MickSlash
/
F95API
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Merge branch 'master' of https://github.com/MillenniumEarl/F95API

pull/8/head
MillenniumEarl 2020-10-07 08:28:34 +02:00
parent 5920563516 a84f8c2908
commit 8a2fd49308
3 changed files with 255 additions and 210 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
README.md
View File

@ -1,4 +1,7 @@
# F95API # F95API
Unofficial Node JS module for scraping F95Zone platform
Unofficial Node JS module for scraping F95Zone platform
[![FOSSA Status](https://app.fossa.com/api/projects/git%2Bgithub.com%2FMillenniumEarl%2FF95API.svg?type=shield)](https://app.fossa.com/projects/git%2Bgithub.com%2FMillenniumEarl%2FF95API?ref=badge_shield) [![FOSSA Status](https://app.fossa.com/api/projects/git%2Bgithub.com%2FMillenniumEarl%2FF95API.svg?type=shield)](https://app.fossa.com/projects/git%2Bgithub.com%2FMillenniumEarl%2FF95API?ref=badge_shield)
[![DeepSource](https://deepsource.io/gh/MillenniumEarl/F95API.svg/?label=active+issues&show_trend=true)](https://deepsource.io/gh/MillenniumEarl/F95API/?ref=repository-badge)
[![codecov](https://codecov.io/gh/MillenniumEarl/F95API/branch/master/graph/badge.svg?token=KHN1TNIH7D)](undefined)

173
app/scripts/game-scraper.js
View File

@ -1,17 +1,17 @@
'use strict'; "use strict";
// Public modules from npm // Public modules from npm
const HTMLParser = require('node-html-parser'); const HTMLParser = require("node-html-parser");
const puppeteer = require('puppeteer'); const puppeteer = require("puppeteer");
const urlExist = require('url-exist'); const urlExist = require("url-exist");
// Modules from file // Modules from file
const shared = require('./shared.js'); const shared = require("./shared.js");
const selectors = require('./costants/css-selectors.js'); const selectors = require("./costants/css-selectors.js");
const { preparePage } = require('./puppeteer-helper.js'); const { preparePage } = require("./puppeteer-helper.js");
const GameDownload = require('./classes/game-download.js'); const GameDownload = require("./classes/game-download.js");
const GameInfo = require('./classes/game-info.js'); const GameInfo = require("./classes/game-info.js");
const { isStringAValidURL, isF95URL } = require('./urls-helper.js'); const { isStringAValidURL, isF95URL } = require("./urls-helper.js");
/** /**
* @protected * @protected
@ -20,18 +20,18 @@ const { isStringAValidURL, isF95URL } = require('./urls-helper.js');
* @param {URL} url URL of the game/mod to extract data from * @param {URL} url URL of the game/mod to extract data from
* @return {Promise<GameInfo>} Complete information about the game you are looking for * @return {Promise<GameInfo>} Complete information about the game you are looking for
*/ */
module.exports.getGameInfo = async function(browser, url) { module.exports.getGameInfo = async function (browser, url) {
if (shared.debug) console.log('Obtaining game info'); if (shared.debug) console.log("Obtaining game info");
// Verify the correctness of the URL // Verify the correctness of the URL
if (!isF95URL(url)) throw url + ' is not a valid F95Zone URL'; if (!isF95URL(url)) throw url + " is not a valid F95Zone URL";
let exists = await urlExist(url.toString()); let exists = await urlExist(url.toString());
if (!exists) return new GameInfo(); if (!exists) return new GameInfo();
let page = await preparePage(browser); // Set new isolated page let page = await preparePage(browser); // Set new isolated page
await page.setCookie(...shared.cookies); // Set cookies to avoid login await page.setCookie(...shared.cookies); // Set cookies to avoid login
await page.goto(url.toString(), { await page.goto(url.toString(), {
waitUntil: shared.WAIT_STATEMENT waitUntil: shared.WAIT_STATEMENT,
}); // Go to the game page and wait until it loads }); // Go to the game page and wait until it loads
// It asynchronously searches for the elements and // It asynchronously searches for the elements and
@ -53,15 +53,19 @@ module.exports.getGameInfo = async function(browser, url) {
info.overview = overview; info.overview = overview;
info.tags = await tags; info.tags = await tags;
info.f95url = url; info.f95url = url;
info.version = info.isMod ? parsedInfos['MOD VERSION'] : parsedInfos['VERSION']; info.version = info.isMod
info.lastUpdate = info.isMod ? parsedInfos['UPDATED'] : parsedInfos['THREAD UPDATED']; ? parsedInfos["MOD VERSION"]
: parsedInfos["VERSION"];
info.lastUpdate = info.isMod
? parsedInfos["UPDATED"]
: parsedInfos["THREAD UPDATED"];
info.previewSource = await previewSource; info.previewSource = await previewSource;
info.downloadInfo = await downloadData; info.downloadInfo = await downloadData;
await page.close(); // Close the page await page.close(); // Close the page
if (shared.debug) console.log('Founded data for ' + info.name); if (shared.debug) console.log("Founded data for " + info.name);
return info; return info;
} };
//#region Private methods //#region Private methods
/** /**
@ -75,9 +79,9 @@ module.exports.getGameInfo = async function(browser, url) {
function getOverview(text, isMod) { function getOverview(text, isMod) {
// Get overview (different parsing for game and mod) // Get overview (different parsing for game and mod)
let overviewEndIndex; let overviewEndIndex;
if (isMod) overviewEndIndex = text.indexOf('Updated'); if (isMod) overviewEndIndex = text.indexOf("Updated");
else overviewEndIndex = text.indexOf('Thread Updated'); else overviewEndIndex = text.indexOf("Thread Updated");
return text.substring(0, overviewEndIndex).replace('Overview:\n', '').trim(); return text.substring(0, overviewEndIndex).replace("Overview:\n", "").trim();
} }
/** /**
@ -88,12 +92,15 @@ function getOverview(text, isMod) {
* @returns {Promise<String>} Structured text * @returns {Promise<String>} Structured text
*/ */
async function getMainPostStructuredText(page) { async function getMainPostStructuredText(page) {
// Gets the first post, where are listed all the game's informations // Gets the first post, where are listed all the game's informations
let post = (await page.$$(selectors.THREAD_POSTS))[0]; let post = (await page.$$(selectors.THREAD_POSTS))[0];
// The info are plain text so we need to parse the HTML code // The info are plain text so we need to parse the HTML code
let bodyHTML = await page.evaluate( /* istanbul ignore next */ (mainPost) => mainPost.innerHTML, post); let bodyHTML = await page.evaluate(
return HTMLParser.parse(bodyHTML).structuredText; /* istanbul ignore next */ (mainPost) => mainPost.innerHTML,
post
);
return HTMLParser.parse(bodyHTML).structuredText;
} }
/** /**
@ -104,16 +111,18 @@ return HTMLParser.parse(bodyHTML).structuredText;
*/ */
async function getGameAuthor(page) { async function getGameAuthor(page) {
// Get the game/mod name (without square brackets) // Get the game/mod name (without square brackets)
let titleHTML = await page.evaluate( /* istanbul ignore next */ (selector) => let titleHTML = await page.evaluate(
/* istanbul ignore next */ (selector) =>
document.querySelector(selector).innerHTML, document.querySelector(selector).innerHTML,
selectors.GAME_TITLE); selectors.GAME_TITLE
);
let structuredTitle = HTMLParser.parse(titleHTML); let structuredTitle = HTMLParser.parse(titleHTML);
// The last element **shoud be** the title without prefixes (engines, status, other...) // The last element **shoud be** the title without prefixes (engines, status, other...)
let gameTitle = structuredTitle.childNodes.pop().rawText; let gameTitle = structuredTitle.childNodes.pop().rawText;
// The last square brackets contain the author // The last square brackets contain the author
let startTitleIndex = gameTitle.lastIndexOf('[') + 1; let startTitleIndex = gameTitle.lastIndexOf("[") + 1;
return gameTitle.substring(startTitleIndex, gameTitle.length - 1).trim(); return gameTitle.substring(startTitleIndex, gameTitle.length - 1).trim();
} }
@ -128,18 +137,17 @@ function parseConversationPage(text) {
let dataPairs = {}; let dataPairs = {};
// The information searched in the game post are one per line // The information searched in the game post are one per line
let splittedText = text.split('\n'); let splittedText = text.split("\n");
for (let line of splittedText) { for (let line of splittedText) {
if (!line.includes(":")) continue;
if (!line.includes(':')) continue;
// Create pair key/value // Create pair key/value
let splitted = line.split(':'); let splitted = line.split(":");
let key = splitted[0].trim().toUpperCase(); // Uppercase to avoid mismatch let key = splitted[0].trim().toUpperCase(); // Uppercase to avoid mismatch
let value = splitted[1].trim(); let value = splitted[1].trim();
// Add pair to the dict if valid // Add pair to the dict if valid
if (value != '') dataPairs[key] = value; if (value != "") dataPairs[key] = value;
} }
return dataPairs; return dataPairs;
@ -152,13 +160,16 @@ function parseConversationPage(text) {
* @returns {Promise<URL>} URL of the image or null if failed to get it * @returns {Promise<URL>} URL of the image or null if failed to get it
*/ */
async function getGamePreviewSource(page) { async function getGamePreviewSource(page) {
let src = await page.evaluate( /* istanbul ignore next */ (selector) => { let src = await page.evaluate(
/* istanbul ignore next */ (selector) => {
// Get the firs image available // Get the firs image available
let img = document.querySelector(selector); let img = document.querySelector(selector);
if (img) return img.getAttribute('src'); if (img) return img.getAttribute("src");
else return null; else return null;
}, selectors.GAME_IMAGES); },
selectors.GAME_IMAGES
);
// Check if the URL is valid // Check if the URL is valid
return isStringAValidURL(src) ? new URL(src) : null; return isStringAValidURL(src) ? new URL(src) : null;
@ -172,14 +183,16 @@ async function getGamePreviewSource(page) {
*/ */
async function getGameTitle(page) { async function getGameTitle(page) {
// Get the game/mod name (without square brackets) // Get the game/mod name (without square brackets)
let titleHTML = await page.evaluate( /* istanbul ignore next */ (selector) => let titleHTML = await page.evaluate(
/* istanbul ignore next */ (selector) =>
document.querySelector(selector).innerHTML, document.querySelector(selector).innerHTML,
selectors.GAME_TITLE); selectors.GAME_TITLE
);
let structuredTitle = HTMLParser.parse(titleHTML); let structuredTitle = HTMLParser.parse(titleHTML);
// The last element **shoud be** the title without prefixes (engines, status, other...) // The last element **shoud be** the title without prefixes (engines, status, other...)
let gameTitle = structuredTitle.childNodes.pop().rawText; let gameTitle = structuredTitle.childNodes.pop().rawText;
let endTitleIndex = gameTitle.indexOf('['); let endTitleIndex = gameTitle.indexOf("[");
return gameTitle.substring(0, endTitleIndex).trim(); return gameTitle.substring(0, endTitleIndex).trim();
} }
@ -194,7 +207,10 @@ async function getGameTags(page) {
// Get the game tags // Get the game tags
for (let handle of await page.$$(selectors.GAME_TAGS)) { for (let handle of await page.$$(selectors.GAME_TAGS)) {
let tag = await page.evaluate( /* istanbul ignore next */ (element) => element.innerText, handle); let tag = await page.evaluate(
/* istanbul ignore next */ (element) => element.innerText,
handle
);
tags.push(tag.toUpperCase()); tags.push(tag.toUpperCase());
} }
return tags.sort(); return tags.sort();
@ -209,20 +225,22 @@ async function getGameTags(page) {
* @returns {Promise<GameInfo>} GameInfo object passed in to which the identified information has been added * @returns {Promise<GameInfo>} GameInfo object passed in to which the identified information has been added
*/ */
async function parsePrefixes(page, info) { async function parsePrefixes(page, info) {
const MOD_PREFIX = 'MOD'; const MOD_PREFIX = "MOD";
// The 'Ongoing' status is not specified, only 'Abandoned'/'OnHold'/'Complete' // The 'Ongoing' status is not specified, only 'Abandoned'/'OnHold'/'Complete'
info.status = 'Ongoing'; info.status = "Ongoing";
for (let handle of await page.$$(selectors.GAME_TITLE_PREFIXES)) { for (let handle of await page.$$(selectors.GAME_TITLE_PREFIXES)) {
let value = await page.evaluate( /* istanbul ignore next */ (element) => element.innerText, handle); let value = await page.evaluate(
/* istanbul ignore next */ (element) => element.innerText,
handle
);
// Clean the prefix // Clean the prefix
let prefix = value.toUpperCase().replace('[', '').replace(']', '').trim(); let prefix = value.toUpperCase().replace("[", "").replace("]", "").trim();
// Getting infos... // Getting infos...
if (shared.statuses.includes(prefix)) info.status = prefix; if (shared.statuses.includes(prefix)) info.status = prefix;
else if (shared.engines.includes(prefix)) info.engine = prefix; else if (shared.engines.includes(prefix)) info.engine = prefix;
// This is not a game but a mod // This is not a game but a mod
else if (prefix === MOD_PREFIX) info.isMod = true; else if (prefix === MOD_PREFIX) info.isMod = true;
} }
@ -237,39 +255,57 @@ async function parsePrefixes(page, info) {
*/ */
async function getGameDownloadLink(page) { async function getGameDownloadLink(page) {
// Most used hosting platforms // Most used hosting platforms
let hostingPlatforms = ['MEGA', 'NOPY', 'FILESUPLOAD', 'MIXDROP', 'UPLOADHAVEN', 'PIXELDRAIN', 'FILESFM']; let hostingPlatforms = [
"MEGA",
"NOPY",
"FILESUPLOAD",
"MIXDROP",
"UPLOADHAVEN",
"PIXELDRAIN",
"FILESFM",
];
// Supported OS platforms // Supported OS platforms
let platformOS = ['WIN', 'LINUX', 'MAC', 'ALL'] let platformOS = ["WIN", "LINUX", "MAC", "ALL"];
// Gets the <span> which contains the download links // Gets the <span> which contains the download links
let temp = await page.$$(selectors.DOWNLOAD_LINKS_CONTAINER); let temp = await page.$$(selectors.DOWNLOAD_LINKS_CONTAINER);
if(temp.length === 0) return []; if (temp.length === 0) return [];
// Look for the container that contains the links // Look for the container that contains the links
// It is necessary because the same css selector // It is necessary because the same css selector
// also identifies other elements on the page // also identifies other elements on the page
let container = null; let container = null;
for(let candidate of temp) { for (let candidate of temp) {
if (container !== null) break; if (container !== null) break;
let upperText = (await page.evaluate( /* istanbul ignore next */ (e) => e.innerText, candidate)).toUpperCase(); let upperText = (
await page.evaluate(
/* istanbul ignore next */ (e) => e.innerText,
candidate
)
).toUpperCase();
// Search if the container contains the name of a hosting platform // Search if the container contains the name of a hosting platform
for (let p of hostingPlatforms) { for (let p of hostingPlatforms) {
if(upperText.includes(p)) { if (upperText.includes(p)) {
container = candidate; container = candidate;
break; break;
} }
} }
} }
if(container === null) return []; if (container === null) return [];
// Extract the HTML text from the container // Extract the HTML text from the container
let searchText = (await page.evaluate( /* istanbul ignore next */ (e) => e.innerHTML, container)).toLowerCase(); let searchText = (
await page.evaluate(
/* istanbul ignore next */ (e) => e.innerHTML,
container
)
).toLowerCase();
// Parse the download links // Parse the download links
let downloadData = []; let downloadData = [];
for(let platform of platformOS) { for (let platform of platformOS) {
let data = extractGameHostingData(platform, searchText); let data = extractGameHostingData(platform, searchText);
downloadData.push(...data); downloadData.push(...data);
} }
@ -285,34 +321,33 @@ async function getGameDownloadLink(page) {
* @returns {GameDownload[]} List of game download links for the selected platform * @returns {GameDownload[]} List of game download links for the selected platform
*/ */
function extractGameHostingData(platform, text) { function extractGameHostingData(platform, text) {
const PLATFORM_BOLD_OPEN = '<b>'; const PLATFORM_BOLD_OPEN = "<b>";
const CONTAINER_SPAN_CLOSE = '</span>'; const CONTAINER_SPAN_CLOSE = "</span>";
const LINK_OPEN = '<a'; const LINK_OPEN = "<a";
const LINK_CLOSE = '</a>'; const LINK_CLOSE = "</a>";
const HREF_START = 'href="'; const HREF_START = 'href="';
const HREF_END = '"'; const HREF_END = '"';
const TAG_CLOSE = '>'; const TAG_CLOSE = ">";
// Identify the individual platforms // Identify the individual platforms
let startIndex = text.indexOf(platform.toLowerCase()); let startIndex = text.indexOf(platform.toLowerCase());
if(startIndex === -1) return []; if (startIndex === -1) return [];
else startIndex += platform.length else startIndex += platform.length;
// Find the <b>platform</b> // Find the <b>platform</b>
let endIndex = text.indexOf( let endIndex =
PLATFORM_BOLD_OPEN, text.indexOf(PLATFORM_BOLD_OPEN, startIndex) + PLATFORM_BOLD_OPEN.length;
startIndex) + PLATFORM_BOLD_OPEN.length;
// Find the end of the container // Find the end of the container
if (endIndex === -1) text.indexOf( if (endIndex === -1)
CONTAINER_SPAN_CLOSE, text.indexOf(CONTAINER_SPAN_CLOSE, startIndex) +
startIndex) + CONTAINER_SPAN_CLOSE.length; CONTAINER_SPAN_CLOSE.length;
text = text.substring(startIndex, endIndex); text = text.substring(startIndex, endIndex);
let downloadData = []; let downloadData = [];
let linkTags = text.split(LINK_OPEN); let linkTags = text.split(LINK_OPEN);
for(let tag of linkTags) { for (let tag of linkTags) {
// Ignore non-link string // Ignore non-link string
if (!tag.includes(HREF_START)) continue; if (!tag.includes(HREF_START)) continue;

9
test/test.js
View File

@ -1,4 +1,11 @@
const { debug, login, getGameData, loadF95BaseData, getUserData, logout } = require("../app/index"); const {
debug,
login,
getGameData,
loadF95BaseData,
getUserData,
logout,
} = require("../app/index");
//debug(true); //debug(true);
main(); main();