Type senseid as str in Spanish extractor

tatuylonen · Jan 5, 2024 · 1f7b3f1 · 1f7b3f1
1 parent b976575
commit 1f7b3f1
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 5 deletions.
diff --git a/src/wiktextract/extractor/es/gloss.py b/src/wiktextract/extractor/es/gloss.py
@@ -2,6 +2,7 @@
 
 from wikitextprocessor import NodeKind, WikiNode
 from wikitextprocessor.parser import WikiNodeChildrenList
+
 from wiktextract.extractor.es.models import Sense, WordEntry
 from wiktextract.extractor.es.sense_data import process_sense_data_list
 from wiktextract.page import clean_node
@@ -38,7 +39,7 @@ def extract_gloss(
         match = re.match(r"^(\d+)", gloss_note)
 
         if match:
-            gloss_data.senseid = int(match.group(1))
+            gloss_data.senseid = match.group(1)
             tag_string = gloss_note[len(match.group(1)) :].strip()
         else:
             tag_string = gloss_note.strip()

diff --git a/src/wiktextract/extractor/es/models.py b/src/wiktextract/extractor/es/models.py
@@ -93,7 +93,7 @@ class Sense(BaseModelWrap):
     # subsenses: list["Sense"] = Field(
     #     default=[], description="List of subsenses"
     # )
-    senseid: Optional[int] = Field(
+    senseid: Optional[str] = Field(
         default=None, description="Sense number used in Wiktionary"
     )
     antonyms: Optional[list[Linkage]] = []

diff --git a/tests/test_es_gloss.py b/tests/test_es_gloss.py
@@ -44,13 +44,13 @@ def test_es_extract_glosses(self):
                     "glosses": [
                         "Contribuir esfuerzo o recursos para la realización de algo."
                     ],
-                    "senseid": 1,
+                    "senseid": "1",
                 },
                 {
                     "glosses": [
                         "Por antonomasia, cooperar a que alguno salga de una situación dificultosa"
                     ],
-                    "senseid": 2,
+                    "senseid": "2",
                 },
             ],
         )
@@ -80,7 +80,7 @@ def test_es_extract_gloss_categories(self):
                     "glosses": [
                         "Sentimiento afectivo de atracción, unión y afinidad que se experimenta hacia una persona, animal o cosa"
                     ],
-                    "senseid": 1,
+                    "senseid": "1",
                     "tags": ["Humanidades"],
                     "categories": ["ES:Sentimientos"],
                 }