Examples with AnalyzedToken org.languagetool.AnalyzedToken used on opensource projects

Example 11 with AnalyzedToken

use of org.languagetool.AnalyzedToken in project languagetool by languagetool-org.

the class LanguageToolFilter method incrementToken.

@Override
public boolean incrementToken() throws IOException {
    if (posStack.size() > 0) {
        String pop = posStack.pop();
        restoreState(current);
        termAtt.append(pop);
        posIncrAtt.setPositionIncrement(0);
        typeAtt.setType("pos");
        return true;
    }
    if (tokenIter == null || !tokenIter.hasNext()) {
        // there are no remaining tokens from the current sentence... are there more sentences?
        if (input.incrementToken()) {
            // a new sentence is available: process it.
            String sentenceStr = termAtt.toString();
            collectedInput.append(sentenceStr);
            if (sentenceStr.length() >= 255) {
                // later. See https://github.com/languagetool-org/languagetool/issues/364
                return true;
            } else {
                sentenceStr = collectedInput.toString();
                collectedInput.setLength(0);
            }
            AnalyzedSentence sentence = languageTool.getAnalyzedSentence(sentenceStr);
            List<AnalyzedTokenReadings> tokenBuffer = Arrays.asList(sentence.getTokens());
            tokenIter = tokenBuffer.iterator();
            /*
         * it should not be possible to have a sentence with 0 words, check just in case. returning
         * EOS isn't the best either, but it's the behavior of the original code.
         */
            if (!tokenIter.hasNext()) {
                return false;
            }
        } else {
            // no more sentences, end of stream!
            return false;
        }
    }
    // It must clear attributes, as it is creating new tokens.
    clearAttributes();
    AnalyzedTokenReadings tr = tokenIter.next();
    // add POS tag for sentence start.
    if (tr.isSentenceStart()) {
        // TODO: would be needed so negated tokens can match on something (see testNegatedMatchAtSentenceStart())
        // but breaks other cases:
        //termAtt.append("SENT_START");
        typeAtt.setType("pos");
        String posTag = tr.getAnalyzedToken(0).getPOSTag();
        String lemma = tr.getAnalyzedToken(0).getLemma();
        if (toLowerCase) {
            termAtt.append(POS_PREFIX.toLowerCase()).append(posTag.toLowerCase());
            if (lemma != null) {
                termAtt.append(LEMMA_PREFIX.toLowerCase()).append(lemma.toLowerCase());
            }
        } else {
            termAtt.append(POS_PREFIX).append(posTag);
            if (lemma != null) {
                termAtt.append(LEMMA_PREFIX).append(lemma);
            }
        }
        return true;
    }
    // by pass the white spaces.
    if (tr.isWhitespace()) {
        return this.incrementToken();
    }
    offsetAtt.setOffset(tr.getStartPos(), tr.getEndPos());
    for (AnalyzedToken token : tr) {
        if (token.getPOSTag() != null) {
            if (toLowerCase) {
                posStack.push(POS_PREFIX.toLowerCase() + token.getPOSTag().toLowerCase());
            } else {
                posStack.push(POS_PREFIX + token.getPOSTag());
            }
        }
        if (token.getLemma() != null) {
            if (toLowerCase) {
                posStack.push(LEMMA_PREFIX.toLowerCase() + token.getLemma().toLowerCase());
            } else {
                // chances are good this is the same for all loop iterations, store it anyway...
                posStack.push(LEMMA_PREFIX + token.getLemma());
            }
        }
    }
    current = captureState();
    if (toLowerCase) {
        termAtt.append(tr.getAnalyzedToken(0).getToken().toLowerCase());
    } else {
        termAtt.append(tr.getAnalyzedToken(0).getToken());
    }
    return true;
}

Example 12 with AnalyzedToken

use of org.languagetool.AnalyzedToken in project languagetool by languagetool-org.

the class AbstractCompoundRule method match.

@Override
public RuleMatch[] match(AnalyzedSentence sentence) {
    List<RuleMatch> ruleMatches = new ArrayList<>();
    AnalyzedTokenReadings[] tokens = getSentenceWithImmunization(sentence).getTokensWithoutWhitespace();
    RuleMatch prevRuleMatch = null;
    Queue<AnalyzedTokenReadings> prevTokens = new ArrayBlockingQueue<>(MAX_TERMS);
    for (int i = 0; i < tokens.length + MAX_TERMS - 1; i++) {
        AnalyzedTokenReadings token;
        // we need to extend the token list so we find matches at the end of the original list:
        if (i >= tokens.length) {
            token = new AnalyzedTokenReadings(new AnalyzedToken("", "", null), prevTokens.peek().getStartPos());
        } else {
            token = tokens[i];
        }
        if (i == 0) {
            addToQueue(token, prevTokens);
            continue;
        }
        if (token.isImmunized()) {
            continue;
        }
        AnalyzedTokenReadings firstMatchToken = prevTokens.peek();
        List<String> stringsToCheck = new ArrayList<>();
        // original upper/lowercase spelling
        List<String> origStringsToCheck = new ArrayList<>();
        Map<String, AnalyzedTokenReadings> stringToToken = getStringToTokenMap(prevTokens, stringsToCheck, origStringsToCheck);
        // sure we match longer strings first:
        for (int k = stringsToCheck.size() - 1; k >= 0; k--) {
            String stringToCheck = stringsToCheck.get(k);
            String origStringToCheck = origStringsToCheck.get(k);
            if (getCompoundRuleData().getIncorrectCompounds().contains(stringToCheck)) {
                AnalyzedTokenReadings atr = stringToToken.get(stringToCheck);
                String msg = null;
                List<String> replacement = new ArrayList<>();
                if (!getCompoundRuleData().getNoDashSuggestion().contains(stringToCheck)) {
                    replacement.add(origStringToCheck.replace(' ', '-'));
                    msg = withHyphenMessage;
                }
                if (isNotAllUppercase(origStringToCheck) && !getCompoundRuleData().getOnlyDashSuggestion().contains(stringToCheck)) {
                    replacement.add(mergeCompound(origStringToCheck));
                    msg = withoutHyphenMessage;
                }
                String[] parts = stringToCheck.split(" ");
                if (parts.length > 0 && parts[0].length() == 1) {
                    replacement.clear();
                    replacement.add(origStringToCheck.replace(' ', '-'));
                    msg = withHyphenMessage;
                } else if (replacement.isEmpty() || replacement.size() == 2) {
                    // isEmpty shouldn't happen
                    msg = withOrWithoutHyphenMessage;
                }
                RuleMatch ruleMatch = new RuleMatch(this, firstMatchToken.getStartPos(), atr.getEndPos(), msg, shortDesc);
                ruleMatch.setSuggestedReplacements(replacement);
                // avoid duplicate matches:
                if (prevRuleMatch != null && prevRuleMatch.getFromPos() == ruleMatch.getFromPos()) {
                    prevRuleMatch = ruleMatch;
                    break;
                }
                prevRuleMatch = ruleMatch;
                ruleMatches.add(ruleMatch);
                break;
            }
        }
        addToQueue(token, prevTokens);
    }
    return toRuleMatchArray(ruleMatches);
}

Example 13 with AnalyzedToken

use of org.languagetool.AnalyzedToken in project languagetool by languagetool-org.

the class AbstractSimpleReplaceRule method match.

@Override
public RuleMatch[] match(AnalyzedSentence sentence) {
    List<RuleMatch> ruleMatches = new ArrayList<>();
    AnalyzedTokenReadings[] tokens = sentence.getTokensWithoutWhitespace();
    for (AnalyzedTokenReadings tokenReadings : tokens) {
        // short for SENT_START
        if (JLanguageTool.SENTENCE_START_TAGNAME.equals(tokenReadings.getAnalyzedToken(0).getPOSTag()))
            continue;
        // and speller-ignorable rules
        if (tokenReadings.isImmunized() || tokenReadings.isIgnoredBySpeller()) {
            continue;
        }
        String originalTokenStr = tokenReadings.getToken();
        if (ignoreTaggedWords && isTagged(tokenReadings)) {
            continue;
        }
        String tokenString = cleanup(originalTokenStr);
        // try first with the original word, then with the all lower-case version
        List<String> possibleReplacements = getWrongWords().get(originalTokenStr);
        if (possibleReplacements == null) {
            possibleReplacements = getWrongWords().get(tokenString);
        }
        if (possibleReplacements == null && checkLemmas) {
            possibleReplacements = new ArrayList<>();
            List<String> lemmas = new ArrayList<>();
            for (AnalyzedToken analyzedToken : tokenReadings.getReadings()) {
                String lemma = analyzedToken.getLemma();
                if (lemma != null && getWrongWords().containsKey(lemma) && !lemmas.contains(lemma)) {
                    lemmas.add(cleanup(lemma));
                }
            }
            for (String lemma : lemmas) {
                List<String> replacements = getWrongWords().get(lemma);
                if (replacements != null) {
                    possibleReplacements.addAll(replacements);
                }
            }
            possibleReplacements = possibleReplacements.stream().distinct().collect(Collectors.toList());
        }
        if (possibleReplacements != null && possibleReplacements.size() > 0) {
            List<String> replacements = new ArrayList<>();
            replacements.addAll(possibleReplacements);
            if (replacements.contains(originalTokenStr)) {
                replacements.remove(originalTokenStr);
            }
            if (replacements.size() > 0) {
                RuleMatch potentialRuleMatch = createRuleMatch(tokenReadings, replacements);
                ruleMatches.add(potentialRuleMatch);
            }
        }
    }
    return toRuleMatchArray(ruleMatches);
}

Example 14 with AnalyzedToken

use of org.languagetool.AnalyzedToken in project languagetool by languagetool-org.

the class AbstractWordCoherencyRule method match.

@Override
public RuleMatch[] match(List<AnalyzedSentence> sentences) {
    List<RuleMatch> ruleMatches = new ArrayList<>();
    // e.g. aufwändig -> RuleMatch of aufwendig
    Map<String, RuleMatch> shouldNotAppearWord = new HashMap<>();
    int pos = 0;
    for (AnalyzedSentence sentence : sentences) {
        AnalyzedTokenReadings[] tokens = sentence.getTokensWithoutWhitespace();
        for (AnalyzedTokenReadings tmpToken : tokens) {
            String token = tmpToken.getToken();
            List<AnalyzedToken> readings = tmpToken.getReadings();
            // TODO: in theory we need to care about the other readings, too (affects e.g. German "Schenke" as a noun):
            if (readings.size() > 0) {
                String baseform = readings.get(0).getLemma();
                if (baseform != null) {
                    token = baseform;
                }
            }
            if (shouldNotAppearWord.containsKey(token)) {
                RuleMatch otherMatch = shouldNotAppearWord.get(token);
                String otherSpelling = otherMatch.getMessage();
                String msg = getMessage(token, otherSpelling);
                RuleMatch ruleMatch = new RuleMatch(this, pos + tmpToken.getStartPos(), pos + tmpToken.getEndPos(), msg);
                ruleMatch.setSuggestedReplacement(otherSpelling);
                ruleMatches.add(ruleMatch);
            } else if (getWordMap().containsKey(token)) {
                String shouldNotAppear = getWordMap().get(token);
                RuleMatch potentialRuleMatch = new RuleMatch(this, pos + tmpToken.getStartPos(), pos + tmpToken.getEndPos(), token);
                shouldNotAppearWord.put(shouldNotAppear, potentialRuleMatch);
            }
        }
        pos += sentence.getText().length();
    }
    return toRuleMatchArray(ruleMatches);
}

Example 15 with AnalyzedToken

use of org.languagetool.AnalyzedToken in project languagetool by languagetool-org.

the class AgreementRule method getAgreementCategories.

/** Return Kasus, Numerus, Genus of those forms with a determiner. */
private Set<String> getAgreementCategories(AnalyzedTokenReadings aToken, Set<GrammarCategory> omit, boolean skipSol) {
    Set<String> set = new HashSet<>();
    List<AnalyzedToken> readings = aToken.getReadings();
    for (AnalyzedToken tmpReading : readings) {
        if (skipSol && tmpReading.getPOSTag() != null && tmpReading.getPOSTag().endsWith(":SOL")) {
            // SOL = alleinstehend - needs to be skipped so we find errors like "An der roter Ampel."
            continue;
        }
        AnalyzedGermanToken reading = new AnalyzedGermanToken(tmpReading);
        if (reading.getCasus() == null && reading.getNumerus() == null && reading.getGenus() == null) {
            continue;
        }
        if (reading.getGenus() == GermanToken.Genus.ALLGEMEIN && tmpReading.getPOSTag() != null && // STV: stellvertretend (!= begleitend)
        !tmpReading.getPOSTag().endsWith(":STV") && !possessiveSpecialCase(aToken, tmpReading)) {
            // e.g. "Ich Arbeiter" doesn't get flagged as incorrect:
            if (reading.getDetermination() == null) {
                // Nouns don't have the determination property (definite/indefinite), and as we don't want to
                // introduce a special case for that, we just pretend they always fulfill both properties:
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.MASKULINUM, GermanToken.Determination.DEFINITE, omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.MASKULINUM, GermanToken.Determination.INDEFINITE, omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.FEMININUM, GermanToken.Determination.DEFINITE, omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.FEMININUM, GermanToken.Determination.INDEFINITE, omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.NEUTRUM, GermanToken.Determination.DEFINITE, omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.NEUTRUM, GermanToken.Determination.INDEFINITE, omit));
            } else {
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.MASKULINUM, reading.getDetermination(), omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.FEMININUM, reading.getDetermination(), omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), GermanToken.Genus.NEUTRUM, reading.getDetermination(), omit));
            }
        } else {
            if (reading.getDetermination() == null || "jed".equals(tmpReading.getLemma()) || "manch".equals(tmpReading.getLemma())) {
                // "jeder" etc. needs a special case to avoid false alarm
                set.add(makeString(reading.getCasus(), reading.getNumerus(), reading.getGenus(), GermanToken.Determination.DEFINITE, omit));
                set.add(makeString(reading.getCasus(), reading.getNumerus(), reading.getGenus(), GermanToken.Determination.INDEFINITE, omit));
            } else {
                set.add(makeString(reading.getCasus(), reading.getNumerus(), reading.getGenus(), reading.getDetermination(), omit));
            }
        }
    }
    return set;
}