Examples with StemAndEnding zemberek.morphology.structure.StemAndEnding used on opensource projects

Example 1 with StemAndEnding

use of zemberek.morphology.structure.StemAndEnding in project zemberek-nlp by ahmetaa.

the class TurkishMorphology method analyzeWordsWithApostrophe.

private List<WordAnalysis> analyzeWordsWithApostrophe(String word) {
    int index = word.indexOf('\'');
    if (index >= 0) {
        if (index == 0 || index == word.length() - 1) {
            return Collections.emptyList();
        }
        StemAndEnding se = new StemAndEnding(word.substring(0, index), word.substring(index + 1));
        String stem = TurkishAlphabet.INSTANCE.normalize(se.stem);
        String withoutQuote = word.replaceAll("'", "");
        List<WordAnalysis> noQuotesParses = wordAnalyzer.analyze(withoutQuote);
        if (noQuotesParses.size() == 0) {
            return Collections.emptyList();
        }
        return noQuotesParses.stream().filter(noQuotesParse -> noQuotesParse.getStems().contains(stem)).collect(Collectors.toList());
    } else {
        return Collections.emptyList();
    }
}

Example 2 with StemAndEnding

use of zemberek.morphology.structure.StemAndEnding in project zemberek-nlp by ahmetaa.

the class UnidentifiedTokenAnalyzer method analyze.

public synchronized List<WordAnalysis> analyze(String word) {
    if (word.contains("?")) {
        return Collections.emptyList();
    }
    if (!Strings.containsNone(word, "0123456789")) {
        return parseNumeral(word);
    }
    int index = word.indexOf('\'');
    if (index >= 0) {
        if (index == 0 || index == word.length() - 1) {
            return Collections.emptyList();
        }
        StemAndEnding se = new StemAndEnding(word.substring(0, index), word.substring(index + 1));
        String stem = TurkishAlphabet.INSTANCE.normalize(se.stem);
        String ending = TurkishAlphabet.INSTANCE.normalize(se.ending);
        String pronunciation = guessPronunciation(stem);
        DictionaryItem itemProp = new DictionaryItem(Turkish.capitalize(stem), stem, pronunciation, PrimaryPos.Noun, SecondaryPos.ProperNoun);
        itemProp.attributes.add(RootAttribute.Runtime);
        graph.addDictionaryItem(itemProp);
        String toParse = stem + ending;
        List<WordAnalysis> properResults = parser.analyze(toParse);
        graph.removeDictionaryItem(itemProp);
        return properResults;
    } else if (Character.isUpperCase(word.charAt(0))) {
        String normalized = TurkishAlphabet.INSTANCE.normalize(word);
        String pronunciation = guessPronunciation(normalized);
        DictionaryItem itemProp = new DictionaryItem(Turkish.capitalize(normalized), normalized, pronunciation, PrimaryPos.Noun, SecondaryPos.ProperNoun);
        itemProp.attributes.add(RootAttribute.Runtime);
        graph.addDictionaryItem(itemProp);
        // TODO eliminate gross code duplication
        List<WordAnalysis> properResults = parser.analyze(normalized);
        graph.removeDictionaryItem(itemProp);
        return properResults;
    }
    return Collections.emptyList();
}

Example 3 with StemAndEnding

use of zemberek.morphology.structure.StemAndEnding in project zemberek-nlp by ahmetaa.

the class UnidentifiedTokenAnalyzer method parseNumeral.

public List<WordAnalysis> parseNumeral(String s) {
    StemAndEnding se = getFromNumeral(s);
    String lemma;
    if (se.stem.endsWith(".")) {
        String ss = se.stem.substring(0, se.stem.length() - 1);
        lemma = numeralEndingMachine.find(ss);
        lemma = ordinalMap.get(lemma);
    } else {
        lemma = numeralEndingMachine.find(se.stem);
    }
    List<WordAnalysis> results = Lists.newArrayListWithCapacity(1);
    for (TurkishDictionaryLoader.Digit digit : TurkishDictionaryLoader.Digit.values()) {
        Matcher m = digit.pattern.matcher(se.stem);
        if (m.find()) {
            String toParse;
            if (se.ending.length() > 0 && lemma.equals("dört") && TurkishAlphabet.INSTANCE.isVowel(se.ending.charAt(0))) {
                toParse = "dörd" + se.ending;
            } else {
                toParse = lemma + se.ending;
            }
            List<WordAnalysis> res = turkishParser.getWordAnalyzer().analyze(toParse);
            for (WordAnalysis re : res) {
                if (re.dictionaryItem.primaryPos != PrimaryPos.Numeral) {
                    continue;
                }
                re.dictionaryItem = new DictionaryItem(se.stem, se.stem, s + lemma, PrimaryPos.Numeral, digit.secondaryPos);
                re.dictionaryItem.attributes.add(RootAttribute.Runtime);
                re.root = se.stem;
                results.add(re);
            }
        }
    }
    return results;
}