Merge pull request #1048 from xxyzz/ku

[ku] fix Lua error and improve linkage section code
tatuylonen · Feb 24, 2025 · 5624056 · 5624056
2 parents 8a9b3fc + 3e22526
commit 5624056
Show file tree

Hide file tree

Showing 3 changed files with 36 additions and 2 deletions.
diff --git a/src/wiktextract/data/ku/config.json b/src/wiktextract/data/ku/config.json
@@ -1,3 +1,9 @@
 {
-  "save_ns_names": ["Main", "Template", "Module", "Tewandin"]
+  "save_ns_names": [
+    "Main",
+    "Template",
+    "Module",
+    "Tewandin",
+    "Nimînok"
+  ]
 }
diff --git a/src/wiktextract/extractor/ku/linkage.py b/src/wiktextract/extractor/ku/linkage.py
@@ -222,8 +222,11 @@ def extract_linkage_list_item(
     shared_tags: list[str] = [],
 ) -> None:
     raw_tags = []
+    forms = []
     for node in list_item.children:
-        if isinstance(node, WikiNode) and node.kind == NodeKind.LINK:
+        if (
+            isinstance(node, WikiNode) and node.kind == NodeKind.LINK
+        ) or isinstance(node, str):
             word = clean_node(wxr, None, node)
             if word != "":
                 if linkage_type != "":
@@ -233,10 +236,12 @@ def extract_linkage_list_item(
                         raw_tags=raw_tags,
                         tags=shared_tags,
                     )
+                    forms.append(l_data)
                     translate_raw_tags(l_data)
                     getattr(word_entry, linkage_type).append(l_data)
                 else:
                     form = Form(form=word, raw_tags=raw_tags, tags=shared_tags)
+                    forms.append(l_data)
                     translate_raw_tags(form)
                     word_entry.forms.append(form)
         elif isinstance(node, TemplateNode):
@@ -268,6 +273,9 @@ def extract_linkage_list_item(
                 raw_tag = clean_node(wxr, None, node).strip("() ")
                 if raw_tag != "":
                     raw_tags.append(raw_tag)
+                    for form in forms:
+                        form.raw_tags.append(raw_tag)
+                        translate_raw_tags(form)
 
 
 def extract_stûn_template(

diff --git a/tests/test_ku_linkage.py b/tests/test_ku_linkage.py
@@ -149,3 +149,23 @@ def test_kol_tag(self):
         self.assertEqual(
             page_data[0]["derived"], [{"word": "aqil kirin", "tags": ["verb"]}]
         )
+
+    def test_kol_mj(self):
+        self.wxr.wtp.add_page("Şablon:ziman", 10, "Îngilîzî")
+        self.wxr.wtp.add_page("Şablon:mj", 10, "(xweşbêjî)")
+        page_data = parse_page(
+            self.wxr,
+            "go the way of all flesh",
+            """== {{ziman|en}} ==
+=== Biwêj ===
+# [[texsîrî]]
+==== Jê ====
+==== Hevmane ====
+{{kol3|en|cure=Hevmane
+|exit {{mj|{{nimînok|xweşbêjî}}}}
+}}""",
+        )
+        self.assertEqual(
+            page_data[0]["synonyms"],
+            [{"word": "exit", "tags": ["euphemistic"]}],
+        )