Direct Upload to Amazon S3 With CORS
Easily upload files directly to Amazon S3. Multi file uploading supported by jquery-fileupload.
Code extracted from Ryan Bates' gallery-jquery-fileupload.
Add this line to your application's Gemfile:
gem 's3_direct_upload'
Then add a new initalizer with your AWS credentials:
config/initializers/s3directupload.rb
ruby S3DirectUpload.config do |c| c.access_key_id = "" # your access key id c.secret_access_key = "" # your secret access key c.bucket = "" # your bucket name c.region = nil # region prefix of your bucket url. This is _required_ for the non-default AWS region, eg. "s3-eu-west-1" c.url = nil # S3 API endpoint (optional), eg. "https://#{c.bucket}.s3.amazonaws.com/" end
Make sure your AWS S3 CORS settings for your bucket look something like this:
xml http://0.0.0.0:3000 GET POST PUT 3000 *In production the AllowedOrigin key should be your domain.
Add the following js and css to your asset pipeline:
application.js.coffee ```coffeescript
**application.css** ```css //= require s3_direct_upload_progress_bars
Use the
s3_uploader_formhelper to add an s3 upload file field to your view:
ruby
Then in your application.js.coffee, call the S3Uploader jQuery plugin on the element you created above:
coffeescript jQuery -> $("#s3-uploader").S3Uploader()
Optionally, you can also place this template in the same view for the progress bars:
js+erb
callback_url:No default. The url that is POST'd to after file is uploaded to S3. If you don't specify this option, no callback to the server will be made after the file has uploaded to S3.
callback_method:Defaults to
POST. Use PUT and remove the multiple option from your file field to update a model.
callback_param:Defaults to
file. Parameter key for the POST to
callback_urlthe value will be the full s3 url of the file. If for example this is set to "model[imageurl]" then the data posted would be `model[imageurl] : http://bucketname.s3.amazonws.com/filename.ext`
server_side_encryption:Default to nothing. Specifies whether Server-Side encryption should be used to secure uploaded file. See
key:Defaults to
uploads/{timestamp}-{unique_id}-#{SecureRandom.hex}/${filename}. It is the key, or filename used on s3.
{timestamp},
{unique_id},
{extension}and
{cleaned_filename}are special substitution strings that will be populated by javascript with values for the current upload. {cleanedfilename} is the original filename with special characters removed.
${filename}is a special s3 string that will be populated with the original uploaded file name. Needs to be at least
"${filename}"or `"${cleanedfilename}"
. It is highly recommended to use both{unique_id}
, which will prevent collisions when uploading files with the same name (such as from a mobile device, where every photo is named image.jpg), and a server-generated random value such as#{SecureRandom.hex}`, which adds further collision protection with other uploaders.
key_starts_with:Defaults to
uploads/. Constraint on the key on s3. if you change the
keyoption, make sure this starts with what you put there. If you set this as a blank string the upload path to s3 can be anything - not recommended!
acl:Defaults to
public-read. The AWS acl for files uploaded to s3.
max_file_size:Defaults to
500.megabytes. Maximum file size allowed.
id:Optional html id for the form, its recommended that you give the form an id so you can reference with the jQuery plugin.
class:Optional html class for the form.
data:Optional html data attribute hash.
bucket:Optional (defaults to bucket used in config).
:val} do %>
It is recommended that you persist the url that is sent via the POST request (to the url given to the
callback_urloption and as the key given in the
callback_paramoption).
One way to do this is to make sure you have
resources modelin your routes file, and add a
s3_url(or something similar) attribute to your model. Then make sure you have the create action in your controller for that model that saves the url from the callback_param.
You could then have your create action render a javascript file like this: create.js.erb
rubySo that javascript code would be executed after the model instance is created, without a page refresh. See @rbates's gallery-jquery-fileupload) for an example of that method.alert("Failed to upload model: ");
$("#container").append("");
Note: the POST request to the rails app also includes the following parameters
filesize,
filetype,
filenameand
filepath.
Feel free to override the styling for the progress bars in s3directuploadprogressbars.css, look at the source for inspiration.
Also feel free to write your own js to interface with jquery-file-upload. You might want to do this to do custom validations on the files before it is sent to S3 for example. To do this remove
s3_direct_uploadfrom your application.js and include the necessary jquery-file-upload scripts in your asset pipeline (they are included in this gem automatically): ```cofeescript
Use the javascript in `s3_direct_upload` as a guide.Options for S3Upload jQuery Plugin
path:
manual path for the files on your s3 bucket. Example: path/to/my/files/on/s3
Note: Your path MUST start with the option you put in your form builder for key_starts_with
, or else you will get S3 permission errors. The file path in your s3 bucket will be path + key
.additional_data:
You can send additional data to your rails app in the persistence POST request. This would be accessible in your params hash as params[:key][:value]
Example: {key: value}
remove_completed_progress_bar:
By default, the progress bar will be removed once the file has been successfully uploaded. You can set this to false
if you want to keep the progress bar.remove_failed_progress_bar:
By default, the progress bar will not be removed when uploads fail. You can set this to true
if you want to remove the progress bar.before_add:
Callback function that executes before a file is added to the queue. It is passed file object and expects true
or false
to be returned. This could be useful if you would like to validate the filenames of files to be uploaded for example. If true is returned file will be uploaded as normal, false will cancel the upload.progress_bar_target:
The jQuery selector for the element where you want the progress bars to be appended to. Default is the form element.click_submit_target:
The jQuery selector for the element you wish to add a click handler to do the submitting instead of submiting on file open.```coffeescript jQuery -> $("#myS3Uploader").S3Uploader path: 'path/to/my/files/on/s3' additional_data: {key: 'value'} remove_completed_progress_bar: false before_add: myCallBackFunction # must return true or false if set progress_bar_target: $('.js-progress-bars') click_submit_target: $('.submit-target')
This demonstrates how to use progressbartarget and allowmultiplefiles (only works with false option - single file) to show only one progress bar without script template.
jQuery -> $("#myS3Uploader").S3Uploader progress_bar_target: $('.js-progress-bars') allow_multiple_files: false
Target for progress bar
You can change the settings on your form later on by accessing the jQuery instance:
jQuery -> v = $("#myS3Uploader").S3Uploader() ... v.path("new/path/") #only works when the key_starts_with option is blank. Not recommended. v.additional_data("newdata")
s3_uploads_startis fired once when any batch of uploads is starting.
coffeescript $('#myS3Uploader').bind 's3_uploads_start', (e) -> alert("Uploads have started")
When a file has been successfully uploaded to S3, the
s3_upload_completeis triggered on the form. A
contentobject is passed along with the following attributes :
urlThe full URL to the uploaded file on S3.
filenameThe original name of the uploaded file.
filepathThe path to the file (without the filename or domain)
filesizeThe size of the uploaded file.
filetypeThe type of the uploaded file.
This hook could be used for example to fill a form hidden field with the returned S3 url :
coffeescript $('#myS3Uploader').bind "s3_upload_complete", (e, content) -> $('#someHiddenField').val(content.url)
When an error occured during the transferm the
s3_upload_failedis triggered on the form with the same
contentobject is passed for the successful upload with the addition of the
error_thrownattribute. The most basic way to handle this error would be to display an alert message to the user in case the upload fails :
coffeescript $('#myS3Uploader').bind "s3_upload_failed", (e, content) -> alert("#{content.filename} failed to upload : #{content.error_thrown}")
When all uploads finish in a batch an
s3_uploads_completeevent will be triggered on
document, so you could do something like:
coffeescript $(document).bind 's3_uploads_complete', -> alert("All Uploads completed")
In addition, the regular rails ajax callbacks will trigger on the form with regards to the POST to the server.
$('#myS3Uploader').bind "ajax:success", (e, data) -> alert("server was notified of new file on S3; responded with '#{data}")
You may be processing the files upon upload and reuploading them to another bucket or directory. If so you can remove the originali files by running a rake task.
First, add the fog gem to your
Gemfileand run
bundle:
ruby gem 'fog'
Then, run the rake task to delete uploads older than 2 days:
$ rake s3_direct_upload:clean_remote_uploads Deleted file with key: "uploads/20121210T2139Z_03846cb0329b6a8eba481ec689135701/06 - PCR_RYA014-25.jpg" Deleted file with key: "uploads/20121210T2139Z_03846cb0329b6a8eba481ec689135701/05 - PCR_RYA014-24.jpg" $
Optionally customize the prefix used for cleaning (default is
uploads/#{2.days.ago.strftime('%Y%m%d')}): config/initalizers/s3directupload.rb
ruby S3DirectUpload.config do |c| # ... c.prefix_to_clean = "my_path/#{1.week.ago.strftime('%y%m%d')}" end
Alternately, if you'd prefer for S3 to delete your old uploads automatically, you can do so by setting your bucket's Lifecycle Configuration.
IE file uploads are working but with a couple caveats.
But IE should still upload your files fine.
This is just a simple gem that only really provides some javascript and a form helper. This gem could go all sorts of ways based on what people want and how people contribute. Ideas: * More specs! * More options to control file types, ability to batch upload. * More convention over configuration on rails side * Create generators. * Model methods. * Model method to delete files from s3
This gem is basically a small wrapper around code that Ryan Bates wrote for Railscast#383. Most of the code in this gem was extracted from gallery-jquery-fileupload.
Thank you Ryan Bates!
This code also uses the excellent jQuery-File-Upload, which is included in this gem by its rails counterpart jquery-fileupload-rails