pull parser vs push parser

When designing parsers as a reusable library (rather than a bespoke parser for a single purpose), we have the choice to make push parser and pull parser.

Push Parsers—Events are generated by the API in the form of callback functions like startDocument() and endDocument() and are beyond the programmer’s control. We as programmers could handle the events, but the generation of events is beyond our control.

Pull Parsers - Events are generated when we call some API. Example shown below. So we, as programmers, can decide when to generate events. ¹

Push:

if (myChar == '(')
    handler.handleOpenParen(); // push the open paren to the handler

Pull:

Token token = parser.next(); // pull the next token from the parser

Push vs pull can be seen as a special instance of internal vs external iterator.

Pros and cons

Some advantages of a pull parser include loose coupling, less memory usage, and ease of use from the user side. ²

References

Footnotes

1. java - What is ‘Push Approach’ and ‘Pull Approach’ to parsing? - Stack Overflow ↩

2. GitHub - pulldown-cmark/pulldown-cmark: An efficient, reliable parser for CommonMark, a standard dialect of Markdown ↩