sema-core 2.0.10 → 2.0.11

package/dist/util/microcompact.js

@@ -0,0 +1,194 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.MICRO_COMPACT_PROTECT_EXTRA_BLOCKS = exports.MICRO_COMPACT_MIN_CONTENT_CHARS = exports.MICRO_COMPACT_PLACEHOLDER_PREFIX = void 0;
+exports.estimateTokensFromText = estimateTokensFromText;
+exports.microCompactMessages = microCompactMessages;
+/**
+ * Micro 压缩：把模型已消费过的旧 tool_result 块内容替换为占位符，
+ * 在昂贵的全量摘要之前廉价释放上下文空间。
+ *
+ * 纯函数模块：无 IO、无事件、无配置读取，集成逻辑见 compact.ts 的 applyMicroCompact。
+ *
+ * 设计约束：
+ * - 只替换 tool_result 块的 content，不删除任何消息；
+ *   uuid / tool_use_id / toolUseResult / controlSignal / checkpointSeq 全部保留，
+ *   tool_use 与 tool_result 的配对关系结构上不可能被破坏；
+ * - 用户输入（含用户粘贴的图片）与 assistant 消息永不触碰；
+ * - 命中的消息用 spread 重建，绝不原地 mutate（messages 元素与 StateManager 共享引用）。
+ */
+// 幂等判据 + 协议常量：勿改措辞、勿本地化（旧代码眼中只是普通文本，revert 安全）
+exports.MICRO_COMPACT_PLACEHOLDER_PREFIX = '<cleared>';
+const PLACEHOLDER = '<cleared>Old tool result removed to free context. Re-run the tool if this output is needed again.</cleared>';
+const IMAGE_PLACEHOLDER = '<cleared>Old tool result removed to free context (an image was removed). Re-run the tool if this output is needed again.</cleared>';
+// 文本清理门槛：≈500 token，相对占位符收益比 ≥16:1
+exports.MICRO_COMPACT_MIN_CONTENT_CHARS = 2000;
+// 未消费批次（最后一条 assistant 之后的全部块）之外，额外保护的尾部块数
+exports.MICRO_COMPACT_PROTECT_EXTRA_BLOCKS = 3;
+// 图片块固定估算（勿用 base64 长度折算，会严重高估）
+const IMAGE_BLOCK_TOKEN_ESTIMATE = 1500;
+// 占位符自身的 token 开销，从节省估算中扣除（偏保守）
+const PLACEHOLDER_TOKEN_ESTIMATE = 30;
+function isCjkCharCode(code) {
+    return ((code >= 0x2e80 && code <= 0x9fff) || // CJK 部首/日文假名/CJK 统一表意
+        (code >= 0xf900 && code <= 0xfaff) || // CJK 兼容表意
+        (code >= 0xff00 && code <= 0xffef) || // 全角符号
+        (code >= 0xac00 && code <= 0xd7af) // 谚文音节
+    );
+}
+/**
+ * 估算文本 token 数：CJK 字符 ≈ 1 token/字符，其余按 4 字符/token。
+ * 误差方向刻意偏向低估节省（宁可多做一次全量摘要，不可少做）。
+ */
+function estimateTokensFromText(text) {
+    let cjkCount = 0;
+    for (let i = 0; i < text.length; i++) {
+        if (isCjkCharCode(text.charCodeAt(i)))
+            cjkCount++;
+    }
+    return cjkCount + Math.ceil((text.length - cjkCount) / 4);
+}
+function isClearedContent(content) {
+    if (typeof content === 'string') {
+        return content.startsWith(exports.MICRO_COMPACT_PLACEHOLDER_PREFIX);
+    }
+    if (Array.isArray(content)) {
+        const first = content[0];
+        return (content.length === 1 &&
+            first?.type === 'text' &&
+            first.text.startsWith(exports.MICRO_COMPACT_PLACEHOLDER_PREFIX));
+    }
+    return false;
+}
+function analyzeBlock(block) {
+    // 报错结果永不清理（通常很短，且对后续纠错有持续价值）
+    if (block.is_error === true)
+        return { clearable: false };
+    // 幂等：已清理过的块跳过
+    if (isClearedContent(block.content))
+        return { clearable: false };
+    const content = block.content;
+    if (typeof content === 'string') {
+        if (content.length < exports.MICRO_COMPACT_MIN_CONTENT_CHARS)
+            return { clearable: false };
+        return {
+            clearable: true,
+            savedTokens: Math.max(0, estimateTokensFromText(content) - PLACEHOLDER_TOKEN_ESTIMATE),
+            placeholder: PLACEHOLDER,
+        };
+    }
+    if (Array.isArray(content)) {
+        let textChars = 0;
+        let textTokens = 0;
+        let imageCount = 0;
+        for (const item of content) {
+            if (item.type === 'text') {
+                textChars += item.text.length;
+                textTokens += estimateTokensFromText(item.text);
+            }
+            else if (item.type === 'image') {
+                imageCount++;
+            }
+            else {
+                // 含未知块类型的结果不动，保守优先
+                return { clearable: false };
+            }
+        }
+        // 图片块无条件清（单张 ≈1500 token，不设体积门槛）；纯文本按门槛判断
+        if (imageCount === 0 && textChars < exports.MICRO_COMPACT_MIN_CONTENT_CHARS) {
+            return { clearable: false };
+        }
+        return {
+            clearable: true,
+            savedTokens: Math.max(0, textTokens + imageCount * IMAGE_BLOCK_TOKEN_ESTIMATE - PLACEHOLDER_TOKEN_ESTIMATE),
+            placeholder: imageCount > 0 ? IMAGE_PLACEHOLDER : PLACEHOLDER,
+        };
+    }
+    return { clearable: false };
+}
+/**
+ * 扫描全部历史（不分用户轮次，含当前轮内部），清理保护区之外符合条件的 tool_result 块。
+ *
+ * 保护区（原封不动）：
+ * - 最后一条 assistant 消息之后的全部 tool_result 块：模型刚发起调用、尚未消费的
+ *   最新一批（数量 = 上一步并行调用数），整批无条件保护；
+ * - 再往前 MICRO_COMPACT_PROTECT_EXTRA_BLOCKS 个块。
+ */
+function microCompactMessages(messages) {
+    const unchanged = {
+        changed: false,
+        messages,
+        clearedCount: 0,
+        estimatedSavedTokens: 0,
+    };
+    // 没有 assistant 消息则不存在"已消费"的结果，全部视为保护
+    let lastAssistantIdx = -1;
+    for (let i = messages.length - 1; i >= 0; i--) {
+        if (messages[i].type === 'assistant') {
+            lastAssistantIdx = i;
+            break;
+        }
+    }
+    if (lastAssistantIdx === -1)
+        return unchanged;
+    // 收集最后一条 assistant 之前的全部 tool_result 块位置（其后的块天然不进候选 = 未消费批次整批保护）
+    const candidates = [];
+    for (let i = 0; i < lastAssistantIdx; i++) {
+        const msg = messages[i];
+        if (msg.type !== 'user' || !Array.isArray(msg.message.content))
+            continue;
+        const content = msg.message.content;
+        for (let blockIdx = 0; blockIdx < content.length; blockIdx++) {
+            if (content[blockIdx].type === 'tool_result') {
+                candidates.push({ msgIdx: i, blockIdx });
+            }
+        }
+    }
+    // 尾部再额外保护 N 个块
+    const clearableCandidates = candidates.slice(0, Math.max(0, candidates.length - exports.MICRO_COMPACT_PROTECT_EXTRA_BLOCKS));
+    if (clearableCandidates.length === 0)
+        return unchanged;
+    // 逐块判定，生成清理计划：msgIdx -> (blockIdx -> 占位符)
+    const clearPlan = new Map();
+    let clearedCount = 0;
+    let estimatedSavedTokens = 0;
+    for (const { msgIdx, blockIdx } of clearableCandidates) {
+        const content = messages[msgIdx].message.content;
+        const verdict = analyzeBlock(content[blockIdx]);
+        if (!verdict.clearable)
+            continue;
+        let blockMap = clearPlan.get(msgIdx);
+        if (!blockMap) {
+            blockMap = new Map();
+            clearPlan.set(msgIdx, blockMap);
+        }
+        blockMap.set(blockIdx, verdict.placeholder);
+        clearedCount++;
+        estimatedSavedTokens += verdict.savedTokens;
+    }
+    if (clearedCount === 0)
+        return unchanged;
+    // spread 重建命中的消息，未命中的复用原引用
+    const nextMessages = messages.map((msg, msgIdx) => {
+        const blockMap = clearPlan.get(msgIdx);
+        if (!blockMap)
+            return msg;
+        const userMsg = msg;
+        const content = userMsg.message.content.map((block, blockIdx) => {
+            const placeholder = blockMap.get(blockIdx);
+            if (placeholder === undefined)
+                return block;
+            return { ...block, content: placeholder };
+        });
+        return {
+            ...userMsg,
+            message: { ...userMsg.message, content },
+        };
+    });
+    return {
+        changed: true,
+        messages: nextMessages,
+        clearedCount,
+        estimatedSavedTokens,
+    };
+}
+//# sourceMappingURL=microcompact.js.map

package/dist/util/microcompact.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"microcompact.js","sourceRoot":"","sources":["../../src/util/microcompact.ts"],"names":[],"mappings":";;;AAsDA,wDAMC;AAgFD,oDAqFC;AA9ND;;;;;;;;;;;;GAYG;AAEH,+CAA+C;AAClC,QAAA,gCAAgC,GAAG,WAAW,CAAA;AAE3D,MAAM,WAAW,GACf,6GAA6G,CAAA;AAC/G,MAAM,iBAAiB,GACrB,oIAAoI,CAAA;AAEtI,mCAAmC;AACtB,QAAA,+BAA+B,GAAG,IAAI,CAAA;AACnD,2CAA2C;AAC9B,QAAA,kCAAkC,GAAG,CAAC,CAAA;AACnD,gCAAgC;AAChC,MAAM,0BAA0B,GAAG,IAAI,CAAA;AACvC,gCAAgC;AAChC,MAAM,0BAA0B,GAAG,EAAE,CAAA;AASrC,SAAS,aAAa,CAAC,IAAY;IACjC,OAAO,CACL,CAAC,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,IAAI,uBAAuB;QAC7D,CAAC,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,IAAI,WAAW;QACjD,CAAC,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,IAAI,OAAO;QAC7C,CAAC,IAAI,IAAI,MAAM,IAAI,IAAI,IAAI,MAAM,CAAC,CAAI,OAAO;KAC9C,CAAA;AACH,CAAC;AAED;;;GAGG;AACH,SAAgB,sBAAsB,CAAC,IAAY;IACjD,IAAI,QAAQ,GAAG,CAAC,CAAA;IAChB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACrC,IAAI,aAAa,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC;YAAE,QAAQ,EAAE,CAAA;IACnD,CAAC;IACD,OAAO,QAAQ,GAAG,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,MAAM,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAA;AAC3D,CAAC;AAID,SAAS,gBAAgB,CAAC,OAAmC;IAC3D,IAAI,OAAO,OAAO,KAAK,QAAQ,EAAE,CAAC;QAChC,OAAO,OAAO,CAAC,UAAU,CAAC,wCAAgC,CAAC,CAAA;IAC7D,CAAC;IACD,IAAI,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,MAAM,KAAK,GAAG,OAAO,CAAC,CAAC,CAA4C,CAAA;QACnE,OAAO,CACL,OAAO,CAAC,MAAM,KAAK,CAAC;YACpB,KAAK,EAAE,IAAI,KAAK,MAAM;YACtB,KAAK,CAAC,IAAI,CAAC,UAAU,CAAC,wCAAgC,CAAC,CACxD,CAAA;IACH,CAAC;IACD,OAAO,KAAK,CAAA;AACd,CAAC;AAMD,SAAS,YAAY,CAAC,KAAsB;IAC1C,6BAA6B;IAC7B,IAAI,KAAK,CAAC,QAAQ,KAAK,IAAI;QAAE,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;IACxD,cAAc;IACd,IAAI,gBAAgB,CAAC,KAAK,CAAC,OAAO,CAAC;QAAE,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;IAEhE,MAAM,OAAO,GAAG,KAAK,CAAC,OAAO,CAAA;IAE7B,IAAI,OAAO,OAAO,KAAK,QAAQ,EAAE,CAAC;QAChC,IAAI,OAAO,CAAC,MAAM,GAAG,uCAA+B;YAAE,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;QACjF,OAAO;YACL,SAAS,EAAE,IAAI;YACf,WAAW,EAAE,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,sBAAsB,CAAC,OAAO,CAAC,GAAG,0BAA0B,CAAC;YACtF,WAAW,EAAE,WAAW;SACzB,CAAA;IACH,CAAC;IAED,IAAI,KAAK,CAAC,OAAO,CAAC,OAAO,CAAC,EAAE,CAAC;QAC3B,IAAI,SAAS,GAAG,CAAC,CAAA;QACjB,IAAI,UAAU,GAAG,CAAC,CAAA;QAClB,IAAI,UAAU,GAAG,CAAC,CAAA;QAClB,KAAK,MAAM,IAAI,IAAI,OAAwC,EAAE,CAAC;YAC5D,IAAI,IAAI,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;gBACzB,SAAS,IAAI,IAAI,CAAC,IAAI,CAAC,MAAM,CAAA;gBAC7B,UAAU,IAAI,sBAAsB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAA;YACjD,CAAC;iBAAM,IAAI,IAAI,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;gBACjC,UAAU,EAAE,CAAA;YACd,CAAC;iBAAM,CAAC;gBACN,mBAAmB;gBACnB,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;YAC7B,CAAC;QACH,CAAC;QACD,0CAA0C;QAC1C,IAAI,UAAU,KAAK,CAAC,IAAI,SAAS,GAAG,uCAA+B,EAAE,CAAC;YACpE,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;QAC7B,CAAC;QACD,OAAO;YACL,SAAS,EAAE,IAAI;YACf,WAAW,EAAE,IAAI,CAAC,GAAG,CACnB,CAAC,EACD,UAAU,GAAG,UAAU,GAAG,0BAA0B,GAAG,0BAA0B,CAClF;YACD,WAAW,EAAE,UAAU,GAAG,CAAC,CAAC,CAAC,CAAC,iBAAiB,CAAC,CAAC,CAAC,WAAW;SAC9D,CAAA;IACH,CAAC;IAED,OAAO,EAAE,SAAS,EAAE,KAAK,EAAE,CAAA;AAC7B,CAAC;AAED;;;;;;;GAOG;AACH,SAAgB,oBAAoB,CAAC,QAAmB;IACtD,MAAM,SAAS,GAAuB;QACpC,OAAO,EAAE,KAAK;QACd,QAAQ;QACR,YAAY,EAAE,CAAC;QACf,oBAAoB,EAAE,CAAC;KACxB,CAAA;IAED,qCAAqC;IACrC,IAAI,gBAAgB,GAAG,CAAC,CAAC,CAAA;IACzB,KAAK,IAAI,CAAC,GAAG,QAAQ,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;QAC9C,IAAI,QAAQ,CAAC,CAAC,CAAC,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;YACrC,gBAAgB,GAAG,CAAC,CAAA;YACpB,MAAK;QACP,CAAC;IACH,CAAC;IACD,IAAI,gBAAgB,KAAK,CAAC,CAAC;QAAE,OAAO,SAAS,CAAA;IAE7C,iEAAiE;IACjE,MAAM,UAAU,GAAgD,EAAE,CAAA;IAClE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,gBAAgB,EAAE,CAAC,EAAE,EAAE,CAAC;QAC1C,MAAM,GAAG,GAAG,QAAQ,CAAC,CAAC,CAAC,CAAA;QACvB,IAAI,GAAG,CAAC,IAAI,KAAK,MAAM,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC;YAAE,SAAQ;QACxE,MAAM,OAAO,GAAG,GAAG,CAAC,OAAO,CAAC,OAAwC,CAAA;QACpE,KAAK,IAAI,QAAQ,GAAG,CAAC,EAAE,QAAQ,GAAG,OAAO,CAAC,MAAM,EAAE,QAAQ,EAAE,EAAE,CAAC;YAC7D,IAAI,OAAO,CAAC,QAAQ,CAAC,CAAC,IAAI,KAAK,aAAa,EAAE,CAAC;gBAC7C,UAAU,CAAC,IAAI,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,QAAQ,EAAE,CAAC,CAAA;YAC1C,CAAC;QACH,CAAC;IACH,CAAC;IAED,eAAe;IACf,MAAM,mBAAmB,GAAG,UAAU,CAAC,KAAK,CAC1C,CAAC,EACD,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,UAAU,CAAC,MAAM,GAAG,0CAAkC,CAAC,CACpE,CAAA;IACD,IAAI,mBAAmB,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,SAAS,CAAA;IAEtD,0CAA0C;IAC1C,MAAM,SAAS,GAAG,IAAI,GAAG,EAA+B,CAAA;IACxD,IAAI,YAAY,GAAG,CAAC,CAAA;IACpB,IAAI,oBAAoB,GAAG,CAAC,CAAA;IAE5B,KAAK,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,IAAI,mBAAmB,EAAE,CAAC;QACvD,MAAM,OAAO,GAAG,QAAQ,CAAC,MAAM,CAAC,CAAC,OAAO,CAAC,OAAwC,CAAA;QACjF,MAAM,OAAO,GAAG,YAAY,CAAC,OAAO,CAAC,QAAQ,CAAoB,CAAC,CAAA;QAClE,IAAI,CAAC,OAAO,CAAC,SAAS;YAAE,SAAQ;QAEhC,IAAI,QAAQ,GAAG,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;QACpC,IAAI,CAAC,QAAQ,EAAE,CAAC;YACd,QAAQ,GAAG,IAAI,GAAG,EAAkB,CAAA;YACpC,SAAS,CAAC,GAAG,CAAC,MAAM,EAAE,QAAQ,CAAC,CAAA;QACjC,CAAC;QACD,QAAQ,CAAC,GAAG,CAAC,QAAQ,EAAE,OAAO,CAAC,WAAW,CAAC,CAAA;QAC3C,YAAY,EAAE,CAAA;QACd,oBAAoB,IAAI,OAAO,CAAC,WAAW,CAAA;IAC7C,CAAC;IAED,IAAI,YAAY,KAAK,CAAC;QAAE,OAAO,SAAS,CAAA;IAExC,2BAA2B;IAC3B,MAAM,YAAY,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC,GAAG,EAAE,MAAM,EAAE,EAAE;QAChD,MAAM,QAAQ,GAAG,SAAS,CAAC,GAAG,CAAC,MAAM,CAAC,CAAA;QACtC,IAAI,CAAC,QAAQ;YAAE,OAAO,GAAG,CAAA;QAEzB,MAAM,OAAO,GAAG,GAAc,CAAA;QAC9B,MAAM,OAAO,GAAI,OAAO,CAAC,OAAO,CAAC,OAAyC,CAAC,GAAG,CAC5E,CAAC,KAAK,EAAE,QAAQ,EAAE,EAAE;YAClB,MAAM,WAAW,GAAG,QAAQ,CAAC,GAAG,CAAC,QAAQ,CAAC,CAAA;YAC1C,IAAI,WAAW,KAAK,SAAS;gBAAE,OAAO,KAAK,CAAA;YAC3C,OAAO,EAAE,GAAI,KAAyB,EAAE,OAAO,EAAE,WAAW,EAAE,CAAA;QAChE,CAAC,CACF,CAAA;QACD,OAAO;YACL,GAAG,OAAO;YACV,OAAO,EAAE,EAAE,GAAG,OAAO,CAAC,OAAO,EAAE,OAAO,EAAE;SAC9B,CAAA;IACd,CAAC,CAAC,CAAA;IAEF,OAAO;QACL,OAAO,EAAE,IAAI;QACb,QAAQ,EAAE,YAAY;QACtB,YAAY;QACZ,oBAAoB;KACrB,CAAA;AACH,CAAC"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "sema-core",
-  "version": "2.0.10",
+  "version": "2.0.11",
   "description": "Event-driven core library for Sema AI assistant",
   "main": "dist/index.js",
   "types": "dist/index.d.ts",