问题
I have memory problems with the webapp responsible from parsing XML event and pushing them to our RabbitMQ bus. This webapp receives XML event in a servlet, we decode it and push it to our bus and then we acknowledge the sender. We are curently using org.xml.sax.helpers.DefaultHandler in java 5, but it seems that Stax (with XMLStreamReader) is much better and still easy to read, but we would need to migrate our code to java6 (see this question on SO). Does it worth it ?
What is the most memory efficient method to parse XML in java ? We are looking for a fast, memory efficient and easy to write/read code. Does this exist ?
回答1:
Event-driven parsers don't build up ASTs and are therefore fast and memory-efficient. The standard for XML would be defined by the SAX standard.
The Xerces implementation claims to be fast.
回答2:
NanoXML is a very small and lightweight XML parser which supports SAX parsing. It is a good alternative to Xerces.
回答3:
Beware of the answers mentioning Sax, as it is at the expense of significantly less usefulness. The most memory efficient sax parsing w/o sacrificing usability is vtd-xml. Here is the proof
Processing XML with Java – A Performance Benchmark http://sdiwc.us/digitlib/journal_paper.php?paper=00000582.pdf
来源：https://stackoverflow.com/questions/7332163/what-is-the-most-memory-efficient-method-to-parse-xml-in-java