Merge pull request #438 from xxyzz/fr

Extract links in "Dérivés dans d’autres langues" section
tatuylonen · Dec 27, 2023 · 82fc4d5 · 82fc4d5
2 parents ae805ae + 34712de
commit 82fc4d5
Show file tree

Hide file tree

Showing 2 changed files with 24 additions and 8 deletions.
diff --git a/src/wiktextract/extractor/fr/linkage.py b/src/wiktextract/extractor/fr/linkage.py
@@ -33,12 +33,14 @@ def process_derives_autres_list(
     for list_item in level_node.find_child_recursively(NodeKind.LIST_ITEM):
         lang_code = ""
         lang_name = ""
-        for template_node in list_item.find_child(NodeKind.TEMPLATE):
-            if template_node.template_name == "L":
-                lang_code = template_node.template_parameters.get(1)
-                lang_name = clean_node(wxr, None, template_node)
-            elif template_node.template_name == "lien":
-                word = clean_node(wxr, None, template_node)
+        for node in list_item.find_child(NodeKind.TEMPLATE | NodeKind.LINK):
+            if isinstance(node, TemplateNode) and node.template_name == "L":
+                lang_code = node.template_parameters.get(1)
+                lang_name = clean_node(wxr, None, node)
+            elif node.kind == NodeKind.LINK or (
+                isinstance(node, TemplateNode) and node.template_name == "lien"
+            ):
+                word = clean_node(wxr, None, node)
                 page_data[-1].derived.append(
                     Linkage(lang_code=lang_code, lang_name=lang_name, word=word)
                 )

diff --git a/tests/test_fr_linkage.py b/tests/test_fr_linkage.py
@@ -143,14 +143,23 @@ def test_sense(self):
 
     def test_derives_autres_langues_section(self):
         # https://fr.wiktionary.org/wiki/eau#Dérivés_dans_d’autres_langues
+        # https://fr.wiktionary.org/wiki/caligineux#Dérivés_dans_d’autres_langues
         self.wxr.wtp.add_page("Modèle:lien", 10, body="{{{1}}}")
-        self.wxr.wtp.add_page("Modèle:L", 10, body="Karipúna")
+        self.wxr.wtp.add_page(
+            "Modèle:L",
+            10,
+            body="""{{#switch: {{{1}}}
+| kmv = Karipúna
+| en = Anglais
+}}""",
+        )
         page_data = [
             WordEntry(word="test", lang_code="fr", lang_name="Français")
         ]
         self.wxr.wtp.start_page("eau")
         root = self.wxr.wtp.parse(
-            "* {{L|kmv}} : {{lien|dlo|kmv}}, {{lien|djilo|kmv}}"
+            """* {{L|kmv}} : {{lien|dlo|kmv}}, {{lien|djilo|kmv}}
+* {{L|en}} : [[caliginous#en|caliginous]]"""
         )
         extract_linkage(self.wxr, page_data, root, "dérivés autres langues")
         self.assertEqual(
@@ -169,6 +178,11 @@ def test_derives_autres_langues_section(self):
                     "lang_code": "kmv",
                     "lang_name": "Karipúna",
                 },
+                {
+                    "word": "caliginous",
+                    "lang_code": "en",
+                    "lang_name": "Anglais",
+                },
             ],
         )