public WordsToSentencesAnnotator(
boolean verbose,
String boundaryTokenRegex,
Set<String> boundaryToDiscard,
Set<String> htmlElementsToDiscard,
String newlineIsSentenceBreak) {
this(
verbose,
false,
new WordToSentenceProcessor<CoreLabel>(
boundaryTokenRegex,
boundaryToDiscard,
htmlElementsToDiscard,
WordToSentenceProcessor.stringToNewlineIsSentenceBreak(newlineIsSentenceBreak)));
}
public WordsToSentencesAnnotator(
boolean verbose,
String boundaryTokenRegex,
Set<String> boundaryToDiscard,
Set<String> htmlElementsToDiscard,
String newlineIsSentenceBreak,
String boundaryMultiTokenRegex,
Set<String> tokenRegexesToDiscard) {
this(
verbose,
false,
new WordToSentenceProcessor<CoreLabel>(
boundaryTokenRegex,
boundaryToDiscard,
htmlElementsToDiscard,
WordToSentenceProcessor.stringToNewlineIsSentenceBreak(newlineIsSentenceBreak),
(boundaryMultiTokenRegex != null)
? TokenSequencePattern.compile(boundaryMultiTokenRegex)
: null,
tokenRegexesToDiscard));
}