CODEWITHSUNDEEP
bashamazon-web-servicesamazon-s3
J. Swaelen
J. Swaelen

Reputation: 818

AWS S3: How to check if a file exists in a bucket using bash

I'd like to know if it's possible to check if there are certain files in a certain bucket.

This is what I've found:

Checking if a file is in a S3 bucket using the s3cmd

It should fix my problem, but for some reason it keeps returning that the file doesn't exist, while it does. This solution is also a little dated and doesn't use the doesObjectExist method.

Summary of all the methods that can be used in the Amazon S3 web service

This gives the syntax of how to use this method, but I can't seem to make it work.

Do they expect you to make a boolean variable to save the status of the method, or does the function directly give you an output / throw an error?

This is the code I'm currently using in my bash script:

existBool=doesObjectExist(${BucketName}, backup_${DomainName}_${CurrentDate}.zip)

if $existBool ; then
        echo 'No worries, the file exists.'
fi

I tested it using only the name of the file, instead of giving the full path. But since the error I'm getting is a syntax error, I'm probably just using it wrong.

Hopefully someone can help me out and tell me what I'm doing wrong.

!Edit

I ended up looking for another way to do this since using doesObjectExist isn't the fastest or easiest.

Upvotes: 68

Views: 134103

Answers (11)

ItayB
ItayB

Reputation: 11337

Following to @DaveMaple & @MichaelGlenn answers, here is the condition I'm using:

aws s3api head-object --bucket <some_bucket> --key <some_key> || not_exist=true
if [ $not_exist ]; then
  echo "it does not exist"
else
  echo "it exists"
fi

EDIT (Dec-23):

for hiding the other messages (of aws) you can send both STDOUT/STDERR to /dev/null as suggested in the comments:

aws s3api head-object --bucket <some_bucket> --key <some_key> > /dev/null 2>&1 || not_exist=true
if [ $not_exist ]; then
  echo "it does not exist"
else
  echo "it exists"
fi

Upvotes: 51

Sandy
Sandy

Reputation: 1054

From awscli, we do a ls along with a grep, example

aws s3 ls s3://<bucket_name> | grep 'filename'

This can be included in the bash script.

Upvotes: 2

Damian Kober
Damian Kober

Reputation: 1

the cheapest way I found is

if aws s3 ls s3://mybucket
then  
    echo "exists"
else
    echo "does not exist"
fi

Upvotes: -2

Elifarley
Elifarley

Reputation: 1420

Here's a simple POSIX shell function (so it also works in Bash) based on @Dmitri Orgonov's answer:

s3_key_exists() {
  aws >/dev/null 2>&1 s3api head-object --bucket "$1" --key "$2"
  test $? != 254
}

And here's how to use it:

s3_key_exists myBucket path/to/my/file.txt \
  && echo "It's there!" \
  || echo "Not found..."

Now, if what you have is an S3 path instead of a bucket and a key:

s3_file_exists() {
  local bucketAndKey="$(s3_bucket_and_key "$1")"
  s3_key_exists "${bucketAndKey%:*}" "${bucketAndKey#*:}"
}
s3_bucket_and_key() {
  local input="${1#/}"; local bucket="${input%%/*}"; local key="${input#$bucket}"
  echo "$bucket:${key#/}"
}

And here's a usage example:

s3_file_exists /myBucket/path/to/my/file.txt \
  && echo "It's there!" \
  || echo "Not found..."

Or...

s3_file_exists myBucket/path/to/my/other-file.txt \
  && echo "It's there too!" \
  || echo "Not found either..."

Upvotes: 0

Soundararajan
Soundararajan

Reputation: 2194

A simpler solution, but not as sophisticated as other aws s3 api's is to use the exit code

aws s3 ls <full path to object>

Returns a non-zero return code if the object doesn't exist. 0 if it exists.

Upvotes: 4

icpp-pro
icpp-pro

Reputation: 1456

This statement will return a true or false response:

aws s3api list-objects-v2 \
  --bucket <bucket_name> \
  --query "contains(Contents[].Key, '<object_name>')"

So, in case of the example provided in the question:

aws s3api list-objects-v2 \
  --bucket ${BucketName} \
  --query "contains(Contents[].Key, 'backup_${DomainName}_${CurrentDate}.zip')"

I like this approach, because:

  • The --query option uses the JMESPath syntax for client-side filtering and it is well documented here how to use it.

  • Since the --query option is build into the aws cli, no additional dependencies need to be installed.

  • You can first run the command without the --query option, like:

      aws s3api list-objects-v2 --bucket <bucket_name>

    That returns a nicely formatted JSON, something like:

{
    "Contents": [
        {
            "Key": "my_file_1.tar.gz",
            "LastModified": "----",
            "ETag": "\"-----\"",
            "Size": -----,
            "StorageClass": "------"
        },
        {
            "Key": "my_file_2.txt",
            "LastModified": "----",
            "ETag": "\"----\"",
            "Size": ----,
            "StorageClass": "----"
        },
        ...
    ]
}

  • This then allows you to design an appropriate query. In this case you want to check if the JSON contains a list Contents and that an item in that list has a Key equal to your file (object) name:

    --query "contains(Contents[].Key, '<object_name>')"

Upvotes: 6

Colin
Colin

Reputation: 1142

Inspired by the answers above, I use this to also check the file size, because my bucket was trashed by some script with a 404 answers. It requires jq tho.

minsize=100
s3objhead=$(aws s3api head-object \
  --bucket "$BUCKET" --key "$KEY" 
  --output json || echo '{"ContentLength": 0}')

if [ $(printf "%s" "$s3objhead" | jq '.ContentLength') -lt "$minsize" ]; then
  # missing or small
else
  # exist and big
fi

Upvotes: 2

Amri
Amri

Reputation: 1100

I usually use set -eufo pipefail and the following works better for me because I do not need to worry about unset variables or the entire script exiting.

object_exists=$(aws s3api head-object --bucket $bucket --key $key || true)
if [ -z "$object_exists" ]; then
  echo "it does not exist"
else
  echo "it exists"
fi

Upvotes: 8

Michael Glenn
Michael Glenn

Reputation: 421

Note that "aws s3 ls" does not quite work, even though the answer was accepted. It searches by prefix, not by a specific object key. I found this out the hard way when someone renamed a file by adding a '1' to the end of the filename, and the existence check would still return True.

(Tried to add this as a comment, but do not have enough rep yet.)

Upvotes: 29

traceformula
traceformula

Reputation: 391

One simple way is using aws s3 ls

exists=$(aws s3 ls $path_to_file)
if [ -z "$exists" ]; then
  echo "it does not exist"
else
  echo "it exists"
fi

Upvotes: 8

Dave Maple
Dave Maple

Reputation: 8402

Last time I saw performance comparisons getObjectMetadata was the fastest way to check if an object exists. Using the AWS cli that would be the head-object method, example:

aws s3api head-object --bucket www.codeengine.com --key index.html

which returns:

{
    "AcceptRanges": "bytes",
    "ContentType": "text/html; charset=utf-8",
    "LastModified": "Sun, 08 Jan 2017 22:49:19 GMT",
    "ContentLength": 38106,
    "ContentEncoding": "gzip",
    "ETag": "\"bda80810592763dcaa8627d44c2bf8bb\"",
    "StorageClass": "REDUCED_REDUNDANCY",
    "CacheControl": "no-cache, no-store",
    "Metadata": {}
}

Upvotes: 75

Related Questions