Examples with NullSuffixForm zemberek.morphology.lexicon.NullSuffixForm used on opensource projects

Example 1 with NullSuffixForm

use of zemberek.morphology.lexicon.NullSuffixForm in project zemberek-nlp by ahmetaa.

the class DynamicSuffixProviderTest method testRegister.

@Test
public void testRegister() {
    DynamicSuffixProvider provider = new TestSuffixProvider();
    Suffix sf1 = new Suffix("sf1");
    Suffix sf2 = new Suffix("sf2");
    Suffix sf3 = new Suffix("sf3");
    Suffix sf4 = new Suffix("sf4");
    Suffix sf5 = new Suffix("sf4");
    Suffix sf6 = new Suffix("sf6");
    Suffix sf7 = new Suffix("sf6");
    Suffix sf8 = new Suffix("sf6");
    Suffix sf9 = new Suffix("sf6");
    SuffixForm frm1 = provider.getForm("fs1", sf1, "abc");
    SuffixForm frm2_1 = provider.getForm("fs2-1", sf2, "ali");
    SuffixForm frm2_2 = provider.getForm("fs2-2", sf2, "kaan");
    SuffixForm frm4 = provider.getForm("fs4", sf4, "akin");
    SuffixForm frm5 = provider.getForm("frm5", sf6, "dadada");
    SuffixForm frm6 = provider.getForm("frm6", sf9, "aguagu");
    SuffixFormTemplate tmp1 = provider.getTemplate("tmp1", sf3, TerminationType.TRANSFER);
    SuffixFormTemplate tmp2 = provider.getTemplate("tmp2", sf5, TerminationType.TRANSFER);
    SuffixFormTemplate tmp3 = provider.getTemplate("tmp3", sf7, TerminationType.TRANSFER);
    SuffixFormTemplate tmp4 = provider.getTemplate("tmp4", sf8, TerminationType.TRANSFER);
    // 
    // /--frm5.............
    // /         \         \
    // -->frm4- -.........->frm2_2
    // /          \ \          /
    // tmp2-->frm1--->tmp1------
    // \      \      /        \
    // \      .....C......-> frm2_1        --- Direct link  ... indirect link.
    // \........ /........../
    // \
    // \---- tmp3----tmp4--- frm6
    // \............./......./
    // 
    tmp2.connections.add(frm4, frm1, frm5, tmp3);
    tmp2.indirectConnections.add(tmp1, frm2_2, frm2_1, tmp4, frm6);
    frm1.connections.add(tmp1);
    frm1.indirectConnections.add(frm2_2);
    frm4.connections.add(tmp1);
    frm4.indirectConnections.add(frm2_1);
    frm5.connections.add(tmp1);
    frm5.indirectConnections.add(frm2_2);
    tmp1.connections.add(frm2_1, frm2_2);
    tmp3.connections.add(tmp4);
    tmp3.indirectConnections.add(frm6);
    tmp4.connections.add(frm6);
    // register tmp2. It should not effect the graph.
    provider.registerForm(tmp2);
    Assert.assertEquals(0, provider.getFormCount());
    // after registering frm1, there should be 2 forms registered internally. frm1 and a nullForm from tmp1
    provider.registerForm(frm1);
    Assert.assertEquals(2, provider.getFormCount());
    // after registering frm4, there should be 4 forms registered internally. frm4 and another nullForm from tmp1 with different connections.
    provider.registerForm(frm4);
    Assert.assertEquals(4, provider.getFormCount());
    // if we attempt to re-register, should not effect the graph.
    provider.registerForm(frm4);
    Assert.assertEquals(4, provider.getFormCount());
    // we register frm5. it should not generate a null morpheme from tmp1 because it was already generated when frm4 is registered.
    provider.registerForm(frm5);
    Assert.assertEquals(5, provider.getFormCount());
    provider.registerForms(frm2_1, frm2_2);
    Assert.assertEquals(7, provider.getFormCount());
    // now we generate a nullmorpheme from tmp2 and register it. We apply a constraint so frm1 is out of connections
    SuffixData constraint = new SuffixData(tmp2.connections).remove(frm1).add(tmp2.indirectConnections);
    NullSuffixForm null2_1 = provider.generateNullFormFromTemplate(tmp2, constraint);
    Assert.assertFalse(null2_1.connections.contains(frm1));
    Assert.assertTrue(null2_1.connections.contains(frm4));
    Assert.assertTrue(null2_1.connections.contains(frm5));
    provider.registerForm(null2_1.copy());
    // null morphemes for tmp3 and tm4 also will be registered internally.
    Assert.assertEquals(10, provider.getFormCount());
    // we try to do the same. it should not effect the graph.
    SuffixData constraint2 = new SuffixData(tmp2.connections).remove(frm1).add(tmp2.indirectConnections);
    NullSuffixForm null2_2 = provider.generateNullFormFromTemplate(tmp2, constraint2);
    Assert.assertEquals(null2_1, null2_2);
    provider.registerForm(null2_2.copy());
    Assert.assertEquals(10, provider.getFormCount());
    SuffixData constraint3 = new SuffixData(tmp2.allConnections());
    NullSuffixForm null2_3 = provider.generateNullFormFromTemplate(tmp2, constraint3);
    provider.registerForm(null2_3.copy());
    Assert.assertNotSame(null2_3, null2_2);
    Assert.assertEquals(11, provider.getFormCount());
}

Example 2 with NullSuffixForm

use of zemberek.morphology.lexicon.NullSuffixForm in project zemberek-nlp by ahmetaa.

the class DynamicSuffixProvider method generateNullFormFromTemplate.

protected NullSuffixForm generateNullFormFromTemplate(SuffixFormTemplate templateForm, SuffixData constraints) {
    NullSuffixForm nullForm = new NullSuffixForm(-1, "", templateForm);
    nullForm.connections = new SuffixData(templateForm.connections).retain(constraints);
    nullForm.indirectConnections = new SuffixData(templateForm.indirectConnections).retain(constraints);
    if (nullFormsUnprocessed.containsKey(nullForm)) {
        return nullFormsUnprocessed.get(nullForm);
    } else {
        nullForm.index = getNewIndex();
        nullForm.id = idMaker.get(templateForm.id);
        nullFormsUnprocessed.put(nullForm, nullForm);
        return nullForm;
    }
}

Example 3 with NullSuffixForm

use of zemberek.morphology.lexicon.NullSuffixForm in project zemberek-nlp by ahmetaa.

the class DynamicSuffixProvider method registerForm.

protected void registerForm(SuffixForm formSet) {
    // duplicates of newly generated FormSets.
    if (formSet instanceof SuffixFormTemplate) {
        formLookupByName.put(formSet.getId(), formSet);
        return;
    }
    if (suffixForms.containsKey(formSet)) {
        return;
    }
    SuffixData allConnections = formSet.allConnections();
    List<SuffixForm> templateFormsToRemove = new ArrayList<>();
    List<SuffixForm> nullFormsToRegister = new ArrayList<>();
    for (SuffixForm connection : formSet.connections) {
        if (connection instanceof SuffixFormTemplate) {
            NullSuffixForm nullForm = generateNullFormFromTemplate((SuffixFormTemplate) connection, new SuffixData(allConnections)).copy();
            nullFormsToRegister.add(nullForm);
            templateFormsToRemove.add(connection);
        }
    }
    formSet.connections.remove(templateFormsToRemove);
    // we dont need indirect connection data anymore.
    formSet.indirectConnections.clear();
    formSet.connections.add(nullFormsToRegister);
    if (formSet.index != -1) {
        formSet.index = getNewIndex();
    }
    suffixForms.put(formSet, formSet);
    formLookupByName.put(formSet.getId(), formSet);
    for (SuffixForm form : nullFormsToRegister) {
        registerForm(form);
    }
}

Example 4 with NullSuffixForm

use of zemberek.morphology.lexicon.NullSuffixForm in project zemberek-nlp by ahmetaa.

the class TurkishSuffixes method getRootSet.

@Override
public SuffixForm getRootSet(DictionaryItem item, SuffixData suffixConstraint) {
    if (suffixConstraint.isEmpty()) {
        switch(item.primaryPos) {
            case Noun:
                if (item.hasAttribute(RootAttribute.CompoundP3sg)) {
                    return Noun_Comp_P3sg;
                }
                if (item.hasAttribute(RootAttribute.CompoundP3sgRoot)) {
                    return Noun_Comp_P3sg_Root;
                }
                switch(item.secondaryPos) {
                    case ProperNoun:
                        return ProperNoun_Default;
                    case Time:
                        return Noun_Time_Default;
                    default:
                        return Noun_Default;
                }
            case Adjective:
                return Adj_Default;
            case Verb:
                return Verb_Default;
            case Adverb:
                return Adv_Default;
            case Numeral:
                return Numeral_Default;
            case Interjection:
                return Interj_Default;
            case Question:
                return Ques_Default;
            case Conjunction:
                return Conj_Default;
            case PostPositive:
                return Postp_Default;
            case Punctuation:
                return Punc_Default;
            case Determiner:
                return Det_Default;
            case Duplicator:
                return Dup_Default;
            case Pronoun:
                switch(item.secondaryPos) {
                    case DemonstrativePron:
                        return DemonsPron_Default;
                    case QuantitivePron:
                        return QuantPron_Default;
                    case QuestionPron:
                        return QuesPron_Default;
                    case ReflexivePron:
                        return ReflexPron_Default;
                    default:
                        return PersPron_Default;
                }
            default:
                return Noun_Default;
        }
    } else {
        SuffixFormTemplate template;
        switch(item.primaryPos) {
            case Noun:
                template = Noun_TEMPLATE;
                break;
            case Adjective:
                template = Adj_TEMPLATE;
                break;
            case Verb:
                template = Verb_TEMPLATE;
                break;
            case PostPositive:
                template = Postp_Template;
                break;
            case Pronoun:
                if (item.secondaryPos == SecondaryPos.DemonstrativePron) {
                    template = DemonsPron_TEMPLATE;
                } else if (item.secondaryPos == SecondaryPos.QuantitivePron) {
                    template = QuantPron_TEMPLATE;
                } else if (item.secondaryPos == SecondaryPos.QuestionPron) {
                    template = QuesPron_TEMPLATE;
                } else {
                    return PersPron_TEMPLATE;
                }
                break;
            default:
                template = Noun_TEMPLATE;
        }
        NullSuffixForm copy = generateNullFormFromTemplate(template, suffixConstraint).copy();
        registerForm(copy);
        return copy;
    }
}

Example 5 with NullSuffixForm

use of zemberek.morphology.lexicon.NullSuffixForm in project zemberek-nlp by ahmetaa.

the class ZemberekNlpScripts method generateSuffixNames.

@Test
@Ignore("Not a Test.")
public void generateSuffixNames() throws IOException {
    TurkishSuffixes suffixes = new TurkishSuffixes();
    List<SuffixForm> forms = new ArrayList<>();
    for (SuffixForm form : suffixes.getAllForms()) {
        if (form instanceof NullSuffixForm) {
            continue;
        }
        forms.add(form);
    }
    forms.sort(Comparator.comparing(SuffixForm::getId));
    List<String> result = forms.stream().map(s -> s.id).collect(Collectors.toList());
    Files.write(Paths.get("suffix-list"), result);
}