☰

Wayback Machine Small Bug Fixes

Fixes encoded ampersands on Wayback Machine's captures graph and problems that arise when trailing slashes are missing in an URL and other small issues universally present in all crawled sites

As of 2016-02-15. See the latest version.

  1. // ==UserScript==
  2. // @name          Wayback Machine Small Bug Fixes
  3. // @namespace     DoomTay
  4. // @description   Fixes encoded ampersands on Wayback Machine's captures graph and problems that arise when trailing slashes are missing in an URL and other small issues universally present in all crawled sites
  5. // @version       1.2.8
  6. // @include       http://web.archive.org/web/*
  7. // @include       http://wayback.archive.org/web/*
  8. // @include       https://web.archive.org/web/*
  9. // @include       https://wayback.archive.org/web/*
  10. // @run-at        document-start
  11. // @exclude       /\*/
  12. // @grant         none
  13.  
  14. // ==/UserScript==
  15.  
  16. var toolbarNav = document.getElementById("wm-graph-anchor");
  17. var lastFolder = window.location.href.substring(window.location.href.lastIndexOf("/") + 1);
  18. var pics = document.images;
  19. var backgrounds = document.querySelectorAll("[background]");
  20. var shouldHaveTrailingSlash = (window.location.href.lastIndexOf(".") < window.location.href.lastIndexOf("/") || window.location.href.substring(window.location.href.lastIndexOf("//") + 2) == lastFolder) && lastFolder.indexOf("?") == -1;
  21. var hasTrailingSlash = window.location.href.lastIndexOf("/") == window.location.href.length - 1;
  22. var domain = window.location.href.substring(0,window.location.href.indexOf("/",window.location.href.lastIndexOf("//") + 2));
  23.  
  24. function fixToolbar()
  25. {
  26. 	while(toolbarNav.href.indexOf("&amp;") > -1) toolbarNav.href = toolbarNav.href.replace("&amp;","&");
  27. }
  28.  
  29. //Fix cases of &amp; in the capture graph
  30. if(toolbarNav) fixToolbar();
  31.  
  32. if(!document.getElementsByTagName("base")[0])
  33. {
  34. 	var base = document.createElement("base");
  35. 	if(shouldHaveTrailingSlash && !hasTrailingSlash) base.href = window.location.href + "/";
  36. 	else if((!hasTrailingSlash && !shouldHaveTrailingSlash) || hasTrailingSlash) base.href = document.baseURI;
  37. 	else base.href = domain + "/";
  38. 	document.head.appendChild(base);
  39. }
  40. 	
  41.  
  42. for(var i = 0; i < pics.length; i++)
  43. {
  44. 	//Skip over stuff related to the Wayback Machine toolbar and data URIs
  45. 	if((document.getElementById("wm-ipp") && document.getElementById("wm-ipp").contains(pics[i])) || pics[i].src.indexOf("data:") > -1) continue;
  46. 	//Refresh images in case the "base url" had to be modified.
  47. 	pics[i].src = pics[i].src;
  48. 	//For whatever reason, some images will point to within Internet Archive's "main" servers, instead of the crawled site. This attempts to fix that.
  49. 	if(pics[i].src.indexOf(document.domain + "/web") == -1) pics[i].src = fixURL(pics[i].src);
  50. }
  51.  
  52. for(var b = 0; b < backgrounds.length; b++)
  53. {
  54. 	var bg = backgrounds[b].background || backgrounds[b].getAttribute("background");
  55. 	//Skip over stuff related to the Wayback Machine toolbar and data URIs
  56. 	if((document.getElementById("wm-ipp") && document.getElementById("wm-ipp").contains(backgrounds[b])) || bg.indexOf("data:") > -1) continue;
  57. 	//Refresh images in case the "base url" had to be modified.
  58. 	changeBackground(backgrounds[b],bg);
  59. 	//For whatever reason, some images will point to within Internet Archive's "main" servers, instead of the crawled site. This attempts to fix that.
  60. 	if(relativeToAbsolute(bg).indexOf(document.domain + "/web") == -1) 
  61. 	{
  62. 		var absoluteBG = relativeToAbsolute(bg)
  63. 		changeBackground(backgrounds[b],fixURL(absoluteBG));
  64. 	}
  65. }
  66.  
  67. function relativeToAbsolute(bgURL)
  68. {
  69. 	var img = new Image();
  70. 	img.src = bgURL;
  71. 	return img.src;
  72. }
  73.  
  74. function fixURL(URL)
  75. {
  76. 	if(URL.indexOf(document.domain) > -1) return domain + URL.substring(URL.indexOf("/",URL.lastIndexOf("//") + 2));
  77. 	else return domain.substring(0,domain.indexOf("/http") + 1) + URL.substring(URL.indexOf("/http") + 1);
  78. }
  79.  
  80. function changeBackground(node, newBackground)
  81. {
  82. 	if(node.background) node.background = newBackground;
  83. 	else if(node.getAttribute("background")) node.setAttribute("background",newBackground);
  84. }
  85.  
  86. var observer = new MutationObserver(function(mutations) {
  87. 	mutations.forEach(function(mutation) {
  88. 		if(mutation.target.id == "wm-graph-anchor") toolbarNav = mutation.target;
  89. 		if(mutation.type == "attributes" && mutation.target == toolbarNav) fixToolbar();
  90. 		if(mutation.target.nodeType == 1) checkMutations(mutation.target);
  91. 		for(var i = 0; i < mutation.addedNodes.length; i++) 
  92. 		{
  93. 			if(mutation.addedNodes[i].nodeType == 1) checkMutations(mutation.addedNodes[i]);
  94. 		}
  95. 	});    
  96. });
  97.  
  98. function checkMutations(node)
  99. {
  100. 	if(document.getElementById("wm-ipp") && document.getElementById("wm-ipp").contains(node)) return;
  101. 	if(node.nodeName == "IMG" && node.src.indexOf(document.domain + "/web") == -1)
  102. 	{
  103. 		node.src = fixURL(node.src);
  104. 	}
  105. 	if((node.getAttribute("background") || node.background) && (node.getAttribute("background") || node.background).indexOf(document.domain + "/web") == -1)
  106. 	{
  107. 		var bg = node.background || node.getAttribute("background");
  108. 		var absoluteBG = relativeToAbsolute(bg);
  109. 		changeBackground(node,fixURL(absoluteBG));
  110. 	}
  111. 	for(var n = 0; n < node.childNodes.length; n++)
  112. 	{
  113. 		if(node.childNodes[n].nodeType == 1) checkMutations(node.childNodes[n]);
  114. 	}
  115. }
  116.  
  117. var config = { attributes: true, childList: true, characterData: true, subtree: true };
  118. observer.observe(document.body || document, config);