bug: table-key-properties in metadata extra does not populate schema key_properties for tap-bigquery

[ReubenFrankel]

Singer SDK Version

0.39.1

Is this a regression?

• Yes

Python Version

3.9

Bug scope

Taps (catalog, state, etc.)

Operating System

Ubuntu 22.04 LTS

Description

I've been trying to use table-key-properties to configure key properties for a BigQuery stream:

plugins:
  extractors:
  - name: tap-bigquery
    variant: meltanolabs
    pip_url: git+https://github.com/Matatika/tap-bigquery.git
    metadata:
      '*-events_*':
        table-key-properties:
        - event_timestamp
        - event_name
        - event_bundle_sequence_id

In this case, key_properties is empty so when loading to a database target no primary keys are configured:

Generated catalog

{
  "streams": [
    {
      "tap_stream_id": "analytics_376142012-events_20240825",
      "table_name": "events_20240825",
      "replication_method": "",
      "key_properties": [],
      "schema": ...,
      "is_view": false,
      "stream": "analytics_376142012-events_20240825",
      "metadata": [
        ...,
        {
          "breadcrumb": [],
          "metadata": {
            "inclusion": "available",
            "table-key-properties": [
              "event_timestamp",
              "event_name",
              "event_bundle_sequence_id"
            ],
            "forced-replication-method": "",
            "schema-name": "analytics_376142012",
            "selected": true
          }
        },
        ...
      ],
      "selected": true,
      "table_key_properties": [
        "event_timestamp",
        "event_name",
        "event_bundle_sequence_id"
      ]
    }
  ]
}

Primary keys are configured when using key-properties/key_properties (not documented here):

    metadata:
      '*-events_*':
        key-properties:
        - event_timestamp
        - event_name
        - event_bundle_sequence_id

Generated catalog

{
  "streams": [
    {
      "tap_stream_id": "analytics_376142012-events_20240825",
      "table_name": "events_20240825",
      "replication_method": "",
      "key_properties": [
        "event_timestamp",
        "event_name",
        "event_bundle_sequence_id"
      ],
      "schema": ...,
      "is_view": false,
      "stream": "analytics_376142012-events_20240825",
      "metadata": [
        ...,
        {
          "breadcrumb": [],
          "metadata": {
            "inclusion": "available",
            "table-key-properties": [],
            "forced-replication-method": "",
            "schema-name": "analytics_376142012",
            "selected": true,
            "key-properties": [
              "event_timestamp",
              "event_name",
              "event_bundle_sequence_id"
            ]
          }
        },
        ...
      ],
      "selected": true
    }
  ]
}

    metadata:
      '*-events_*':
        key_properties:
        - event_timestamp
        - event_name
        - event_bundle_sequence_id

Generated catalog

{
  "streams": [
    {
      "tap_stream_id": "analytics_376142012-events_20240825",
      "table_name": "events_20240825",
      "replication_method": "",
      "key_properties": [
        "event_timestamp",
        "event_name",
        "event_bundle_sequence_id"
      ],
      "schema": ...,
      "is_view": false,
      "stream": "analytics_376142012-events_20240825",
      "metadata": [
        ...,
        {
          "breadcrumb": [],
          "metadata": {
            "inclusion": "available",
            "table-key-properties": [],
            "forced-replication-method": "",
            "schema-name": "analytics_376142012",
            "selected": true,
            "key_properties": [
              "event_timestamp",
              "event_name",
              "event_bundle_sequence_id"
            ]
          }
        },
        ...
      ],
      "selected": true
    }
  ]
}

Why does this work? What is the correct way to do this? I assume table-key-properties is the intended approach, so I don't know if I'm going to run into some undefined behaviour down the line.

---

Related:

• https://discuss.meltano.com/t/16277188/hi-all-when-trying-to-specify-custom-pks-in-my-meltano-yml-b
• https://discuss.meltano.com/t/16276147/so-i-have-a-problem-i-need-to-replicate-table-from-postgres-
• https://discuss.meltano.com/t/16272691/need-advice-i-m-trying-to-develop-tap-bigquery-existing-from

Code

No response

Link to Slack/Linen

https://meltano.slack.com/archives/C069CQNHDNF/p1725976166433339

[ReubenFrankel]

Worth noting that you can also set key properties via stream maps:

    config:
      stream_maps:
        '*-events_*':
          __key_properties__:
          - event_timestamp
          - event_name
          - event_bundle_sequence_id