Merge pull request #13 from bretttolbert/catalan-contd

improved support for Catalan

pyproject.toml

@@ -1,3 +1,7 @@
+[build-system]
+requires = ["setuptools"]
+build-backend = "setuptools.build_meta"
+
 [project]
 name = "verbecc"
 version = "1.9.0"
@@ -60,3 +64,6 @@ Changelog = "https://github.com/bretttolbert/verbecc/blob/master/CHANGELOG.md"
 
 [tool.setuptools.package-data]
 trained_models = ['verbecc/data/models/*']
+
+[tool.setuptools.packages]
+find = {}

tests/test_conjugator.py

@@ -8,15 +8,15 @@
 
 cg = conjugator.Conjugator(lang='fr')
 
-def test_get_verbs_list():
-    verbs_list = cg.get_verbs_list()
-    assert len(verbs_list) > 7000
-    assert 'parler' in verbs_list
+def test_get_infinitives():
+    infinitives = cg.get_infinitives()
+    assert len(infinitives) > 7000
+    assert 'parler' in infinitives
 
-def test_get_templates_list():
-    templates_list = cg.get_templates_list()
-    assert len(templates_list) >= 146
-    assert 'aim:er' in templates_list
+def test_get_template_names():
+    template_names = cg.get_template_names()
+    assert len(template_names) >= 146
+    assert 'aim:er' in template_names
 
 test_verbs = [
     (u"manger"), 
@@ -28,11 +28,10 @@ def test_get_templates_list():
     (u"pleuvoir")
 ]
 
-@pytest.mark.parametrize("infinitif", test_verbs)
-def test_conjugator_conjugate(infinitive):
-    for infinitive in test_verbs:
-        output = cg.conjugate(infinitive)
-        assert output
+@pytest.mark.parametrize("infinitive", test_verbs)
+def test_conjugator_conjugate_basic(infinitive):
+    output = cg.conjugate(infinitive)
+    assert output
 
 def test_conjugator_predict_conjugation_er_verb_indicative_present():
     if config.ml:

tests/test_inflector_ca.py

@@ -8,6 +8,22 @@
 
 cg = Conjugator(lang='ca')
 
+@pytest.mark.skip("known failure")
+def test_all_verbs_have_templates():
+    """Have not finished adding templates for all verbs, so this should fail"""
+    verbs = cg.get_verbs()
+    template_names = cg.get_template_names()
+    missing_templates = set()
+    for verb in verbs:
+        if verb.template not in template_names:
+            missing_templates.add(verb.template)
+    assert len(missing_templates) == 0
+
+def test_find_verb_by_infinitive():
+    v = cg.find_verb_by_infinitive('abandonar')
+    assert v.infinitive == 'abandonar'
+    assert v.template == 'cant:ar'
+
 test_ca_conjugate_mood_tense_data = [
     ('ser', 'indicatiu', 'present', 
         ['jo sóc', 'tu ets', 'ell és', 'nosaltres som', 'vosaltres sou', 'ells són']),
@@ -67,6 +83,20 @@
         ['jo hagi', 'tu hagis', 'ell hagi', 'nosaltres hàgim', 'vosaltres hàgiu', 'ells hagin']),
     ('tenir', 'indicatiu', 'present', 
         ['jo tinc', 'tu tens', 'ell té', 'nosaltres tenim', 'vosaltres teniu', 'ells tenen']),
+    ('fer', 'indicatiu', 'present', 
+        ['jo faig', 'tu fas', 'ell fa', 'nosaltres fem', 'vosaltres feu', 'ells fan']),
+    ('fer', 'indicatiu', 'imperfet', 
+        ['jo feia', 'tu feies', 'ell feia', 'nosaltres fèiem', 'vosaltres fèieu', 'ells feien']),
+    ('servir', 'indicatiu', 'present', 
+        ['jo serveixo', 'tu serveixes', 'ell serveix', 'nosaltres servim', 'vosaltres serviu', 'ells serveixen']),
+    ('veure', 'indicatiu', 'present', 
+        ['jo veig', 'tu veus', 'ell veu', 'nosaltres veiem', 'vosaltres veieu', 'ells veuen']),
+    ('abandonar', 'indicatiu', 'present', 
+        ['jo abandono', 'tu abandones', 'ell abandona', 'nosaltres abandonem', 'vosaltres abandoneu', 'ells abandonen']),
+    ('rebre', 'indicatiu', 'present', 
+        ['jo rebo', 'tu reps', 'ell rep', 'nosaltres rebem', 'vosaltres rebeu', 'ells reben']),
+    ('cabre', 'indicatiu', 'present', 
+        ['jo cabo', 'tu caps', 'ell cap', 'nosaltres cabem', 'vosaltres cabeu', 'ells caben']),
 ]
 
 @pytest.mark.parametrize("infinitive,mood,tense,expected_result",
@@ -78,11 +108,22 @@ def test_inflector_ca_get_conj_obs():
     co = cg._inflector._get_conj_obs('parlar')
     assert co.verb.infinitive == "parlar"
     assert co.verb_stem == "parl"
+    assert co.template.name == "cant:ar"
+
+def test_inflector_ca_get_conj_obs_2():
+    co = cg._inflector._get_conj_obs('abandonar')
+    assert co.verb.infinitive == "abandonar"
+    assert co.verb_stem == "abandon"
+    assert co.template.name == "cant:ar"
 
 def test_inflector_ca_get_verb_stem():
     verb_stem = cg._inflector._get_verb_stem(u"parlar", u"cant:ar")
     assert verb_stem == u"parl"
 
+def test_inflector_ca_get_verb_stem_2():
+    verb_stem = cg._inflector._get_verb_stem(u"abandonar", u"cant:ar")
+    assert verb_stem == u"abandon"
+
 def test_inflector_ca_conjugate_simple_mood_tense():
     verb_stem = u"parl"
     tense_elem = etree.fromstring(

tests/test_inflector_es.py

@@ -8,6 +8,15 @@
 
 cg = Conjugator(lang='es')
 
+def test_all_verbs_have_templates():
+    verbs = cg.get_verbs()
+    template_names = cg.get_template_names()
+    missing_templates = set()
+    for verb in verbs:
+        if verb.template not in template_names:
+            missing_templates.add(verb.template)
+    assert len(missing_templates) == 0
+
 # presente = Subjunctive Present (yo haya)
 # pretérito-perfecto = Subjunctive Perfect (yo haya habido)
 # pretérito-imperfecto-1 = Subjunctive Past 1 (yo hubiera)

tests/test_inflector_fr.py

@@ -3,10 +3,22 @@
 import pytest
 from lxml import etree
 
+from verbecc import Conjugator
 from verbecc import inflector_fr
 from verbecc.tense_template import TenseTemplate
 from verbecc.exceptions import ConjugatorError
 
+cg = Conjugator(lang='fr')
+
+def test_all_verbs_have_templates():
+    verbs = cg.get_verbs()
+    template_names = cg.get_template_names()
+    missing_templates = set()
+    for verb in verbs:
+        if verb.template not in template_names:
+            missing_templates.add(verb.template)
+    assert len(missing_templates) == 0
+
 inf = inflector_fr.InflectorFr()
 
 def test_add_subjunctive_relative_prounoun():

tests/test_inflector_it.py

@@ -6,6 +6,15 @@
 
 cg = Conjugator(lang='it')
 
+def test_all_verbs_have_templates():
+    verbs = cg.get_verbs()
+    template_names = cg.get_template_names()
+    missing_templates = set()
+    for verb in verbs:
+        if verb.template not in template_names:
+            missing_templates.add(verb.template)
+    assert len(missing_templates) == 0
+
 test_it_conjugate_mood_tense_data = [
     ('avere', 'indicativo', 'presente',
         ['io ho', 'tu hai', 'lui ha', 'noi abbiamo', 'voi avete', 'loro hanno']),

tests/test_inflector_pt.py

@@ -6,6 +6,15 @@
 
 cg = Conjugator(lang='pt')
 
+def test_all_verbs_have_templates():
+    verbs = cg.get_verbs()
+    template_names = cg.get_template_names()
+    missing_templates = set()
+    for verb in verbs:
+        if verb.template not in template_names:
+            missing_templates.add(verb.template)
+    assert len(missing_templates) == 0
+
 test_pt_conjugate_mood_tense_data = [
     ('ter', 'indicativo', 'presente',
         ['eu tenho', 'tu tens', 'ele tem', 'nós temos', 'vós tendes', 'eles têm']),

tests/test_inflector_ro.py

@@ -6,6 +6,15 @@
 
 cg = Conjugator(lang='ro')
 
+def test_all_verbs_have_templates():
+    verbs = cg.get_verbs()
+    template_names = cg.get_template_names()
+    missing_templates = set()
+    for verb in verbs:
+        if verb.template not in template_names:
+            missing_templates.add(verb.template)
+    assert len(missing_templates) == 0
+
 test_ro_conjugate_mood_tense_data = [
     ('avea', 'participiu', 'participiu', False,
         ['avut']),

verbecc/conjugations_parser.py

@@ -13,7 +13,7 @@
 class ConjugationsParser:
     def __init__(self, lang: str='fr'):
         self.templates: List[conjugation_template.ConjugationTemplate] = []
-        parser = etree.XMLParser(dtd_validation=True, encoding='utf-8')
+        parser = etree.XMLParser(dtd_validation=True, encoding='utf-8', remove_comments=True)
         tree = etree.parse(
             resource_filename("verbecc",
                               "data/conjugations-{}.xml".format(lang)),

verbecc/conjugator.py

@@ -2,6 +2,7 @@
 
 from typing import Dict, List
 
+from verbecc import verb
 from verbecc.exceptions import InvalidLangError
 from verbecc import conjugation_template
 
@@ -53,13 +54,19 @@ def conjugate_mood_tense(self, infinitive: str, mood_name: str, tense_name: str,
         return self._inflector.conjugate_mood_tense( 
             infinitive, mood_name, tense_name, alternate)
 
-    def get_verbs_list(self) -> List[str]:
-        return self._inflector.get_verbs_list()
+    def get_verbs(self) -> List[verb.Verb]:
+        return self._inflector.get_verbs()
 
-    def get_templates_list(self):
-        return self._inflector.get_templates_list()
+    def get_infinitives(self) -> List[str]:
+        return self._inflector.get_infinitives()
 
-    def find_verb_by_infinitive(self, infinitive: str):
+    def get_templates(self) -> List[conjugation_template.ConjugationTemplate]:
+        return self._inflector.get_templates()
+
+    def get_template_names(self) -> List[str]:
+        return self._inflector.get_template_names()
+
+    def find_verb_by_infinitive(self, infinitive: str) -> verb.Verb:
         return self._inflector.find_verb_by_infinitive(infinitive)
 
     def find_template(self, name: str) -> conjugation_template.ConjugationTemplate: