doc = getLatestPage(topicID);
	const latest = latestPost(doc);
	Logger.log(latest);

	if (scriptProperties.getProperty("topic_"+topicID+"_latest_post") != latest) {
	scriptProperties.setProperty("topic_"+topicID+"_latest_post", latest);
	return latest;
	}
	return null;
	}

	function main() {
	const errors = {};
	const updated = [];
	Logger.log("The rutracker topic watcher starting...");

	for (const topicID of TRACKED_TOPICS) {
	try {
	const res = processTopic(topicID);
	if (res != null) {
	updated.push([topicID, res]);
	}
	} catch(err) {
	errors[topicID] = err;
	}
	}

	if (updated.length) {
	notifyOwner(updated);
	}

	Logger.log("updated=%s", updated);
	Logger.log("errors=%s", errors);
	if (errors.length) {
	throw errors;
	}
	Logger.log("rutracker topic watcher run finished.");
	}

	function trigger(e) {
	Logger.log("Trigger activated: %s", JSON.stringify(e));
	main();
	}