Home Page
Archive > Posts > 2014

Facebook career resume upload+import erasing fix
Features causing unintended consequences

So I was applying for a job that requires using Facebook’s career section, and I found it immensely helpful that it automatically filled in most of the form (Skills, Education History, Experience, etc) via your profile, as that information can take well over half an hour to fill in every time. However, an uploaded resume is also required on the form, and when you upload the file, Facebook tries to import the contents from it, thereby deleting all information already filled in!

The following is the solution to this problem
  1. Make sure all information you have already filled in (the non automatic stuff) is saved elsewhere to fill in again. If this includes Skills, Education History or Experience, I highly recommend you fill this information directly into your Facebook profile. It can be made private if you do not wish it to be shared.
  2. Upload your resume, which will erase everything already on the forms.
  3. After you have uploaded your resume, run the following javascript (available via your browser’s javacript/developer console):
    and store the result
  4. Do a complete/hard refresh (ctrl+f5) of the page. While this will erase the resume, it will again fill in all the other automatic information.
  5. Run the following javascript to restore the resume (replacing the ### with the value received from above):
    The change will not show on the page, but it works
Pulling HTML from Github markdown for external use
Although, converting to markdown is a time consuming pain

So I started getting on the Github bandwagon FINALLY. I figured that while I was going to the trouble of remaking readme files for the projects into github markdown files, I might as well duplicate the compiled HTML for my website.

The below code is a simple PHP script to pull in the converted HTML from Github’s API and then do some more modifications to facilitate directly inserting it into a website.

  • The variables that can be updated are all at the top of the file.
  • The script will always output the finished result to the user’s browser, but can also optionally save it to an external file by setting the $SaveFileName variable.
  • Stylesheet:
    • The script automatically includes a specified stylesheet from the $StylesheetLocation variable.
    • The stylesheet I used is from https://gist.github.com/somebox/1082608. I’m not too happy with its coloring scheme, but it’ll do for now.
    • The required modifications that need to be made to the css are to change “body” to “.GHMarkdown”, and then add “.GHMarkdown” before all other rules.
    • This is the one I am currently using for my website, but it also has a few modifications made specifically for my layouts.
  • Modifications
    • In my markdowns, I like to link to internal sections by first creating a bookmark as “<div name="BOOKMARK_NAME">...</div>” and then linking via “[LinkName](#BOOKMARK_NAME)”. While this works on github, the bookmark’s names are actually changed to something like “user-content-BOOKMARK-NAME”, which is not useable outside of github. The first $RegexModifications item therefore updates the bookmarks back to their original name, and turns them into <span>s (which github does not support).
    • The second rule just removes the “aria-hidden” attributes, which my W3C checking scripts throw a warning on.
  • Note that sometimes, the script may return an error of “transfer closed with XXX bytes remaining to read”. This means that github denied the request (probably due to too many requests in too short a timespan), but the input is too large so github prematurely terminated the connection. If this happens, try sending a tiny input and see if you get back a proper error.

$SaveFileName='Output.html'; //Optionally save output to a file. Comment out to not save
        '/<div name="user-content-(.*?)"(.*?)<\/div>/s'=>'<span id="$1"$2</span>', //Change <div name="user-contentXXX ---TO--- <span name="XXX
        '/ ?aria-hidden="true"/'=>'' //Remove aria-hidden attribute

//Set the curl options
$CurlHandle=curl_init(); //Init curl
curl_setopt_array($CurlHandle, Array(
        CURLOPT_URL=>           'https://api.github.com/markdown/raw', //Markdown/raw takes and returns plain text input and output
        CURLOPT_FAILONERROR=>   false,
        CURLOPT_RETURNTRANSFER=>1, //Return result as a string
        CURLOPT_TIMEOUT=>       300,
        CURLOPT_POST=>          1,
        CURLOPT_POSTFIELDS=>    file_get_contents($InputFile), //Pull in the requested file
        CURLOPT_HTTPHEADER=>    Array('Content-type: text/plain'), //Github expects the given data to be plaintext
        CURLOPT_SSL_VERIFYPEER=>0, //In case there are problems with the PHP ssl chain (often the case in Windows), ignore the error
        CURLOPT_USERAGENT=>     'Curl/PHP' //Github now requires a useragent to process the request

//Pull in the html converted markdown from Github
if(curl_errno($CurlHandle)) //Check for error

//Make regex modifications
$Return=preg_replace(array_keys($RegexModifications), array_values($RegexModifications), $Return);

//Generate the final HTML. It will also be output here if not saving to a file
header('Content-Type: text/html; charset=utf-8');
if(isset($SaveFileName)) //If saving to a file, buffer output
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<title>Markdown pull</title>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
<link href="<?=$StylesheetLocation?>" rel=stylesheet type="text/css">
</head><body><div class=GHMarkdown>
//Save to a file if requested
        file_put_contents($SaveFileName, ob_get_flush()); //Actual output happens here too when saving to a file
Format Text [to HTML] Script

After writing the documentation in plaintext format for DSQL just now, I needed to convert it into HTML for the project’s page. I’ve done this before manually and it’s always very daunting, so I decided to really quickly write a script to do most of the work for me, which can be downloaded here, or the code seen below.

It has the following:
  • Input text box with HTML data that is instantly shown as HTML in a below section when modified.
    Both sections take up half the vertical screen space
  • Undo/redo buffer for the text box (very primitive functionality)
  • “Open in new page” button, which opens a new window with just the HTML data (useful for validation [W3C or whatnot]).
    This is disabled by default because it is a dangerous option (XSS exploitable, so the script would need to be secured/password protected if this was on)
  • “Escape HTML” escapes HTML characters so they are not improperly interpreted (e.x. “<” becomes “&lt;”)
  • Listize:
    • Turns tabbed lists into HTML
    • For example:
      would become:
      • 2
      • 3
        • 4

I realized while making the script that I should probably instead just start making my documentation in a markup (like GitHub’s) and then have that converted to HTML and text files. Oh well.

<? header('Content-Type: text/html; charset=utf-8'); ?>
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<title>Format Text</title>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
$AllowRenderText=true; //Set to true only if this is in a secure environment, as directly outputting a given value can lead to XSS
	return print '</head><body>'.($AllowRenderText ? $_REQUEST['RenderText'] : 'Rendering of text not allowed').'</body></html>';
<style type="text/css">
html, body { width:100%; height:100%; margin:0; padding:0; }
.HalfScreen { display:block; width:calc(100% - 2px); height:calc(50% - 2px - 30px/2); margin:0; border:1px solid black; }
#RenderForm { overflow:hidden; }
#RenderText { margin:0; border:0; width:100%; height:100%; }
#RenderHTML { overflow-x:hidden; overflow-y:scroll; }
.TopBar { height:30px; background-color:grey; }
.Hide { position:absolute; visibility:hidden; top:-10000px; }

<script type="text/javascript" src="https://code.jquery.com/jquery-2.1.1.min.js"></script>
<script type="text/javascript">$(document).ready(function() {

//History for undoing
var UndoBuf=[], RedoBuf=[];
function Undo()
function Redo()
$('#Undo').click(function(e) { e.preventDefault(); Undo(); });
$('#Redo').click(function(e) { e.preventDefault(); Redo(); });

//Render HTML
function Render() {
	//Do the render
	var MyVal=$('#RenderText').val();

	//Save current value to the history
	//*Better history functionality here would be real nice (using smart currentTarget.selectionStart/End calculations), along with an undo/redo button, but not within the scope of this project
	if(RedoBuf.length) //Empty redo buffer
	if(UndoBuf.length>100) //Limit history buffer
$('#RenderText').on('keypress paste', function(e) { setTimeout(Render, 1); }); //Automatic update on paste requires a timeout

//Open in new page
$('#OpenInNewPage').click(function(e) {

//Escape HTML
$('#EscapeHTML').click(function(e) {
	$('#RenderText').val(function(index, value) {
		$.each({"&amp;":/&/g, "&lt;":/</g, "&gt;":/>/g, "&quot;":/"/g, "&#039;":/'/g}, function(HTMLStr, ReplStr) {
			value=value.replace(ReplStr, HTMLStr); });
		return value;

//Listize based on tabbing
//If a successive line is tabbed over beyond the current, it is made inside a new nested list.
//Tabbing over more than once on a successive line will create multiple nests
//Having @@@ at the beginning of a line will include it in the previous line item, no matter the tabbing
//Make sure to have @@@ blank lines tabbed over to the proper nested level
$('#Listize').click(function(e) {
	//Get the text to replace
	var T=$('#RenderText').val();

	//Go over each line and if the next line is tabbed beyond it, make it a new nested list. Blank
	var CurTabLevel=0, NewLines=[]; //NewLines is 2 items per line: the original string and the new html tags
	$.each(T.split(/\r?\n/), function(Index, Str) {
		//Check for a continued line item
		if(Str.substr(0, 3)=='@@@')
			return NewLines.push('<br>', Str.substr(3));

		//In/de-dent as needed
		var Tags='';
		var NewTabLevel=/^\t*/.exec(Str)[0].length, PreLevel=CurTabLevel; //Get the nested level

		//Fill out the rest of the line
		if(NewTabLevel==0) //Breaks between top level new lines
			Tags+=(Index && PreLevel==0 ? '<br>' : '');
		else if(PreLevel>=NewTabLevel) //If previous item needs to be ended (new level is not greater and not 0)

		NewLines.push(Tags, Str);

	//Finish de-dent as needed
	var Final=[NewLines.shift()];
	var EndLine='';

	//Combine each line with the tags
	for(var i=0;i<NewLines.length;i+=2)

	//Update from the replaced text


	<div class=TopBar>
		<input type=button id=EscapeHTML value="Escape HTML">
		<input type=button id=Listize value="Listize">
		<? if($AllowRenderText) { ?> <input type=button id=OpenInNewPage value="Open In New Page"> <? } ?>
		<input type=button id=Undo value="Undo">
		<input type=button id=Redo value="Redo">
	<form action="FormatText.php" method=post id=RenderForm target="_blank" class=HalfScreen>
		<textarea id=RenderText name=RenderText></textarea>
		<input type=submit class=Hide>
	<div id=RenderHTML class=HalfScreen></div>