feat(wrapped): 年度关键词卡升级为年度常用语词云

- 新增常用语短句过滤与统计，按重复短句（非分词）生成词云数据。

- 引入常用语扫描元数据（scannedCandidates/matchedCandidates/uniquePhrases 等）。

- 示例语句抽样改为唯一优先，不足补齐重复，提升短句命中率。

- wrapped cache version 升级到 24，并补充常用语相关单元测试。

tests/test_wrapped_keywords_wordcloud.py

@@ -8,6 +8,63 @@ sys.path.insert(0, str(ROOT / "src"))
 
 
 class TestWrappedKeywordsWordCloud(unittest.TestCase):
+    def test_weflow_common_phrase_filter(self):
+        from wechat_decrypt_tool.wrapped.cards.card_05_keywords_wordcloud import _weflow_common_phrase_or_empty
+
+        self.assertEqual(_weflow_common_phrase_or_empty("  在吗  "), "在吗")
+        self.assertEqual(_weflow_common_phrase_or_empty("ok"), "ok")
+        self.assertEqual(_weflow_common_phrase_or_empty("a"), "")  # too short
+        self.assertEqual(_weflow_common_phrase_or_empty("x" * 21), "")  # too long
+        self.assertEqual(_weflow_common_phrase_or_empty("看看 http://x.com"), "")  # contains http
+        self.assertEqual(_weflow_common_phrase_or_empty("<msg>xml</msg>"), "")  # contains "<"
+        self.assertEqual(_weflow_common_phrase_or_empty("[捂脸]"), "")  # bracketed payload
+        self.assertEqual(_weflow_common_phrase_or_empty("<?xml version='1.0'?>"), "")  # xml payload
+
+    def test_build_common_phrases_payload_structure(self):
+        from collections import Counter
+
+        from wechat_decrypt_tool.wrapped.cards.card_05_keywords_wordcloud import build_common_phrases_payload
+
+        phrase_counts = Counter({"好的": 5, "在吗": 2, "movie": 2, "单次": 1})
+        example_texts = [
+            "好的收到",
+            "好的好的，明白了",
+            "你好的呀",
+            "在吗宝贝",
+            "movie night is fun",
+            "MOVIE time now",
+        ]
+        payload = build_common_phrases_payload(
+            phrase_counts=phrase_counts,
+            seed=123456,
+            top_n=32,
+            bubble_limit=50,
+            example_texts=example_texts,
+            examples_per_word=3,
+        )
+
+        self.assertIn("keywords", payload)
+        self.assertIn("bubbleMessages", payload)
+        self.assertIn("examples", payload)
+        self.assertIn("topKeyword", payload)
+
+        self.assertEqual(payload["topKeyword"]["word"], "好的")
+        self.assertEqual(int(payload["topKeyword"]["count"]), 5)
+
+        self.assertTrue(all(int(x.get("count") or 0) >= 2 for x in payload["keywords"]))
+        self.assertTrue(all(isinstance(x.get("word"), str) and x.get("word") for x in payload["keywords"]))
+
+        # Examples should contain real message samples with an upper bound.
+        for ex in payload["examples"]:
+            msgs = ex.get("messages") or []
+            self.assertGreaterEqual(len(msgs), 1)
+            self.assertLessEqual(len(msgs), 3)
+            word = str(ex.get("word") or "")
+            if any("\u4e00" <= ch <= "\u9fff" for ch in word):
+                self.assertTrue(any(word in str(m) for m in msgs))
+            else:
+                self.assertTrue(any(word.lower() in str(m).lower() for m in msgs))
+
     def test_extract_keywords_jieba_basic(self):
         from wechat_decrypt_tool.wrapped.cards.card_05_keywords_wordcloud import extract_keywords_jieba
 
@@ -96,6 +153,24 @@ class TestWrappedKeywordsWordCloud(unittest.TestCase):
         m_movie = next(x for x in out if x["word"] == "movie")
         self.assertTrue(all("movie" in m.lower() for m in m_movie["messages"]))
 
+    def test_pick_examples_short_phrase_can_fill_three(self):
+        from wechat_decrypt_tool.wrapped.cards.card_05_keywords_wordcloud import pick_examples
+
+        keywords = [{"word": "在吗", "count": 9, "weight": 1.0}]
+        pool = [
+            "在吗",
+            "在吗",
+            "在吗",
+            "在吗？",
+            "ok",
+        ]
+
+        out = pick_examples(keywords, pool, per_word=3)
+        self.assertEqual(len(out), 1)
+        msgs = out[0]["messages"]
+        self.assertEqual(len(msgs), 3)
+        self.assertTrue(all("在吗" in m for m in msgs))
+
     def test_build_keywords_payload_structure(self):
         from wechat_decrypt_tool.wrapped.cards.card_05_keywords_wordcloud import build_keywords_payload