Merge pull request #441 from xxyzz/fr

Extract sense text and index data from description list(`;`)
tatuylonen · Dec 28, 2023 · 2bdf0f7 · 2bdf0f7
2 parents a570899 + 54c733e
commit 2bdf0f7
Show file tree

Hide file tree

Showing 2 changed files with 25 additions and 1 deletion.
diff --git a/src/wiktextract/extractor/fr/linkage.py b/src/wiktextract/extractor/fr/linkage.py
@@ -1,3 +1,5 @@
+import re
+
 from wikitextprocessor import NodeKind, WikiNode
 from wikitextprocessor.parser import TemplateNode
 from wiktextract.page import clean_node
@@ -71,6 +73,18 @@ def process_linkage_list(
             if sense_index_text.isdigit():
                 sense_index = int(sense_index_text)
             continue
+        # sense could also be in ";" description list
+        if (
+            template_or_list_node.kind == NodeKind.LIST_ITEM
+            and template_or_list_node.sarg == ";"
+        ):
+            sense_text = clean_node(wxr, None, template_or_list_node.children)
+            index_pattern = r"\s*\((?:sens\s*)?(\d+)\)$"
+            m = re.search(index_pattern, sense_text)
+            if m is not None:
+                sense_text = re.sub(index_pattern, "", sense_text)
+                sense_index = int(m.group(1))
+            continue
 
         linkage_data = Linkage()
         if len(sense_text) > 0:

diff --git a/tests/test_fr_linkage.py b/tests/test_fr_linkage.py
@@ -117,13 +117,18 @@ def test_sub_list(self):
         )
 
     def test_sense(self):
+        # https://fr.wiktionary.org/wiki/autrice
+        # https://fr.wiktionary.org/wiki/embouteillage
         page_data = [
-            WordEntry(word="test", lang_code="fr", lang_name="Français")
+            WordEntry(word="autrice", lang_code="fr", lang_name="Français")
         ]
         self.wxr.wtp.start_page("autrice")
         root = self.wxr.wtp.parse(
             """{{(|Celle qui est à l’origine de quelque chose|1}}
 * [[artisane]]
+
+; Mise en bouteille (sens 1)
+* [[bouchonnerie]]
 """
         )
         extract_linkage(self.wxr, page_data, root, "synonymes")
@@ -138,6 +143,11 @@ def test_sense(self):
                     "sense": "Celle qui est à l’origine de quelque chose",
                     "sense_index": 1,
                 },
+                {
+                    "word": "bouchonnerie",
+                    "sense": "Mise en bouteille",
+                    "sense_index": 1,
+                },
             ],
         )