AWS Ruby SDK 核心：将文件上传到 S3

Question

I want to upload a file (any file, could be a .txt, .mp4, .mp3, .zip, .tar ...etc) to AWS S3 using AWS-SDK-CORE ruby SDK

Here is my code:

require 'aws-sdk-core'
Aws.config = {
  :access_key_id => MY_ACCESS_KEY
  :secret_access_key => MY_SECRET_KEY,
  :region => 'us-west-2'
}

s3 = Aws::S3.new
resp = s3.put_object(
  :bucket => "mybucket",
  :key => "myfolder/upload_me.sql",
  :body => "./upload_me.sql"
)

Now, Above code runs and creates a key myfolder/upload_me.sql which has only one line written and that is ./upload_me.sql which is wrong. The file upload_me.sql has several lines.

Expected behaviour is to upload the file upload_me.sql on S3 as mybucket/myfolder/upload_me.sql . But instead it just writes one line to mybucket/myfolder/upload_me.sql and that is ./upload_me.sql

Now, If I omit the :body part as below:

s3 = Aws::S3.new
resp = s3.put_object(
  :bucket => "mybucket",
  :key => "myfolder/upload_me.sql",
)

Then it just creates and empty key called mybucket/myfolder/upload_me.sql which is not even downloadable (well, even if it gets downloaded, it is useless)

Could you point me where I am going wrong?

Here is ruby-SDK-core documentation for put_object Method: http://docs.aws.amazon.com/sdkforruby/api/Aws/S3/V20060301.html#put_object-instance_method

UPDATE: If I try to upload the same file using AWS-CLI, it gets uploaded fine. Here is the command:

aws s3api put-object --bucket mybucket --key myfolder/upload_me.sql --body ./upload_me.sql

Answer 1

So, After spending a frustrating sunday afternoon on htis issue, I finally cracked it. What I really needed is :body => IO.read("./upload_me.sql")

So my code looks like below:

s3 = Aws::S3.new
resp = s3.put_object(
  :bucket => "mybucket",
  :key => "myfolder/upload_me.sql",
  :body => IO.read("./upload_me.sql")
)

Answer 2

The body variable is the contents that will be written to S3. So if you send a file to S3 you need to manually load by using File.read("upload_me.sql") something similar.

s3 = Aws::S3.new
resp = s3.put_object(
  :bucket => "mybucket",
  :key => "myfolder/upload_me.sql",
  :body => File.read("./upload_me.sql")
)

According to the documentation another way to do this is to use write on the bucket.

s3 = AWS::S3.new
key = File.basename(file_name)
s3.buckets["mybucket"].objects[key].write(:file => "upload_me.sql")

Answer 3

Another way would be

AWS.config(
    :access_key_id => 'MY_ACCESS_KEY',
    :secret_access_key => 'MY_SECRET_KEY',
)

#Set the filename
file_name = 'filename.txt'

#Set the bucket name
s3_bucket_name = 'my bucket name'

#If file has to go in some specific folder
bucket_directory = 'key or folder'

begin
  s3 = AWS::S3.new

  #Check if directory name has provided and Make an object in your bucket for your upload
  if bucket_directory == ''
    bucket_obj = s3.buckets[s3_bucket_name].objects[bucket_directory]
  else
    bucket_obj = s3.buckets[s3_bucket_name].objects["#{bucket_directory}/#{file_name}"]
  end

  # Upload the file
  bucket_obj.write(:file => file_name)
  puts "File was successfully uploaded : #{bucket_obj}"
rescue Exception => e
  puts "There was an error in uploading file: #{e}"
end

Working Example

Reference

Answer 4

Probably the file wasn't found as the path is relative. This is a strange behavior, where the interface try to make too many decisions.

I can assure you this works (v3):

client = Aws::S3::Client.new(...)
client.put_object(
  body: './existing_file.txt',
  bucket: 'kick-it',
  key: 'test1.txt'
) # kick-it:/test1.txt contains the same as the contents of existing_file.txt

client.put_object(
  body: './non_existing_file.txt',
  bucket: 'kick-it',
  key: 'test2.txt'
) # kick-it:/test2.txt contains just the string './non_existing_file.txt'

Using body for both cases is a bad decision, if you ask me.