Download to read offline
![Basic Java types
• java.lang.Number (JSON number)
• java.util.List (JSON array)
• Java array (e.g., int[]) (JSON array)
• java.lang.CharSequence (JSON String)
• java.util.Map (JSON object)](https://image.slidesharecdn.com/jparse-230217024226-26961b85/85/JParse-Fast-JSON-Parser-16-320.jpg)
![Many characters sources
InputStream, CharBu
ff
er, char[], String, Reader, File,
CharSequence (StringBuilder) etc.](https://image.slidesharecdn.com/jparse-230217024226-26961b85/85/JParse-Fast-JSON-Parser-23-320.jpg)
![Basic Java types
• java.lang.Number (JSON number)
• java.util.List (JSON array)
• Java array (e.g., int[]) (JSON array)
• java.lang.CharSequence (JSON String)
• java.util.Map (JSON object)](https://image.slidesharecdn.com/jparse-230217024226-26961b85/75/JParse-Fast-JSON-Parser-16-2048.jpg)
![Many characters sources
InputStream, CharBu
ff
er, char[], String, Reader, File,
CharSequence (StringBuilder) etc.](https://image.slidesharecdn.com/jparse-230217024226-26961b85/75/JParse-Fast-JSON-Parser-23-2048.jpg)
Uploaded byRick Hightower
JParse Fast JSON Parser
Jparse is a lightweight JSON parser developed as a simpler alternative to the larger utility library Boon, containing about 4200 lines of code and designed for performance without unsafe operations. It supports a subset of JSONPath and focuses solely on parsing, lacking advanced features such as JavaBean serialization or extended mappings. While benchmarks indicate that Jparse is fast, with a small footprint, it still requires improvement in error handling and adheres strictly to the JSON specification.
More Related Content
Similar to JParse Fast JSON Parser
JParse Fast JSON Parser
- 1.
- 2. • Wrote Boon •Java utility lib, that became a JSON Parser through a $100 bet • Wrote QBit, Crank, worked a lot with Vert.x and Akka About Me
- 3. Boon got someattention • InfoQ article • Adopted as part of Groovy • GSON hash attack benchmark • Benchmarks and response to bench marks
- 4. Boon’s day inthe sun 4/30/2014
- 5. Why a newParser? • Why not just update Boon? • Large, was meant to be a utility lib and became a JSON parser • 90,000 LoC in (just Java not including test classes) • Does too many things that no one uses • Uses index overlay but was not originally designed to use Index Overlay • Uses Unsafe which is not allowed in new JVMs
- 6. JParse • Just aJSON parser plus a small subset of JSONPath • Small (currently 4200 lines of code) • Status: • Perf tuned and worked on usable, fl uent API • Does not use Unsafe • Very fast, uses an index overlay from the ground up • Does not do JavaBean serialization but can serialize into basic Java types
- 7. Is JParse fast? BenchmarkMode Cnt Score Error Units BenchMark.simpleDeserializeJJsonWebXML thrpt 2 237334.575 ops/s BenchMark.simpleDeserializeJacksonWebXML thrpt 2 114135.958 ops/s YES!
- 8. Caveat on Benchmarks •JParse is being actively developed, as we add features or make improvements, it can impact benchmarks • JParse is focused only on Parse and mini JSONPath • JParse does not have very many features • JParse only supports JSON as de fi ned by json.org • No extensions • No advanced mappings
- 9. Benchmark Mode CntScore Error Units BenchMark.deserializeIntoMapJParseAndGetPathThenSerialize thrpt 2 788400.189 ops/s BenchMark.deserializeIntoMapJacksonAndGetPathThenSerialize thrpt 2 285313.787 ops/s
- 10.
- 11. Conclusion for Benchmarks •Parsers got a lot faster • Jackson got a lot faster in 9 years • Other parsers are fast too • Not as easy to win as it was 9 years ago
- 12.
- 13. Slicing and dicingJSON Supports JSONPath subset
- 14. Easy to accessproperties
- 15. No special API,just Java types
- 16. Basic Java types •java.lang.Number (JSON number) • java.util.List (JSON array) • Java array (e.g., int[]) (JSON array) • java.lang.CharSequence (JSON String) • java.util.Map (JSON object)
- 17. Easy to combineJava functional programming JSON Array is just a Java list so you can do stream/functional programming
- 18. Helper functions tomake Java functional programming easier with JParse Versus this!
- 19. Easy to usefunctional support For custom mapping
- 20. Easy to usefunctional support For custom mapping part 2
- 21. Easy to pulloff properties
- 22. You can alsowork with nodes directly for more conversions and such
- 23. Many characters sources InputStream,CharBu ff er, char[], String, Reader, File, CharSequence (StringBuilder) etc.
- 24. Nodes are allbasic Java types • ArrayNode is a java.util.List • ObjectNode is a Map • All nodes are CharSequences • NumberNode is java.lang.Number (toInt, toFloat, toDouble, toShort, etc.) • StringNode is a CharSequence and works well with Strings (toUnencodedString, toString, toEncodedString) Easy to learn!
- 25.
- 26.
- 27. Continue to useJackson • Used in a lot of projects • Support for other data formats • Advanced Mappings • But large
- 28. When to useJParse • You want a small footprint JSON parser that is fast • You don’t need advanced mappings and are willing and able to map using Functional programming
- 29. JParse • Not doneyet (BenchMark will change) • Still needs to improve error handling • Virtually none so no real useful messages if their are errors • Boon had really excellent error messages and JParse will to • Does not support any extras (by design) • Does not support NaN, +In fi nity, -In fi nity because that is not in the JSON org spec. • Does not support any attribute keys but Strings (JSON org spec.) • Just strict JSON no minimal JSON