def __getitem__()

in utils.py [0:0]

• 14 lines of code
• 10 McCabe index (conditional complexity)

    def __getitem__(self, index):
        # grab a uri and label
        uri = self.uris[index]
        label = self.labels[index]

        # check if URI looks like an s3 key.  if so, download key from s3.
        if uri[:5] == "s3://":
            # init s3 client if not yet done
            if self.s3_client is None:
                self.s3_client = boto3.Session().client('s3')

            # split s3 uri into bucket and key
            uri_parsed = urlparse(uri, allow_fragments=False)
            bucket, key = uri_parsed.netloc, uri_parsed.path[1:]

            # download sample from s3
            sample_bytes = self.s3_client.get_object(Bucket=bucket, Key=key, RequestPayer='requester')["Body"].read()
        # otherwise, open from filesystem
        else:
            sample_bytes = open(uri, "rb").read()

        # if the sample is gzip'd, decompress it.
        if sample_bytes[:2] == b"\x1f\x8b":
            sample_bytes = gzip.decompress(sample_bytes)

        return sample_bytes, uri, float(label)