Process K templates in German Wiktionary glosses

This work is a contribution to the EWOK project, which receives funding from LABEX ASLAN (ANR–10–LABX–0081) at the Université de Lyon, as part of the "Investissements d'Avenir" program initiated and overseen by the Agence Nationale de la Recherche (ANR) in France.

src/wiktextract/extractor/de/gloss.py

@@ -63,7 +63,7 @@ def process_gloss_list_item(
             raw_gloss = clean_node(wxr, {}, list_item_node.children)
             gloss_data["raw_glosses"] = [raw_gloss]
 
-            extract_categories_from_gloss_node(wxr, gloss_data, list_item_node)
+            process_K_template(wxr, gloss_data, list_item_node)
 
             gloss_text = clean_node(wxr, gloss_data, list_item_node.children)
 
@@ -82,9 +82,8 @@ def process_gloss_list_item(
                     sortid="extractor/de/glosses/extract_glosses/28",
                 )
 
-            gloss_text = extract_categories_from_gloss_text(
-                gloss_data, gloss_text
-            )
+            # XXX: Extract tags from nodes instead using Italic and Template
+            gloss_text = extract_tags_from_gloss_text(gloss_data, gloss_text)
 
             if gloss_text or not sub_glosses_list_nodes:
                 gloss_data["glosses"] = [gloss_text]
@@ -117,34 +116,50 @@ def handle_sense_modifier(wxr, list_item_node):
     pass
 
 
-def extract_categories_from_gloss_node(
+def process_K_template(
     wxr: WiktextractContext,
     gloss_data: defaultdict(list),
     list_item_node: NodeKind.LIST_ITEM,
 ) -> None:
     for template_node in list_item_node.find_child(NodeKind.TEMPLATE):
         if template_node.template_name == "K":
-            categories = template_node.template_parameters.values()
-
-            categories = [clean_node(wxr, {}, [c]) for c in categories]
+            text = clean_node(wxr, gloss_data, template_node).removesuffix(":")
+            tags = re.split(r";|,", text)
+            gloss_data["tags"] = [t.strip() for t in tags]
+
+            # Prepositional and case information is sometimes only expanded to
+            # category links and not present in cleaned node. We still want it
+            # as a tag.
+            prep = template_node.template_parameters.get("Prä")
+            case = template_node.template_parameters.get("Kas")
+            category = (prep if prep else "") + (" + " + case if case else "")
+            if category:
+                gloss_data["tags"].append(category)
+
+            # XXX: Investigate better ways to handle free text in K template
+            ft = template_node.template_parameters.get("ft")
+            if ft:
+                wxr.wtp.debug(
+                    f"Found ft '{ft}' in K template which could be considered part of the gloss. Moved to tags for now.",
+                    sortid="extractor/de/glosses/extract_glosses/63",
+                )
 
+            # Remove the template_node from the children of list_item_node
             list_item_node.children = [
                 c for c in list_item_node.children if c != template_node
             ]
 
-            gloss_data["categories"].extend(categories)
-
 
-def extract_categories_from_gloss_text(
+def extract_tags_from_gloss_text(
     gloss_data: defaultdict(list), gloss_text: str
 ) -> None:
     parts = gloss_text.split(":", 1)
     if len(parts) > 1:
-        categories_part = parts[0].strip()
+        tags_part = parts[0].strip()
 
-        categories = [c.strip() for c in re.split(",|and", categories_part)]
+        categories = [c.strip() for c in re.split(",", tags_part)]
         if all(c.isalnum() for c in categories):
-            gloss_data["categories"].extend(categories)
+            gloss_data["tags"].extend(categories)
             return parts[1].strip()
 
     return gloss_text