Divide French Wiktionary linkage list item words with slash

tatuylonen · Oct 30, 2023 · 1fd8dd9 · 1fd8dd9
1 parent f6a6412
commit 1fd8dd9
Show file tree

Hide file tree

Showing 2 changed files with 20 additions and 1 deletion.
diff --git a/src/wiktextract/extractor/fr/linkage.py b/src/wiktextract/extractor/fr/linkage.py
@@ -106,7 +106,7 @@ def process_linkage_list(
                 ) and tag_text.strip().endswith(")"):
                     tag_text = pending_tag + tag_text
                     pending_tag = ""
-                elif tag_text.strip() == ",":
+                elif tag_text.strip() in {",", "/"}:
                     # list item has more than one word
                     page_data[-1][linkage_type].append(linkage_data)
                     linkage_data = defaultdict(list)

diff --git a/tests/test_fr_linkage.py b/tests/test_fr_linkage.py
@@ -179,3 +179,22 @@ def test_derives_autres_langues_section(self):
                 }
             ],
         )
+
+    def test_words_divided_by_slash(self):
+        page_data = [defaultdict(list)]
+        self.wxr.wtp.start_page("eau")
+        root = self.wxr.wtp.parse(
+            "* [[benoîte d’eau]] / [[benoite d’eau]]"
+        )
+        extract_linkage(self.wxr, page_data, root, "dérivés")
+        self.assertEqual(
+            page_data,
+            [
+                {
+                    "derived": [
+                        {"word": "benoîte d’eau"},
+                        {"word": "benoite d’eau"},
+                    ]
+                }
+            ],
+        )