A source point is the root of a Kafka Streams topology. It consumes event records from Kafka topics, turns their byte-array keys and values into application objects, and forwards those records to downstream processing nodes. A topology can contain processing nodes and sink nodes after its source; a minimal application has a source that reads a topic and a sink that writes results to Kafka. section 6.1 section 6.7
Create a DSL source with `StreamsBuilder.stream(...)`, supplying a topic and, when needed, `Consumed` settings. `Consumed.with(...)` selects key and value Serdes, whose deserializers let the stream process typed records instead of broker bytes. `Consumed` can also configure a timestamp extractor, offset reset, and operator name. A source may subscribe to one topic, a collection of topics, or a topic-name pattern. section 6.3.1
Listing 6.1 Defining the source for the stream
KStream<String, String> simpleFirstStream = builder.stream("src-topic",
Consumed.with(Serdes.String(), Serdes.String()));
This minimal DSL call creates a `KStream<String, String>` source for `src-topic` and assigns String Serdes to both record fields. section 6.3.1
Read, transform, and write a stream
A Yelling app starts at `src-topic`, maps each value to uppercase, and sends the resulting records to `out-topic`. The source is the entry point; `mapValues` is its child processing step, and `to` creates the sink side of the pipeline. section 6.3 section 6.3.1
The Processor API exposes the source more directly with `Topology.addSource(...)`. Unlike the DSL, it requires a node name so Kafka Streams can connect child nodes to their parent. It takes a topic name or pattern and can take a timestamp extractor plus separate key and value deserializers; only the processor name and topic selection are required. section 10.1.1
Mask retail purchases at the source
A retail stream reads `RetailPurchase` records from the `transactions` topic using a String key Serde and custom purchase Serde, then immediately applies `mapValues` to mask credit-card numbers. If a transformation needs the key while retaining it, use `ValueMapperWithKey`; use `map` when producing a new key and value. section 6.4.1
