refactor(handler): implement defensive token update strategy and extract cache creation token utility

- Add extract_cache_creation_tokens utility to handle new/old cache creation token formats
- Implement defensive update strategy in StreamContext to prevent zero values overwriting valid data
- Simplify cache creation token parsing in Claude handler using new utility
- Add comprehensive test suite for cache creation token extraction
- Improve type hints in handler classes

tests/__init__.py

@@ -0,0 +1 @@
+"""测试模块"""

tests/api/handlers/base/test_utils.py

@@ -0,0 +1,90 @@
+"""测试 handler 基础工具函数"""
+
+import pytest
+
+from src.api.handlers.base.utils import extract_cache_creation_tokens
+
+
+class TestExtractCacheCreationTokens:
+    """测试 extract_cache_creation_tokens 函数"""
+
+    def test_new_format_only(self) -> None:
+        """测试只有新格式字段"""
+        usage = {
+            "claude_cache_creation_5_m_tokens": 100,
+            "claude_cache_creation_1_h_tokens": 200,
+        }
+        assert extract_cache_creation_tokens(usage) == 300
+
+    def test_new_format_5m_only(self) -> None:
+        """测试只有 5 分钟缓存"""
+        usage = {
+            "claude_cache_creation_5_m_tokens": 150,
+            "claude_cache_creation_1_h_tokens": 0,
+        }
+        assert extract_cache_creation_tokens(usage) == 150
+
+    def test_new_format_1h_only(self) -> None:
+        """测试只有 1 小时缓存"""
+        usage = {
+            "claude_cache_creation_5_m_tokens": 0,
+            "claude_cache_creation_1_h_tokens": 250,
+        }
+        assert extract_cache_creation_tokens(usage) == 250
+
+    def test_old_format_only(self) -> None:
+        """测试只有旧格式字段"""
+        usage = {
+            "cache_creation_input_tokens": 500,
+        }
+        assert extract_cache_creation_tokens(usage) == 500
+
+    def test_both_formats_prefers_new(self) -> None:
+        """测试同时存在时优先使用新格式"""
+        usage = {
+            "claude_cache_creation_5_m_tokens": 100,
+            "claude_cache_creation_1_h_tokens": 200,
+            "cache_creation_input_tokens": 999,  # 应该被忽略
+        }
+        assert extract_cache_creation_tokens(usage) == 300
+
+    def test_empty_usage(self) -> None:
+        """测试空字典"""
+        usage = {}
+        assert extract_cache_creation_tokens(usage) == 0
+
+    def test_all_zeros(self) -> None:
+        """测试所有字段都为 0"""
+        usage = {
+            "claude_cache_creation_5_m_tokens": 0,
+            "claude_cache_creation_1_h_tokens": 0,
+            "cache_creation_input_tokens": 0,
+        }
+        assert extract_cache_creation_tokens(usage) == 0
+
+    def test_partial_new_format_with_old_format_fallback(self) -> None:
+        """测试新格式字段不存在时回退到旧格式"""
+        usage = {
+            "cache_creation_input_tokens": 123,
+        }
+        assert extract_cache_creation_tokens(usage) == 123
+
+    def test_new_format_zero_fallback_to_old(self) -> None:
+        """测试新格式为 0 时回退到旧格式"""
+        usage = {
+            "claude_cache_creation_5_m_tokens": 0,
+            "claude_cache_creation_1_h_tokens": 0,
+            "cache_creation_input_tokens": 456,
+        }
+        assert extract_cache_creation_tokens(usage) == 456
+
+    def test_unrelated_fields_ignored(self) -> None:
+        """测试忽略无关字段"""
+        usage = {
+            "input_tokens": 1000,
+            "output_tokens": 2000,
+            "cache_read_input_tokens": 300,
+            "claude_cache_creation_5_m_tokens": 50,
+            "claude_cache_creation_1_h_tokens": 75,
+        }
+        assert extract_cache_creation_tokens(usage) == 125