W21: anthropic Stream+Tools + --prompt batch + sanitizer fix + plugin unit tests (W21.1-W21.6)

- W21.1: ci-sanitize preset 独立 Linux-clang + ci-threadsan (TSan) - W21.2: anthropic tool_use content_block 解析 + configure 缓存 tools_json - W21.3: --prompt 非交互批处理模式 - W21.4: session auto-save 失败告警 + 当前目录 fallback - W21.5: smoke 补 tool_calls 边界用例 4 块 12 断言 - W21.6: anthropic 11 块 78 CHECK + deepseek 12 块 78 CHECK Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-05-27 20:40:58 +08:00
parent 20ead86e88
commit b2b381b9b3
14 changed files with 1671 additions and 34 deletions
--- a/plugins/anthropic/src/anthropic_plugin.cpp
+++ b/plugins/anthropic/src/anthropic_plugin.cpp
@@ -6,6 +6,7 @@
 #include <atomic>
 #include <cstring>
 #include <string>
+#include <vector>

 namespace json = boost::json;

@@ -28,6 +29,7 @@ struct PluginConfig {
    double temperature = 0.7;
 };
 static PluginConfig g_cfg;
+static std::string g_tools_json;  // W21.2: cached by configure(), consumed by chat/chat_stream

 // ============================================================================
 // 安全擦除：用 volatile 写零循环防止编译器优化
@@ -79,6 +81,7 @@ static std::string build_headers_json()
 static std::string build_request_json(
    const dstalk_message_t* history, int history_len,
    const std::string& user_input,
+    const std::string& tools_json,
    bool stream)
 {
    json::object root;
@@ -121,6 +124,11 @@ static std::string build_request_json(
        root["temperature"] = g_cfg.temperature;
    }

+    // W21.2: tools 定义传递给 API
+    if (!tools_json.empty()) {
+        root["tools"] = json::parse(tools_json);
+    }
+
    return json::serialize(root);
 }

@@ -161,21 +169,51 @@ static void parse_response(const char* body, int http_status,
        auto obj = jv.as_object();
        auto content = obj["content"].as_array();
        if (!content.empty()) {
-            // 取第一个 text block
+            // W21.2: 提取 text 和 tool_use content blocks
+            std::string text_content;
+            json::array tool_use_blocks;
+
            for (const auto& block : content) {
                auto bobj = block.as_object();
-                if (bobj.contains("type") &&
-                    json::value_to<std::string>(bobj["type"]) == "text") {
-                    std::string text = json::value_to<std::string>(bobj["text"]);
-                    r.content = h->strdup(text.c_str());
-                    r.ok = 1;
-                    r.error = nullptr;
-                    r.tool_calls_json = nullptr;
-                    return;
+                if (!bobj.contains("type")) continue;
+
+                std::string btype = json::value_to<std::string>(bobj["type"]);
+                if (btype == "text") {
+                    text_content = json::value_to<std::string>(bobj["text"]);
+                } else if (btype == "tool_use") {
+                    // 转换为 OpenAI 兼容格式: {id, type:"function", function:{name, arguments}}
+                    json::object tc;
+                    tc["id"] = bobj["id"];
+                    tc["type"] = "function";
+                    json::object func;
+                    func["name"] = bobj["name"];
+                    func["arguments"] = json::serialize(bobj["input"]);
+                    tc["function"] = func;
+                    tool_use_blocks.push_back(std::move(tc));
                }
            }
+
+            if (!tool_use_blocks.empty()) {
+                r.tool_calls_json = h->strdup(
+                    json::serialize(tool_use_blocks).c_str());
+            } else {
+                r.tool_calls_json = nullptr;
+            }
+
+            if (!text_content.empty()) {
+                r.content = h->strdup(text_content.c_str());
+                r.ok = 1;
+                r.error = nullptr;
+                return;
+            } else if (!tool_use_blocks.empty()) {
+                // tool-only 响应
+                r.content = nullptr;
+                r.ok = 1;
+                r.error = nullptr;
+                return;
+            }
            r.ok = 0;
-            r.error = h->strdup("no text content block found");
+            r.error = h->strdup("no text or tool_use content block found");
        } else {
            r.ok = 0;
            r.error = h->strdup("empty response");
@@ -200,9 +238,26 @@ static void parse_response(const char* body, int http_status,
 // SSE 事件解析（Anthropic 格式: event/content_block_delta)
 // ============================================================================

-// 状态机：记录当前正在处理的事件类型
-// 简化版：直接从 data: 行解析，不依赖 event: 行
-static bool parse_sse_data(const std::string& data, std::string& token_out)
+// W21.2: 按 content_block index 累积 Anthropic tool_use 增量
+struct ToolCallAccum {
+    int index = -1;
+    std::string id;
+    std::string name;
+    std::string arguments;  // 从 input_json_delta.partial_json 累积
+};
+
+struct StreamContext {
+    const dstalk_host_api_t* host;
+    dstalk_stream_cb user_cb;
+    void* userdata;
+    std::string accumulated;
+    bool saw_data_line = false;
+    std::vector<ToolCallAccum> tool_calls;  // W21.2: 按 index 累积 tool_use content blocks
+};
+
+// W21.2: 解析 Anthropic SSE 事件，含 tool_use content_block 增量解析
+static bool parse_sse_data(const std::string& data, std::string& token_out,
+                           StreamContext* ctx)
 {
    try {
        auto jv = json::parse(data);
@@ -212,6 +267,34 @@ static bool parse_sse_data(const std::string& data, std::string& token_out)
        if (!type_ptr || !type_ptr->is_string()) return false;
        std::string type = json::value_to<std::string>(*type_ptr);

+        if (type == "content_block_start") {
+            // content_block_start 可能为 tool_use
+            auto* cb = obj.if_contains("content_block");
+            if (!cb || !cb->is_object()) return false;
+            auto& cb_obj = cb->as_object();
+            auto* cb_type = cb_obj.if_contains("type");
+            if (!cb_type || !cb_type->is_string()) return false;
+            std::string cb_type_str = json::value_to<std::string>(*cb_type);
+
+            if (cb_type_str == "tool_use" && ctx) {
+                auto* idx_ptr = obj.if_contains("index");
+                int idx = idx_ptr ? static_cast<int>(
+                    json::value_to<int64_t>(*idx_ptr)) : -1;
+                if (idx < 0) return false;
+
+                while (static_cast<int>(ctx->tool_calls.size()) <= idx) {
+                    ctx->tool_calls.push_back({});
+                }
+                auto& acc = ctx->tool_calls[idx];
+                acc.index = idx;
+                if (cb_obj.contains("id") && cb_obj["id"].is_string())
+                    acc.id = json::value_to<std::string>(cb_obj["id"]);
+                if (cb_obj.contains("name") && cb_obj["name"].is_string())
+                    acc.name = json::value_to<std::string>(cb_obj["name"]);
+            }
+            return false;
+        }
+
        if (type == "content_block_delta") {
            auto* delta = obj.if_contains("delta");
            if (!delta || !delta->is_object()) return false;
@@ -227,12 +310,25 @@ static bool parse_sse_data(const std::string& data, std::string& token_out)
                    token_out = json::value_to<std::string>(*text);
                    return true;
                }
+            } else if (delta_type == "input_json_delta" && ctx) {
+                // W21.2: 累积 tool_use arguments 分片
+                auto* pj = dobj.if_contains("partial_json");
+                if (pj && pj->is_string()) {
+                    auto* idx_ptr = obj.if_contains("index");
+                    int idx = idx_ptr ? static_cast<int>(
+                        json::value_to<int64_t>(*idx_ptr)) : -1;
+                    if (idx >= 0 && idx < static_cast<int>(ctx->tool_calls.size())) {
+                        ctx->tool_calls[idx].arguments +=
+                            json::value_to<std::string>(*pj);
+                    }
+                }
+                return false;
            }
        } else if (type == "message_stop") {
            token_out.clear();
            return true;  // 流结束
        }
-        // 忽略: message_start, content_block_start, content_block_stop, ping, message_delta
+        // 忽略: message_start, content_block_stop, ping, message_delta
    } catch (...) {
        // 解析失败忽略
    }
@@ -256,6 +352,17 @@ static int my_configure(const char* provider, const char* base_url,

        const auto* h = g_host.load(std::memory_order_acquire);
        if (h) {
+            // W21.2: 从 tools service 缓存 tools_json，供 chat/chat_stream 复用
+            auto* tools_svc = reinterpret_cast<const dstalk_tools_service_t*>(
+                h->query_service("tools", 1));
+            if (tools_svc && tools_svc->get_tools_json) {
+                char* json = tools_svc->get_tools_json();
+                if (json) {
+                    g_tools_json = json;
+                    h->free(json);
+                }
+            }
+
            h->log(DSTALK_LOG_INFO,
                "[anthropic] configured: model=%s base_url=%s max_tokens=%d temperature=%.2f",
                g_cfg.model.c_str(), g_cfg.base_url.c_str(),
@@ -279,7 +386,7 @@ static int my_configure(const char* provider, const char* base_url,
 static dstalk_chat_result_t my_chat(
    const dstalk_message_t* history, int history_len,
    const char* user_input,
-    const char* /*tools_json*/)
+    const char* tools_json)
 {
    try {
        dstalk_chat_result_t r = {};
@@ -298,7 +405,8 @@ static dstalk_chat_result_t my_chat(
        std::string target_path = target + "/v1/messages";

        std::string body = build_request_json(history, history_len,
-            user_input ? user_input : "", false);
+            user_input ? user_input : "",
+            tools_json ? tools_json : g_tools_json, false);

        std::string headers_json = build_headers_json();

@@ -342,14 +450,6 @@ static dstalk_chat_result_t my_chat(
 // chat_stream
 // ============================================================================

-struct StreamContext {
-    const dstalk_host_api_t* host;
-    dstalk_stream_cb user_cb;
-    void* userdata;
-    std::string accumulated;
-    bool saw_data_line = false;
-};
-
 // 行回调
 static int sse_line_callback(const char* line, void* userdata)
 {
@@ -363,7 +463,7 @@ static int sse_line_callback(const char* line, void* userdata)
        if (line_str.rfind("data: ", 0) == 0) {
            std::string data = line_str.substr(6);
            std::string token;
-            if (parse_sse_data(data, token)) {
+            if (parse_sse_data(data, token, ctx)) {
                ctx->saw_data_line = true;
                if (token.empty()) {
                    // message_stop
@@ -410,7 +510,7 @@ static dstalk_chat_result_t my_chat_stream(
        std::string target_path = target + "/v1/messages";

        std::string body = build_request_json(history, history_len,
-            user_input ? user_input : "", true);
+            user_input ? user_input : "", g_tools_json, true);

        std::string headers_json = build_headers_json();

@@ -460,7 +560,11 @@ static dstalk_chat_result_t my_chat_stream(

        if (response_body) host->free(response_body);

-        if (ctx.accumulated.empty() && !ctx.saw_data_line) {
+        // W21.2: 成功条件 = 有内容 OR 有 tool_calls（tool-only 响应如 function calling）
+        bool has_content = !ctx.accumulated.empty();
+        bool has_tool_calls = !ctx.tool_calls.empty();
+
+        if (!has_content && !has_tool_calls) {
            r.ok = 0;
            r.error = host->strdup("no content received");
            r.content = nullptr;
@@ -468,8 +572,28 @@ static dstalk_chat_result_t my_chat_stream(
        } else {
            r.ok = 1;
            r.error = nullptr;
-            r.content = host->strdup(ctx.accumulated.c_str());
-            r.tool_calls_json = nullptr;
+            r.content = has_content
+                ? host->strdup(ctx.accumulated.c_str()) : nullptr;
+
+            // W21.2: 序列化累积的 tool_calls 为 JSON（兼容 OpenAI tool_calls 格式）
+            if (has_tool_calls) {
+                json::array tc_array;
+                for (auto& tc : ctx.tool_calls) {
+                    json::object tc_obj;
+                    tc_obj["index"] = tc.index;
+                    if (!tc.id.empty())   tc_obj["id"]   = tc.id;
+                    tc_obj["type"] = "function";
+                    json::object func;
+                    if (!tc.name.empty()) func["name"] = tc.name;
+                    func["arguments"] = tc.arguments;
+                    tc_obj["function"] = func;
+                    tc_array.push_back(std::move(tc_obj));
+                }
+                std::string tc_json = json::serialize(tc_array);
+                r.tool_calls_json = host ? host->strdup(tc_json.c_str()) : nullptr;
+            } else {
+                r.tool_calls_json = nullptr;
+            }
        }
        return r;
    } catch (const std::exception& e) {