One of two languages with same confidence value is erroneously removed

[mmedek]

There is a potential bug in the computeLanguageConfidenceValues method in LanguageDetector class when the same confidence is computed.

The return value SortedMap<Language, Double> uses comparator by prob-value, this means that when any two languages have the same probability then only one of them could remain in the map. So for example, when the original distribution is cs=0.4, de=0.4, en=0.2 then this method will return only de=0.4, en=0.2.

lingua/src/main/kotlin/com/github/pemistahl/lingua/api/LanguageDetector.kt

Line 140 in 1245654

fun computeLanguageConfidenceValues(text: String): SortedMap<Language, Double> {

[pemistahl]

Good catch @mmedek, thank you. I wasn't aware of that. I will write a custom comparator that fixes the issue.