Merge pull request #567 from xxyzz/de

Handle de edition's "Ähnlichkeiten Umschrift" template

src/wiktextract/extractor/de/models.py

@@ -167,3 +167,4 @@ class WordEntry(BaseModelWrap):
     tags: list[str] = []
     raw_tags: list[str] = []
     categories: list[str] = []
+    redirects: list[str] = []

src/wiktextract/extractor/de/page.py

@@ -3,7 +3,7 @@
 
 from mediawiki_langcodes import name_to_code
 from wikitextprocessor import NodeKind, WikiNode
-from wikitextprocessor.parser import LevelNode
+from wikitextprocessor.parser import LevelNode, TemplateNode
 from wiktextract.page import clean_node
 from wiktextract.wxr_context import WiktextractContext
 
@@ -199,8 +199,36 @@ def parse_page(
                 )
                 for level3_node in level2_node.find_child(NodeKind.LEVEL3):
                     parse_section(wxr, page_data, base_data, level3_node)
+                for template_node in level2_node.find_child(NodeKind.TEMPLATE):
+                    if template_node.template_name == "Ähnlichkeiten Umschrift":
+                        process_umschrift_template(
+                            wxr, page_data, base_data, template_node
+                        )
 
     for data in page_data:
         if len(data.senses) == 0:
             data.senses.append(Sense(tags=["no-gloss"]))
     return [d.model_dump(exclude_defaults=True) for d in page_data]
+
+
+def process_umschrift_template(
+    wxr: WiktextractContext,
+    page_data: list[WordEntry],
+    base_data: WordEntry,
+    template_node: TemplateNode,
+) -> None:
+    # https://de.wiktionary.org/wiki/Vorlage:Ähnlichkeiten_Umschrift
+    # soft-redirect template, similar to en edition's "zh-see"
+    data = base_data.model_copy(deep=True)
+    data.pos = "soft-redirect"
+    for key, value in template_node.template_parameters.items():
+        if isinstance(key, int):
+            redirect_page = clean_node(wxr, None, value)
+            link_arg = template_node.template_parameters.get(f"link{key}", "")
+            link_text = clean_node(wxr, None, link_arg)
+            if len(link_text) > 0:
+                redirect_page = link_text
+            if len(redirect_page) > 0:
+                data.redirects.append(redirect_page)
+    if len(data.redirects) > 0:
+        page_data.append(data)

tests/test_de_page.py

@@ -102,3 +102,29 @@ def test_multiple_pos(self):
                 }
             ],
         )
+
+    def test_umschrift(self):
+        self.wxr.wtp.add_page("Vorlage:Sprache", 10, "{{{1}}}")
+        self.wxr.wtp.start_page("iku")
+        self.assertEqual(
+            parse_page(
+                self.wxr,
+                "iku",
+                """== hiki ({{Sprache|Umschrift}}) ==
+{{Ähnlichkeiten Umschrift
+|1=行く|spr1=ja
+|2=幾|spr2=ja
+|3=𒃷#𒃷 (iku) (Sumerisch)|spr3=sux|link3=𒃷
+}}""",
+            ),
+            [
+                {
+                    "lang_code": "",
+                    "lang": "Umschrift",
+                    "pos": "soft-redirect",
+                    "redirects": ["行く", "幾", "𒃷"],
+                    "senses": [{"tags": ["no-gloss"]}],
+                    "word": "iku",
+                }
+            ],
+        )