Why does file uploaded to S3 have content type application/octet-stream unless I name the file .html?

Question

Even though I set content type to text/html it ends up as application/octet-stream on S3.

ByteArrayInputStream contentsAsStream = new ByteArrayInputStream(contentAsBytes); ObjectMetadata md = new ObjectMetadata(); md.setContentLength(contentAsBytes.length); md.setContentType("text/html"); s3.putObject(new PutObjectRequest(ARTIST_BUCKET_NAME, artistId, contentsAsStream, md));             

If however I name the file so that it ends up with .html

s3.putObject(new PutObjectRequest(ARTIST_BUCKET_NAME, artistId + ".html", contentsAsStream, md)); 

then it works.

Is my md object just being ignored? How can I get round this programmatically as over time I need to upload thousands of files so cannot just go into S3 UI and manually fix the contentType.

Answer 1

You must be doing something else in your code. I just tried your code example using the 1.9.6 S3 SDK and the file gets the "text/html" content type.

Here's the exact (Groovy) code:

class S3Test {     static void main(String[] args) {          def s3 = new AmazonS3Client()          def random = new Random()         def bucketName = "raniz-playground"         def keyName = "content-type-test"          byte[] contentAsBytes = new byte[1024]         random.nextBytes(contentAsBytes)          ByteArrayInputStream contentsAsStream = new ByteArrayInputStream(contentAsBytes);         ObjectMetadata md = new ObjectMetadata();         md.setContentLength(contentAsBytes.length);         md.setContentType("text/html");         s3.putObject(new PutObjectRequest(bucketName, keyName, contentsAsStream, md))          def object = s3.getObject(bucketName, keyName)         println(object.objectMetadata.contentType)         object.close()     } } 

The program prints

text/html

And the S3 metadata says the same:

Here are the communication sent over the net (courtesy of Apache HTTP Commons debug logging):

>> PUT /content-type-test HTTP/1.1 >> Host: raniz-playground.s3.amazonaws.com >> Authorization: AWS <nope> >> User-Agent: aws-sdk-java/1.9.6 Linux/3.2.0-84-generic Java_HotSpot(TM)_64-Bit_Server_VM/25.45-b02/1.8.0_45 >> Date: Fri, 12 Jun 2015 02:11:16 GMT >> Content-Type: text/html >> Content-Length: 1024 >> Connection: Keep-Alive >> Expect: 100-continue << HTTP/1.1 200 OK << x-amz-id-2: mOsmhYGkW+SxipF6S2+CnmiqOhwJ62WfWUkmZk4zU3rzkWCEH9P/bT1hUz27apmO << x-amz-request-id: 8706AE3BE8597644 << Date: Fri, 12 Jun 2015 02:11:23 GMT << ETag: "6c53debeb28f1d12f7ad388b27c9036d" << Content-Length: 0 << Server: AmazonS3  >> GET /content-type-test HTTP/1.1 >> Host: raniz-playground.s3.amazonaws.com >> Authorization: AWS <nope> >> User-Agent: aws-sdk-java/1.9.6 Linux/3.2.0-84-generic Java_HotSpot(TM)_64-Bit_Server_VM/25.45-b02/1.8.0_45 >> Date: Fri, 12 Jun 2015 02:11:23 GMT >> Content-Type: application/x-www-form-urlencoded; charset=utf-8 >> Connection: Keep-Alive << HTTP/1.1 200 OK << x-amz-id-2: 9U1CQ8yIYBKYyadKi4syaAsr+7BV76Q+5UAGj2w1zDiPC2qZN0NzUCQNv6pWGu7n << x-amz-request-id: 6777433366DB6436 << Date: Fri, 12 Jun 2015 02:11:24 GMT << Last-Modified: Fri, 12 Jun 2015 02:11:23 GMT << ETag: "6c53debeb28f1d12f7ad388b27c9036d" << Accept-Ranges: bytes << Content-Type: text/html << Content-Length: 1024 << Server: AmazonS3 

And this is also the behaviour that looking at the source code shows us - if you set the content type the SDK won't override it.