Alibaba Cloud OSS SDK for Ruby is a Ruby client program for convenient access to Alibaba Cloud OSS (Object Storage Service) RESTful APIs. For more information about OSS, visit the OSS official website.

• Ruby 2.0 or above. For Ruby 1.9, please use v0.5.0.
• Windows, Linux or OS X system that supports Ruby.

To learn how to install Ruby, refer to: ruby-lang.

Log onto the official website and click Activate Now. Follow the prompts to activate OSS. After the service is activated, go to Console to view your AccessKeyId and AccessKeySecret. These two information items are required when you use Alibaba Cloud OSS SDK.

gem install aliyun-sdk

Include the following in your project or 'irb' command:

require 'aliyun/oss'

Note:

• Some gems on which the SDK depends are local extensions, and you need to install ruby-dev to compile locally extended gems after you install Ruby.
• The environment for running the SDK-dependent gem (nokogiri) for processing XML must have the zlib library.

The following method is used to install the preceding dependencies taking Ubuntu as an example:

sudo apt-get install ruby-dev
sudo apt-get install zlib1g-dev

The practices for other systems are similar.

client = Aliyun::OSS::Client.new(
  :endpoint => 'endpoint',
  :access_key_id => 'access_key_id',
  :access_key_secret => 'access_key_secret')

In specific, the endpoint is the OSS service address. The address may vary based on different regions for the node. For example, the address for a Hangzhou node is: http://oss-cn-hangzhou.aliyuncs.com. For addresses for other nodes, see: Node List.

The access_key_id and access_key_secret are credentials for your service. You can view them in Console on the official website. Please keep your AccessKeySecret safe. Disclosing the AccessKeySecret may compromise your data security.

OSS supports binding a custom domain name and allows you to direct your domain name to the OSS service address (CNAME) of Alibaba Cloud. In this way, you don't need to change the resource path in your app when migrating your data to the OSS. The bound domain name points to a bucket in the OSS. The domain name binding operation can only be carried out in the OSS console. For more information about binding a custom domain name, visit the official website: Binding Custom Domain Names in OSS.

After you have bound a custom domain name, you can use the standard OSS service address as the specified endpoint of the OSS, or use the bound domain name:

client = Aliyun::OSS::Client.new(
  :endpoint => 'http://img.my-domain.com',
  :access_key_id => 'access_key_id',
  :access_key_secret => 'access_key_secret',
  :cname => true)

Note:

• You must set the cname to true when initializing the client.
• The custom domain name is bound to a bucket of the OSS, so the client created in this method does not support List_buckets operations.
• You still need to specify the bucket name during the {Aliyun::OSS::Client#get_bucket} operation and the bucket name should be the same as that bound to the domain name.

OSS supports access via STS. For more information about STS, refer to Alibaba Cloud STS. Before using STS, you must apply for a temporary token from the STS. Alibaba Cloud Ruby SDK contains the STS SDK, and you only need to require 'aliyun/sts' for usage:

require 'aliyun/sts'
sts = Aliyun::STS::Client.new(
  access_key_id: 'access_key_id',
  access_key_secret: 'access_key_secret')

token = sts.assume_role('role-arn', 'my-app')

client = Aliyun::OSS::Client.new(
  :endpoint => 'http://oss-cn-hangzhou.aliyuncs.com',
  :access_key_id => token.access_key_id,
  :access_key_secret => token.access_key_secret,
  :sts_token => token.security_token)

Note: the :sts_token parameter must be specified for using STS. You can also apply for a token with a policy through STS::Client, for details, refer to API Documentation.

buckets = client.list_buckets
buckets.each{ |b| puts b.name }

The list_buckets command returns an iterator for you to get the information of each bucket in order. Bucket For the object structure, see {Aliyun::OSS::Bucket} in the API documentation.

bucket = client.create_bucket('my-bucket')

bucket = client.get_bucket('my-bucket')
objects = bucket.list_objects
objects.each{ |o| puts o.key }

The list_objects command returns an iterator for you to get the information of each object in order. Object For the object structure, see {Aliyun::OSS::Object} in the API documentation.

bucket.put_object(object_key){ |stream| stream << 'hello world' }

You can also create an object by uploading a local file:

bucket.put_object(object_key, :file => local_file)

bucket.get_object(object_key){ |content| puts content }

You can also download the object to a local file:

bucket.get_object(object_key, :file => local_file)

bucket.copy_object(from_key, to_key)

bucket.object_exists?(object_key)

For more operations on buckets, refer to {Aliyun::OSS::Bucket} in the API documentation.

OSS is a storage service for objects and does not support the directory structure. All objects are flatly structured. But you can simulate the directory structure by setting the object key in the format "foo/bar/file". Suppose there are several objects as follows:

foo/x
foo/bar/f1
foo/bar/dir/file
foo/hello/file

Listing all the objects under the "foo/" directory means to perform the list_objects operation with "foo/" as the prefix. But this method will also list all the objects under "foo/bar/". That's why we need the delimiter parameter. This parameter means to stop processing at the first delimiter after the prefix. The key during the process acts as the common prefix of objects, objects with the prefix will be included in the list_objects result.

objs = bucket.list_objects(:prefix => 'foo/', :delimiter => '/')
objs.each do |i|
  if i.is_a?(Aliyun::OSS::Object) # a object
    puts "object: #{i.key}"
  else
    puts "common prefix: #{i}"
  end
end
# output
object: foo/x
common prefix: foo/bar/
common prefix: foo/hello/

Common prefixes free you from traversing all the objects (the number of objects may be huge) to determine the prefix, and is quite helpful for simulating the directory structure.

You can specify a callback for put_object and resumable_upload so that after the file is successfully uploaded to the OSS, the OSS will initiate an HTTP POST request to the server address you provided to notify you that the corresponding event has occurred. You can perform desired actions after receiving the notification, such as updating the database and making statistics. For more details about upload callback, refer to OSS Upload Callback.

The example below demonstrates how to use the upload callback:

callback = Aliyun::OSS::Callback.new(
  url: 'http://10.101.168.94:1234/callback',
  query: {user: 'put_object'},
  body: 'bucket=${bucket}&object=${object}'
)

begin
  bucket.put_object('files/hello', callback: callback)
rescue Aliyun::OSS::CallbackError => e
  puts "Callback failed: #{e.message}"
end

Note:

• The callback URL must not contain the query string which must be specified in the :query parameter.
• In the event that the file is successfully uploaded but callback execution fails, the client will throw CallbackError. To ignore the error, you need to explicitly catch the exception.
• For detailed examples, refer to callback.rb.
• For servers that support callback, refer to callback_server.rb.

OSS supports the storage of large objects. If the upload/download task of a large object is interrupted (due to network transient disconnections, program crashes, or machine power-off), the re-upload/re-download is taxing on system resources. The OSS supports multipart upload/download to divide a large object into multiple parts for upload/download. Alibaba Cloud OSS SDK provides the resumable upload/download feature based on this principle. If an interruption occurs, you can resume the upload/download task beginning with the interrupted part. Resumable upload/download is recommended for objects larger than 100MB.

bucket.resumable_upload(object_key, local_file, :cpt_file => cpt_file)

In specific, :cpt_file specifies the location of the checkpoint object which stores the intermediate state of the upload. If no object is specified, the SDK will generate a local_file.cpt in the directory of the local_file. After the upload interruption, you only need to provide the same cpt object for the upload task to resume from the interrupted part. The typical upload code is:

retry_times = 5
retry_times.times do
  begin
    bucket.resumable_upload(object_key, local_file)
  rescue => e
    logger.error(e.message)
  end
end

Notes:

• The SDK records the upload intermediate states in the cpt object. Therefore, ensure that you have write permission on the cpt object.
• The cpt object records the intermediate state information of the upload and has a self-checking function. You cannot edit the object. Upload will fail if the cpt object is corrupted. When the upload is completed, the checkpoint file will be deleted.

bucket.resumable_download(object_key, local_file, :cpt_file => cpt_file)

In specific, :cpt_file specifies the location of the checkpoint object which stores the intermediate state of the download. If no object is specified, the SDK will generate a local_file.cpt in the directory of the local_file. After the download interruption, you only need to provide the same cpt object for the download task to resume from the interrupted part. The typical download code is:

retry_times = 5
retry_times.times do
  begin
    bucket.resumable_download(object_key, local_file)
  rescue => e
    logger.error(e.message)
  end
end

Notes:

• During the download process, a temporary object of local_file.part.N will be generated in the directory of the local_file for each part downloaded. When the download is completed, the objects will be deleted. You cannot edit or delete the part objects, otherwise the download will not proceed.
• The SDK records the download intermediate states in the cpt object; therefore, ensure that you have write permission on the cpt object.
• The cpt object records the intermediate state information of the download and has a self-checking function. You cannot edit the object. Download will fail if the cpt object is corrupted. When the download is completed, the checkpoint object will be deleted.

Objects in Alibaba Cloud OSS can be divided into two types: Normal and Appendable.

• A normal object functions as a whole for every upload. If an object already exists, the later uploaded object will overwrite the previous object with the same key.
• An appendable object is created through append_object for the first time. The later uploaded object through append_object will not overwrite the previous one, but will append content to the end of the object.
• You cannot append content to a normal object.
• You cannot copy an appendable object.

bucket.append_object(object_key, 0){ |stream| stream << "hello world" }

The second parameter indicates the position to append the content. This parameter is 0 for the first append to the object. In later append operations, the value of this parameter is the length of the object before the append.

Of course, you can also read the appended content from the object:

bucket.append_object(object_key, 0, :file => local_file)

pos = bucket.get_object(object_key).size
next_pos = bucket.append_object(object_key, pos, :file => local_file)

During the first append, you can use {Aliyun::OSS::Bucket#get_object} to get the object length. For later append operations, you can refer to the response of {Aliyun::OSS::Bucket#append_object} to determine the length value for next append.

Note: Concurrent append_object and next_pos operations do not always produce correct results.

Besides the object content, the OSS also allows you to set some meta information for the object during object uploading. The meta information is a key-value pair to identify the specific attributes of the object. The meta information will be stored together with the object and returned to users in get_object and get_object_meta operations.

bucket.put_object(object_key, :file => local_file,
                  :metas => {
                    'key1' => 'value1',
                    'key2' => 'value2'})

obj = bucket.get_object(object_key, :file => localfile)
puts obj.metas

Note:

• The key and value of the meta information can only be simple ASCII non-newline characters and the total size must not exceed 8KB.
• In the copy object operation, the meta information of the source object will be copied by default. If you don't want this, explicitly set the :meta_directive to {Aliyun::OSS::MetaDirective::REPLACE}.

OSS allows you to set access permissions for buckets and objects respectively, so that you can conveniently control external access to your resources. A bucket is enabled with three types of access permissions:

• public-read-write: Anonymous users are allowed to create/retrieve/delete objects in the bucket.
• public-read: Anonymous users are allowed to retrieve objects in the bucket.
• private: Anonymous users are not allowed to access the bucket. Signature is required for all accesses.

When a bucket is created, the private permission applies by default. You can use 'bucket.acl=' to set the ACL of the bucket.

bucket.acl = Aliyun::OSS::ACL::PUBLIC_READ
puts bucket.acl # public-read

An object is enabled with four types of access permissions:

• default: The object inherits the access permissions of the bucket it belongs to, that is, the access permission of the object is the same as that of the bucket where the object is stored.
• public-read-write: Anonymous users are allowed to read/write the object.
• public-read: Anonymous users are allowed to read the object.
• private: Anonymous users are not allowed to access the object. Signature is required for all accesses.

When an object is created, the default permission applies by default. You can use 'bucket.set_object_acl' to configure the ACL of the object.

acl = bucket.get_object_acl(object_key)
puts acl # default
bucket.set_object_acl(object_key, Aliyun::OSS::ACL::PUBLIC_READ)
acl = bucket.get_object_acl(object_key)
puts acl # public-read

Notes:

• If an object is configured with an ACL policy, the object ACL takes priority during permission authentication when the object is accessed. The bucket ACL will be ignored.

• If anonymous access is allowed (public-read or public-read-write is configured for the object), you can directly access the object using a browser. For example,

    http://bucket-name.oss-cn-hangzhou.aliyuncs.com/object.jpg
  

• A bucket or an object with the public permission can be accessed by an anonymous client which is created with the following code:

    # If access_key_id and access_key_secret are not specified, an anonymous client will be created. The client can only access 
    # the buckets and objects with the public permission.
    client = Client.new(:endpoint => 'oss-cn-hangzhou.aliyuncs.com')
    bucket = client.get_bucket('public-bucket')
    obj = bucket.get_object('public-object', :file => local_file)
  

Some example projects are provided in the examples/ directory of the SDK to demonstrate the SDK features. You can run the examples after some configuration. The permission information and the bucket information required by the examples are available in the ~/.oss.yml configuration file under the HOME directory. The information should include the following fields (Note the space after the colon):

endpoint: oss-cn-hangzhou.aliyuncs.com
cname: false
access_key_id: <ACCESS KEY ID>
access_key_secret: <ACCESS KEY SECRET>
bucket: <BUCKET NAME>

You need to create (if not in existence) or modify the content and run the example project:

ruby examples/aliyun/oss/bucket.rb

bundle exec rake spec

export RUBY_SDK_OSS_ENDPOINT=endpoint
export RUBY_SDK_OSS_ID=AccessKeyId
export RUBY_SDK_OSS_KEY=AccessKeySecret
export RUBY_SDK_OSS_BUCKET=bucket-name

bundle exec rake test

• MIT

For more documentation, see:

• Alibaba Cloud OSS Ruby SDK documentation.
• Alibaba Cloud OSS documentation.