Merge pull request #38 from bretttolbert/parser-refactoring

Cont. Refactoring

Author: bretttolbert

CHANGELOG.md

@@ -2,6 +2,7 @@
 
 - 1.11.7 [TBR]
   - Cont. refactoring
+  - Separating XML parsing code from logical types
 
 - 1.11.6 [26 October 2025]
   - Fixed Voseo conjugation for irregular verb `ser` for the subjuntivo (no vowel accents)

tests/test_inflectors/test_inflector_ca.py

@@ -4,8 +4,9 @@
 from verbecc.src.defs.types.lang_code import LangCodeISO639_1 as Lang
 from verbecc.src.defs.types.gender import Gender
 from verbecc.src.defs.types.person import Person
+from verbecc.src.defs.types.data.tense_template import TenseTemplate
 from verbecc.src.conjugator.conjugator import Conjugator
-from verbecc.src.parsers.tense_template import TenseTemplate
+from verbecc.src.parsers.tense_template_parser import TenseTemplateParser
 from verbecc.src.defs.types.alternates_behavior import AlternatesBehavior
 
 
@@ -2985,7 +2986,7 @@ def test_inflector_ca_conjugate_simple_mood_tense(cg):
         </present>""",
         parser=None,
     )
-    tense_template = TenseTemplate(Lang.ca, mood, tense_elem)
+    tense_template = TenseTemplateParser(Lang.ca, mood).parse(tense_elem)
     out = cg._conjugate_simple_mood_tense(verb_stem, mood, tense, tense_template)
     assert len(out) == 6
     assert out == [

tests/test_inflectors/test_inflector_es.py

@@ -12,9 +12,9 @@
 )
 from verbecc.src.defs.types.lang.es.voseo_options import VoseoOptions
 from verbecc.src.conjugator.conjugator import Conjugator
-from verbecc.src.parsers.tense_template import TenseTemplate
+from verbecc.src.parsers.tense_template_parser import TenseTemplateParser
 
-cg = Conjugator(lang="es")
+cg = Conjugator(lang=Lang.es)
 
 
 def test_all_verbs_have_templates():
@@ -635,7 +635,7 @@ def test_inflector_es_conjugate_simple_mood_tense():
         </presente>""",
         parser=None,
     )
-    tense_template = TenseTemplate(Lang.es, mood, tense_elem)
+    tense_template = TenseTemplateParser(Lang.es, mood).parse(tense_elem)
     out = cg._conjugate_simple_mood_tense(verb_stem, mood, tense, tense_template)
     assert len(out) == 6
     assert out == [

tests/test_inflectors/test_inflector_fr.py

@@ -6,7 +6,7 @@
 from verbecc.src.defs.types.gender import Gender
 from verbecc.src.defs.types.person import Person
 from verbecc.src.conjugator.conjugator import Conjugator, AlternatesBehavior
-from verbecc.src.parsers.tense_template import TenseTemplate
+from verbecc.src.parsers.tense_template_parser import TenseTemplateParser
 from verbecc.src.defs.types.exceptions import ConjugatorError
 from verbecc.src.defs.types.conjugation import MoodsConjugation
 
@@ -68,7 +68,7 @@ def test_inflector_frverb_can_be_reflexive(cg, infinitive, expected_result):
 def test_inflector_fr_impersonal_verbs(cg):
     impersonal_verbs = [
         v.infinitive
-        for v in cg._inflector._verb_parser.verbs
+        for v in cg._inflector._verbs
         if cg._inflector._is_impersonal_verb(v.infinitive)
     ]
     assert set(impersonal_verbs) == set(
@@ -111,7 +111,7 @@ def test_inflector_fr_conjugate_simple_mood_tense(cg):
         </présent>""",
         parser=None,
     )
-    tense_template = TenseTemplate(Lang.fr, mood, tense_elem)
+    tense_template = TenseTemplateParser(Lang.fr, mood).parse(tense_elem)
     out = cg._conjugate_simple_mood_tense(verb_stem, mood, tense, tense_template)
     assert len(out) == 6
     assert out == [

tests/test_inflectors/test_inflector_ro.py

@@ -1,8 +1,8 @@
 import pytest
 
+from verbecc.src.conjugator.conjugator import Conjugator
 from verbecc.src.defs.types.gender import Gender
 from verbecc.src.defs.types.person import Person
-from verbecc.src.conjugator.conjugator import Conjugator
 from verbecc.src.defs.types.alternates_behavior import AlternatesBehavior
 
 

tests/test_mlconjug.py

@@ -8,7 +8,7 @@
 @pytest.fixture(scope="module")
 def verb_template_pairs():
     inf = InflectorFr()
-    yield [(v.infinitive, v.template) for v in inf._verb_parser.verbs]
+    yield [(v.infinitive, v.template) for v in inf._verbs]
 
 
 def test_extract_verb_features():

tests/test_parser/test_conjugation_template.py

@@ -2,7 +2,7 @@
 
 import pytest
 
-from verbecc.src.parsers.conjugation_template import ConjugationTemplate
+from verbecc.src.parsers.conjugation_template_parser import ConjugationTemplateParser
 from verbecc.src.defs.types.exceptions import ConjugationTemplateError
 from verbecc.src.defs.types.lang_code import LangCodeISO639_1 as Lang
 
@@ -11,11 +11,11 @@
 def test_template_invalid_tag_template(mock_template_elem):
     mock_template_elem.tag.return_value = "not-template"
     with pytest.raises(ConjugationTemplateError):
-        template = ConjugationTemplate(Lang.fr, mock_template_elem)
+        template = ConjugationTemplateParser(Lang.fr).parse(mock_template_elem)
 
 
 @patch("lxml.etree._Element")
 def test_template_invalid_tag_name(mock_template_elem):
     mock_template_elem.get.return_value = "not-name"
     with pytest.raises(ConjugationTemplateError):
-        template = ConjugationTemplate(Lang.ca, mock_template_elem)
+        template = ConjugationTemplateParser(Lang.ca).parse(mock_template_elem)

tests/test_parser/test_tense_and_person.py

@@ -1,6 +1,7 @@
 from lxml import etree
 
-from verbecc.src.parsers.tense_template import TenseTemplate
+from verbecc.src.parsers.tense_template_parser import TenseTemplateParser
+from verbecc.src.defs.types.data.tense_template import TenseTemplate
 from verbecc.src.defs.types.lang_code import LangCodeISO639_1 as Lang
 from verbecc.src.defs.types.tense import Tense
 
@@ -17,9 +18,9 @@ def test_tense_and_person():
         </présent>"""
     tense_elem: etree._Element = etree.fromstring(tense_elem_str)
     tense = Tense.fr.Présent
-    tense_template = TenseTemplate(Lang.fr, mood, tense_elem)
+    tense_template = TenseTemplateParser(Lang.fr, mood).parse(tense_elem)
     assert tense_template.mood == mood
-    assert tense_template.name == str(tense.value)
+    assert tense_template.tense == tense
     assert tense_template.person_endings[0].get_ending() == "ie"
     assert tense_template.person_endings[0].get_alternate_ending_if_available() == "ye"
     assert tense_template.person_endings[0].get_person() == "1s"

tests/test_parser/test_verb_parser.py

@@ -0,0 +1,15 @@
+from unittest.mock import patch
+
+import pytest
+
+from verbecc.src.parsers.verb_parser import VerbParser
+from verbecc.src.defs.types.exceptions import VerbsParserError
+from verbecc.src.defs.types.lang_code import LangCodeISO639_1 as Lang
+
+
+@patch("lxml.etree._Element")
+def test_verb_invalid_xml(mock_v_elem):
+    v = VerbParser()
+    mock_v_elem.tag.return_value = "not-v"
+    with pytest.raises(VerbsParserError):
+        v.parse(mock_v_elem)

tests/test_parser/test_verbs_parser.py

@@ -1,48 +1,8 @@
-from unittest.mock import patch
-
-import pytest
-
 from verbecc.src.parsers.verbs_parser import VerbsParser
-from verbecc.src.parsers.verb import Verb
-from verbecc.src.defs.types.exceptions import VerbsParserError
+from verbecc.src.defs.types.lang_code import LangCodeISO639_1 as Lang
 
 
 def test_verbs_parser():
-    vp = VerbsParser()
-    assert len(vp.verbs) >= 7000
-
-
-def test_verb():
-    vp = VerbsParser()
-    verb = vp.find_verb_by_infinitive("manger")
-    assert verb.infinitive == "manger"
-    assert verb.template == "man:ger"
-    assert verb.translation_en == "eat"
-
-
-def test_verb_two():
-    vp = VerbsParser()
-    verb = vp.find_verb_by_infinitive("abattre")
-    assert verb.infinitive == "abattre"
-    assert verb.template == "bat:tre"
-    assert verb.translation_en == "tear down"
-
-
-@patch("lxml.etree._Element")
-def test_verb_invalid_xml(mock_v_elem):
-    mock_v_elem.tag.return_value = "not-v"
-    with pytest.raises(VerbsParserError):
-        v = Verb(mock_v_elem)
-
-
-@pytest.mark.parametrize(
-    "query,expected_matches",
-    [
-        ("mang", ["mangeotter", "manger"]),
-        ("Mang", ["mangeotter", "manger"]),
-    ],
-)
-def test_get_verbs_that_start_with(query, expected_matches):
-    vp = VerbsParser()
-    matches = vp.get_verbs_that_start_with(query)
-    assert matches == expected_matches
+    vp = VerbsParser(Lang.fr)
+    verbs = vp.parse()
+    assert len(verbs) >= 7000