Translate "qualifier" template tags in linkage list

tatuylonen · Mar 1, 2024 · 7296a7e · 7296a7e
1 parent 63d6a7a
commit 7296a7e
Showing 4 changed files with 29 additions and 1 deletion.
diff --git a/src/wiktextract/extractor/zh/gloss.py b/src/wiktextract/extractor/zh/gloss.py
@@ -1,4 +1,3 @@
-
 from wikitextprocessor import NodeKind, WikiNode
 from wikitextprocessor.parser import TemplateNode
 from wiktextract.page import clean_node

diff --git a/src/wiktextract/extractor/zh/linkage.py b/src/wiktextract/extractor/zh/linkage.py
@@ -13,6 +13,7 @@
 )
 from .descendant import DESCENDANT_TEMPLATES, extract_descendant_list_item
 from .models import Linkage, WordEntry
+from .tags import translate_raw_tags
 
 
 def extract_linkages(
@@ -54,6 +55,7 @@ def extract_linkages(
                             linkage_data.raw_tags.append(
                                 clean_node(wxr, None, item_child).strip("()")
                             )
+                            translate_raw_tags(linkage_data)
                         elif template_name.lower() in DESCENDANT_TEMPLATES:
                             not_term_indexes.add(index)
                             extract_descendant_list_item(

diff --git a/src/wiktextract/extractor/zh/tags.py b/src/wiktextract/extractor/zh/tags.py
@@ -84,12 +84,22 @@
 # https://zh.wiktionary.org/wiki/Template:注释
 LABEL_TAGS = {
     "棄用": "obsolete",
+    "非標準": "nonstandard",
+    "非正式": "informal",
+    "古舊": "dated",
+    "新詞": "neologism",
+    "定語": "attributive",
+    "書面": "literary",
+    "貶義": "derogatory",
+    "比喻": "figuratively",
+    "俗語": "slang",
     "古": "archaic",
     "陽": "masculine",
     "陰": "feminine",
     "喻": "figuratively",
     "書": "literary",
     "口": "colloquial",
+    "口語": "colloquial",
     "俚": "slang",
     "俗": "slang",
     "方": "dialectal",

diff --git a/tests/test_zh_linkage.py b/tests/test_zh_linkage.py
@@ -67,3 +67,20 @@ def test_ja_r_template(self):
                 "word": "家主",
             },
         )
+
+    def test_qual_tag(self):
+        page_data = [WordEntry(lang="漢語", lang_code="zh", word="駱駝")]
+        self.wxr.wtp.add_page("Template:qual", 10, "({{{1}}})")
+        self.wxr.wtp.add_page("Template:zh-l", 10, "{{{1}}}")
+        self.wxr.wtp.start_page("駱駝")
+        node = self.wxr.wtp.parse("* {{qual|比喻}} {{zh-l|沙漠之舟}}")
+        extract_linkages(self.wxr, page_data, node.children, "synonyms", "")
+        self.assertEqual(
+            [
+                s.model_dump(exclude_defaults=True)
+                for s in page_data[0].synonyms
+            ],
+            [
+                {"tags": ["figuratively"], "word": "沙漠之舟"},
+            ],
+        )