Merge pull request #878 from xxyzz/ko

[ko] extract gloss label template

src/wiktextract/clean.py

@@ -1521,7 +1521,7 @@ def repl_1_syntaxhighlight(m: re.Match) -> str:
     # XXX "Category" -> config variable for portability
     category_ns_data = wxr.wtp.NAMESPACE_DATA.get("Category", {})  # type: ignore[typeddict-item]
     # Fail if we received empty dict from .get()
-    category_ns_names = {category_ns_data["name"]} | set(
+    category_ns_names = {"Category", category_ns_data["name"]} | set(
         category_ns_data["aliases"]
     )
     category_names_pattern = rf"(?:{'|'.join(category_ns_names)})"

src/wiktextract/extractor/ko/example.py

@@ -28,6 +28,7 @@ def extract_example_list_item(
         elif isinstance(node, TemplateNode) and node.template_name in [
             "예문",
             "ux",
+            "uxi",
         ]:
             extract_ux_template(wxr, sense, example, node)
             break

src/wiktextract/extractor/ko/pos.py

@@ -13,6 +13,7 @@
 from .models import AltForm, Sense, WordEntry
 from .section_titles import LINKAGE_SECTIONS, POS_DATA
 from .sound import SOUND_TEMPLATES, extract_sound_template
+from .tags import translate_raw_tags
 from .translation import extract_translation_template
 
 
@@ -83,12 +84,22 @@ def extract_gloss_list_item(
         ):
             extract_form_of_template(wxr, sense, node)
             gloss_nodes.append(node)
+        elif isinstance(node, TemplateNode) and node.template_name == "라벨":
+            sense.raw_tags.extend(
+                [
+                    raw_tag.strip()
+                    for raw_tag in clean_node(wxr, sense, node)
+                    .strip("()")
+                    .split(",")
+                ]
+            )
         else:
             gloss_nodes.append(node)
 
     gloss_text = clean_node(wxr, sense, gloss_nodes)
     if len(gloss_text) > 0:
         sense.glosses.append(gloss_text)
+        translate_raw_tags(sense)
         word_entry.senses.append(sense)
 
 

src/wiktextract/extractor/ko/tags.py

@@ -0,0 +1,26 @@
+from .models import WordEntry
+
+# https://ko.wiktionary.org/wiki/모듈:labels/data/topical
+# https://ko.wiktionary.org/wiki/모듈:labels/data
+GLOSS_TAGS = {
+    "고어": "archaic",
+    "자동사": "intransitive",
+}
+
+TAGS = {**GLOSS_TAGS}
+
+TOPICS = {
+    "금융": "finance",
+}
+
+
+def translate_raw_tags(data: WordEntry) -> None:
+    raw_tags = []
+    for raw_tag in data.raw_tags:
+        if raw_tag in TAGS:
+            data.tags.append(TAGS[raw_tag])
+        elif raw_tag in TOPICS:
+            data.topics.append(TOPICS[raw_tag])
+        else:
+            raw_tags.append(raw_tag)
+    data.raw_tags = raw_tags

tests/test_ko_gloss.py

@@ -96,9 +96,8 @@ def test_form_of_template(self):
         self.wxr.wtp.add_page(
             "틀:ko-hanja form of",
             10,
-            """<span class="form-of-definition"><i class="None mention" lang="ko">[[전화#한국어|전화]]</i> <span class="mention-gloss-paren annotation-paren">(</span><span class="mention-gloss-double-quote">“</span><span class="mention-gloss">전화기로 말을 주고받는 일</span><span class="mention-gloss-double-quote">”</span><span class="mention-gloss-paren annotation-paren">)</span>의 [[한자#한국어|한자]] 형태.</span>""",
+            """<span class="form-of-definition"><i class="None mention" lang="ko">[[전화#한국어|전화]]</i> <span class="mention-gloss-paren annotation-paren">(</span><span class="mention-gloss-double-quote">“</span><span class="mention-gloss">전화기로 말을 주고받는 일</span><span class="mention-gloss-double-quote">”</span><span class="mention-gloss-paren annotation-paren">)</span>[[Category:한국어 비표준 문자가 포함된 낱말 (링크)|電話]]의 [[한자#한국어|한자]] 형태.</span>""",
         )
-
         data = parse_page(
             self.wxr,
             "電話",
@@ -109,9 +108,31 @@ def test_form_of_template(self):
         self.assertEqual(
             data[0]["senses"][0],
             {
-                # "categories": ["한국어 비표준 문자가 포함된 낱말 (링크)"],
+                "categories": ["한국어 비표준 문자가 포함된 낱말 (링크)"],
                 "form_of": [{"word": "전화"}],
                 "tags": ["form-of"],
                 "glosses": ["전화 (“전화기로 말을 주고받는 일”)의 한자 형태."],
             },
         )
+
+    def test_label_template(self):
+        self.wxr.wtp.add_page(
+            "틀:라벨",
+            10,
+            """<span class="usage-label-sense"><span class="ib-brac">(</span><span class="ib-content">[[부록:용어사전#잘 쓰이지 않는 표현과 그 정도|고어]][[Category:한국어 고어|열다]]<span class="ib-comma">,</span>&#32;[[부록:용어사전#자동사|자동사]][[Category:한국어 자동사|열다]]</span><span class="ib-brac">)</span></span>""",
+        )
+        data = parse_page(
+            self.wxr,
+            "열다",
+            """== 한국어 ==
+=== 명사 ===
+# {{라벨|ko|고어|자동사}} [[열매가 맺히다]]""",
+        )
+        self.assertEqual(
+            data[0]["senses"][0],
+            {
+                "categories": ["한국어 고어", "한국어 자동사"],
+                "tags": ["archaic", "intransitive"],
+                "glosses": ["열매가 맺히다"],
+            },
+        )