public void analyze(Document document) {
List<Sentence> sentences = document.getSentences();
for (Sentence sentence : sentences) {
List<Token> tokens = sentence.getTokens();
for (int i = 0; i < tokens.size(); i++) {
String tag = tokens.get(i).getPOSTag();
String word = tokens.get(i).getLexeme();
String[] lemmas = dict.getLemmas(word, tag);
if (lemmas == null || lemmas.length == 0) {
lemmas = dict.getLemmas(word.toLowerCase(), tag);
}
tokens.get(i).setLemmas(lemmas);
}
}
}
public void analyze(Document document) {
List<Sentence> sentences = document.getSentences();
for (Sentence sentence : sentences) {
Span[] contractionsSpan;
synchronized (this.contractionFinder) {
contractionsSpan = contractionFinder.find(TextUtils.tokensToString(sentence.getTokens()));
}
List<Token> newTokens = sentence.getTokens();
for (int i = contractionsSpan.length - 1; i >= 0; i--) {
int start = contractionsSpan[i].getStart();
String lexeme = sentence.getTokens().get(start).getLexeme();
String[] contractions = ContractionUtility.expand(lexeme);
Token original = newTokens.remove(start);
if (contractions != null) {
for (int j = contractions.length - 1; j >= 0; j--) {
Token token = new TokenImpl(original.getStart(), original.getEnd(), contractions[j]);
newTokens.add(start, token);
String caze = null;
if (j == 0) caze = "B";
else if (j == contractions.length - 1) caze = "E";
else caze = "I";
token.addContext(Analyzers.CONTRACTION_FINDER, caze);
}
} else {
LOGGER.debug("Missing contraction: " + lexeme);
}
}
sentence.setTokens(newTokens);
}
}
public List<Mistake> check(Sentence sentence) {
List<Mistake> mistakes = new ArrayList<Mistake>();
if (sentence.getTokens().size() < 2) {
return mistakes;
}
for (int i = 0; i < sentence.getTokens().size(); i++) {
Token originalToken = sentence.getTokens().get(i);
String wanted = originalToken.getLexeme();
String wantedLowerCase = wanted.toLowerCase();
if (map.containsKey(wantedLowerCase)) {
String candidate = RuleUtils.useCasedString(wanted, map.get(wantedLowerCase));
String sentenceText = sentence.getText();
String alternativeText =
sentenceText.substring(0, originalToken.getStart())
+ candidate
+ sentenceText.substring(originalToken.getEnd());
if (LOGGER.isDebugEnabled()) {
LOGGER.debug("\n****** Sentença alternativa **********:\n" + alternativeText);
}
Document alternative = new DocumentImpl(alternativeText);
this.analyzer.analyze(alternative);
Sentence alternativeSentence = alternative.getSentences().get(0);
if (alternativeSentence.getTokensProb() - sentence.getTokensProb() > 0.1) {
if (LOGGER.isDebugEnabled()) {
LOGGER.debug("Prob original: " + sentence.getTokensProb());
LOGGER.debug("Prob alternat: " + alternativeSentence.getTokensProb());
LOGGER.debug(
"\n****** Possível correção **********:\n"
+ sentenceText
+ " -> "
+ alternativeText);
}
Token alternativeToken = alternativeSentence.getTokens().get(i);
String[] suggestions = {candidate};
String[] longMsgArgs = {wanted, translate(alternativeToken.getPOSTag()), candidate};
String[] shortMsgArgs = {wanted, candidate};
mistakes.add(
createMistake(
ID,
longMsgArgs,
shortMsgArgs,
suggestions,
originalToken.getStart(),
originalToken.getEnd(),
sentence.getText()));
}
}
}
return mistakes;
}