coding-proxy 0.3.0a3__tar.gz → 0.3.1a1__tar.gz

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/AGENTS.md

@@ -47,10 +47,11 @@
   3. **Link Validity**: 确保所有引用的 URL 可访问且具备明确的上下文价值；
   4. **Git Commit**: 在需要提交变更到 Git 时，一律使用 Shell 调用 Claude Code 的自定义 Slash Command: `/commit` 进行 git commit 操作（若环境中未安装 Claude Code，则直接读取 `~/.claude/commands/commit.md`，按照其中的规则进行 git commit 操作）。不要执行 Rebase。
   5. **Pre-commit Hooks**: 克隆仓库后执行 `uv run pre-commit install` 激活本地 Git hooks，使 Ruff lint（含 auto-fix）、Ruff format 及通用代码卫生检查在每次 commit 前自动运行。若 hooks 自动修复了问题，提交会被中断，执行 `git add -p` 审阅修复内容后重新提交即可。
+  6. **Issue**: 在 docs/issue.md 中维护你处理过的 Issue 摘要（问题描述、表因根因、处理方式、后续防范、同类问题影响与处理注意实现等），便于同类问题的跨上下文处理；注意识别相同 Issue，不要同 Issue 多处维护。
 - **Package Management Standardization (包管理规范)**:
   1. **Python**: 严禁使用 pip/poetry，**必须**统一使用 `uv` 进行包管理与脚本执行（如 `uv run`）；
   2. **JavaScript/TypeScript**: 严禁使用 npm/yarn，**必须**统一使用 `pnpm` 进行包管理与脚本执行。
-- **Database Management**: 谨慎操作，数据迁移、测试等操作严禁将现有数据删除。
+- **Database Management**: 谨慎操作，数据迁移、测试等操作严禁将现有数据删除，谨慎操作数据迁移的回滚，防止数据被清理。
 
 ## Documentation Standards (文档规范)
 

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/CHANGELOG.md

@@ -4,16 +4,25 @@
 
 ## [Unreleased]
 
-- feat(server): 默认监听端口 `8046` → `3392`，避免与常见企业端口段冲突；BREAKING: 已将端口固化进客户端 `ANTHROPIC_BASE_URL` / SDK `base_url` / 监控看板 URL 的用户需同步更新，或显式通过 `--port 8046` / `server.port: 8046` 回退到旧值；
-- feat(native-api): 原生 API 透传三家 provider (`openai` / `gemini` / `anthropic`) `enabled` 默认由 `false` → `true` 开箱即用；`config.default.yaml` 顶层追加 `native_api:` 实体配置块（首次启动自动拷贝至 `~/.coding-proxy/config.yaml`，用户直接编辑 `base_url` 即可将上游切换至第三方代理）；同时新增 `NATIVE_OPENAI_BASE_URL` / `NATIVE_GEMINI_BASE_URL` / `NATIVE_ANTHROPIC_BASE_URL` 三枚环境变量覆写通道（空串/纯空白视作未设置），三级优先级 **env > yaml > 内置默认** 由 `NativeApiConfig._apply_env_overrides` `@model_validator(mode="after")` 统一注入；BREAKING: 升级后 `/api/{openai,gemini,anthropic}/**` 默认暴露，proxy 仍不保管凭据（`Authorization` / `x-api-key` / `?key=` 均由客户端透传），如需关闭显式设置 `native_api.<provider>.enabled: false`；
-- feat(native-api): 新增 `/api/{openai,gemini,anthropic}/**` 原生 LLM API 全量 catch-all 透传通道——客户端只需改 SDK `base_url` 即可复用 proxy 链路访问 OpenAI / Gemini / Anthropic 官方 API，认证完全透传不保管凭据；核心由 `NativeProxyHandler` + `OperationClassifier` + `NativeUsageExtractor` Registry 三件套组成，与既有 `/v1/messages` Claude Code 链路正交共存、零回归；首版覆盖 chat / completions / responses / embeddings / audio / image / count_tokens / moderations / cachedContents / messages / batches 等全量端点；
-- feat(usage): `usage_log` 新增 `client_category` / `operation` / `endpoint` / `extra_usage_json` 四列（全部 `DEFAULT`、幂等迁移），区分 Claude Code 场景（`'cc'`）与原生 API 场景（`'api'`），承载规范化操作名与非规范 token 字段（reasoning / audio / thoughts / server_tool_use 等）；`query_usage` 支持按新列过滤，`_PERIOD_SQL` 聚合追加 `client_category, operation` 维度；
-- feat(usage-parser): `parse_usage_from_chunk` 扩展识别 Gemini SSE `usageMetadata.*`（promptTokenCount / candidatesTokenCount / cachedContentTokenCount / thoughtsTokenCount / toolUsePromptTokenCount），新增 `gemini_usage_metadata` evidence kind，既有 Anthropic/OpenAI 分支行为零变更；
-- refactor(vendor-channels): 彻底收敛跨供应商兼容性逻辑——删除 `server/request_normalizer.py` 入口通用规范化层，将 `srvtoolu_*` ID 重写、`server_tool_use_delta` 私有块剥离全部迁入源→目标绑定通道（`prepare_zhipu_to_anthropic`、`prepare_zhipu_to_copilot`）；新增 `infer_source_vendor_from_body` 内容感知源推断，在无会话状态的首次请求场景下兜底识别源供应商；`_RouteExecutor._determine_source_vendor` 扩充为三级优先级（failed_tier → session_state → body inference），确保未注册转换对不触发任何清洗；
-- refactor(count-tokens): `/v1/messages/count_tokens` 端点移除无条件 `strip_thinking_blocks` 过度防御，改为基于 `infer_source_vendor_from_body` + `get_transition_channel` 的按需通道清洗，语义与 `/v1/messages` 对齐；
-- fix(request-normalizer): 重设计 zhipu→anthropic 跨供应商 tool_use/tool_result 配对修复——以单遍自包含 `enforce_anthropic_tool_pairing` 替代原有多步串联管线（剥离→重定位→孤儿修复），消除步骤间隐式依赖导致的孤儿 tool_use 漏修问题，彻底根治 `tool_use ids were found without tool_result blocks` 400 异常;
-- refactor(vendor-channels): 将供应商转换通道从「目标 vendor 专属」重构为「源→目标绑定」模型——注册表键从 `target_vendor` 改为 `(source, target)` 二元组，通道函数从 `prepare_for_X` 重命名为 `prepare_X_to_Y`，触发逻辑从 `_needs_vendor_channel` 替换为 `_determine_source_vendor`（基于请求内 `failed_tier_name` 和会话历史推断源 vendor），未注册的转换对（如 anthropic→zhipu）不触发任何通道;
-- feat(vendor-channels): 新增 zhipu→anthropic、zhipu→copilot、copilot→zhipu 三条源→目标绑定转换通道，在跨供应商故障转移时自动清理源 vendor 产物（thinking 块、cache_control 字段、thinking 参数、tool_use/tool_result 配对），消除 `likely format incompatibility (400 + tool_results)` 错误;
+## [v0.3.0](https://github.com/ThreeFish-AI/coding-proxy/releases/tag/v0.3.0) — 2026-04-20
+
+> [!IMPORTANT]
+>
+> **🚀 OpenAI、Anthropic、Gemini 原生 API 进驻 Coding Proxy！**
+>
+> 服务对象不在局限 Claude Code，凡兼容 OpenAI、Anthropic、Gemini 三巨头 API 协议的客户端，出口 LLM 流量可统一收敛到 Coding Proxy。
+
+### ✨ 核心亮点
+
+- feat(native-api): 新增 `/api/{openai,gemini,anthropic}/**` 原生 LLM API 全量 catch-all 透传通道；
+- feat(dashboard): 新增实时 Web Dashboard 页面，聚合展示流量与用量统计;
+- feat(usage): `usage` 区分 Claude Code 场景（`'cc'`）与原生 API 场景（`'api'`）；
+- refactor(vendor-channels): 将供应商转换通道从目标专属重构为源→目标绑定模型;
+- docs(user-guide): 补充 POST /v1/messages 完整 API 参考文档;
+
+### 🔧 更多特性
+
+- feat(server): 默认监听端口 `8046` → `3392`，规范化 [Negentropy](https://github.com/ThreeFish-AI/negentropy) 体系端口；
 
 ## [v0.2.3](https://github.com/ThreeFish-AI/coding-proxy/releases/tag/v0.2.3) — 2026-04-16
 

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: coding-proxy
-Version: 0.3.0a3
+Version: 0.3.1a1
 Summary: A High-Availability, Transparent, and Smart Multi-Vendor Proxy for Claude Code. Support Claude Plans, GitHub Copilot, Google Antigravity, ZAI/GLM, MiniMax, Qwen, Xiaomi, Kimi, Doubao...
 Project-URL: Source Code, https://github.com/ThreeFish-AI/coding-proxy
 Project-URL: User Guide, https://github.com/ThreeFish-AI/coding-proxy/blob/master/docs/user-guide.md

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "coding-proxy"
-version = "0.3.0a3"
+version = "0.3.1a1"
 description = "A High-Availability, Transparent, and Smart Multi-Vendor Proxy for Claude Code. Support Claude Plans, GitHub Copilot, Google Antigravity, ZAI/GLM, MiniMax, Qwen, Xiaomi, Kimi, Doubao..."
 readme = "README.md"
 requires-python = ">=3.12"

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/src/coding/proxy/config/config.default.yaml

@@ -485,6 +485,9 @@ pricing:
   - vendor: gemini
     model: text-embedding-004
     input_cost_per_mtok: $0.10
+  - vendor: gemini
+    model: text-embedding-005
+    input_cost_per_mtok: $0.10
   # ── Google Antigravity（参照 Anthropic 定价估算） ──
   - vendor: antigravity
     model: claude-opus-4-6-thinking

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/src/coding/proxy/routing/model_mapper.py

@@ -62,12 +62,6 @@ class ModelMapper:
                 continue
             if not rule.is_regex and "*" not in rule.pattern:
                 if rule.pattern == model:
-                    logger.debug(
-                        "Model mapped: %s -> %s (vendor=%s exact)",
-                        model,
-                        rule.target,
-                        display_name,
-                    )
                     return rule.target
 
         # 2. 通配符/正则匹配
@@ -77,23 +71,9 @@ class ModelMapper:
             if rule.is_regex:
                 compiled = self._compiled[rule.pattern]
                 if compiled.fullmatch(model):
-                    logger.debug(
-                        "Model mapped: %s -> %s (vendor=%s regex=%s)",
-                        model,
-                        rule.target,
-                        display_name,
-                        rule.pattern,
-                    )
                     return rule.target
             elif "*" in rule.pattern:
                 if fnmatch.fnmatch(model, rule.pattern):
-                    logger.debug(
-                        "Model mapped: %s -> %s (vendor=%s glob=%s)",
-                        model,
-                        rule.target,
-                        display_name,
-                        rule.pattern,
-                    )
                     return rule.target
 
         # 3. 默认值

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/src/coding/proxy/routing/usage_parser.py

@@ -172,10 +172,6 @@ def parse_usage_from_chunk(
         if isinstance(msg, dict) and "usage" in msg:
             u = msg["usage"]
             input_tokens = u.get("input_tokens", 0) or u.get("prompt_tokens", 0)
-            if input_tokens > 0:
-                logger.debug(
-                    "Extracted input tokens from message.usage: %d", input_tokens
-                )
             _set_if_nonzero(usage, "input_tokens", input_tokens)
             _set_if_nonzero(
                 usage, "cache_creation_tokens", u.get("cache_creation_input_tokens", 0)
@@ -204,18 +200,6 @@ def parse_usage_from_chunk(
             cache_creation_tokens = u.get("cache_creation_input_tokens", 0)
             cache_read_tokens = u.get("cache_read_input_tokens", 0)
 
-            _label = f" ({vendor_label})" if vendor_label else ""
-            if output_tokens > 0:
-                logger.debug(
-                    "Extracted output tokens from data.usage: %d%s",
-                    output_tokens,
-                    _label,
-                )
-            if input_tokens > 0:
-                logger.debug(
-                    "Extracted input tokens from data.usage: %d%s", input_tokens, _label
-                )
-
             _set_if_nonzero(usage, "output_tokens", output_tokens)
             _set_if_nonzero(usage, "input_tokens", input_tokens)
             _set_if_nonzero(usage, "cache_creation_tokens", cache_creation_tokens)
@@ -241,20 +225,6 @@ def parse_usage_from_chunk(
                 thoughts_tc = int(um.get("thoughtsTokenCount", 0) or 0)
                 tool_use_tc = int(um.get("toolUsePromptTokenCount", 0) or 0)
 
-                _label = f" ({vendor_label})" if vendor_label else ""
-                if prompt_tc > 0:
-                    logger.debug(
-                        "Extracted Gemini prompt tokens from usageMetadata: %d%s",
-                        prompt_tc,
-                        _label,
-                    )
-                if cand_tc > 0:
-                    logger.debug(
-                        "Extracted Gemini candidates tokens from usageMetadata: %d%s",
-                        cand_tc,
-                        _label,
-                    )
-
                 _set_if_nonzero(usage, "input_tokens", prompt_tc)
                 _set_if_nonzero(usage, "output_tokens", cand_tc)
                 _set_if_nonzero(usage, "cache_read_tokens", cached_tc)

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/src/coding/proxy/server/dashboard.py

@@ -699,14 +699,15 @@ function makeExternalTooltipHandler(opts = {}) {
 
     const titleLines = tooltip.title || [];
     const dataPoints = tooltip.dataPoints || [];
+    const visiblePoints = dataPoints.filter(dp => dp.raw);
 
     let html = '';
     if (titleLines.length) {
       html += '<div id="chart-tooltip-title">' + titleLines.join('<br>') + '</div>';
     }
-    if (dataPoints.length) {
+    if (visiblePoints.length) {
       html += '<div id="chart-tooltip-items">';
-      dataPoints.forEach(dp => {
+      visiblePoints.forEach(dp => {
         // 折线图: borderColor 为字符串；甜甜圈图: backgroundColor 为数组
         const bg = dp.dataset.backgroundColor;
         const color = dp.dataset.borderColor
@@ -722,8 +723,8 @@ function makeExternalTooltipHandler(opts = {}) {
       });
       html += '</div>';
     }
-    if (showTotal && dataPoints.length > 1) {
-      const total = dataPoints.reduce((s, dp) => s + (dp.raw || 0), 0);
+    if (showTotal && visiblePoints.length > 1) {
+      const total = visiblePoints.reduce((s, dp) => s + (dp.raw || 0), 0);
       if (total > 0) {
         html += '<div id="chart-tooltip-footer">合计: ' + fmtTotalFn(total) + '</div>';
       }

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/tests/test_model_mapper.py

@@ -1,7 +1,5 @@
 """模型映射器单元测试."""
 
-import logging
-
 from coding.proxy.config.schema import ModelMappingRule
 from coding.proxy.routing.model_mapper import ModelMapper
 
@@ -95,19 +93,3 @@ def test_legacy_rule_only_applies_to_fallback():
         )
         == "claude-sonnet-4-20250514"
     )
-
-
-def test_zhipu_vendor_logs_with_original_name(caplog):
-    """zhipu 供应商传入时，日志应显示 vendor=zhipu 而非 vendor=fallback."""
-    caplog.set_level(logging.DEBUG, logger="coding.proxy.routing.model_mapper")
-    mapper = _make_mapper(
-        [
-            ModelMappingRule(pattern="claude-sonnet-*", target="glm-5.1"),
-        ]
-    )
-    result = mapper.map("claude-sonnet-4-20250514", vendor="zhipu")
-    assert result == "glm-5.1"
-
-    # 日志中应包含 vendor=zhipu 而非 vendor=fallback
-    assert any("vendor=zhipu" in r.message for r in caplog.records)
-    assert not any("vendor=fallback" in r.message for r in caplog.records)

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/tests/test_parse_usage.py

@@ -1,7 +1,5 @@
 """SSE chunk 用量解析单元测试 — 覆盖 Anthropic / OpenAI(Zhipu) / 混合格式."""
 
-import logging
-
 from coding.proxy.routing.usage_parser import _set_if_nonzero, parse_usage_from_chunk
 
 
@@ -267,63 +265,3 @@ def test_multiple_sse_lines_in_single_chunk():
     parse_usage_from_chunk(chunk, usage)
     assert usage["input_tokens"] == 80
     assert usage["output_tokens"] == 20
-
-
-# --- vendor_label 日志标签 ---
-
-
-def test_vendor_label_anthropic(caplog):
-    """传入 vendor_label=Anthropic 时，日志应标注 (Anthropic)."""
-    caplog.set_level(logging.DEBUG, logger="coding.proxy.routing.usage_parser")
-    usage: dict = {}
-    parse_usage_from_chunk(
-        _sse(
-            '{"type":"message_delta","delta":{},"usage":{"output_tokens":42,"input_tokens":10}}'
-        ),
-        usage,
-        vendor_label="Anthropic",
-    )
-    assert any("(Anthropic)" in r.message for r in caplog.records)
-
-
-def test_vendor_label_openai(caplog):
-    """传入 vendor_label=OpenAI 时，日志应标注 (OpenAI)."""
-    caplog.set_level(logging.DEBUG, logger="coding.proxy.routing.usage_parser")
-    usage: dict = {}
-    parse_usage_from_chunk(
-        _sse('{"usage":{"prompt_tokens":100,"completion_tokens":25}}'),
-        usage,
-        vendor_label="OpenAI",
-    )
-    assert any("(OpenAI)" in r.message for r in caplog.records)
-
-
-def test_vendor_label_gemini(caplog):
-    """传入 vendor_label=Gemini 时，日志应标注 (Gemini)."""
-    caplog.set_level(logging.DEBUG, logger="coding.proxy.routing.usage_parser")
-    usage: dict = {}
-    parse_usage_from_chunk(
-        _sse(
-            '{"type":"message_delta","delta":{},"usage":{"output_tokens":30,"input_tokens":200}}'
-        ),
-        usage,
-        vendor_label="Gemini",
-    )
-    assert any("(Gemini)" in r.message for r in caplog.records)
-
-
-def test_no_vendor_label_omits_parenthesis(caplog):
-    """不传 vendor_label 时，日志不应包含括号标签."""
-    caplog.set_level(logging.DEBUG, logger="coding.proxy.routing.usage_parser")
-    usage: dict = {}
-    parse_usage_from_chunk(
-        _sse('{"type":"message_delta","delta":{},"usage":{"output_tokens":10}}'),
-        usage,
-    )
-    # 不应有任何带括号的 vendor 标签
-    for record in caplog.records:
-        assert (
-            "(" not in record.message
-            or "Extracted" not in record.message
-            or ") " not in record.message[record.message.index("(") :]
-        )

{coding_proxy-0.3.0a3 → coding_proxy-0.3.1a1}/uv.lock

@@ -74,7 +74,7 @@ wheels = [
 
 [[package]]
 name = "coding-proxy"
-version = "0.3.0a3"
+version = "0.3.1a1"
 source = { editable = "." }
 dependencies = [
     { name = "aiosqlite" },