☰

HIT Scraper WITH EXPORT

Snag HITs.

目前为 2014-06-06 提交的版本。查看 最新版本。

  1. // ==UserScript==
  2. // @name        HIT Scraper WITH EXPORT
  3. // @author      Kerek and TJ
  4. // @description Snag HITs.
  5. //              Based in part on code from mmmturkeybacon Export Mturk History and mmmturkeybacon Color Coded Search with Checkpoints
  6. // @namespace   http://userscripts.org/users/536998
  7. // @match       https://www.mturk.com/mturk/findhits?match=true#hit_scraper*
  8. // @match       https://www.mturk.com/mturk/findhits?match=true?hit_scraper*
  9. // @version     1.3.0.9
  10. // @grant       GM_xmlhttpRequest
  11. // @grant       GM_getValue
  12. // @grant       GM_setValue
  13. // @grant	GM_deleteValue
  14. // @require     http://code.jquery.com/jquery-latest.min.js
  15. // ==/UserScript==
  16.  
  17. //alter the requester ignore last as you desire, case insensitive
  18. var ignore_list = ["oscar smith", "jonathon weber", "jerry torres", "Crowdsource", "we-pay-you-fast", "turk experiment", "jon brelig"];
  19.  
  20. //this searches extra pages if you skip too much, helps fill out results if you hit a chunk of ignored HITs.  Change to true for this behavior.
  21. var correct_for_skips = false;
  22.  
  23. //weight the four TO ratings for the coloring. Default has pay twice as important as fairness and nothing for communication and fast.
  24. var COMM_WEIGHT = 0;
  25. var PAY_WEIGHT  = 10;
  26. var FAIR_WEIGHT = 5;
  27. var FAST_WEIGHT = 0;
  28.  
  29. //display your hitdb records if applicable
  30. var check_hitDB = true;
  31.  
  32. //default text size
  33. var default_text_size=11;
  34.  
  35.  
  36.  
  37. var HITStorage = {};
  38. var indexedDB = window.indexedDB || window.webkitIndexedDB ||
  39.                 window.mozIndexedDB;
  40. window.IDBTransaction = window.IDBTransaction || window.webkitIDBTransaction || window.mozIDBTransaction;
  41. window.IDBKeyRange = window.IDBKeyRange || window.webkitIDBKeyRange || window.mozIDBKeyRange;
  42. HITStorage.IDBTransactionModes = { "READ_ONLY": "readonly", "READ_WRITE": "readwrite", "VERSION_CHANGE": "versionchange" };
  43. var IDBKeyRange = window.IDBKeyRange;
  44.  
  45. HITStorage.indexedDB = {};
  46. HITStorage.indexedDB = {};
  47. HITStorage.indexedDB.db = null;
  48.  
  49. HITStorage.indexedDB.onerror = function(e) {
  50.   console.log(e);
  51. };
  52.  
  53. var v=4;
  54.  
  55. HITStorage.indexedDB.checkTitle = function(title,button) {
  56.   var request = indexedDB.open("HITDB", v);
  57. 	request.onsuccess = function(e) {
  58. 		HITStorage.indexedDB.db = e.target.result;
  59. 		var db = HITStorage.indexedDB.db;
  60.     if (!db.objectStoreNames.contains("HIT"))
  61.     {
  62.       db.close();
  63.       return;
  64.     }
  65. 		var trans = db.transaction(["HIT"], HITStorage.IDBTransactionModes.READ_ONLY);
  66. 		var store = trans.objectStore("HIT");
  67.  
  68. 		var index = store.index("title");
  69.     index.get(title).onsuccess = function(event)
  70.     {
  71.       if (event.target.result === undefined)
  72.       {
  73.              console.log(title + ' not found');
  74.         history[button].titledb=false;
  75.       }
  76.       else
  77.       {
  78.           console.log(title + ' found');
  79.         history[button].titledb=true;
  80.       }
  81.       
  82.       db.close();
  83.     };
  84. 	};
  85. 	request.onerror = HITStorage.indexedDB.onerror;
  86. };
  87.  
  88. HITStorage.indexedDB.checkRequester = function(id,button) {
  89.   var request = indexedDB.open("HITDB", v);
  90. 	request.onsuccess = function(e) {
  91. 		HITStorage.indexedDB.db = e.target.result;
  92. 		var db = HITStorage.indexedDB.db;
  93.     if (!db.objectStoreNames.contains("HIT"))
  94.     {
  95.       db.close();
  96.       return;
  97.     }
  98. 		var trans = db.transaction(["HIT"], HITStorage.IDBTransactionModes.READ_ONLY);
  99. 		var store = trans.objectStore("HIT");
  100.  
  101. 		var index = store.index("requesterId");
  102.     index.get(id).onsuccess = function(event)
  103.     {
  104.       if (event.target.result === undefined)
  105.       {history[button].reqdb=false;
  106.         console.log(id + ' not found');
  107.       }
  108.       else
  109.       {
  110.             history[button].reqdb=true;
  111.         console.log(id + ' found');
  112.       }
  113.       db.close();
  114.     };
  115. 	};
  116. 	request.onerror = HITStorage.indexedDB.onerror;
  117. };
  118.  
  119. var PAGES_TO_SCRAPE = 3;
  120. var MINIMUM_HITS = 100;
  121. var SEARCH_REFRESH=0;
  122. var URL_BASE = "/mturk/searchbar?searchWords=&selectedSearchType=hitgroups";
  123. var initial_url = URL_BASE;
  124. var TO_REQ_URL = "http://turkopticon.ucsd.edu/reports?id=";
  125. var found_key_list=[];
  126. var last_clear_time = new Date().getTime();
  127. var searched_once = false;
  128. var save_new_results_time = 120;
  129. var save_results_time = 3600;
  130. var default_type = 0;
  131. var cur_loc = window.location.href;
  132. var time_input = document.createElement("INPUT");
  133. time_input.value = 0;
  134. var page_input = document.createElement("INPUT");
  135. page_input.value = 3;
  136. var min_input = document.createElement("INPUT");
  137. var new_time_display_input = document.createElement("INPUT");
  138. new_time_display_input.value = 300;
  139. var reward_input = document.createElement("INPUT");
  140. var qual_input = document.createElement("INPUT");
  141. qual_input.type = "checkbox";
  142. qual_input.checked = true;
  143. var masters_input = document.createElement("INPUT");
  144. masters_input.type = "checkbox";
  145. var sort_input1 = document.createElement("INPUT");
  146. sort_input1.type = "radio";
  147. sort_input1.name = "sort_type";
  148. sort_input1.value = "latest";
  149. sort_input1.checked = true;
  150. var sort_input2 = document.createElement("INPUT");
  151. sort_input2.type = "radio";
  152. sort_input2.name = "sort_type";
  153. sort_input2.value = "most";
  154. var sort_input3 = document.createElement("INPUT");
  155. sort_input3.type = "radio";
  156. sort_input3.name = "sort_type";
  157. sort_input3.value = "amount";
  158.  
  159. var search_input = document.createElement("INPUT");
  160.  
  161. var LINK_BASE = "https://www.mturk.com";
  162. var BACKGROUND_COLOR = "rgb(19, 19, 19)";
  163. var STATUSDETAIL_DELAY = 250;
  164. var MPRE_DELAY = 3000;
  165.  
  166. var next_page = 1;
  167.  
  168. var GREEN   = '#66CC66'; //  > 4
  169. var LIGHTGREEN = '#ADFF2F'; // > 3  GREEN YELLOW
  170. var YELLOW = '#FFD700';
  171. var ORANGE  = '#FF9900'; //  > 2
  172. var RED     = '#FF3030'; // <= 2
  173. var BLUE    = '#C0D9D9'; // no TO
  174. var GREY = 'lightGrey';
  175. var BROWN = '#94704D';
  176. var DARKGREY = '#9F9F9F';
  177. $('body').css('background', BACKGROUND_COLOR);
  178.  
  179. var API_PROXY_BASE = 'https://api.turkopticon.istrack.in/';
  180. var API_MULTI_ATTRS_URL = API_PROXY_BASE + 'multi-attrs.php?ids=';
  181. var REVIEWS_BASE = 'http://turkopticon.ucsd.edu/';
  182.  
  183. var control_panel_HTML = '<div id="control_panel" style="margin: 0 auto 0 auto;' +
  184.                          'border-bottom: 1px solid #000000; margin-bottom: 5px; ' +
  185.                          'background-color: ' + BACKGROUND_COLOR + ';"></div>';
  186. $('body > :not(#control_panel)').hide(); //hide all nodes directly under the body
  187. $('body').prepend(control_panel_HTML);
  188.  
  189. var control_panel = document.getElementById("control_panel");
  190. var big_red_button = document.createElement("BUTTON");
  191. var progress_report = document.createTextNode("Stopped");
  192. var text_area = document.createElement("TABLE");
  193. big_red_button.textContent = "Show Interface";
  194. big_red_button.onclick =  function(){show_interface();};
  195. control_panel.appendChild(big_red_button);
  196.  
  197. show_interface();
  198.  
  199. var global_run = false;
  200. var statusdetail_loop_finished = false;
  201. var date_header = "";
  202. var history = {};
  203. var wait_loop;
  204.  
  205. function set_progress_report(text, force)
  206. {
  207.     if (global_run == true || force == true)
  208.     {
  209.         progress_report.textContent = text;
  210.     }
  211. }
  212.  
  213. function get_progress_report()
  214. {
  215.     return progress_report.textContent;
  216. }
  217.  
  218. function wait_until_stopped()
  219. {
  220.     if (global_run == true)
  221.     {
  222.         if (statusdetail_loop_finished == true)
  223.         {
  224.             big_red_button.textContent = "Start";
  225.             set_progress_report("Finished", false);
  226.         }
  227.         else
  228.         {
  229.             setTimeout(function(){wait_until_stopped();}, 500);
  230.         }
  231.     }
  232. }
  233.  
  234. function display_wait_time(wait_time)
  235. {
  236.     if (global_run == true)
  237.     {
  238.         var current_progress = get_progress_report();
  239.         if (current_progress.indexOf("Searching again in")!==-1)
  240.         {
  241.             set_progress_report(current_progress.replace(/Searching again in \d+ seconds/ , "Searching again in " + wait_time + " seconds"),false);
  242.         }
  243.         else
  244.             set_progress_report(current_progress + " Searching again in " + wait_time + " seconds.", false);
  245.         if (wait_time>1)
  246.             setTimeout(function(){display_wait_time(wait_time-1);}, 1000);
  247.     }
  248. }
  249.  
  250. function dispArr(ar)
  251. {
  252.     var disp = "";
  253.     for (var z = 0; z < ar.length; z++)
  254.     {
  255.         disp += "id " + z + " is " + ar[z] + " ";
  256.     }
  257.     console.log(disp);
  258. }
  259.  
  260. function scrape($src)
  261. {
  262.     var $requester = $src.find('a[href^="/mturk/searchbar?selectedSearchType=hitgroups&requester"]');
  263.     var $title = $src.find('a[class="capsulelink"]');
  264.     var $reward = $src.find('span[class="reward"]');
  265.     var $preview = $src.find('a[href^="/mturk/preview?"]');
  266.     var $qualified = $src.find('a[href^="/mturk/notqualified?"]');
  267.     var $times = $src.find('a[id^="duration_to_complete"]');
  268.     var $descriptions = $src.find('a[id^="description"]');
  269.     var not_qualified_group_IDs=[];
  270.     var $quals = $src.find('a[id^="qualificationsRequired"]');
  271.     $qualified.each(function(){
  272.         var groupy = $(this).attr('href');
  273.         groupy = groupy.replace("/mturk/notqualified?hitId=","");
  274.         not_qualified_group_IDs.push(groupy);
  275.     });
  276.     var $mixed =  $src.find('a[href^="/mturk/preview?"],a[href^="/mturk/notqualified?"]');
  277.     var listy =[];
  278.     $mixed.each(function(){
  279.         var groupy = $(this).attr('href');
  280.         groupy = groupy.replace("/mturk/notqualified?hitId=","");
  281.         groupy = groupy.replace("/mturk/preview?groupId=","");
  282.         listy.push(groupy);
  283.     });
  284.     listy = listy.filter(function(elem, pos) {
  285.         return listy.indexOf(elem) == pos;
  286.     });
  287.  
  288.     for (var j = 0; j < $requester.length; j++)
  289.     {
  290.         var $hits = $requester.eq(j).parent().parent().parent().parent().parent().parent().find('td[class="capsule_field_text"]');
  291.         var requester_name = $requester.eq(j).text().trim();
  292.         var requester_link = $requester.eq(j).attr('href');
  293.         var group_ID=listy[j];
  294.         var preview_link = "/mturk/preview?groupId=" + group_ID;
  295.         var title = $title.eq(j).text().trim();
  296.         var reward = $reward.eq(j).text().trim();
  297.         var hits = $hits.eq(4).text().trim();
  298.         var time = $times.eq(j).parent()[0].nextSibling.nextSibling.innerHTML;
  299.         var description = $descriptions.eq(j).parent()[0].nextSibling.nextSibling.innerHTML;
  300.         //console.log(description);
  301.         var requester_id = requester_link.replace('/mturk/searchbar?selectedSearchType=hitgroups&requesterId=','');
  302.         var accept_link;
  303.         accept_link = preview_link.replace('preview','previewandaccept');
  304.         
  305.         /*HIT SCRAPER ADDITION*/
  306.         var qElements = $quals.eq(j).parent().parent().parent().find('tr');
  307.         //console.log(qElements);
  308.  
  309.         var qualifications = [];
  310.         for (var i = 1; i < qElements.length; i++) {
  311.             qualifications.push((qElements[i].childNodes[1].textContent.trim().replace(/\s+/g, ' ').indexOf("Masters") != -1 ? "[color=red][b]"+qElements[i].childNodes[1].textContent.trim().replace(/\s+/g, ' ')+"[/b][/color]" : qElements[i].childNodes[1].textContent.trim().replace(/\s+/g, ' ')));
  312.         }
  313.         var qualList = (qualifications.join(', ') ? qualifications.join(', ') : "None");
  314.  
  315.         key = requester_name+title+reward+group_ID;
  316.         found_key_list.push(key);
  317.         if (history[key] == undefined)
  318.         {
  319.             history[key] = {requester:"", title:"", description:"", reward:"", hits:"", req_link:"", quals:"", prev_link:"", rid:"", acc_link:"", new_result:"", qualified:"", found_this_time:"", initial_time:"", reqdb:"",titledb:"",time:""};
  320.             history[key].req_link = requester_link;
  321.             history[key].prev_link = preview_link;
  322.             history[key].requester = requester_name;
  323.             history[key].title = title;
  324.             history[key].reward = reward;
  325.             history[key].hits = hits;
  326.             history[key].rid = requester_id;
  327.             history[key].acc_link = accept_link;
  328.             history[key].time = time;
  329.             history[key].quals = qualList;
  330.             history[key].description = description;
  331. 			HITStorage.indexedDB.checkRequester(requester_id,key);
  332. 			HITStorage.indexedDB.checkTitle(title,key);
  333.             if (searched_once)
  334.             {
  335.                 history[key].initial_time = new Date().getTime();//-1000*(save_new_results_time - SEARCH_REFRESH);
  336.                 history[key].new_result = 0;
  337.             }
  338.             else
  339.             {
  340.                 history[key].initial_time = new Date().getTime()-1000*save_new_results_time;
  341.                 history[key].new_result = 1000*save_new_results_time;
  342.             }
  343.             if (not_qualified_group_IDs.indexOf(group_ID)!==-1)
  344.                 history[key].qualified = false;
  345.             else
  346.                 history[key].qualified = true;
  347.  
  348.             history[key].found_this_time = true;
  349.         }
  350.         else
  351.         {
  352.             history[key].new_result = new Date().getTime() - history[key].initial_time;
  353.             history[key].found_this_time = true;
  354.             history[key].hits = hits;
  355.         }
  356.     }
  357. }
  358.  
  359. function statusdetail_loop(next_URL)
  360. {
  361.     if (global_run == true)
  362.     {
  363.         if (next_URL.length != 0)
  364.         {
  365.             $.get(next_URL, function(data)
  366.             {
  367.                 var $src = $(data);
  368.                 var maxpagerate = $src.find('td[class="error_title"]:contains("You have exceeded the maximum allowed page request rate for this website.")');
  369.                 if (maxpagerate.length == 0)
  370.                 {
  371.                     set_progress_report("Processing page " + next_page, false);
  372.                     scrape($src);
  373.         
  374.                     $next_URL = $src.find('a[href^="/mturk/viewsearchbar"]:contains("Next")');
  375.                     next_URL = ($next_URL.length != 0) ? $next_URL.attr("href") : "";
  376.                     next_page++;
  377.                     if (default_type == 1)
  378.                     {
  379.                         var hmin = MINIMUM_HITS+1;
  380.                         for (j = 0; j < found_key_list.length; j++)
  381.                         {
  382.                             if (history[found_key_list[j]].hits < hmin)
  383.                             {
  384.                                 next_URL = "";
  385.                                 next_page = -1;
  386.                                 break;
  387.                             }
  388.                         }
  389.                     }
  390.         
  391.                     else if (next_page > PAGES_TO_SCRAPE && correct_for_skips)
  392.                     {
  393.                         var skipped_hits = 0;
  394.                         var added_pages = 0;
  395.                         for (j = 0; j < found_key_list.length; j++)
  396.                         {
  397.                             var obj = history[found_key_list[j]];
  398.                             if (! ignore_check(obj.requester,obj.title))
  399.                                 skipped_hits++;
  400.                         }
  401.                         added_pages = Math.floor(skipped_hits/10);
  402.                         if (skipped_hits%10 >6)
  403.                             added_pages++;
  404.                         if (next_page > PAGES_TO_SCRAPE + added_pages)
  405.                         {
  406.                             next_URL = "";
  407.                             next_page = -1;
  408.                         }
  409.             
  410.                     }
  411.                     else if (next_page > PAGES_TO_SCRAPE)
  412.                     {
  413.                         next_URL = "";
  414.                         next_page = -1;
  415.                     }
  416.                     
  417.                     setTimeout(function(){statusdetail_loop(next_URL);}, STATUSDETAIL_DELAY);
  418.                 }
  419.                 else
  420.                 {
  421.                     console.log("MPRE");
  422.                     setTimeout(function(){statusdetail_loop(next_URL);}, MPRE_DELAY);
  423.                 }
  424.             });
  425.         }
  426.         else
  427.         {
  428.             searched_once = true;
  429.             var found_hits = found_key_list.length;
  430.             var shown_hits = 0;
  431.             var new_hits = 0;
  432.             var url = API_MULTI_ATTRS_URL;
  433.             var rids = [];
  434.             var lastRow = text_area.rows.length - 1;
  435.             for (i = lastRow; i>0; i--)
  436.                 text_area.deleteRow(i);
  437.             for (j = 0; j < found_key_list.length; j++)
  438.             {
  439.                 //(function(url,rids,j) {
  440.                     var obj = history[found_key_list[j]];
  441.                     if (ignore_check(obj.requester,obj.title) && obj.found_this_time){
  442.                         ++shown_hits;
  443.                         //console.log(obj);
  444.                         //hit export will update col_heads[1]
  445.                         var col_heads = ["<a href='"+ LINK_BASE+obj.req_link +"' target='_blank'>" + obj.requester + "</a>","<a href='"+ LINK_BASE+obj.prev_link +"' target='_blank' title='"+ obj.description +"'>" + obj.title + "</a>",obj.reward,obj.hits,"TO down","<a href='"+ LINK_BASE+obj.acc_link +"' target='_blank'>Accept</a>"];
  446.                         var row = text_area.insertRow(text_area.rows.length);
  447.                         url += obj.rid + ',';
  448.                         rids.push(obj.rid);
  449.                         if (check_hitDB)
  450.                         {
  451.                             col_heads.push("R");
  452.                             col_heads.push("T");
  453.                         }
  454.                         if (!obj.qualified)
  455.                         {
  456.                             col_heads.push("Not Qualified");
  457.                         }
  458.                         for (i=0; i<col_heads.length; i++)
  459.                         {
  460.                             var this_cell = row.insertCell(i);
  461.                             row.cells[i].style.fontSize = default_text_size;
  462.                             this_cell.innerHTML = col_heads[i];
  463.                             if(i>1)
  464.                                 this_cell.style.textAlign = 'center';
  465.                             if (check_hitDB)
  466.                             {
  467.                                 if (i==6)
  468.                                 {
  469.                                     if (obj.reqdb)
  470.                                         this_cell.style.backgroundColor = GREEN;
  471.                                     else
  472.                                         this_cell.style.backgroundColor = RED;
  473.                                 }
  474.                                 else if (i==7)
  475.                                 {
  476.                                     if (obj.titledb)
  477.                                         this_cell.style.backgroundColor = GREEN;
  478.                                     else
  479.                                         this_cell.style.backgroundColor = RED;
  480.                                 }
  481.                                 else if (i==8)
  482.                                     this_cell.style.backgroundColor = DARKGREY;
  483.                             }
  484.                             else if (i==6)
  485.                                 this_cell.style.backgroundColor = DARKGREY;
  486.                         }
  487.                         if (Object.keys(history).length>0)
  488.                         {
  489.                             if (obj.new_result < 1000*save_new_results_time)
  490.                             {
  491.                                 new_hits++;
  492.                                 for (i in col_heads)
  493.                                 {
  494.                                     row.cells[i].style.fontSize = default_text_size + 1;
  495.                                     row.cells[i].style.fontWeight = "bold";
  496.                                 }
  497.                             }
  498.                         }
  499.                         button = document.createElement('button'); //HIT SCRAPER ADDITION
  500.                         button.textContent = 'vB';
  501.                         button.title = 'Export this HIT description as vBulletin formatted text';
  502.                         
  503.                         button.style.height = '14px';
  504.                         button.style.width = '30px';
  505.                         button.style.fontSize = '8px';
  506.                         button.style.border = '1px solid';
  507.                         button.style.padding = '0px';
  508.                         button.style.backgroundColor = 'transparent';
  509.                         
  510.                         //button.addEventListener("click", function() {export_func_deleg(j);}.bind(null,j), false);
  511.                         button.addEventListener("click", (function (obj,j) { return function() {export_func_deleg(obj,j);}})(obj,j));
  512.                         row.cells[1].appendChild(button);
  513.                     }
  514.                 //});
  515.                 
  516.             }
  517.             set_progress_report("Scrape complete. " + shown_hits + " HITs found (" + new_hits + " new results). " + (found_hits - shown_hits) + " HITs ignored.", false);
  518.             url = url.substring(0,url.length - 1);
  519.             //console.log(url);
  520.             var success_flag = false;
  521.             GM_xmlhttpRequest(
  522.             {
  523.                 method: "GET",
  524.                 url: url,
  525.                 onload: function (results)
  526.                 {
  527.                     //console.log(results.responseText);
  528.                     rdata = $.parseJSON(results.responseText);
  529.                     for (i = 0; i < rids.length; i++)
  530.                     {
  531.                         text_area.rows[i+1].style.backgroundColor = GREY;
  532.                         if (rdata[rids[i]])
  533.                         {
  534.                             var pay = rdata[rids[i]].attrs.pay
  535.                             var reviews = rdata[rids[i]].reviews
  536.                             var average = 0;
  537.                             var sum = 0;
  538.                             var divisor = 0;
  539.                             var comm = rdata[rids[i]].attrs.comm;
  540.                             var fair = rdata[rids[i]].attrs.fair;
  541.                             var fast = rdata[rids[i]].attrs.fast;
  542.                             if (comm > 0)
  543.                             {
  544.                                 sum += COMM_WEIGHT*comm;
  545.                                 divisor += COMM_WEIGHT;
  546.                             }
  547.                             if (pay > 0)
  548.                             {
  549.                                 sum += PAY_WEIGHT*pay;
  550.                                 divisor += PAY_WEIGHT;
  551.                             }
  552.                             if (fair > 0)
  553.                             {
  554.                                 sum += FAIR_WEIGHT*fair;
  555.                                 divisor += FAIR_WEIGHT;
  556.                             }
  557.                             if (fast > 0)
  558.                             {
  559.                                 sum += FAST_WEIGHT*fast;
  560.                                 divisor += FAST_WEIGHT;
  561.                             }
  562.                             if (divisor > 0)
  563.                             {
  564.                                 average = sum/divisor;
  565.                             }
  566.                             text_area.rows[i+1].cells[4].innerHTML = "<a href='"+ TO_REQ_URL+rids[i] +"' target='_blank'>" + pay + "</a>";
  567.                             if (reviews > 4)
  568.                             {
  569.                                 if (average > 4.49)
  570.                                     text_area.rows[i+1].style.backgroundColor = GREEN;
  571.                                 else if (average > 3.49)
  572.                                     text_area.rows[i+1].style.backgroundColor = LIGHTGREEN;
  573.                                 //else if (average > 2.99)
  574.                                  //   text_area.rows[i+1].style.backgroundColor = YELLOW;
  575.                                 else if (average > 1.99)
  576.                                     text_area.rows[i+1].style.backgroundColor = ORANGE;
  577.                                 else if (average > 0)
  578.                                     text_area.rows[i+1].style.backgroundColor = RED;
  579.                              }
  580.                         }
  581.                         else
  582.                         {
  583.                             text_area.rows[i+1].cells[4].innerHTML = "No data";
  584.                         }
  585.                     }
  586.                     success_flag = true;
  587.                  }
  588.             });
  589.             if (!success_flag)
  590.                 for (i = 0; i < rids.length; i++) text_area.rows[i+1].style.backgroundColor = GREY;
  591.             
  592.             statusdetail_loop_finished = true;
  593.             if (SEARCH_REFRESH>0)
  594.             {
  595.                 wait_loop = setTimeout(function(){if (global_run) start_it();}, 1000*SEARCH_REFRESH);
  596.                 display_wait_time(SEARCH_REFRESH);
  597.             }
  598.             else
  599.             {
  600.                 global_run = false;
  601.                 big_red_button.textContent = "Start";
  602.             }
  603.         }
  604.     }
  605. }
  606.  
  607. function ignore_check(r,t){
  608.     return -1 == ignore_list.map(function(item) { return item.toLowerCase(); }).indexOf(r.toLowerCase());
  609. }
  610.  
  611. function start_running()
  612. {
  613.     if (big_red_button.textContent == "Start")
  614.     {
  615.         global_run = true;
  616.         initial_url = URL_BASE;
  617.         if (search_input.value.length>0)
  618.         {
  619.             initial_url = initial_url.replace("searchWords=", "searchWords=" + search_input.value);
  620.         }
  621.         if (time_input.value.replace(/[^0-9]+/g,"") != "")
  622.         {
  623.             SEARCH_REFRESH = Number(time_input.value);
  624.         }
  625.         if (page_input.value.replace(/[^0-9]+/g,"") != "")
  626.         {
  627.             PAGES_TO_SCRAPE = Number(page_input.value);
  628.         }
  629.         if (min_input.value.replace(/[^0-9]+/g,"") != "")
  630.         {
  631.             MINIMUM_HITS = Number(min_input.value);
  632.         }
  633.         if (new_time_display_input.value.replace(/[^0-9]+/g,"") != "")
  634.         {
  635.             save_new_results_time = Number(new_time_display_input.value);
  636.         }
  637.         if (reward_input.value.replace(/[^0-9]+/g,"") != "")
  638.         {
  639.             initial_url += "&minReward=" + reward_input.value;
  640.         }
  641.         else
  642.         {
  643.             initial_url += "&minReward=0.00";
  644.         }
  645.         if (qual_input.checked)
  646.         {
  647.             initial_url += "&qualifiedFor=on"
  648.         }
  649.         else
  650.         {
  651.             initial_url += "&qualifiedFor=off"
  652.         }
  653. 		if (masters_input.checked)
  654.         {
  655.             initial_url += "&requiresMasterQual=on"
  656.         }
  657.         if (sort_input1.checked)
  658.         {
  659.             initial_url+= "&sortType=LastUpdatedTime%3A1";
  660.             default_type = 0;
  661.         }
  662.         else if (sort_input2.checked)
  663.         {
  664.             initial_url+= "&sortType=NumHITs%3A1";
  665.             default_type = 1;
  666.         }
  667.         else if (sort_input3.checked)
  668.         {
  669.             initial_url+= "&sortType=Reward%3A1";
  670.             default_type = 0;
  671.         }
  672.         
  673.         initial_url+="&pageNumber=1&searchSpec=HITGroupSearch"
  674.         start_it();
  675.     }
  676.     else
  677.     {
  678.         global_run = false;
  679.         clearTimeout(wait_loop);
  680.         big_red_button.textContent = "Start";
  681.         set_progress_report("Stopped", true);
  682.     }
  683. }
  684.  
  685. function start_it()
  686. {
  687.     statusdetail_loop_finished = false;
  688.     big_red_button.textContent = "Stop";
  689.     found_key_list=[];
  690.     var ctime = new Date().getTime()
  691.     if (ctime - last_clear_time > save_results_time*666)
  692.     {
  693.         var last_history=history;
  694.         history = {};
  695.         for (var key in last_history)
  696.         {
  697.             if (last_history[key].new_result<save_results_time*1000)
  698.             {
  699.                 history[key]=last_history[key];
  700.                 if (last_history[key].found_this_time)
  701.                 {
  702.                     last_history[key].found_this_time = false;
  703.                     if (last_history[key].new_result>save_new_results_time*1000)
  704.                         last_history[key].initial_time = ctime-1000*save_new_results_time;
  705.                 }
  706.             }
  707.  
  708.         }
  709.         last_clear_time = ctime;
  710.     }
  711.     next_page = 1;
  712.     statusdetail_loop(initial_url);
  713. }
  714.  
  715.  
  716. function show_interface()
  717. {
  718.     control_panel.style.color = BROWN;
  719.     control_panel.style.fontSize = 14;
  720.     control_panel.removeChild(big_red_button);
  721.     control_panel.appendChild(document.createTextNode("Auto-refresh delay: "));
  722.     time_input.onkeydown = function(event){if (event.keyCode == 13){start_running();}};
  723.     time_input.title = "Enter search refresh delay in seconds\n" + "Enter 0 for no auto-refresh\n" + "Default is 0 (no auto-refresh)";
  724.     time_input.size = 3;
  725.     control_panel.appendChild(time_input);
  726.     
  727.     control_panel.appendChild(document.createTextNode("   "));
  728.     
  729.     control_panel.appendChild(document.createTextNode("Pages to scrape: "));
  730.     page_input.onkeydown = function(event){if (event.keyCode == 13){start_running();}};
  731.     page_input.title = "Enter number of pages to scrape\n" + "Default is 4";
  732.     page_input.size = 3;
  733.     control_panel.appendChild(page_input);
  734.     
  735.     control_panel.appendChild(document.createTextNode("   "));
  736.     
  737.     control_panel.appendChild(document.createTextNode("Minimum batch size: "));
  738.     min_input.onkeydown = function(event){if (event.keyCode == 13){start_running();}};
  739.     min_input.title = "Enter minimum HITs for batch search\n" + "Default is 100";
  740.     min_input.size = 3;
  741.     control_panel.appendChild(min_input);
  742.     control_panel.appendChild(document.createTextNode("   "));
  743.     
  744.     control_panel.appendChild(document.createTextNode("New HIT highlighting: "));
  745.     new_time_display_input.onkeydown = function(event){if (event.keyCode == 13){start_running();}};
  746.     new_time_display_input.title = "Enter time (in seconds) to keep new HITs highlighted\n" + "Default is 300 (5 minutes)";
  747.     new_time_display_input.size = 6;
  748.     control_panel.appendChild(new_time_display_input);
  749.     
  750.     control_panel.appendChild(document.createElement("P"));
  751.     control_panel.appendChild(document.createTextNode("Minimum reward: "));
  752.     reward_input.size = 6;
  753.     control_panel.appendChild(reward_input);
  754.     control_panel.appendChild(document.createTextNode("   "));
  755.  
  756.     control_panel.appendChild(document.createTextNode("Qualified"));
  757.     control_panel.appendChild(qual_input);
  758.     control_panel.appendChild(document.createTextNode("     "));
  759. 	control_panel.appendChild(document.createTextNode("Masters"));
  760.     control_panel.appendChild(masters_input);
  761.     control_panel.appendChild(document.createTextNode("     "));
  762.     control_panel.appendChild(document.createTextNode("Sort types:   "));
  763.     control_panel.appendChild(sort_input1);
  764.     control_panel.appendChild(document.createTextNode("Latest"));
  765.     control_panel.appendChild(sort_input2);
  766.     control_panel.appendChild(document.createTextNode("Most Available"));
  767.     control_panel.appendChild(sort_input3);
  768.     control_panel.appendChild(document.createTextNode("Amount"));
  769.        
  770.     control_panel.appendChild(document.createElement("P"));
  771.     
  772.     control_panel.appendChild(search_input);
  773.     search_input.size = 20;
  774.     search_input.title = "Enter a search term to include\n" + "Default is blank (no included terms)";
  775.     search_input.placeholder="Enter search terms here";
  776.     
  777.     control_panel.appendChild(document.createTextNode("   "));
  778.     
  779.     big_red_button.textContent = "Start";
  780.     big_red_button.onclick = function(){start_running();};
  781.     
  782.     control_panel.appendChild(big_red_button);
  783.    
  784.     control_panel.appendChild(document.createTextNode("   "));
  785.     control_panel.appendChild(progress_report);
  786.      
  787.     control_panel.appendChild(document.createElement("P"));
  788.     
  789.     text_area.style.fontWeight = 400;
  790.     text_area.createCaption().innerHTML = "HITs";
  791.     var col_heads = ['Requester','Title','Reward','HITs Available','TO pay',"Accept HIT"];
  792.     var row = text_area.createTHead().insertRow(0);
  793.     text_area.caption.style.fontWeight = 800;
  794.     text_area.caption.style.color = BROWN;
  795. 	if (default_text_size > 10)
  796. 		text_area.cellPadding=Math.min(Math.max(1,Math.floor((default_text_size-10)/2)),5);
  797.     //console.log(text_area.cellPadding);
  798.     //text_area.cellPadding=2;
  799.     text_area.caption.style.fontSize = 28;
  800.     text_area.rows[0].style.fontWeight = 800;
  801.     text_area.rows[0].style.color = BROWN;
  802.     for (i=0; i<col_heads.length; i++)
  803.     {
  804.         var this_cell = row.insertCell(i);
  805.         this_cell.innerHTML = col_heads[i];
  806.         this_cell.style.fontSize = 14;
  807.         if (i > 1)
  808.             this_cell.style.textAlign = 'center';
  809.     }
  810.     
  811.     control_panel.appendChild(text_area);
  812. }
  813.  
  814.  
  815. /********HIT EXPORT ADDITIONS*****/
  816.  
  817. 	var EDIT = false;
  818.     var HIT;
  819.  
  820.     var TO_BASE = "http://turkopticon.ucsd.edu/";
  821.     var API_BASE = "https://api.turkopticon.istrack.in/";
  822.     var API_URL = API_BASE + "multi-attrs.php?ids=";
  823.     
  824.     DEFAULT_TEMPLATE = '[table][tr][td][b]Title:[/b] [url={prev_link}][COLOR=blue]{title}[/COLOR][/url]\n';
  825.     DEFAULT_TEMPLATE += '[b]Requester:[/b] [url=https://www.mturk.com/mturk/searchbar?selectedSearchType=hitgroups&requesterId={rid}][COLOR=blue]{requester}[/COLOR][/url]';
  826.     DEFAULT_TEMPLATE += ' [{rid}] ([url='+TO_BASE+'{rid}][COLOR=blue]TO[/COLOR][/url])';
  827.     DEFAULT_TEMPLATE += '\n[b]TO Ratings:[/b]{to_stuff}';
  828.     DEFAULT_TEMPLATE += '\n[b]Description:[/b] {description}';
  829.     DEFAULT_TEMPLATE += '\n[b]Time:[/b] {time}';
  830.     DEFAULT_TEMPLATE += '\n[b]Reward:[/b] [COLOR=green][b]{reward}[/b][/COLOR]';
  831.     DEFAULT_TEMPLATE += '\n[b]Qualifications:[/b] {quals}[/td][/tr][/table]';
  832.  
  833.     var TEMPLATE;
  834.     var EASYLINK;
  835.  
  836.     if (typeof GM_getValue === 'undefined')
  837.         TEMPLATE = null;
  838.     else {
  839.         TEMPLATE = GM_getValue('HITScraper Template');
  840.         EASYLINK = GM_getValue('HITScraper Easylink');
  841.     }
  842.     if (TEMPLATE == null) {
  843.         TEMPLATE = DEFAULT_TEMPLATE;
  844.     }
  845.  
  846.     function buildXhrUrl(rai) {
  847.         var url = API_URL;
  848.         var ri = rai;
  849.         url += rai;
  850.         return url;
  851.     }
  852.  
  853.     function makeXhrQuery(url) {
  854.         var xhr = new XMLHttpRequest();
  855.         try{
  856.         	xhr.open('GET', url, false);
  857.             xhr.send(null);
  858.             return $.parseJSON(xhr.response);
  859.         }
  860.         catch(err){
  861.             return "TO DOWN";
  862.         }
  863.     }
  864.  
  865.     function getNamesForEmptyResponses(rai, resp) {
  866.         for (var rid in rai) {
  867.             if (rai.hasOwnProperty(rid) && resp[rid] == "") {
  868.                 resp[rid] = $.parseJSON('{"name": "' + rai[rid][0].innerHTML + '"}');
  869.             }
  870.         }
  871.         return resp;
  872.     }
  873.  
  874.     function getKeys(obj) {
  875.         var keys = [];
  876.         for (var key in obj) {
  877.             keys.push(key);
  878.         }
  879.         return keys;
  880.     }
  881.  
  882. function export_func_deleg(item,index) {
  883. 	//console.log(item);
  884.     export_func(item);
  885.     }
  886.  
  887. function export_func(item) {
  888.     HIT = item;
  889.     edit_button.textContent = 'Edit Template';
  890.     apply_template(item);
  891.     div.style.display = 'block';
  892.     textarea.select();
  893. }
  894.  
  895. function apply_template(hit_data) {
  896.         var txt = TEMPLATE;
  897.  
  898.         var vars = ['title', 'requester', 'rid', 'description', 'reward', 'quals', 'prev_link', 'time', 'hits', 'to_stuff', 'to_text'];
  899.  
  900.         var resp = null;
  901.         if (txt.indexOf('{to_text}') >= 0 || txt.indexOf('{to_stuff}') >= 0){
  902.             var url = buildXhrUrl(hit_data["rid"]);
  903.             resp = makeXhrQuery(url);
  904.             //console.log(resp);
  905.         }
  906.         var toText = "";
  907.         var toStuff = "";
  908.         var toData = "";
  909.         var numResp = (resp == null || resp == "TO DOWN" ? "n/a" : resp[hit_data["rid"]].reviews);
  910.         if (resp == "TO DOWN"){
  911.             toStuff = " [URL=\""+TO_BASE+hit_data['rid']+"\"]TO down.[/URL]";
  912.             toText = toStuff;
  913.         }
  914.         else if (resp == null || resp[hit_data["rid"]].attrs == null && resp != "TO DOWN") {
  915.             toStuff = " No TO ";
  916.             toText = " No TO ";
  917.             toStuff += "[URL=\""+TO_BASE+"report?requester[amzn_id]=" + hit_data['rid'] + "&requester[amzn_name]=" + hit_data['requester'] + "\"]";
  918.             toStuff += "(Submit a new TO rating for this requester)[/URL]";
  919.         }
  920.         else {
  921.             for (var key in resp[hit_data["rid"]].attrs) {
  922.                 //toText += "\n[*]"+key+": "+resp[hit_data["requesterId"]].attrs[key]+"\n";
  923.                 var i = 0;
  924.                 var color = "green";
  925.                 var name = key;
  926.                 var num = Math.floor(resp[hit_data["rid"]].attrs[key]);
  927.                 switch (key){
  928.                     case "comm":
  929.                         name = "Communicativity";
  930.                         break;
  931.                     case "pay":
  932.                         name = "Generosity";
  933.                         break;
  934.                     case "fast":
  935.                         name = "Promptness";
  936.                         break;
  937.                     case "fair":
  938.                         name = "Fairness";
  939.                         break;
  940.                     default:
  941.                         name = key;
  942.                         break;
  943.                 }
  944.                 switch (num){
  945.                     case 0:
  946.                         color = "red";
  947.                         break;
  948.                     case 1:
  949.                         color = "red";
  950.                         break;
  951.                     case 2:
  952.                         color = "orange";
  953.                         break;
  954.                     case 3:
  955.                         color = "yellow";
  956.                         break;
  957.                     default:
  958.                         break;
  959.                 }
  960.                 toText += (num > 0 ? "\n[color="+color+"]" : "\n");
  961.                 for (i; i < num; i++){
  962.                     toText += "[b]☢[/b]"
  963.                 }
  964.                 toText += (num > 0 ? "[/color]" : "")
  965.                 if (i < 5){
  966.                     toText += "[color=white]";
  967.                     for (i; i < 5; i++)
  968.                         toText += "[b]☢[/b]";
  969.                     toText += "[/color]";
  970.                 }
  971.                 toText += " "+Number(resp[hit_data["rid"]].attrs[key]).toFixed(2)+" "+name;
  972.                 toData += Number(resp[hit_data["rid"]].attrs[key]).toFixed(2) + ",";
  973.             }
  974.             //toText += "[/list]";
  975.             toText += (txt.indexOf('{to_stuff}') >= 0 ? "" : "\nNumber of Reviews: "+numResp+"\n[URL=\""+TO_BASE+"report?requester[amzn_id]=" + hit_data['rid'] + "&requester[amzn_name]=" + hit_data['requester'] + "\"](Submit a new TO rating for this requester)[/URL]");
  976.             toStuff = '\n[img]http://data.istrack.in/to/' + toData.slice(0,-1) + '.png[/img]';
  977.             toStuff += (txt.indexOf('{to_stuff}') >= 0 ? (txt.indexOf('{to_text}') >= 0 ? "" : toText) : "");
  978.             toStuff += "\nNumber of Reviews: "+numResp;
  979.             toStuff += "[URL=\""+TO_BASE+"report?requester[amzn_id]=" + hit_data['rid'] + "&requester[amzn_name]=" + hit_data['requester'] + "\"]";
  980.             toStuff += "\n(Submit a new TO rating for this requester)[/URL]";
  981.         }
  982.         
  983.         for (var i = 0; i < vars.length; i++) {
  984.             t = new RegExp('\{' + vars[i] + '\}', 'g');
  985.             if (vars[i] == "to_stuff") {
  986.                 txt = txt.replace(t, toStuff);
  987.             }
  988.             else if (vars[i] == "to_text"){
  989.                 txt = txt.replace(t, toText);
  990.             }
  991.             else if (vars[i] == "prev_link"){
  992.                 txt = txt.replace(t,"https://www.mturk.com"+hit_data[vars[i]]);
  993.             }
  994.             else if (vars[i] == "acc_link"){
  995.                 txt = txt.replace(t,"https://www.mturk.com"+hit_data[vars[i]]);
  996.             }
  997.             else
  998.                 txt = txt.replace(t, hit_data[vars[i]]);
  999.         }
  1000.         textarea.value = txt;
  1001.     }
  1002.  
  1003. function hide_func(div) {
  1004.     if (EDIT == false)
  1005.         div.style.display = 'none';
  1006.     }
  1007.  
  1008. function edit_func() {
  1009.     if (EDIT == true) {
  1010.         EDIT = false;
  1011.         TEMPLATE = textarea.value;
  1012.         edit_button.textContent = 'Edit Template';
  1013.         apply_template(HIT);
  1014.     }
  1015.     else {
  1016.         console.log("Editing");
  1017.         EDIT = true;
  1018.         edit_button.textContent = 'Show Changes';
  1019.         save_button.disabled = false;
  1020.         textarea.value = TEMPLATE;
  1021.     }
  1022.     }
  1023.  
  1024. function default_func() {
  1025.     GM_deleteValue('HITScraper Template');
  1026.     TEMPLATE = DEFAULT_TEMPLATE;
  1027.     EDIT = false;
  1028.     edit_button.textContent = 'Edit Template';
  1029.     apply_template(HIT);
  1030.     }
  1031.  
  1032.     function save_func() {
  1033.         if (EDIT)
  1034.             TEMPLATE = textarea.value;
  1035.         GM_setValue('HITScraper Template', TEMPLATE);
  1036.     }
  1037.  
  1038. 	var div = document.createElement('div');
  1039.     var textarea = document.createElement('textarea');
  1040.     var div2 = document.createElement('label');
  1041.  
  1042.     div.style.position = 'fixed';
  1043.     div.style.width = '500px';
  1044.     div.style.height = '235px';
  1045.     div.style.left = '50%';
  1046.     div.style.right = '50%';
  1047.     div.style.margin = '-250px 0px 0px -250px';
  1048.     div.style.top = '300px';
  1049.     div.style.padding = '5px';
  1050.     div.style.border = '2px';
  1051.     div.style.backgroundColor = 'black';
  1052.     div.style.color = 'white';
  1053.     div.style.zIndex = '100';
  1054.  
  1055.     textarea.style.padding = '2px';
  1056.     textarea.style.width = '500px';
  1057.     textarea.style.height = '200px';
  1058.     textarea.title = '{title}\n{requester}\n{rid}\n{description}\n{reward}\n{quals}\n{prev_link}\n{time}\n{hit}\n{to_stuff}\n{to_text}';
  1059.  
  1060.     div.textContent = 'Press Ctrl+C to copy to clipboard. Click textarea to close';
  1061.     div.style.fontSize = '12px';
  1062.     div.appendChild(textarea);
  1063.  
  1064.     var edit_button = document.createElement('button');
  1065.     var save_button = document.createElement('button');
  1066.     var default_button = document.createElement('button');
  1067.     var easy_button = document.createElement('button');
  1068.  
  1069.     edit_button.textContent = 'Edit Template';
  1070.     edit_button.setAttribute('id', 'edit_button');
  1071.     edit_button.style.height = '18px';
  1072.     edit_button.style.width = '100px';
  1073.     edit_button.style.fontSize = '10px';
  1074.     edit_button.style.paddingLeft = '3px';
  1075.     edit_button.style.paddingRight = '3px';
  1076.     edit_button.style.backgroundColor = 'white';
  1077.  
  1078.     save_button.textContent = 'Save Template';
  1079.     save_button.setAttribute('id', 'save_button');
  1080.     save_button.style.height = '18px';
  1081.     save_button.style.width = '100px';
  1082.     save_button.style.fontSize = '10px';
  1083.     save_button.style.paddingLeft = '3px';
  1084.     save_button.style.paddingRight = '3px';
  1085.     save_button.style.backgroundColor = 'white';
  1086.     save_button.style.marginLeft = '5px';
  1087.  
  1088.     easy_button.textContent = 'Change Adfly Url';
  1089.     easy_button.setAttribute('id', 'easy_button');
  1090.     easy_button.style.height = '18px';
  1091.     easy_button.style.width = '100px';
  1092.     easy_button.style.fontSize = '10px';
  1093.     easy_button.style.paddingLeft = '3px';
  1094.     
  1095.     default_button.textContent = ' D ';
  1096.     default_button.setAttribute('id', 'default_button');
  1097.     default_button.style.height = '18px';
  1098.     default_button.style.width = '20px';
  1099.     default_button.style.fontSize = '10px';
  1100.     default_button.style.paddingLeft = '3px';
  1101.     default_button.style.paddingRight = '3px';
  1102.     default_button.style.backgroundColor = 'white';
  1103.     default_button.style.marginLeft = '5px';
  1104.     default_button.title = 'Return default template';
  1105.  
  1106.     div.appendChild(edit_button);
  1107.     div.appendChild(save_button);
  1108.     div.appendChild(default_button);
  1109.     div.appendChild(easy_button);
  1110.     save_button.disabled = true;
  1111.  
  1112.     div.style.display = 'none';
  1113.     textarea.addEventListener("click", function() {hide_func(div);}, false);
  1114.     edit_button.addEventListener("click", function() {edit_func();}, false);
  1115.     save_button.addEventListener("click", function() {save_func();}, false);
  1116.     default_button.addEventListener("click", function() {default_func();}, false);
  1117.     document.body.insertBefore(div, document.body.firstChild);