cos_stream_schema_examples.py

# Define input data schema as a proper Python list (not a string)
input_schema = [
    {
        'id': '1',
        'type': 'struct',
        'fields': [
            {
                'name': 'cos_config', 
                'type': 'object', 
                'nullable': False, 
                'metadata': {
                    'description': 'Cloud Object Storage configuration'
                }
            },
            {
                'name': 'source_urls', 
                'type': 'array', 
                'nullable': False, 
                'metadata': {
                    'description': 'URLs of files to download and upload to COS'
                }
            },
            {
                'name': 'prefix', 
                'type': 'string', 
                'nullable': True, 
                'metadata': {
                    'description': 'Optional prefix to add to the file names in COS'
                }
            },
            {
                'name': 'http_method', 
                'type': 'string', 
                'nullable': True, 
                'metadata': {
                    'description': 'HTTP method to use for downloading files'
                }
            }
        ]
    }
]

# Define output data schema as a proper Python list (not a string)
output_schema = [
    {
        'id': '1',
        'type': 'struct',
        'fields': [
            {
                'name': 'status', 
                'type': 'string', 
                'nullable': False, 
                'metadata': {
                    'description': 'Status of the operation (success or error)'
                }
            },
            {
                'name': 'data', 
                'type': 'object', 
                'nullable': True, 
                'metadata': {
                    'description': 'Response data containing upload details'
                }
            },
            {
                'name': 'message', 
                'type': 'string', 
                'nullable': True, 
                'metadata': {
                    'description': 'Error message if status is error'
                }
            }
        ]
    }
]

# Define sample scoring input
sample_input = {
    'input_data': [
        {
            'fields': ['cos_config', 'source_urls', 'prefix', 'http_method'],
            'values': [
                [
                    {
                        'bucket_name': 'my-bucket',
                        'api_key': '<your-cos-api-key>',
                        'instance_id': '<your-cos-instance-id>',
                        'auth_endpoint': 'https://iam.cloud.ibm.com/identity/token',
                        'endpoint_url': 'https://s3.eu-de.cloud-object-storage.appdomain.cloud'
                    },
                    [
                        'https://example.com/sample-file.pdf', 
                        'https://example.com/another-file.csv'
                    ],
                    'uploads/files',
                    'GET'
                ]
            ]
        }
    ]
}