[de] add "pos_title" field

tatuylonen · Jan 27, 2025 · 613a9c7 · 613a9c7
1 parent cf77bca
commit 613a9c7
Show file tree

Hide file tree

Showing 4 changed files with 14 additions and 0 deletions.
diff --git a/src/wiktextract/extractor/de/models.py b/src/wiktextract/extractor/de/models.py
@@ -136,6 +136,7 @@ class WordEntry(BaseModelWrap):
 
     word: str = Field(description="word string")
     pos: str = Field(default="", description="Part of speech type")
+    pos_title: str = ""
     other_pos: list[str] = []
     # pos_title: str = Field(default=None, description="Original POS title")
     lang_code: str = Field(

diff --git a/src/wiktextract/extractor/de/page.py b/src/wiktextract/extractor/de/page.py
@@ -131,9 +131,12 @@ def process_pos_section(
     level_node: LevelNode,
 ) -> None:
     pos_data_list = []
+    pos_title = ""
     for template_node in level_node.find_content(NodeKind.TEMPLATE):
         if template_node.template_name == "Wortart":
             pos_argument = template_node.template_parameters.get(1, "").strip()
+            if pos_title == "":
+                pos_title = pos_argument
             if pos_argument in IGNORE_POS:
                 continue
             elif pos_argument in FORM_POS:
@@ -169,6 +172,7 @@ def process_pos_section(
                 page_data[-1].tags.append(tag)
         if pos_index == 0:
             page_data[-1].pos = pos
+            page_data[-1].pos_title = pos_title
         elif pos != page_data[-1].pos and pos not in page_data[-1].other_pos:
             page_data[-1].other_pos.append(pos)
 

diff --git a/tests/test_de_gloss.py b/tests/test_de_gloss.py
@@ -259,6 +259,7 @@ def test_form_of(self):
                     "lang": "Deutsch",
                     "lang_code": "de",
                     "pos": "adj",
+                    "pos_title": "Deklinierte Form",
                     "senses": [
                         {
                             "form_of": [{"word": "konjugiert"}],
@@ -300,6 +301,7 @@ def test_no_bedeutungen_section(self):
                     "lang": "Litauisch",
                     "lang_code": "lt",
                     "pos": "unknown",
+                    "pos_title": "Deklinierte Form",
                     "senses": [
                         {
                             "form_of": [{"word": "abakas"}],
@@ -330,6 +332,7 @@ def test_grammatische_merkmale_no_form_of_pos_title(self):
                     "lang": "Latein",
                     "lang_code": "la",
                     "pos": "verb",
+                    "pos_title": "Infinitiv",
                     "senses": [
                         {
                             "form_of": [{"word": "abire"}],
@@ -363,6 +366,7 @@ def test_no_gloss_list(self):
                     "lang": "Interlingua",
                     "lang_code": "ia",
                     "pos": "unknown",
+                    "pos_title": "Konjugierte Form",
                     "senses": [
                         {"glosses": ["Indikativ Präsens Aktiv des Verbs amar"]}
                     ],
@@ -390,6 +394,7 @@ def test_unordered_list(self):
                     "lang": "Prußisch",
                     "lang_code": "prg",
                     "pos": "prep",
+                    "pos_title": "Präposition",
                     "senses": [{"glosses": ["Nebenform der Präposition esse"]}],
                     "word": "assa",
                 }
@@ -413,6 +418,7 @@ def test_description_list_plus_unordered_list(self):
                     "lang": "Polnisch",
                     "lang_code": "pl",
                     "pos": "noun",
+                    "pos_title": "Deklinierte Form",
                     "senses": [
                         {
                             "form_of": [{"word": "auto"}],

diff --git a/tests/test_de_page.py b/tests/test_de_page.py
@@ -42,6 +42,7 @@ def test_de_parse_page(self):
                     "lang_code": "de",
                     "word": "Beispiel",
                     "pos": "noun",
+                    "pos_title": "Substantiv",
                     "senses": [{"tags": ["no-gloss"]}],
                 }
             ],
@@ -69,6 +70,7 @@ def test_de_parse_page_skipping_head_templates(self):
                     "lang_code": "de",
                     "word": "Beispiel",
                     "pos": "noun",
+                    "pos_title": "Substantiv",
                     "senses": [{"tags": ["no-gloss"]}],
                 }
             ],
@@ -91,6 +93,7 @@ def test_multiple_pos(self):
                     "lang": "Deutsch",
                     "lang_code": "de",
                     "pos": "noun",
+                    "pos_title": "Substantiv",
                     "other_pos": ["name"],
                     "senses": [
                         {