[ru] rename field etymology_text to etymology_texts (#1546)

Author: xxyzz

src/wiktextract/extractor/ru/etymology.py

@@ -1,22 +1,27 @@
-from wikitextprocessor import WikiNode
-from wikitextprocessor.parser import LEVEL_KIND_FLAGS, TemplateNode
+from wikitextprocessor import LevelNode, NodeKind, TemplateNode, WikiNode
 
 from ...page import clean_node
 from ...wxr_context import WiktextractContext
 from .models import WordEntry
 
 
 def extract_etymology(
-    wxr: WiktextractContext,
-    word_entry: WordEntry,
-    level_node: WikiNode,
-) -> None:
-    etymology_nodes = []
-    for node in level_node.invert_find_child(
-        LEVEL_KIND_FLAGS, include_empty_str=True
-    ):
-        if isinstance(node, TemplateNode) and node.template_name == "improve":
-            # ignore this template
-            continue
-        etymology_nodes.append(node)
-    word_entry.etymology_text = clean_node(wxr, word_entry, etymology_nodes)
+    wxr: WiktextractContext, word_entry: WordEntry, level_node: LevelNode
+):
+    e_nodes = []
+    for node in level_node.children:
+        if isinstance(node, LevelNode):
+            break
+        elif isinstance(node, WikiNode) and node.kind == NodeKind.LIST:
+            for list_item in node.find_child(NodeKind.LIST_ITEM):
+                e_text = clean_node(wxr, word_entry, list_item.children)
+                if e_text != "":
+                    word_entry.etymology_texts.append(e_text)
+        elif not (
+            isinstance(node, TemplateNode) and node.template_name == "improve"
+        ):
+            e_nodes.append(node)
+    if len(e_nodes) > 0:
+        e_str = clean_node(wxr, word_entry, e_nodes)
+        if e_str != "":
+            word_entry.etymology_texts.append(e_str)

src/wiktextract/extractor/ru/models.py

@@ -147,7 +147,7 @@ class WordEntry(BaseModelWrap):
         default=[], description="List of coordinate terms"
     )
     holonyms: list[Linkage] = Field(default=[], description="List of holonyms")
-    etymology_text: str = ""
+    etymology_texts: list[str] = []
     related: list[Linkage] = []
     metagrams: list[Linkage] = []
     proverbs: list[Linkage] = []

tests/test_ru_page.py

@@ -365,3 +365,16 @@ def test_stressed_form(self):
         self.assertEqual(
             data[0]["forms"], [{"form": "коса́", "tags": ["stressed"]}]
         )
+
+    def test_etymology_list(self):
+        data = parse_page(
+            self.wxr,
+            "albast",
+            """= {{-nl-}} =
+=== Семантические свойства ===
+==== Значение ====
+# [[алебастр]]
+=== Этимология ===
+*От [1285] г.""",
+        )
+        self.assertEqual(data[0]["etymology_texts"], ["От [1285] г."])