youtube-comment-scraper
Version:
Scraping comments from Youtube.
256 lines (216 loc) • 9.42 kB
JavaScript
//
// scraper.js
//
// Copyright (c) 2016-2017 Junpei Kawamoto
//
// This software is released under the MIT License.
//
// http://opensource.org/licenses/mit-license.php
//
const cheerio = require("cheerio");
const phantom = require("./phantom-helper");
const BASE_URL = "https://www.youtube.com/watch?v=";
const HTTPS = "https://";
const HTTP = "http://";
const URL_PARAM = "watch?v=";
const URL_CHANNEL = "/channel/";
const SUCCESS = "success";
/*
Check like score and convert to integer if not.
## Args
* value: Like score to be checked.
## Returns
Integer value.
*/
function check_like_score(value) {
if (value != null) {
const res = parseInt(value, 10);
if (!isNaN(res)) {
return res;
}
}
return 0;
};
module.exports = {
/*
Scraping a given Youtube page and return a set of comments.
## Args
* url: URL of the target page of video ID.
## Returns
Promise object. Use "then" method to receive results.
*/
comments(url) {
let id
if (!url.startsWith(HTTPS) && !url.startsWith(HTTP)) {
id = url;
url = BASE_URL + url;
} else {
const sp = url.split("/");
id = sp[sp.length - 1].substring(URL_PARAM.length);
}
return new Promise((resolve, reject) => {
return phantom.get().then((ph) => {
return ph.createPage().then((page) => {
return page.open(url).then((status) => {
if (status != SUCCESS) {
return Promise.reject(`Open url returns ${status}`);
}
console.warn("The video page is opened");
return new Promise((resolve, reject) => {
// Check header information has been loaded.
// If not, wait 1000 msec and try again, until
// loading is finished.
function check_header() {
// PhantomJS doesn't arrow functions.
page.evaluate(function() {
return document.getElementsByClassName("comment-section-header-renderer").length;
}).then((res) => {
if (res != 0) {
resolve();
} else {
setTimeout(check_header, 1000);
}
}).catch(reject);
}
check_header();
});
}).then(() => {
console.warn("Loading hidden pages and comments");
// PhantomJS doesn't support ES2015.
return page.evaluate(function() {
// Load hidden pages. This function clickes a
// load-more-button and wait 2000msec,
// and then do these steps while there are
// load-more-buttons.
function load_hidden_pages() {
var load_btns = document.getElementsByClassName("load-more-button");
if (load_btns.length !== 0) {
load_btns[0].click();
setTimeout(load_hidden_pages, 1500);
}
};
load_hidden_pages();
});
}).then(() => {
return new Promise((resolve, reject) => {
function get_body() {
// PhantomJS doesn't support arrow functions.
page.evaluate(function() {
if (document.getElementsByClassName("load-more-button").length === 0) {
return document.body.innerHTML;
}
}).then((html) => {
if (html) {
resolve(html);
} else {
setTimeout(get_body, 10000);
}
}).catch(reject);
}
get_body();
});
}).then((html) => {
page.close();
// TODO: fix it.
console.warn("Contents are loaded");
const $ = cheerio.load(html);
const res = [];
$(".comment-thread-renderer").each((_, elem) => {
const root = $(elem).children().first();
const children = [];
$(".comment-replies-renderer .comment-replies-renderer-pages .comment-renderer", this).each((_, elem) => {
const author = $(".comment-renderer-header", elem).children().first();
const child = {
comment: $(".comment-renderer-text-content", elem).text(),
author: author.text(),
author_id: author.data("ytid"),
like: check_like_score($(".comment-renderer-like-count.off", elem).text())
};
const receiver = $(".comment-renderer-text-content", elem).find("a").text();
if (receiver !== "") {
child.receiver = receiver;
}
children.push(child);
});
const author = $(".comment-author-text", root);
const comment = {
root: $(".comment-renderer-text-content", root).text(),
author: author.text(),
author_id: author.data("ytid"),
like: check_like_score($(".comment-renderer-like-count.off", root).text())
};
if (children.length !== 0) {
comment.children = children;
}
res.push(comment);
});
const user = $(".yt-user-info a");
resolve({
id: id,
channel: {
id: user.attr("href").substring(URL_CHANNEL.length),
name: user.text()
},
comments: res
});
}).catch((reason) => {
console.error(reason);
page.close();
reject(reason);
});
});
});
});
},
/*
Scraping a Youtube channel page and return a description of the channel.
## Args
* id: channel ID.
## Returns
Promise object. Use "then" method to receive results.
*/
channel(id) {
let url;
if (id.startsWith(HTTP) || id.startsWith(HTTPS)) {
if (!id.endsWith("/about")) {
url = id + "/about";
} else {
url = id;
}
} else {
url = "https://www.youtube.com/channel/" + id + "/about";
}
return new Promise((resolve, reject) => {
return phantom.get().then((ph) => {
return ph.createPage().then((page) => {
return page.open(url).then((status) => {
if (status != SUCCESS) {
return Promise.reject(`Open url returns ${status}.`)
}
// PhantomJS doesn't support ES2015.
return page.evaluate(function() {
return document.body.innerHTML;
});
}).then((html) => {
page.close();
const $ = cheerio.load(html);
resolve({
id: id,
name: $(".qualified-channel-title-text").text(),
description: $(".about-description").text().replace(/^\s+|\s+$/g, "")
});
}).catch((err) => {
console.error(err);
page.close();
reject(reason);
});
});
});
});
},
/*
Close this module.
This function should be called to close PhantomJS processes.
*/
close: phantom.close
};