Versions Compared

Old Version 25

changes.mady.by.user Shikhar Bhushan

Saved on

compared with

New Version 26

changes.mady.by.user Shikhar Bhushan

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

...

If the result of any Transformation.apply() in a chain is null, that record is discarded (not written to Kafka in the case of a source connector, or not provided to sink connector).

Bundled transformations

Criteria: SMTs that are shipped with Kafka Connect should be general enough to apply to many data sources & serialization formats. They should also be simple enough to not cause any additional library dependency to be introduced.

Beyond those being initially included with this KIP, transformations can be adopted for inclusion in future with JIRA/ML discussion to weigh the tradeoffsWIP. Subject to discussion –  not a final list.

Name
Functionality
Rationale
Configuration
Mask{Key,Value}Mask or replace the specified primitive fields, assuming there is a top-level Struct.

Obscure sensitive info like credit card numbers.

  • randomize.fields – fields to replace with random data
  • clobber.fields – map of fields to replacement string/number
InsertIn{Key,Value}Insert specified fields with given name, assuming there is a top-level Struct.Widely applicable to insert certain record metadata.

  • topic – the target field name for record topic

  • partition – the target field name for record partition

  • offset – the target field name for record offset

  • timestamp – the target field name for record timestamp

  • uuid – the target field name for a UUID (caveat: not deterministic, not really record metadata so not sure it belongs here, but implementation-wise it makes sense to have here)

TimestampRouter

Timestamp-based routing.

Useful for temporal data e.g. application log data being indexed to a search system with a sink connector can be routed to a daily index.

  • topic.format – format string which can contain ${topic} and ${timestamp} as placeholders for the original topic and the timestamp, respectively

  • timestamp.format – a format string compatible with SimpleDateFormat
RegexRouterRegex-based routing.There are too many inconsistent configs to route in different connectors.
  • regex
  • replacement

See http://docs.oracle.com/javase/7/docs/api/java/util/regex/Matcher.html#replaceFirst(java.lang.String)

ValueToKey
Create or replace record key with data from record value.

Useful when a source connector does not populate the record key but only the value with a Struct.

  • fields – list of field names to hoist into the record key as a primitive (single field ) / Struct (multiple fields)
  • force.struct – force wrapping in a Struct even when it is a single field
Flatten

Flatten nested Structs inside a top-level Struct, omitting all other non-primitive fields.

Useful for sink connectors that can only deal with flat Structs.
  • delimiter – the delimiter to use when flattening field names
Replace

Filter and rename fields.

Useful for lightweight data munging.
  • whitelist – fields to include
  • blacklist – fields to exclude
  • rename – map of old field names to new field names
NumericCasts

Casting of numeric field to some specified numeric type.

Useful in conjunction with source connectors that don't have enough information and utilize an unnecessarily wide data type.
  • spec – map of field name to type (i.e. boolean, int8, int16, int32, int64, float32, float64)
TimestampConverterConvert datatype of a timestamp field.Timestamps are represented in a ton of different ways, provide a transformation from going between strings, epoch times as longs, and Connect date/time types.
  • field – the field name (optional, can be left out in case of primitive data)
  • type – desired type (i.e. string, long, Date, Time, Timestamp)
  • format – in case converting to or from a string, a SimpleDateFormat-compatible format string
Hoist{Key,Value}ToStruct

Wrap data in a Struct.

 		Useful when a transformation or sink connector expects Struct but the data is a primitive type.
  • schema.name – name for the new Struct schema
  • field – field name for the original data within this Struct
Extract{Key,Value}FromStruct
Extract a specific field from a Struct. The inverse of Hoist{Key,Value}ToStruct
  • field – field name to extract

...

However, the surface area for such a change is much larger - we would need additional REST APIs for creating, updating and validating transformation chain configs. The current proposal does not prevent taking this direction down the line.

Not including any transformations with Connect

In the interest of providing a better out-of-the-box experience and avoiding duplication of effort in the ecosystem, we will be bundling certain transformations with Connect.

One concern here is that we should have a well-defined criteria for what belongs in Connect vs external dependencies, which was addressed.