AWS Lambda, saving S3 file to /tmp directory

Question

I want to copy a set of files over from S3, and put them in the /tmp directory while my lambda function is running, to use and manipulate the contents. The following code excerpt works fine on my PC (which is running windows)

s3 = boto3.resource('s3')
BUCKET_NAME = 'car_sentiment'
keys = ['automated.csv', 'connected_automated.csv', 'connected.csv', 
        'summary.csv']
for KEY in keys: 
    try:
        local_file_name = 'tmp/'+KEY
        s3.Bucket(BUCKET_NAME).download_file(KEY, local_file_name)
    except botocore.exceptions.ClientError as e:
        if e.response['Error']['Code'] == "404":
            continue
        else:
            raise

However, when I try to run on AWS lambda, I get:

{
  "errorMessage": "[Errno 2] No such file or directory: 'tmp/automated.csv.4Bcd0bB9'",
  "errorType": "FileNotFoundError",
  "stackTrace": [
    [
      "/var/task/SentimentForAWS.py",
      28,
      "my_handler",
      "s3.Bucket(BUCKET_NAME).download_file(KEY, local_file_name)"
    ],
    [
      "/var/runtime/boto3/s3/inject.py",
      246,
      "bucket_download_file",
      "ExtraArgs=ExtraArgs, Callback=Callback, Config=Config)"
    ],
    [
      "/var/runtime/boto3/s3/inject.py",
      172,
      "download_file",
      "extra_args=ExtraArgs, callback=Callback)"
    ],
    [
      "/var/runtime/boto3/s3/transfer.py",
      307,
      "download_file",
      "future.result()"
    ],
    [
      "/var/runtime/s3transfer/futures.py",
      73,
      "result",
      "return self._coordinator.result()"
    ],
    [
      "/var/runtime/s3transfer/futures.py",
      233,
      "result",
      "raise self._exception"
    ],
    [
      "/var/runtime/s3transfer/tasks.py",
      126,
      "__call__",
      "return self._execute_main(kwargs)"
    ],
    [
      "/var/runtime/s3transfer/tasks.py",
      150,
      "_execute_main",
      "return_value = self._main(**kwargs)"
    ],
    [
      "/var/runtime/s3transfer/download.py",
      582,
      "_main",
      "fileobj.seek(offset)"
    ],
    [
      "/var/runtime/s3transfer/utils.py",
      335,
      "seek",
      "self._open_if_needed()"
    ],
    [
      "/var/runtime/s3transfer/utils.py",
      318,
      "_open_if_needed",
      "self._fileobj = self._open_function(self._filename, self._mode)"
    ],
    [
      "/var/runtime/s3transfer/utils.py",
      244,
      "open",
      "return open(filename, mode)"
    ]
  ]
}

Why does it think the file name is tmp/automated.csv.4Bcd0bB9 rather than just tmp/automated.csv and how do I fix it? Been pulling my hair out on this one, trying multiple approaches, some of which generate a similar error when running locally on my PC. Thanks!

Answer 1

You should save in /tmp, rather than tmp/.

eg:

local_file_name = '/tmp/' + KEY