I recently had a need to sanatize html entered by users in an app. The
app (www.csepub.com) allowed
professors to enter homework assignments for their ebooks. One type of
homework assignment is a written assignment. The app needed to allow
the user/student to create and submit an assignment. At first we were
using uploaded documents, but with thousands of students, this became a
heap of uploaded word docs. Recently we decided to allow the student to
submit their assignment via html using an html editor (trimmed way down
to just basic text formatting).
My fear was that allowing this many people to enter html into my
database, that something might get wonked up. I'm not a security
expert, but I figured I should be sanitizing the html.
After a bit of research I found the sanitize gem (https://github.com/rgrove/sanitize).
It does 2 basic things: 1) clean the html of unwanted tags and closes
any open tags. Hopefully this won't be a problem for me because I'm
preventing the user from editing actual html with a slimmed down html
editor, but I don't want to take a chance.
The gem is pretty straight forward. Just place it in your gem file and then bundle.
Next, in the AR model you want to store your html in, I added a 'before_save' hook.
The biggest decision you have to make is which tags you want to
whitelist. Sanitize has several pre built levels for you to use.
Sanitize::Config::BASIC looked like the option for me. However, I
realized that I needed to allow a span tag with a inline style
(professors wanted to make notes to the assignments and highlight text).
I didn't want to recreate the pre setup from scratch, so I figured out
Sanitize::Config::BASIC is just a hash so I just slightly modified it.
If you're using rails and have an integer column in database that you want to increment or decrement, rails provides a nice abstraction for you to use. Normally you'd have do something like this:
It allows you to get a unique set of results from an active record
relation.
Ruby provides a way to get unique results from an array like so:
a=[1,2,3,3,4]a.uniq# results [1, 2, 3, 4]
But if you have an active record relation, the normal uniq won't work.
Say you have a list of stores and you want to find their location, but
you only want 1 location from each store/company. You can use:
Location.near("monroe, ga").uniq_by(&:store_id)
And that will give you a nice unique Array from an AR relation!
Several monts ago, John Nunemaker bloged "Stop Googling". It was
about reading the code in an open source project oppsed to Googleing it.
He was a bit funstrated at someone who Googled, polled friends, and
basically searched everywhere besides the source to look for an answer
for their problem.
Today I found out the hard way how right he was. I was trying to access
the object info passed along to a delayed_job record. I googled for 30
or 45 minutes trying to find an answer and couldn't get any bites. I
knew the handler field had the info I needed, but it was a string. I
also knew that somehow that info had to be extracted when my job was
processed. Finally on a whim, I looked to the source code. In the
second file I opened, I found a method called payload_object.
I saw object, deserialize, and handler. That's all I needed to know.
It was exactly what I wanted. It took less than a minute. I tried it
out and that was it.
That's when I remembered John's blog. Note to self: if a quick google
search doesn't display the obvious answer...LOOK AT THE
CODE!
Ever since I started learning Ruby On Rails several years ago, I’ve been
using git. Git is a cool tool you can use for version control, or in
some cases, simple incremental backups. It basically keeps track of
your code and file changes, you commit/save the changes and can revert
back, review history and do all sorts of cool stuff. It additionally
will allow you to setup a remote repository to push changes to. That’s
where this article comes in.
Using git in rails is really a must, but in the static world, FTP mostly
dominates.
I’m really tired of using FTP. I probably hit Command + Shift + Upload
200 times a day. Sometimes I edit files very quickly and hop from file
to file, add images to a images directory and javascript files to the js
directory. How can I remember which files to FTP? It can be as simple
as “git push” and only my changes are updated.
First off, requirements:
web server with shell access
web server with git installed
local machine with a terminal
local machine with git installed
local machine to test (if you can’t run your code on your local
machine or local server, it doesn’t make much sense to setup git...it
would be painful to add, commit, push every tiny change)
Get Started
Server Side
Take care of the server side first. ssh into your web server and find a
place outside your public folder. For example, a cpanel server is
/home/account_name/public_html. I’d store it in /home/account_name.
mkdir account_name.git && cd account_name.git
git init --bare
Now we’ll tell git where to store the actual files:
Now we have to tell git what to do after we push changes. Create and
open a new file:
vim hooks/post-receive
In this file place this code:
#!/bin/sh#tell git to copy over the files
git checkout -f#change ownership of files so your web server can serve them (I’m logging in as root so it may not be necessary if you logging in as account_namechown-R account_name:account_name /home/account_name/public_html/*
Note: This file (post-receive) needs to be executable. To be sure:
chmod 755 hooks/post-receive
Local Side
Navigate to the directory where your website code lives. If you haven’t already setup git:
Your website should now be live. You can now make changes locally, test
and then push. Example:
#...make some changes to your code
# test them (this could be a visual QA test or automated test)
git add .
git commit -m ‘my first changes’
git push #no need for the origin master now