Catalan WIP, fixed issue with strip-accent and infinitive mood

CHANGELOG.md

@@ -3,7 +3,7 @@
 - 1.9.5 [24 December 2023]
   - Improved Catalan Support
     - Added more missing templates (7+)
-    - TODO: Still missing 11 templates for 35 out of 8616 verbs
+    - TODO: Still missing 9 templates for 28 out of 8616 verbs
     - See `test_inflector_ca.test_all_verbs_have_templates`
 
 - 1.9.4 [23 December 2023]

tests/test_inflector_ca.py

@@ -14,7 +14,7 @@ def test_all_verbs_have_templates():
     verbs = cg.get_verbs()
     assert len(verbs) == 8616
     template_names = cg.get_template_names()
-    assert len(template_names) == 60
+    assert len(template_names) == 61
     missing_verbs = []
     missing_templates = set()
     for verb in verbs:
@@ -2498,6 +2498,45 @@ def test_find_verb_by_infinitive():
             "m",
             ["surt", "surti", "sortim", "sortiu", "surtin"],
         ),
+        (
+            "témer",
+            "indicatiu",
+            "present",
+            False,
+            "f",
+            [
+                "jo temo",
+                "tu tems",
+                "ella tem",
+                "nosaltres temem",
+                "vosaltres temeu",
+                "elles temen",
+            ],
+        ),
+        (
+            "témer",
+            "infinitiu",
+            "infinitiu-present",
+            False,
+            "m",
+            ["témer"],
+        ),
+        (
+            "témer",
+            "gerundi",
+            "gerundi",
+            False,
+            "m",
+            ["tement"],
+        ),
+        (
+            "témer",
+            "particip",
+            "particip",
+            False,
+            "f",
+            ["temut", "temuda", "temuts", "temudes"],
+        ),
     ],
 )
 def test_inflector_ca_conjugate_mood_tense(
@@ -2528,10 +2567,9 @@ def test_inflector_ca_get_conj_obs_2():
 def test_inflector_ca_get_conj_obs_3():
     co = cg._inflector._get_conj_obs("pertànyer")
     assert co.verb.infinitive == "pertànyer"
-    assert (
-        co.verb_stem == "pertany"
-    )  # modify-stem="strip-accents" changes "pertàny" to "pertany"
+    assert co.verb_stem == "pertàny"
     assert co.template.name == "pertàny:er"
+    assert co.template.modify_stem == "strip-accents"
 
 
 def test_inflector_ca_get_verb_stem_from_template_name():

verbecc/data/conjugations-ca.xml

@@ -922,6 +922,101 @@ Note: this verb is only conjugated in the third-person.
 	</Particip>
 </template>
 
+<!-- TODO: Inherit from perdre 
+Literally exactly the same except for infinitive ending.
+-->
+<template name="tém:er" modify-stem="strip-accents">
+	<Indicatiu>
+		<present>
+			<p><i>o</i></p>
+			<p><i>s</i></p>
+			<p><i></i></p>
+			<p><i>em</i></p>
+			<p><i>eu</i></p>
+			<p><i>en</i></p>
+		</present>
+		<imperfet>
+			<p><i>ia</i></p>
+			<p><i>ies</i></p>
+			<p><i>ia</i></p>
+			<p><i>íem</i></p>
+			<p><i>íeu</i></p>
+			<p><i>ien</i></p>
+		</imperfet>
+		<passat-simple>
+			<p><i>í</i></p>
+			<p><i>eres</i></p>
+			<p><i>é</i></p>
+			<p><i>érem</i></p>
+			<p><i>éreu</i></p>
+			<p><i>eren</i></p>
+		</passat-simple>
+		<futur>
+			<p><i>é</i></p>
+			<p><i>às</i></p>
+			<p><i>à</i></p>
+			<p><i>em</i></p>
+			<p><i>eu</i></p>
+			<p><i>an</i></p>
+		</futur>
+	</Indicatiu>
+	<Subjuntiu>
+		<present>
+			<p><i>i</i></p>
+			<p><i>is</i></p>
+			<p><i>i</i></p>
+			<p><i>em</i></p>
+			<p><i>eu</i></p>
+			<p><i>in</i></p>
+		</present>
+		<imperfet>
+			<p><i>és</i></p>
+			<p><i>essis</i></p>
+			<p><i>és</i></p>
+			<p><i>éssim</i></p>
+			<p><i>éssiu</i></p>
+			<p><i>essin</i></p>
+		</imperfet>
+	</Subjuntiu>
+	<Imperatiu>
+		<imperatiu-present>
+			<p><i></i></p>
+			<p><i>i</i></p>
+			<p><i>em</i></p>
+			<p><i>eu</i></p>
+			<p><i>in</i></p>
+		</imperatiu-present>
+	</Imperatiu>
+	<Condicional>
+		<present>
+			<p><i>ria</i></p>
+			<p><i>ries</i></p>
+			<p><i>ria</i></p>
+			<p><i>ríem</i></p>
+			<p><i>ríeu</i></p>
+			<p><i>rien</i></p>
+		</present>
+	</Condicional>
+	<Infinitiu>
+		<infinitiu-present>
+			<p><i>er</i></p>
+		</infinitiu-present>
+	</Infinitiu>
+	<Gerundi>
+		<gerundi>
+			<p><i>ent</i></p>
+		</gerundi>
+	</Gerundi>
+	<Particip>
+		<particip>
+			<p><i>ut</i></p>
+			<p><i>uda</i></p>
+			<p><i>uts</i></p>
+			<p><i>udes</i></p>
+		</particip>
+	</Particip>
+</template>
+
 <!-- TODO: Inherit from perd:re -->
 <template name="mou:re" modify-stem="strip-accents">
 	<Indicatiu>

verbecc/data/verbs-ca.xml

@@ -4924,7 +4924,7 @@
 <v><i>esprimatxar</i><t>cant:ar</t><en></en></v>
 <v><i>esprintar</i><t>cant:ar</t><en></en></v>
 <v><i>esprovar</i><t>cant:ar</t><en></en></v>
-<v><i>esprémer</i><t>t:émer</t><en></en></v>
+<v><i>esprémer</i><t>tém:er</t><en></en></v>
 <v><i>espuar</i><t>evacu:ar</t><en></en></v>
 <v><i>espuntar</i><t>cant:ar</t><en></en></v>
 <v><i>espuntegar</i><t>preg:ar</t><en></en></v>
@@ -7047,7 +7047,7 @@
 <v><i>provenir</i><t>prev:enir</t><en></en></v>
 <v><i>provocar</i><t>tren:car</t><en></en></v>
 <v><i>pruir</i><t>entrellu:ir</t><en></en></v>
-<v><i>prémer</i><t>t:émer</t><en></en></v>
+<v><i>prémer</i><t>tém:er</t><en></en></v>
 <v><i>pseudomorfitzar</i><t>cant:ar</t><en></en></v>
 <v><i>psicoanalitzar</i><t>cant:ar</t><en></en></v>
 <v><i>publicar</i><t>tren:car</t><en></en></v>
@@ -8198,7 +8198,7 @@
 <v><i>termofixar</i><t>cant:ar</t><en></en></v>
 <v><i>terrabastejar</i><t>enve:jar</t><en></en></v>
 <v><i>terraplenar</i><t>cant:ar</t><en></en></v>
-<v><i>terratrémer</i><t>t:émer</t><en></en></v>
+<v><i>terratrémer</i><t>tém:er</t><en></en></v>
 <v><i>terrejar</i><t>enve:jar</t><en></en></v>
 <v><i>terrenejar</i><t>enve:jar</t><en></en></v>
 <v><i>terrenyar</i><t>cant:ar</t><en></en></v>
@@ -8416,7 +8416,7 @@
 <v><i>trullar</i><t>cant:ar</t><en></en></v>
 <v><i>trumfar</i><t>cant:ar</t><en></en></v>
 <v><i>truncar</i><t>tren:car</t><en></en></v>
-<v><i>trémer</i><t>t:émer</t><en></en></v>
+<v><i>trémer</i><t>tém:er</t><en></en></v>
 <v><i>tuberculitzar</i><t>cant:ar</t><en></en></v>
 <v><i>tudar</i><t>cant:ar</t><en></en></v>
 <v><i>tuejar</i><t>enve:jar</t><en></en></v>
@@ -8428,7 +8428,7 @@
 <v><i>tustar</i><t>cant:ar</t><en></en></v>
 <v><i>tutejar</i><t>enve:jar</t><en></en></v>
 <v><i>tutelar</i><t>cant:ar</t><en></en></v>
-<v><i>témer</i><t>perd:re</t><en></en></v>
+<v><i>témer</i><t>tém:er</t><en>fear</en></v>
 <v><i>tòrcer</i><t>comen:çar</t><en>twist</en></v>
 <v><i>ubicar</i><t>tren:car</t><en></en></v>
 <v><i>ucar</i><t>tren:car</t><en></en></v>

verbecc/inflector.py

@@ -147,8 +147,6 @@ def _get_conj_obs(self, infinitive: str) -> ConjugationObjects:
         verb_stem = self._get_verb_stem_from_template_name(
             verb.infinitive, template.name
         )
-        if template.modify_stem == "strip-accents":
-            verb_stem = strip_accents(verb_stem)
         return Inflector.ConjugationObjects(
             infinitive, verb, template, verb_stem, is_reflexive
         )
@@ -216,6 +214,9 @@ def _conjugate_mood_tense(
                 is_reflexive=co.is_reflexive,
                 alternate=alternate,
                 gender=gender,
+                modify_stem_strip_accents=bool(
+                    co.template.modify_stem == "strip-accents"
+                ),
             )
 
     def _get_tenses_conjugated_without_pronouns(self) -> List[str]:
@@ -229,12 +230,18 @@ def _get_auxilary_verb(
     def _is_auxilary_verb_inflected(self, auxilary_verb: str):
         return False
 
+    def _get_infinitive_mood_name(self) -> str:
+        return "infinitive"
+
     def _get_indicative_mood_name(self) -> str:
         return "indicative"
 
     def _get_subjunctive_mood_name(self) -> str:
         return "subjunctive"
 
+    def _get_conditional_mood_name(self) -> str:
+        return "conditional"
+
     def _get_participle_mood_name(self) -> str:
         return "partiple"
 
@@ -288,9 +295,9 @@ def _combine_verb_stem_and_ending(self, verb_stem: str, ending: str) -> str:
         modify-stem="strip-accents"
         So the stem "pertàny" becomes "pertany"
 
-        modify-stem="strip-accents" is applied in the constructor of ConjugationObjects,
-        and so the verb_stem argument to _conjugate_simple_mood_tense and consequently
-        this function is presumed to already have applicable modifications applied.
+        modify-stem="strip-accents" is applied in _conjugate_simple_mood_tense,
+        so the verb_stem argument to this function is presumed to already have
+        applicable modifications applied.
 
         Next, the "pertàny:er" template endings will use the new delete operator
         to delete one letter from the end of the stem.
@@ -316,7 +323,10 @@ def _conjugate_simple_mood_tense(
         is_reflexive: bool = False,
         alternate: bool = False,
         gender: str = "m",
+        modify_stem_strip_accents: bool = False,
     ) -> List[str]:
+        if modify_stem_strip_accents and mood_name != self._get_infinitive_mood_name():
+            verb_stem = strip_accents(verb_stem)
         ret = []
         tense_name = tense_template.name
         if tense_name in self._get_tenses_conjugated_without_pronouns():

verbecc/inflector_ca.py

@@ -61,6 +61,18 @@ def _get_auxilary_verb(
     ) -> str:
         return "haver"
 
+    def _get_infinitive_mood_name(self):
+        return "infinitiu"
+
+    def _get_indicative_mood_name(self):
+        return "indicatiu"
+
+    def _get_subjunctive_mood_name(self):
+        return "subjuntiu"
+
+    def _get_conditional_mood_name(self):
+        return "condicional"
+
     def _get_participle_mood_name(self) -> str:
         return "particip"
 

verbecc/inflector_es.py

@@ -61,6 +61,18 @@ def _get_auxilary_verb(
     ) -> str:
         return "haber"
 
+    def _get_infinitive_mood_name(self):
+        return "infinitivo"
+
+    def _get_indicative_mood_name(self):
+        return "indicativo"
+
+    def _get_subjunctive_mood_name(self):
+        return "subjuntivo"
+
+    def _get_conditional_mood_name(self):
+        return "condicional"
+
     def _get_participle_mood_name(self) -> str:
         return "participo"
 

verbecc/inflector_fr.py

@@ -149,9 +149,18 @@ def _get_auxilary_verb(
     def _is_auxilary_verb_inflected(self, auxilary_verb: str) -> bool:
         return auxilary_verb == "être"
 
+    def _get_infinitive_mood_name(self) -> str:
+        return "infinitif"
+
+    def _get_indicative_mood_name(self) -> str:
+        return "indicatif"
+
     def _get_subjunctive_mood_name(self) -> str:
         return "subjonctif"
 
+    def _get_conditional_mood_name(self) -> str:
+        return "conditionnel"
+
     def _get_participle_mood_name(self) -> str:
         return "participe"
 

verbecc/inflector_it.py

@@ -67,9 +67,18 @@ def _get_tenses_conjugated_without_pronouns(self):
     def _get_auxilary_verb(self, co, mood_name, tense_name):
         return "avere"
 
+    def _get_infinitive_mood_name(self):
+        return "infinitivo"
+
+    def _get_indicative_mood_name(self):
+        return "indicativo"
+
     def _get_subjunctive_mood_name(self):
         return "congiuntivo"
 
+    def _get_conditional_mood_name(self):
+        return "condizionale"
+
     def _get_participle_mood_name(self):
         return "participio"
 

verbecc/inflector_pt.py

@@ -108,9 +108,18 @@ def _get_auxilary_verb(
     ) -> str:
         return "ter"
 
+    def _get_infinitive_mood_name(self):
+        return "infinitivo"
+
+    def _get_indicative_mood_name(self):
+        return "indicativo"
+
     def _get_subjunctive_mood_name(self) -> str:
         return "subjuntivo"
 
+    def _get_conditional_mood_name(self):
+        return "condicional"
+
     def _get_participle_mood_name(self) -> str:
         return "particípio"
 

verbecc/inflector_ro.py

@@ -70,12 +70,18 @@ def _get_auxilary_verb(self, co, mood_name, tense_name):
             return co.verb.infinitive
         return "avea"
 
+    def _get_infinitive_mood_name(self):
+        return "infinitiv"
+
     def _get_indicative_mood_name(self):
         return "indicativ"
 
     def _get_subjunctive_mood_name(self):
         return "conjunctiv"
 
+    def _get_conditional_mood_name(self):
+        raise NotImplementedError
+
     def _get_participle_mood_name(self):
         return "participiu"