This release is identical to v1.2.0, except that a JitPack configuration file has been added, so you can now obtain the artifacts from JitPack.

See issue #3.

Changes:

• Uses Kotlin 2.0
• Different versioning scheme: <version>-L<upstream-lingua-commit>
  Includes the upstream Lingua commit instead of version after the L, to allow targeting not yet released upstream Lingua versions.
• Uses upstream Lingua commit pemistahl@44ee38c (future release after Lingua 1.2.2)
• -with-dependencies.jar no longer contains dependencies repackaged under different package name
  Important: The -with-dependencies.jar is mainly intended for usage on command line; it should not be used as regular dependency.

Changes:

• Fixed exceptions and incorrect behavior for multi-language detection on multi-script text
• Fixed thread-safety issue with multi-language detection GUI
• Changed model reloading for withIncreasedDetectionSpeed()
  Previously the models were always reloaded when language detectors had different withIncreasedDetectionSpeed() settings. However, this would have caused creating detectors with different settings reloading each others models. Now the models are only reloaded if withIncreasedDetectionSpeed() is requested and was not used previously, and therefore the models with increased detection speed are used afterwards by all detectors (regardless of withIncreasedDetectionSpeed() setting) until they are explicitly unloaded.

Tiny Lingua version 1.1.0 based on upstream Lingua version 1.2.2.

Changes:

• Fixed ArrayIndexOutOfBoundsException when using Map returned by LanguageDetector.computeLanguageConfidenceValues
• Added LanguageDetectorBuilder.withIncreasedDetectionSpeed which increases detection speed at the cost of higher memory usage
• Experimental: Added initial support for multi language detection, see below

Multi language detection (experimental)

Added intial support for multi language detection, that is, for a given input text detect sections written in different languages.
This functionality is available through LanguageDetector.detectMultiLanguageOf. Language sections are detect based on certain punctuation characters and line breaks, and requires a certain minimum length of the text section to be reliable.

Additionally a GUI is provided which visualizes the results of multi language detection. That GUI is mainly intended to showcase multi langage detection and to aid with development and troubleshooting, it might be changed in the future. The GUI can be started by first launching the -with-dependencies.jar using java -jar ... and then selection option 3 ("launch multi-language detection GUI"):

java -jar tiny-lingua-1.1.0-L1.2.2-with-dependencies.jar

Feedback is highly appreciated!

Tiny Lingua version 1.0.0 based on upstream Lingua version 1.2.2.

Differences with Lingua 1.2.2:

• New function LanguageDetectorBuilder.withExecutor which allows specifying a custom executor
• No special support for SecurityManager usage