[TIKA-2439] Avoid NullPointerException in org.apache.tika.langdetect.OptimaizeLangDetector if models haven't been loaded

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.17
• Fix Version/s: 1.17
• Component/s: languageidentifier
• Assignee: Unassigned
• Reporter: Karl-Philipp Richter
• Created: 02/Aug/17 10:50
• Updated: 02/Aug/17 13:00
• Resolved: 02/Aug/17 11:42

Description

In an intuitive usage of

LanguageDetector languageDetector = LanguageDetector.getDefaultLanguageDetector();
List<LanguageResult> languageResults = languageDetector.detectAll(someNonEmptyString);

`org.apache.tika.langdetect.OptimaizeLangDetector` might/will be chosen as default detector, but since there's no call to `LanguageDetector.loadModels()` `LanguageDetector.detector` is `null` in `detectAll` which causes an unhelpful error situation which one has to investigate in the code. A simple check whether `detector` is `null` and throwing an `IllegalStateException("models haven't been loaded yet (forgot to call loadModels?")` would be much more intuitive.

If that corresponds to the expected behaviour (it's my first week with Tika), I can provide a patch or pull request.

experienced with 1.16-75-g4455a6f08