Skip to main content
Data Pipelines are now the preferred method for loading data into the System. For details, see Load Data.
Incoming records can optionally be filtered by a JMESPath expression added with an additional filter key in a pipeline configuration.
JSON
{
    "transform": {
        "topics": {
            "my_topic": {
                "filter": "<jmespath-expression>",
                "tables": {
                    "my_table": {
                        "filter": "<jmespath-expression>",
                        "columns": {}
                    }
                }
            }
        }
    }
}
Records that pass through the filter (the filter expression evaluates to true) are processed as normal, while records that fail the filter (the filter expression evaluates to false) are dropped. Note that filters can be applied at the topic or table levels. A record needs to pass through all configured filters in order to be loaded to a table. The filter JMESPath expression must evaluate to a boolean value. Examples of this are JMESPath comparators or functions that return a boolean like contains, starts_with, and ends_with.

​
Example

Pipeline:
JSON
{
    "transform": {
        "topics": {
            "topic0": {
                "filter": "integer0 == `2`",
                "tables": {
                    "table0": {
                        "filter": "integer1 == `3`",
                        "columns": {
                            "col0": "integer0",
                            "col1": "integer1"
                        }
                    }
                }
            }
        }
    }
}
Input:
JSON
[
    {
        "integer0": 0,
        "integer1": 3
    },
    {
        "integer0": 2,
        "integer1": 1
    },
    {
        "integer0": 2,
        "integer1": 3
    }
]
Output:
JSON
[
    {
        "integer0": 2,
        "integer1": 3
    }
]
Ingest Data with Legacy LAT Reference
Last modified on May 27, 2026