gliwka/hyperscan-java

✨ Added

• New callback API for efficient custom match processing ([#58](https://github.com/gliwka/hyperscan-java/issues/58))
• Byte-based scanning API for direct operation on byte[] and ByteBuffer without String overhead
• New `hasMatch` methods for quick existence checking that terminate immediately on first match ([#68](https://github.com/gliwka/hyperscan-java/issues/68))

⚡ Performance

• Reduced memory usage for UTF-8 string mapping by dynamically selecting optimal array type

🐛 Fixed

• Fixed upstream vectorscan correctness regression on x86 architecture with targeted patch ([#228](https://github.com/gliwka/hyperscan-java/issues/228), [#231](https://github.com/gliwka/hyperscan-java/issues/231))
• Ensured all native handles are properly reclaimed after database compilation ([#230](https://github.com/gliwka/hyperscan-java/issues/230))
• Reworked UTF-8 position mapping to handle the mapping correctly in edge cases ([#170](https://github.com/gliwka/hyperscan-java/issues/170))

📋 Changed

• Removed 255 thread limit for concurrent scanning operations ([#222](https://github.com/gliwka/hyperscan-java/issues/222), [#229](https://github.com/gliwka/hyperscan-java/issues/229))

💥 Breaking

• Windows support has been dropped due to vectorscan not supporting it

✨ Added

• Support for ARM64 architecture on Linux and macOS (M1/M2/M3 family of chips)

🐛 Fixed

• Database instances not reclaimable by GC ([#161](https://github.com/gliwka/hyperscan-java/issues/161)) - thanks [@mmimica](https://github.com/mmimica)!
• Race condition during tracking of native references on multiple threads ([#158](https://github.com/gliwka/hyperscan-java/issues/158)) - thanks [@mmimica](https://github.com/mmimica)!
• Expression IDs now can have arbitrary space between them without consuming additional memory ([#163](https://github.com/gliwka/hyperscan-java/issues/163)) - thanks [@mmimica](https://github.com/mmimica)!
• Removed superflous duplicate call during mapping of expressions in PatternFilter ([#205](https://github.com/gliwka/hyperscan-java/pull/205)) - thanks [@Jiar](https://github.com/Jiar)!

✨ Added

• New PatternFilter allowing for prefiltering of java regex patterns similar to chimera
• Windows support
• Possibility to manually specify expression ids

📋 Changed

• Moved access to native library from JNA to JavaCPP
• Removed context object from expressions

🐛 Fixed

• Lock contention while scanning with high concurrency ([#89](https://github.com/gliwka/hyperscan-java/issues/89))

📋 Changes

• Hyperscan v5.1.1 binaries (#48)
• Java 11 compatibility (#54)
• Logical combinations (thanks to @swapnilnawale @digitalreasoning for the contribution) (#55)
• Change from checked to unchecked exceptions to clean up method signature (#53)

📋 Changes

• Enforce UTF-8 encoding in the direct mapping library (#43) - thanks to @eliaslevy for this PR!

📋 Changes

• Segmentation fault when passing null value as expression (#34)

📋 Changes

• Error due to reading of structs after call to free (#31)

📋 Changes

• Support for 32-bit Linux
• New hyperscan binaries (v4.6.0)
• Updated JUnit and Surefire plugins for testing (works now with the most recent IntelliJ release)

📋 Changes

• Add context object to Expression

📋 Changes

• Removed "System.out.println" which has been left in accidentally

📋 Changes

• Closeable interface implemented in Database and Scanner
• New constructor to create expressions without flags
• Avoid some work in the fast path of Scanner.scan. (#24 - thanks to @eliaslevy)
• Bugfix for edge cases causing an ArrayIndexOutOfBoundException (#23)

📋 Changes

• call time speed up due to JNA direct mapping (PR #17) - thanks to @eliaslevy for this contribution.

📋 Changes

• Enforce UTF-8 encoding on every string (java and jna - see #14) - thanks to @eliaslevy for this PR!
• new version of the libhs.so shared linux library - has been updated to v4.5.1 and linked against older libraries to ensure it works on more conservative distributions (see #13)

📋 Changes

• add test scope to the JUnit dependency and pin it to a version that is not RELEASE (#11)

📋 Changes

• Adds a new method, Scanner.allocScratch, to preallocate the scratch space outside the scan code path and removes the allocation in Scanner.scan. (#9)

📋 Changes

• Occasional segmentation faults during high memory pressure (#8)

📋 Changes

• API method names were refactored to start with a lowercase letter to match the offical java style
• The endOfMatch value now matches the the string index (zero based) of the end of the match (issue #6)

🐛 Bugfixes

• Fixed wrong matched string getting returned in case of utf8 (#5)

📋 Changes

• Includes precompiled x86 64-bit shared hyperscan library for most modern linux distributions (#3)

🐛 Bugfixes

• Wrong string length in case of utf8 characters (bug #2)