Merge branch 'feature/1.x.x'

# Conflicts:
#	docs/issue.md
#	pyproject.toml
#	uv.lock

Author: ThreeFish-AI

docs/issue.md

@@ -4,6 +4,50 @@
 
 ---
 
+## streaming usage parse failed: 'NoneType' object has no attribute 'get'
+
+**问题描述**
+
+OpenAI 兼容 SSE 流式响应过程中，单次请求日志反复刷出数十条 WARNING：
+
+```
+WARNING streaming usage parse failed: 'NoneType' object has no attribute 'get'
+```
+
+警告本身被上层 `try/except` 吞掉不影响主链路，但日志噪声严重，且每帧都丢失了 usage 累加。
+
+**表因**
+
+`StreamingUsageAccumulator.feed` 调用 `parse_usage_from_chunk` 解析 SSE chunk 时抛出 `AttributeError`。
+
+**根因**
+
+`src/coding/proxy/routing/usage_parser.py::parse_usage_from_chunk` 中 Anthropic message_start 与 Anthropic message_delta / OpenAI 两条分支都使用了脆弱的判空模式：
+
+```python
+if "usage" in data:        # 仅判断 key 存在
+    u = data["usage"]      # 但值可能是 null
+    u.get("output_tokens", 0)  # AttributeError
+```
+
+部分上游（含某些 OpenAI 兼容供应商）在中间 chunk 显式发送 `"usage": null` 占位帧，`in` 检查通过但取出的是 `None`。
+
+**处理方式**
+
+将两处 guard 统一改为 `u = container.get("usage"); if isinstance(u, dict):`，既排除缺省也排除 null，并顺手移除内部冗余的 `if isinstance(u, dict):` 包装层（已被外层 guard 覆盖）。同时新增三个回归用例覆盖 `data.usage = null` / `message.usage = null` / null 帧后跟有效帧三种场景。
+
+**后续防范**
+
+- 解析外部 SSE / JSON 结构时, 不要单独使用 `if key in data` 作为安全 guard, 应统一采用 `value = data.get(key); if isinstance(value, dict):` 的双重保护, 同时排除缺省与显式 null。
+- 对 try/except 包裹的 WARNING 路径要保持警觉: 异常被吞不代表无害，重复刷屏的同类警告往往暗示防御性 guard 过窄，需要回溯至根因修复，而非依赖 except 兜底。
+
+**同类问题影响与处理注意事项**
+
+- 本仓库内 `parse_usage_from_chunk` 的 Gemini `usageMetadata` 分支 (line ~219) 已经使用 `isinstance(um, dict)` 防御, 不受影响, 可作为参考实现。
+- 检查其他解析器 (如 routing / vendor adapter 层) 是否还有 `if "key" in data: v = data["key"]; v.get(...)` 这种模式, 必要时同步加固。
+
+---
+
 ## zhipu 自循环 400 + tool_results 偶发降级
 
 **问题描述**

src/coding/proxy/routing/usage_parser.py

@@ -169,8 +169,8 @@ def parse_usage_from_chunk(
 
         # Anthropic 格式: message_start 事件 (data.message.usage)
         msg = data.get("message", {})
-        if isinstance(msg, dict) and "usage" in msg:
-            u = msg["usage"]
+        u = msg.get("usage") if isinstance(msg, dict) else None
+        if isinstance(u, dict):
             input_tokens = u.get("input_tokens", 0) or u.get("prompt_tokens", 0)
             _set_if_nonzero(usage, "input_tokens", input_tokens)
             _set_if_nonzero(
@@ -183,18 +183,17 @@ def parse_usage_from_chunk(
                 usage["request_id"] = msg["id"]
             if "model" in msg:
                 usage["model_served"] = msg["model"]
-            if isinstance(u, dict):
-                _append_usage_evidence(
-                    usage,
-                    evidence_kind="message_usage",
-                    raw_usage=dict(u),
-                    request_id=msg.get("id"),
-                    model_served=msg.get("model"),
-                )
+            _append_usage_evidence(
+                usage,
+                evidence_kind="message_usage",
+                raw_usage=dict(u),
+                request_id=msg.get("id"),
+                model_served=msg.get("model"),
+            )
 
         # Anthropic message_delta / OpenAI 最后一个 chunk (data.usage)
-        if "usage" in data:
-            u = data["usage"]
+        u = data.get("usage")
+        if isinstance(u, dict):
             output_tokens = u.get("output_tokens", 0) or u.get("completion_tokens", 0)
             input_tokens = u.get("input_tokens", 0) or u.get("prompt_tokens", 0)
             cache_creation_tokens = u.get("cache_creation_input_tokens", 0)
@@ -204,14 +203,13 @@ def parse_usage_from_chunk(
             _set_if_nonzero(usage, "input_tokens", input_tokens)
             _set_if_nonzero(usage, "cache_creation_tokens", cache_creation_tokens)
             _set_if_nonzero(usage, "cache_read_tokens", cache_read_tokens)
-            if isinstance(u, dict):
-                _append_usage_evidence(
-                    usage,
-                    evidence_kind="data_usage",
-                    raw_usage=dict(u),
-                    request_id=data.get("id"),
-                    model_served=data.get("model"),
-                )
+            _append_usage_evidence(
+                usage,
+                evidence_kind="data_usage",
+                raw_usage=dict(u),
+                request_id=data.get("id"),
+                model_served=data.get("model"),
+            )
 
         # Gemini SSE 格式: data.usageMetadata.{promptTokenCount, candidatesTokenCount, cachedContentTokenCount, thoughtsTokenCount, toolUsePromptTokenCount}
         # Gemini 的流式响应在最后一帧（或每一帧）携带 usageMetadata；字段命名与

src/coding/proxy/server/dashboard.py

@@ -415,7 +415,7 @@ def _build_favicon() -> bytes:
       z-index: 1000;
       opacity: 0;
       transition: opacity .15s ease;
-      max-width: 360px;
+      max-width: 520px;
       max-height: 60vh;
       overflow-y: auto;
       backdrop-filter: blur(8px);
@@ -429,8 +429,8 @@ def _build_favicon() -> bytes:
     #chart-tooltip-items { display: flex; flex-direction: column; gap: 3px; }
     .tt-item { display: flex; align-items: center; gap: 8px; line-height: 1.4; }
     .tt-color { width: 8px; height: 8px; border-radius: 50%; flex-shrink: 0; }
-    .tt-label { flex: 1; color: var(--text-primary); }
-    .tt-value { font-family: 'JetBrains Mono', monospace; font-size: 12px; color: var(--text-secondary); }
+    .tt-label { flex: 1; color: var(--text-primary); white-space: nowrap; }
+    .tt-value { font-family: 'JetBrains Mono', monospace; font-size: 12px; color: var(--text-secondary); white-space: nowrap; }
     #chart-tooltip-footer {
       margin-top: 6px; padding-top: 6px; border-top: 1px solid var(--border-subtle);
       font-weight: 500; font-size: 12px; color: var(--text-secondary);

tests/test_parse_usage.py

@@ -265,3 +265,48 @@ def test_multiple_sse_lines_in_single_chunk():
     parse_usage_from_chunk(chunk, usage)
     assert usage["input_tokens"] == 80
     assert usage["output_tokens"] == 20
+
+
+# --- null usage 安全保护（防御上游 SSE 字段为 null 的极端格式） ---
+
+
+def test_null_usage_at_top_level_does_not_raise():
+    """data.usage 显式为 null 时应被静默忽略，不抛异常、不产生 WARNING."""
+    usage: dict = {}
+    parse_usage_from_chunk(
+        _sse('{"id":"chatcmpl-null","choices":[{"delta":{}}],"usage":null}'),
+        usage,
+    )
+    # 不应写入任何 token 字段
+    assert usage.get("input_tokens", 0) == 0
+    assert usage.get("output_tokens", 0) == 0
+
+
+def test_null_usage_in_message_does_not_raise():
+    """message.usage 显式为 null 时应被静默忽略."""
+    usage: dict = {}
+    parse_usage_from_chunk(
+        _sse('{"type":"message_start","message":{"id":"msg_null","usage":null}}'),
+        usage,
+    )
+    assert usage.get("input_tokens", 0) == 0
+
+
+def test_null_usage_does_not_break_subsequent_valid_chunks():
+    """null usage 帧之后到来的有效帧仍能正确解析."""
+    usage: dict = {}
+    # 1. null usage 帧
+    parse_usage_from_chunk(
+        _sse('{"id":"chatcmpl-1","choices":[{"delta":{"content":"hi"}}],"usage":null}'),
+        usage,
+    )
+    # 2. 有效的最终帧
+    parse_usage_from_chunk(
+        _sse(
+            '{"id":"chatcmpl-1","choices":[{"finish_reason":"stop","delta":{}}],'
+            '"usage":{"prompt_tokens":12,"completion_tokens":3}}'
+        ),
+        usage,
+    )
+    assert usage["input_tokens"] == 12
+    assert usage["output_tokens"] == 3