AgentsforBedrock.Client.create_data_source hierarchal chunking configuration bug

[TamaryOmer]

Describe the issue

Hey,

TL;DR: missing parameter in gentsforBedrock.Client.create_data_source.vectorIngestionConfiguration. hierarchicalChunkingConfiguration docs

I tried to used AgentsforBedrock.create_data_source with hierarchal chunking strategy with the default request Syntax and get an error regarding the vectorIngestionConfiguration. I notice that the request requires a token configuration for both father and son chunks, for example:

`

    'chunkingConfiguration': {
        'chunkingStrategy': 'HIERARCHICAL',
        'hierarchicalChunkingConfiguration': {
            'levelConfigurations': [{
                    "maxTokens": 1000
                },
                {
                    "maxTokens": 300
                }
            ],
            'overlapTokens': 100
        }
    }
    },

`

While in the docs the there is a no reference to the child chunks maxTokens:

`

    'chunkingConfiguration': {
        'chunkingStrategy': 'FIXED_SIZE'|'NONE'|'HIERARCHICAL'|'SEMANTIC',
        'fixedSizeChunkingConfiguration': {
            'maxTokens': 123,
            'overlapPercentage': 123
        },
        'hierarchicalChunkingConfiguration': {
            'levelConfigurations': [
                {
                    'maxTokens': 123
                },
            ],
            'overlapTokens': 123
        }...

`

Thanks!

Links

https://boto3.amazonaws.com/v1/documentation/api/latest/reference/services/bedrock-agent/client/create_data_source.html

[amberkushwaha]

describe the issue in it for the particular request.code of conduct in it.

[TamaryOmer]

I'm not sure I fully understand what you mean.

[tim-finnigan]

Thanks for reaching out. (You can ignore comments from amberkushwaha who has been posting spam — I've reported them to GitHub.) The create_data_source command makes a request to the underlying CreateDataSource API. So this issue involves the API rather than the SDK directly. If there's an issue with API behavior, we generally recommend updating to the latest version in order to rule out that the issue isn't already resolved. (The CHANGELOG lists changes in each version.)

Could you share the specific error your getting in order to help us with further investigation? If you can share your debug logs (with any sensitive info redacted) by adding boto3.set_stream_logger('') to your script then that would also help us. And if you can specify the documentation discrepancy you're observing then we can raise that with the Bedrock team.