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

Example 1 with SuffixForm

use of zemberek.morphology.lexicon.SuffixForm 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 SuffixForm

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

the class DynamicSuffixProviderTest method testCausativeMock.

@Test
public void testCausativeMock() {
    DynamicSuffixProvider provider = new TestSuffixProvider();
    Suffix verb = new Suffix("verb");
    Suffix causative = new Suffix("causative");
    Suffix positive = new Suffix("positive");
    Suffix future = new Suffix("future");
    SuffixForm caus_t = provider.getForm("causative-t", causative, "t");
    SuffixForm caus_tir = provider.getForm("causative-tir", causative, "tir");
    SuffixForm future_acak = provider.getForm("future-acak", future, "acak");
    SuffixFormTemplate verb_temp = provider.getTemplate("verb_temp", verb);
    SuffixFormTemplate verb2verb = provider.getTemplate("verb2verb", verb);
    SuffixFormTemplate positive_temp = provider.getTemplate("positive_temp", positive);
    verb_temp.connections.add(positive_temp, verb2verb);
    verb_temp.indirectConnections.add(caus_t, caus_tir, future_acak);
    verb2verb.connections.add(caus_t, caus_tir);
    caus_t.connections.add(positive_temp, verb2verb);
    caus_t.indirectConnections.add(future_acak, caus_tir);
    caus_tir.connections.add(positive_temp, verb2verb);
    caus_tir.indirectConnections.add(future_acak, caus_t);
    positive_temp.connections.add(future_acak);
    provider.registerForms(verb_temp, verb2verb, caus_tir, caus_t, positive_temp, future_acak);
    Assert.assertEquals(6, provider.getFormCount());
    provider.dumpPath(caus_t, 2);
    provider.dumpPath(caus_tir, 2);
}

Example 3 with SuffixForm

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

the class SuffixDataTest method equalityTest.

@Test
public void equalityTest() {
    Suffix suffix = new Suffix("SUFFIX");
    SuffixForm sf1 = new SuffixForm(1, "sf1", suffix, "lAr", TerminationType.TERMINAL);
    SuffixForm sf2 = new SuffixForm(1, "sf1", suffix, "lAr", TerminationType.TERMINAL);
    Assert.assertTrue(sf1.equals(sf2));
    Assert.assertTrue(sf2.equals(sf1));
    SuffixForm sf3 = new SuffixForm(3, "sf3", suffix, "k", TerminationType.TERMINAL);
    Assert.assertFalse(sf1.equals(sf3));
    sf1.connections.add(sf3);
    Assert.assertFalse(sf1.equals(sf2));
    sf2.connections.add(sf3);
    Assert.assertTrue(sf1.equals(sf2));
    sf1.connections.remove(sf3);
    SuffixForm sf4 = new SuffixForm(4, "sf4", suffix, "lAr", TerminationType.NON_TERMINAL);
    Assert.assertFalse(sf4.equals(sf2));
}

Example 4 with SuffixForm

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

the class SuffixFormSetTest method equalityTest.

@Test
public void equalityTest() {
    Suffix suffix = new Suffix("SUFFIX");
    SuffixForm sf1 = new SuffixForm(1, "sf1", suffix, "lAr", TerminationType.TERMINAL);
    SuffixForm sf2 = new SuffixForm(2, "sf2", suffix, "lAr", TerminationType.TERMINAL);
    SuffixForm sf3 = new SuffixForm(3, "sf3", suffix, "k", TerminationType.TERMINAL);
    sf1.index = 0;
    sf2.index = 1;
    sf3.index = 100;
    SuffixData data1 = new SuffixData(sf1, sf2);
    Assert.assertTrue(data1.contains(sf1));
    Assert.assertTrue(data1.contains(sf2));
}

Example 5 with SuffixForm

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

the class DynamicLexiconGraph method generateNodeConnections.

/**
 * This method generates connections of a SuffixSurfaceNode. A SuffixSurfaceNode is surfaceForm of
 * a SuffixForm. (Suffix form ->A1pl_lAr, SuffixSurfaceNode is lar) We already know the
 * morphotactics of SuffixForms. So we get the specific SuffixNodes that can be connected to a
 * particular SuffixSurfaceNode. Such as, SuffixForm P1sg_Im can follow A1pl_lAr. Therefore, the
 * SuffixSurfaceNode lar can only connect to "ım" surfaceNode of the P1sg_Im suffixForm. Here this
 * connection is generated, as the surfaceNode reference in the successor form is added to this
 * surfaceNode. However, if surfaceNode to be connected does not exist, it is generated as well.
 * And once it is generated and connection is provided Recursively connections to that surfaceNode
 * are also generated.
 *
 * @param surfaceNode Node that connections to successive nodes will be generated.
 */
private void generateNodeConnections(SuffixSurfaceNode surfaceNode) {
    // get the successive form sets for this surfaceNode.
    SuffixData successors = surfaceNode.suffixForm.connections;
    // iterate over form sets.
    for (SuffixForm successiveForm : successors) {
        // get the nodes for the  suffix form.
        List<SuffixSurfaceNode> nodesInSuccessor = suffixSurfaceNodeGenerator.generate(surfaceNode.attributes, surfaceNode.expectations, surfaceNode.exclusiveSuffixData, successiveForm);
        for (SuffixSurfaceNode surfaceNodeInSuccessor : nodesInSuccessor) {
            // if there are expectations for the surfaceNode, check if it matches with the attributes of the surfaceNode in successor.
            if (!surfaceNode.expectations.isEmpty()) {
                if (!expectationsMatches(surfaceNode, surfaceNodeInSuccessor)) {
                    continue;
                }
            }
            boolean recurse = false;
            if (!nodeExists(successiveForm, surfaceNodeInSuccessor)) {
                recurse = true;
            }
            surfaceNodeInSuccessor = addOrReturnExisting(successiveForm, surfaceNodeInSuccessor);
            surfaceNode.addSuccessorNode(surfaceNodeInSuccessor);
            if (recurse) {
                generateNodeConnections(surfaceNodeInSuccessor);
            }
        }
    }
}