3.1.0 - April 26, 2020

• Added LanguageModelDataSource protocol for flexibility in defining where language training files are located.
• Added recognizedBlocks(for:) method on SwiftyTesseract for getting box coordinates from Tesseract.
• Deprecated init(language:bundle:engineMode:) and init(languages:bundle:engineMode:) in favor of init(language:dataSource:engineMode:) and init(languages:dataSource:engineMode:).
• Special thank you to Minitour for his hard work bringing these features to the library!

While not directly impacting the public facing API of the library, a number of housekeeping changes have been put in place:

• Added PR Template
• Added checklists to Feature request and Bug report templates
• Updated Contribution Guidelines and Readme to clarify what the project will support.
• Migrated from Travis CI to GitHub Actions
• Utilizing fastlane for build automation and deployment tasks.
• Moved documentation out of the repo. It will now be auto-generated on release and pushed to the shiny new gh-pages branch.
• Added stale-bot to clean up old issues that have had no activity after 30 days, with a 7-day grace period for anyone to comment to keep the issue open.

3.0.0 - April 2, 2020

• Deprecated performOCR(on:completionHandler:)
• Added performOCR(on:) and performOCRPublisher(on:) methods to replace the aforementioned deprecated method
• Added Makefile and aggregate target libtesseract to build dependencies from source

2.2.0 - March 30, 2019

• Added minimumCharacterHeight property to enable SwiftyTesseract to ignore characters below a user-defined threshold.
• Added createPDF(from:) method to create a PDF Data object from an array of UIImages that OCR is performed on.

2.1.0 - February 2, 2019

• Added preserveInterwordSpaces property to allow more than one space between words if desired.
• Updated underlying Tesseract library to 4.0.0 release version.

2.0.0 - September 25, 2018

• CustomData was enum removed in favor of keeping one enum, RecognitionLanguage, that utilizes an associated types as opposed to raw values. This is the only breaking change in 2.0.0 that should only affect users of CustomData.
• The underlying Tesseract library has been updated to 4.0.0-beta.4

1.1.0 - May 5, 2018

• Created CustomData enum to enable the use of custom .traineddata files.
• Added support for iOS 9.0

1.0.3 - April 4, 2018

• Added semaphore to make performOCR(on:completionHandler:) thread-safe

1.0.2 - April 2, 2018

• Fixed documentation

1.0.1 - March 30, 2018

• Fixed issue with disk I/O bug that starting occurring in iOS 11.3 that caused a crash when calling write(to:options:) on an instance of Data to a file name that didn't exist (previously the file would be created if it didn't exist). All conversions from UIImage to Pix are now done in memory, completely bypassing disk I/O altogether.