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

Example 41 with AnalyzedTokenReadings

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

the class EnglishChunkerTest method testAddChunkTagsSingular.

@Test
public void testAddChunkTagsSingular() throws Exception {
    EnglishChunker chunker = new EnglishChunker();
    JLanguageTool lt = new JLanguageTool(new English());
    List<AnalyzedSentence> sentences = lt.analyzeText("The abacus shows how numbers can be stored");
    List<AnalyzedTokenReadings> readingsList = Arrays.asList(sentences.get(0).getTokens());
    chunker.addChunkTags(readingsList);
    // "The abacus":
    assertThat(readingsList.get(1).getChunkTags().toString(), is("[B-NP-singular]"));
    assertThat(readingsList.get(3).getChunkTags().toString(), is("[E-NP-singular]"));
    // "numbers":
    assertThat(readingsList.get(9).getChunkTags().toString(), is("[B-NP-plural, E-NP-plural]"));
}

Example 42 with AnalyzedTokenReadings

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

the class GermanHelperTest method testHasReadingOfType.

@Test
public void testHasReadingOfType() throws Exception {
    AnalyzedTokenReadings readings = new AnalyzedTokenReadings(new AnalyzedToken("der", "ART:DEF:DAT:SIN:FEM", null), 0);
    assertTrue(GermanHelper.hasReadingOfType(readings, GermanToken.POSType.DETERMINER));
    assertFalse(GermanHelper.hasReadingOfType(readings, GermanToken.POSType.NOMEN));
}

Example 43 with AnalyzedTokenReadings

use of org.languagetool.AnalyzedTokenReadings 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 44 with AnalyzedTokenReadings

use of org.languagetool.AnalyzedTokenReadings 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 45 with AnalyzedTokenReadings

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

the class AbstractCompoundRule method getStringToTokenMap.

private Map<String, AnalyzedTokenReadings> getStringToTokenMap(Queue<AnalyzedTokenReadings> prevTokens, List<String> stringsToCheck, List<String> origStringsToCheck) {
    StringBuilder sb = new StringBuilder();
    Map<String, AnalyzedTokenReadings> stringToToken = new HashMap<>();
    int j = 0;
    for (AnalyzedTokenReadings atr : prevTokens) {
        sb.append(' ');
        sb.append(atr.getToken());
        if (j >= 1) {
            String stringToCheck = normalize(sb.toString());
            stringsToCheck.add(stringToCheck);
            origStringsToCheck.add(sb.toString().trim());
            if (!stringToToken.containsKey(stringToCheck)) {
                stringToToken.put(stringToCheck, atr);
            }
        }
        j++;
    }
    return stringToToken;
}