pullparsing

Pull parsing is a technique for reading structured data streams, notably XML, by actively requesting the next event from a parser. In pull parsing, the application drives the parsing process by calling a method such as next() or nextToken() and inspecting the returned event (for example START_TAG, END_TAG, TEXT). This contrasts with push parsers like SAX, which deliver events to the application via callbacks. Pull parsers are typically streaming and forward-only, enabling memory-efficient processing of large documents and enabling fine-grained control over how and when data is consumed.

Common implementations include Java's StAX (Streaming API for XML) with XMLStreamReader, and Android's XmlPullParser, as well as the XmlPull API (org.xmlpull.v1) used in various environments. A pull parser exposes methods to query the current event, element name, text content, and attributes, and to move to the next event. It can handle START_TAG, END_TAG, TEXT, CDATA, COMMENT, PROCESSING_INSTRUCTION.

Benefits and tradeoffs: It provides deterministic flow, easier state management than SAX callbacks, and good for streaming large data; low memory usage compared to DOM. However, it requires more boilerplate to navigate nested structures and may require buffering to access multiple passes.

Use cases include parsing large XML files, streaming data from networks, incremental processing. It is not ideal when random access to the entire document is needed, or when multiple passes over the data are required.

See also: SAX, DOM, StAX, XML Pull Parser, XmlPullParser.