☰

Mturk Qualification Database and Scraper

Scrape, display, sort and search your Mturk qualifications

当前为 2023-01-26 提交的版本，查看 最新版本。

  1. // ==UserScript==
  2. // @name         Mturk Qualification Database and Scraper
  3. // @namespace    https://greasyfork.org/en/users/1004048-elias041
  4. // @version      0.3
  5. // @description  Scrape, display, sort and search your Mturk qualifications
  6. // @author       Elias041
  7. // @match        https://worker.mturk.com/qualifications/assigned
  8. // @match        https://worker.mturk.com/qt
  9. // @require      https://code.jquery.com/jquery-3.6.3.js
  10. // @require      https://code.jquery.com/ui/1.13.1/jquery-ui.min.js
  11. // @require      https://unpkg.com/dexie/dist/dexie.js
  12. // @require      https://unpkg.com/ag-grid-community@29.0.0/dist/ag-grid-community.min.js
  13. // @require      https://cdnjs.cloudflare.com/ajax/libs/moment.js/2.29.4/moment.min.js
  14. // @resource     https://cdn.jsdelivr.net/npm/ag-grid-community/styles/ag-grid.css
  15. // @resource     https://cdn.jsdelivr.net/npm/ag-grid-community/styles/ag-theme-apline.css
  16. // @icon         https://www.google.com/s2/favicons?sz=64&domain=mturk.com
  17. // @license      none
  18. // @grant        none
  19. // ==/UserScript==
  20.  
  21.  
  22. let timeout = 1850;
  23. let counter = " ";
  24. let retry_count = 0;
  25. let page = "https://worker.mturk.com/qualifications/assigned.json?page_size=100";
  26. let timeoutId = undefined;
  27. let scraping = false
  28. window.onload = function ()
  29. {
  30.  
  31.  
  32.  
  33. 	let t = document.getElementsByClassName("col-xs-5 col-md-3 text-xs-right p-l-0")[0],
  34. 		e = t.parentNode,
  35. 		o = document.createElement("div");
  36. 	o.style.color = "#fff";
  37. 	o.style.padding = "10px";
  38. 	o.style.boxShadow = "2px 2px 4px #888888";
  39. 	o.style.background = "#33773A";
  40. 	o.style.opacity = "0.5";
  41.         o.style.cursor = "pointer";
  42. 	o.id = "button";
  43. 	o.innerHTML = "Scrape&nbspQuals";
  44. 	e.insertBefore(o, t);
  45.  
  46. 	let c = document.createElement("div");
  47. 	c.style.color = "#fff";
  48. 	c.style.background = "#C78D99";
  49. 	c.style.padding = "10px";
  50. 	c.style.boxShadow = "2px 2px 4px #888888";
  51. 	c.style.background = "#383c44";
  52. 	c.style.opacity = "0.5";
  53.         c.style.cursor = "pointer";
  54. 	c.innerHTML = "Cancel";
  55. 	c.id = "cancelButton";
  56. 	e.insertBefore(c, t);
  57.  
  58.  
  59.     let d = document.createElement("div");
  60. 	d.style.color = "#fff";
  61. 	d.style.background = "#fc0f03";
  62. 	d.style.padding = "10px";
  63. 	d.style.boxShadow = "2px 2px 4px #888888";
  64. 	d.style.background = "#323552";
  65. 	d.style.opacity = "0.5";
  66.     d.style.cursor = "pointer";
  67. 	d.innerHTML = "Database";
  68. 	d.id = "dbButton";
  69. 	e.insertBefore(d, t);
  70.  
  71. 	let f = document.createElement("div");
  72. 	f.style.color = "#fff";
  73. 	f.style.padding = "10px";
  74. 	f.style.boxShadow = "2px 2px 4px #888888";
  75. 	f.style.background = "#33773A";
  76. 	f.style.opacity = "0.5";
  77. 	f.id = "progress";
  78. 	f.innerHTML = "-";
  79. 	e.insertBefore(f, t);
  80.  
  81. 	document.getElementById("dbButton").addEventListener("click", function e()
  82. 	{
  83. 		window.open("https://worker.mturk.com/qt", "_blank")
  84. 	})
  85.  
  86.  
  87.  
  88. 	document.getElementById("cancelButton").addEventListener("click", function e()
  89. 	{
  90.  
  91. 		scraping = false
  92. 		$("#cancelButton").css('background', '#383c44')
  93. 		$("#button").css('background', '#33773A')
  94. 		$("#progress").html("-")
  95. 	})
  96. 	document.getElementById("button").addEventListener("click", function e()
  97. 		{
  98. 			scraping = true;
  99. 			$("#button").css('background', '#383c44')
  100. 			$("#cancelButton").css('background', '#CE3132')
  101.  
  102.  
  103.  
  104. 			/*init db*/
  105. 			var db = new Dexie("qualifications");
  106. 			db.version(1).stores(
  107. 			{
  108. 				quals: `
  109.         id,
  110.         requester,
  111.         description,
  112.         score,
  113.         date,
  114.         qualName,
  115.         reqURL,
  116.         reqQURL,
  117.         retURL,
  118.         canRetake,
  119.         hasTest,
  120.         canRequest,
  121.         isSystem`
  122. 			});
  123.  
  124. 			/*main loop*/
  125. 			function getAssignedQualifications(nextPageToken = "")
  126. 			{
  127. 				if (!scraping)
  128. 				{
  129. 					return;
  130. 				} //cancel trap
  131. 				counter++
  132. 				$("#progress").html(counter);
  133. 				//$("#progressBar").html("&nbsp&nbsp&nbspProcessing&nbsppage&nbsp" + counter + "&nbsp&nbsp&nbsp");
  134. 				//console.log("scraping")
  135. 				$.getJSON(page)
  136.  
  137. 					.then(function (data)
  138. 					{
  139. 						data.assigned_qualifications.forEach(function (t)
  140. 						{
  141. 							db.quals.bulkAdd([
  142. 							{
  143. 								id: t.request_qualification_url,
  144. 								requester: t.creator_name,
  145. 								description: t.description,
  146. 								canRetake: t.can_retake_test_or_rerequest,
  147. 								retry: t.earliest_retriable_time,
  148. 								score: t.value,
  149. 								date: t.grant_time,
  150. 								qualName: t.name,
  151. 								reqURL: t.creator_url,
  152. 								retURL: t.retake_test_url,
  153. 								isSystem: t.is_system_qualification,
  154. 								canRequest: t.is_requestable,
  155. 								hasTest: t.has_test
  156. 							}])
  157. 						})
  158.  
  159. 						if (data.next_page_token !== null)
  160. 						{
  161. 							timeoutId = setTimeout(() =>
  162. 							{
  163. 								page = `https://worker.mturk.com/qualifications/assigned.json?page_size=100&next_token=${encodeURIComponent(data.next_page_token)}`
  164. 								getAssignedQualifications(data.next_page_token);
  165. 							}, timeout);
  166.  
  167.  
  168. 						}
  169. 						else
  170. 						{
  171. 							console.log("Scraping completed");
  172. 							console.log(counter + "pages");
  173. 							console.log("Timeout" + timeout);
  174. 							console.log(retry_count + "timeouts");
  175. 							$("#cancelButton").css('background', '#CE3132')
  176. 							$("#progress").css('background', '##25dc12')
  177. 							$("#progress").html('✓')
  178.  
  179. 						}
  180. 					})
  181.  
  182. 					.catch(function (error)
  183. 						{ //handle timeouts
  184. 							if (error.status === 429 && retry_count < 5)
  185. 							{
  186. 								retry_count++;
  187. 								timeout += 500;
  188. 								console.log("timed out, incrementing clock to " + timeout + " milliseconds")
  189. 								setTimeout(() =>
  190. 								{
  191. 									getAssignedQualifications(nextPageToken);
  192. 								}, 10000);
  193. 							}
  194. 							else
  195. 							{
  196. 								$("#progressBar").html("Timed&nbspout&nbsp5&nbsptimes,&nbspaborting.&nbsp" + timeout + "&nbspmilliseconds.");
  197. 								console.log("Timed out 5 times, aborting. " + timeout + " milliseconds.");
  198.  
  199. 							}
  200. 							/* $("#button").html("Retry?");
  201. 							 $("#button").css("background-color", "#e80c0f");
  202. 							 document.getElementById("button").addEventListener("click", function e() {
  203. 							     location.reload()
  204.  
  205.  
  206.  
  207. 							 }*/
  208.  
  209. 						}
  210.  
  211. 					)
  212.  
  213. 			}
  214.  
  215. 			getAssignedQualifications();
  216.  
  217. 		}
  218.  
  219.  
  220. 	)
  221.  
  222. };
  223.  
  224. /*ag-grid*/
  225. if (location.href === "https://worker.mturk.com/qt")
  226. {
  227. 	document.body.innerHTML = "";
  228. 	let gridDiv = document.createElement("div");
  229. 	gridDiv.setAttribute("id", "gridDiv");
  230. 	document.body.appendChild(gridDiv);
  231. 	document.title = "Qualifications";
  232.  
  233.  
  234. 	/*init db*/
  235. 	var db = new Dexie("qualifications");
  236. 	db.version(1).stores(
  237. 	{
  238. 		quals: `
  239.         id,
  240.         requester,
  241.         description,
  242.         score,
  243.         date,
  244.         qualName,
  245.         reqURL,
  246.         reqQURL,
  247.         retURL,
  248.         canRetake,
  249.         hasTest,
  250.         canRequest,
  251.         isSystem`
  252. 	});
  253.  
  254. 	gridDiv.innerHTML = `
  255. <div id="myGrid"  class="ag-theme-alpine">
  256. <style>
  257. .ag-theme-alpine {
  258.     --ag-grid-size: 3px;
  259.     width: 100%;
  260.     height: 100%;
  261.     position: absolute;
  262.     top: 0;
  263.     left: 0;
  264.     right: 0;
  265.     bottom: 0;
  266. }
  267. </style>
  268.      </div>`
  269.  
  270.  
  271. 	const gridOptions = {
  272. 		columnDefs: [
  273. 			{
  274. 				headerName: 'Mturk Qualification Database and Scraper',
  275. 				children: [
  276. 				{
  277. 					field: "qualName"
  278. 				},
  279. 				{
  280. 					field: "requester"
  281. 				}]
  282. 			},
  283.  
  284.  
  285. 			{
  286. 				headerName: ' ',
  287. 				children: [
  288. 				{
  289.  
  290.  
  291. 					field: "description",
  292. 					width: 350
  293. 				},
  294. 				{
  295. 					headerName: "Value",
  296. 					field: "score",
  297. 					width: 100
  298. 				},
  299. 				{
  300. 					headerName: "Date",
  301. 					field: "date",
  302. 					width: 100,
  303. 					valueGetter: function (params)
  304. 					{
  305. 						var date = new Date(params.data.date);
  306. 						return (date.getMonth() + 1) + "/" + date.getDate() + "/" + date.getFullYear();
  307. 					},
  308. 					comparator: function (valueA, valueB, nodeA, nodeB, isInverted)
  309. 					{
  310. 						var dateA = new Date(valueA);
  311. 						var dateB = new Date(valueB);
  312. 						return dateA - dateB;
  313. 					},
  314. 					//valueFormatter: function(params) {
  315. 					//	return new Date(params.value).toString().substring(4, 15);
  316. 					//}
  317. 				},
  318. 				{
  319.  
  320. 					headerName: "Requester ID",
  321. 					width: 150,
  322. 					field: "reqURL",
  323. 					valueFormatter: function (params)
  324. 					{
  325. 						var parts = params.value.split("/");
  326. 						return parts[2];
  327.  
  328. 					},
  329.  
  330. 				},
  331. 				{
  332. 					headerName: "Qual ID",
  333. 					field: "id",
  334.  
  335. 					valueFormatter: function (params)
  336. 					{
  337. 						if (!params.value || params.value === '') return '';
  338. 						var parts = params.value.split("/");
  339. 						return parts[2];
  340.  
  341. 					}
  342. 				}]
  343. 			},
  344. 			{
  345. 				headerName: 'More',
  346. 				children: [
  347. 					{
  348. 						headerName: " ",
  349. 						field: " ",
  350. 						width: 100,
  351. 						columnGroupShow: 'closed'
  352. 					},
  353. 					{
  354. 						headerName: "Retake",
  355. 						field: "canRetake",
  356. 						width: 100,
  357. 						columnGroupShow: 'open',
  358. 						suppressMenu: true
  359. 					},
  360. 					{
  361. 						headerName: "hasTest",
  362. 						field: "hasTest",
  363. 						width: 100,
  364. 						columnGroupShow: 'open',
  365. 						suppressMenu: true
  366. 					},
  367. 					{
  368. 						headerName: "canReq",
  369. 						field: "canRequest",
  370. 						width: 100,
  371. 						columnGroupShow: 'open',
  372. 						suppressMenu: true
  373. 					},
  374. 					{
  375. 						headerName: "System",
  376. 						field: "isSystem",
  377. 						width: 100,
  378. 						columnGroupShow: 'open',
  379. 						suppressMenu: true
  380. 					},
  381.  
  382. 				]
  383. 			}
  384. 		],
  385. 		defaultColDef:
  386. 		{
  387. 			sortable: true,
  388. 			filter: true,
  389. 			editable: true,
  390. 			resizable: true,
  391. 		},
  392. 		rowSelection: 'multiple',
  393. 		animateRows: true,
  394. 		rowData: []
  395. 	};
  396.  
  397. 	window.addEventListener('load', function ()
  398. 	{
  399. 		const gridDiv = document.querySelector('#myGrid');
  400. 		db.quals.toArray().then(data =>
  401. 		{
  402.  
  403. 			var filteredData = data.filter(function (row)
  404. 			{
  405. 				return !row.qualName.includes("Exc: [");
  406. 			});
  407. 			gridOptions.rowData = filteredData;
  408. 			new agGrid.Grid(gridDiv, gridOptions);
  409.  
  410. 		})
  411. 	})
  412. };
  413.  
  414.  
  415.