☰

Mturk Qualification Database and Scraper

Scrape, display, sort and search your Mturk qualifications

当前为 2023-02-18 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      0.77
  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;
  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.         retry_count=0;
  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. 				$.getJSON(page)
  134.  
  135. 					.then(function (data)
  136. 					{
  137. 						data.assigned_qualifications.forEach(function (t)
  138. 						{
  139. 							db.quals.bulkAdd([
  140. 							{
  141. 								id: t.request_qualification_url,
  142. 								requester: t.creator_name,
  143. 								description: t.description,
  144. 								canRetake: t.can_retake_test_or_rerequest,
  145. 								retry: t.earliest_retriable_time,
  146. 								score: t.value,
  147. 								date: t.grant_time,
  148. 								qualName: t.name,
  149. 								reqURL: t.creator_url,
  150. 								retURL: t.retake_test_url,
  151. 								isSystem: t.is_system_qualification,
  152. 								canRequest: t.is_requestable,
  153. 								hasTest: t.has_test
  154. 							}])
  155. 						})
  156.  
  157. 						if (data.next_page_token !== null)
  158. 						{
  159. 							timeoutId = setTimeout(() =>
  160. 							{
  161. 								page = `https://worker.mturk.com/qualifications/assigned.json?page_size=100&next_token=${encodeURIComponent(data.next_page_token)}`
  162. 								getAssignedQualifications(data.next_page_token);
  163. 							}, timeout);
  164.  
  165.  
  166. 						}
  167. 						else if (data.next_page_token === null)
  168. 						{
  169. 							console.log("Scraping completed");
  170. 							console.log(counter + "pages");
  171. 							console.log("Timeout" + timeout);
  172. 							console.log(retry_count + "timeouts");
  173. 							$("#cancelButton").css('background', '#383c44');
  174. 							$("#progress").css('background', '#25dc12');
  175. 							$("#progress").html('✓');
  176.                             $("#dbButton").css('background', '#57ab4f');
  177.                             return;
  178.  
  179.  
  180.                                    } else {
  181.                           console.log("Timeout or abort. Clock was " + timeout);
  182.                          $("#progress").css('background', '#FF0000');
  183.                       $("#progress").html('X');
  184.                                        return;
  185.                                }
  186.  
  187. 					})
  188.  
  189. 					.catch(function (error)
  190. 						{ //handle timeouts
  191. 							if (error.status === 429 && retry_count < 10)
  192. 							{
  193. 								retry_count++;
  194. 								timeout += 500;
  195. 								console.log("timed out, incrementing clock to " + timeout + " milliseconds")
  196. 								setTimeout(() =>
  197. 								{
  198. 									getAssignedQualifications(nextPageToken);
  199. 								}, 10000);
  200.                                  } else if (error.status === 503) {
  201.                                  $("#progress").css('background', '#FFFF00');
  202. 							     $("#progress").html('&#33;');
  203.                                if (confirm("Mturk responded with 503: Service Unavailable. Retry?")) {
  204.                                $("#progress").css('background', '#33773A');
  205.                                setTimeout(() => {
  206.                              getAssignedQualifications(nextPageToken);
  207.                               }, 10000);
  208.                               } else {
  209.                                 $("#progress").css('background', '#FF0000');
  210. 							     $("#progress").html('&#88;');
  211.                                 console.log("User declined retry.");
  212.                                   return;
  213.                                  }
  214. 							}
  215. 							else
  216. 							{
  217.  
  218.                              
  219.                                 return;
  220.  
  221. 							}
  222. 							
  223. 						}
  224.  
  225. 					)
  226.  
  227. 			}
  228.  
  229. 			getAssignedQualifications();
  230.  
  231. 		}
  232.  
  233.  
  234. 	)
  235.  
  236. };
  237.  
  238. /*ag-grid*/
  239. if (location.href === "https://worker.mturk.com/qt")
  240. {
  241. 	document.body.innerHTML = "";
  242. 	let gridDiv = document.createElement("div");
  243. 	gridDiv.setAttribute("id", "gridDiv");
  244. 	document.body.appendChild(gridDiv);
  245. 	document.title = "Qualifications";
  246.  
  247.  
  248. 	/*init db*/
  249. 	var db = new Dexie("qualifications");
  250. 	db.version(1).stores(
  251. 	{
  252. 		quals: `
  253.         id,
  254.         requester,
  255.         description,
  256.         score,
  257.         date,
  258.         qualName,
  259.         reqURL,
  260.         reqQURL,
  261.         retURL,
  262.         canRetake,
  263.         hasTest,
  264.         canRequest,
  265.         isSystem`
  266. 	});
  267.  
  268. 	gridDiv.innerHTML = `
  269. <div id="myGrid"  class="ag-theme-alpine">
  270. <style>
  271. .ag-theme-alpine {
  272.     --ag-grid-size: 3px;
  273.     width: 100%;
  274.     height: 100%;
  275.     position: absolute;
  276.     top: 0;
  277.     left: 0;
  278.     right: 0;
  279.     bottom: 0;
  280. }
  281. </style>
  282.      </div>`
  283.  
  284.  
  285. 	const gridOptions = {
  286. 		columnDefs: [
  287. 			{
  288. 				headerName: 'Mturk Qualification Database and Scraper',
  289. 				children: [
  290. 				{
  291. 					field: "qualName",
  292.                   comparator: function (valueA, valueB, nodeA, nodeB, isInverted) {
  293.     return valueA.toLowerCase().localeCompare(valueB.toLowerCase());
  294.                   } },
  295. {
  296.   headerName: "Requester",
  297.   field: "requester",
  298.   comparator: function (valueA, valueB, nodeA, nodeB, isInverted) {
  299.     return valueA.toLowerCase().localeCompare(valueB.toLowerCase());
  300.   }
  301.  
  302. }]
  303. 			},
  304.  
  305.  
  306. 			{
  307. 				headerName: ' ',
  308. 				children: [
  309. 						{field: "description",
  310. width: 350,
  311. cellRenderer: function(params) {
  312.     return '<span title="' + params.value + '">' + params.value + '</span>';
  313. },
  314. comparator: function (valueA, valueB, nodeA, nodeB, isInverted) {
  315.     return valueA.toLowerCase().localeCompare(valueB.toLowerCase());
  316. }},
  317. 				{
  318. 					headerName: "Value",
  319. 					field: "score",
  320. 					width: 100
  321. 				},
  322. 				{
  323. 					headerName: "Date",
  324. 					field: "date",
  325. 					width: 100,
  326. 					valueGetter: function (params)
  327. 					{
  328. 						var date = new Date(params.data.date);
  329. 						return (date.getMonth() + 1) + "/" + date.getDate() + "/" + date.getFullYear();
  330. 					},
  331. 					comparator: function (valueA, valueB, nodeA, nodeB, isInverted)
  332. 					{
  333. 						var dateA = new Date(valueA);
  334. 						var dateB = new Date(valueB);
  335. 						return dateA - dateB;
  336. 					},
  337. 					//valueFormatter: function(params) {
  338. 					//	return new Date(params.value).toString().substring(4, 15);
  339. 					//}
  340. 				},
  341. 				{
  342.  
  343. 					headerName: "Requester ID",
  344. 					width: 150,
  345. 					field: "reqURL",
  346. 					valueFormatter: function (params)
  347. 					{
  348. 						var parts = params.value.split("/");
  349. 						return parts[2];
  350.  
  351. 					},
  352.  
  353. 				},
  354. 				{
  355. 					headerName: "Qual ID",
  356. 					field: "id",
  357.  
  358. 					valueFormatter: function (params)
  359. 					{
  360. 						if (!params.value || params.value === '') return '';
  361. 						var parts = params.value.split("/");
  362. 						return parts[2];
  363.  
  364. 					}
  365. 				}]
  366. 			},
  367. 			{
  368. 				headerName: 'More',
  369. 				children: [
  370. 					{
  371. 						headerName: " ",
  372. 						field: " ",
  373. 						width: 100,
  374. 						columnGroupShow: 'closed'
  375. 					},
  376. 					{
  377. 						headerName: "Retake",
  378. 						field: "canRetake",
  379. 						width: 100,
  380. 						columnGroupShow: 'open',
  381. 						suppressMenu: true
  382. 					},
  383. 					{
  384. 						headerName: "hasTest",
  385. 						field: "hasTest",
  386. 						width: 100,
  387. 						columnGroupShow: 'open',
  388. 						suppressMenu: true
  389. 					},
  390. 					{
  391. 						headerName: "canReq",
  392. 						field: "canRequest",
  393. 						width: 100,
  394. 						columnGroupShow: 'open',
  395. 						suppressMenu: true
  396. 					},
  397. 					{
  398. 						headerName: "System",
  399. 						field: "isSystem",
  400. 						width: 100,
  401. 						columnGroupShow: 'open',
  402. 						suppressMenu: true
  403. 					},
  404.  
  405. 				]
  406. 			}
  407. 		],
  408. 		defaultColDef:
  409. 		{
  410. 			sortable: true,
  411. 			filter: true,
  412. 			editable: true,
  413. 			resizable: true,
  414. 		},
  415. 		rowSelection: 'multiple',
  416. 		animateRows: true,
  417. 		rowData: []
  418. 	};
  419.  
  420. 	window.addEventListener('load', function ()
  421. 	{
  422. 		const gridDiv = document.querySelector('#myGrid');
  423. 		db.quals.toArray().then(data =>
  424. 		{
  425.  
  426. 			var filteredData = data.filter(function (row)
  427. 			{
  428. 				return !row.qualName.includes("Exc: [");
  429. 			});
  430. 			gridOptions.rowData = filteredData;
  431. 			new agGrid.Grid(gridDiv, gridOptions);
  432.  
  433. 		})
  434. 	})
  435. };