@hsingjui/contextweaver 0.0.4 → 0.0.6

package/dist/{SearchService-DYGJT2DZ.js → SearchService-ZI7QP3NE.js}

@@ -1,7 +1,7 @@
 import {
   getIndexer,
   getVectorStore
-} from "./chunk-WWYSLCNZ.js";
+} from "./chunk-TJHS7BN7.js";
 import {
   initDb,
   isChunksFtsInitialized,
@@ -9,11 +9,11 @@ import {
   searchChunksFts,
   searchFilesFts,
   segmentQuery
-} from "./chunk-VW5RACJC.js";
+} from "./chunk-OXPWRE3G.js";
 import {
   isDebugEnabled,
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 import {
   getEmbeddingConfig,
   getRerankerConfig

package/dist/{chunk-O3HDM3CF.js → chunk-2SIQBQ2M.js}

@@ -1,9 +1,9 @@
 import {
   generateProjectId
-} from "./chunk-VW5RACJC.js";
+} from "./chunk-OXPWRE3G.js";
 import {
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 
 // src/mcp/tools/codebaseRetrieval.ts
 import fs from "fs";
@@ -67,9 +67,9 @@ function isProjectIndexed(projectId) {
   const dbPath = path.join(BASE_DIR, projectId, "index.db");
   return fs.existsSync(dbPath);
 }
-async function ensureIndexed(repoPath, projectId) {
-  const { withLock } = await import("./lock-PX2BX2YN.js");
-  const { scan } = await import("./scanner-HYP3L57R.js");
+async function ensureIndexed(repoPath, projectId, onProgress) {
+  const { withLock } = await import("./lock-RC33CJZA.js");
+  const { scan } = await import("./scanner-NM7WPJJE.js");
   await withLock(projectId, "index", async () => {
     const wasIndexed = isProjectIndexed(projectId);
     if (!wasIndexed) {
@@ -77,11 +77,12 @@ async function ensureIndexed(repoPath, projectId) {
         { repoPath, projectId: projectId.slice(0, 10) },
         "\u4EE3\u7801\u5E93\u672A\u521D\u59CB\u5316\uFF0C\u5F00\u59CB\u9996\u6B21\u7D22\u5F15..."
       );
+      onProgress?.(0, 100, "\u4EE3\u7801\u5E93\u672A\u7D22\u5F15\uFF0C\u5F00\u59CB\u9996\u6B21\u7D22\u5F15...");
     } else {
       logger.debug({ projectId: projectId.slice(0, 10) }, "\u6267\u884C\u589E\u91CF\u7D22\u5F15...");
     }
     const startTime = Date.now();
-    const stats = await scan(repoPath, { vectorIndex: true });
+    const stats = await scan(repoPath, { vectorIndex: true, onProgress });
     const elapsed = Date.now() - startTime;
     logger.info(
       {
@@ -98,7 +99,7 @@ async function ensureIndexed(repoPath, projectId) {
     );
   });
 }
-async function handleCodebaseRetrieval(args, configOverride = ZEN_CONFIG_OVERRIDE) {
+async function handleCodebaseRetrieval(args, configOverride = ZEN_CONFIG_OVERRIDE, onProgress) {
   const { repo_path, information_request, technical_terms } = args;
   logger.info(
     {
@@ -118,7 +119,7 @@ async function handleCodebaseRetrieval(args, configOverride = ZEN_CONFIG_OVERRID
     return formatEnvMissingResponse(allMissingVars);
   }
   const projectId = generateProjectId(repo_path);
-  await ensureIndexed(repo_path, projectId);
+  await ensureIndexed(repo_path, projectId, onProgress);
   const query = [information_request, ...technical_terms || []].filter(Boolean).join(" ");
   logger.info(
     {
@@ -128,7 +129,7 @@ async function handleCodebaseRetrieval(args, configOverride = ZEN_CONFIG_OVERRID
     },
     "MCP \u67E5\u8BE2\u6784\u5EFA"
   );
-  const { SearchService } = await import("./SearchService-DYGJT2DZ.js");
+  const { SearchService } = await import("./SearchService-ZI7QP3NE.js");
   const service = new SearchService(projectId, repo_path, configOverride);
   await service.init();
   logger.debug("SearchService \u521D\u59CB\u5316\u5B8C\u6210");
@@ -176,16 +177,16 @@ async function handleCodebaseRetrieval(args, configOverride = ZEN_CONFIG_OVERRID
     },
     "MCP codebase-retrieval \u5B8C\u6210"
   );
-  return formatMcpResponse(contextPack, information_request);
+  return formatMcpResponse(contextPack);
 }
-function formatMcpResponse(pack, originalRequest) {
+function formatMcpResponse(pack) {
   const { files, seeds } = pack;
   const fileBlocks = files.map((file) => {
     const segments = file.segments.map((seg) => formatSegment(seg)).join("\n\n");
     return segments;
   }).join("\n\n---\n\n");
   const summary = [
-    `Found ${seeds.length} relevant code blocks for: "${originalRequest}"`,
+    `Found ${seeds.length} relevant code blocks`,
     `Files: ${files.length}`,
     `Total segments: ${files.reduce((acc, f) => acc + f.segments.length, 0)}`
   ].join(" | ");

package/dist/{chunk-QWQ64TBE.js → chunk-AA3ILFHL.js}

@@ -2,7 +2,7 @@ import {
   closeAllIndexers,
   closeAllVectorStores,
   getIndexer
-} from "./chunk-WWYSLCNZ.js";
+} from "./chunk-TJHS7BN7.js";
 import {
   batchDelete,
   batchUpdateMtime,
@@ -16,10 +16,10 @@ import {
   getStoredEmbeddingDimensions,
   initDb,
   setStoredEmbeddingDimensions
-} from "./chunk-VW5RACJC.js";
+} from "./chunk-OXPWRE3G.js";
 import {
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 import {
   getEmbeddingConfig,
   getExcludePatterns
@@ -165,7 +165,7 @@ import path2 from "path";
 import pLimit from "p-limit";
 
 // src/chunking/ParserPool.ts
-import Parser from "tree-sitter";
+import Parser from "@keqingmoe/tree-sitter";
 var GRAMMAR_MODULES = {
   typescript: "tree-sitter-typescript",
   javascript: "tree-sitter-javascript",
@@ -227,20 +227,28 @@ function isLanguageSupported(language) {
 var LANGUAGE_SPECS = {
   typescript: {
     hierarchy: /* @__PURE__ */ new Set([
-      // 类和接口
       "class_declaration",
       "abstract_class_declaration",
       "interface_declaration",
-      // 函数
       "function_declaration",
       "generator_function_declaration",
       "method_definition",
       "arrow_function",
-      // 模块
       "export_statement",
       "import_statement"
     ]),
-    nameFields: ["name", "id"]
+    nameFields: ["name", "id"],
+    nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "type_identifier", "property_identifier"]),
+    prefixMap: {
+      class_declaration: "class ",
+      abstract_class_declaration: "abstract class ",
+      interface_declaration: "interface ",
+      function_declaration: "fn ",
+      generator_function_declaration: "fn* ",
+      method_definition: "",
+      arrow_function: ""
+    },
+    commentTypes: /* @__PURE__ */ new Set(["comment"])
   },
   javascript: {
     hierarchy: /* @__PURE__ */ new Set([
@@ -250,57 +258,94 @@ var LANGUAGE_SPECS = {
       "method_definition",
       "arrow_function"
     ]),
-    nameFields: ["name", "id"]
+    nameFields: ["name", "id"],
+    nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "property_identifier"]),
+    prefixMap: {
+      class_declaration: "class ",
+      function_declaration: "fn ",
+      generator_function_declaration: "fn* ",
+      method_definition: "",
+      arrow_function: ""
+    },
+    commentTypes: /* @__PURE__ */ new Set(["comment"])
   },
   python: {
     hierarchy: /* @__PURE__ */ new Set(["class_definition", "function_definition", "decorated_definition"]),
-    nameFields: ["name"]
+    nameFields: ["name"],
+    nameNodeTypes: /* @__PURE__ */ new Set(["identifier"]),
+    prefixMap: {
+      class_definition: "class ",
+      function_definition: "def ",
+      decorated_definition: ""
+    },
+    commentTypes: /* @__PURE__ */ new Set(["comment"])
   },
   go: {
     hierarchy: /* @__PURE__ */ new Set([
-      // 函数和方法
       "function_declaration",
       "method_declaration",
-      // 类型定义
       "type_spec",
       "type_declaration",
-      // 结构体和接口
       "struct_type",
       "interface_type"
     ]),
-    nameFields: ["name"]
+    nameFields: ["name"],
+    nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "type_identifier", "field_identifier"]),
+    prefixMap: {
+      function_declaration: "func ",
+      method_declaration: "func ",
+      type_spec: "type ",
+      type_declaration: "type ",
+      struct_type: "struct ",
+      interface_type: "interface "
+    },
+    commentTypes: /* @__PURE__ */ new Set(["comment"])
   },
   rust: {
     hierarchy: /* @__PURE__ */ new Set([
-      // 函数
       "function_item",
-      // 结构体、枚举、trait
       "struct_item",
       "enum_item",
       "trait_item",
-      // impl 块
       "impl_item",
-      // 模块
       "mod_item",
-      // 类型别名
       "type_item"
     ]),
-    nameFields: ["name"]
+    nameFields: ["name"],
+    nameNodeTypes: /* @__PURE__ */ new Set(["identifier", "type_identifier"]),
+    prefixMap: {
+      function_item: "fn ",
+      struct_item: "struct ",
+      enum_item: "enum ",
+      trait_item: "trait ",
+      impl_item: "impl ",
+      mod_item: "mod ",
+      type_item: "type "
+    },
+    commentTypes: /* @__PURE__ */ new Set(["line_comment", "block_comment"])
   },
   java: {
     hierarchy: /* @__PURE__ */ new Set([
-      // 类和接口
       "class_declaration",
       "interface_declaration",
       "enum_declaration",
       "annotation_type_declaration",
-      // 方法和构造函数
       "method_declaration",
       "constructor_declaration",
-      // 记录类型 (Java 14+)
       "record_declaration"
     ]),
-    nameFields: ["name", "identifier"]
+    nameFields: ["name", "identifier"],
+    nameNodeTypes: /* @__PURE__ */ new Set(["identifier"]),
+    prefixMap: {
+      class_declaration: "class ",
+      interface_declaration: "interface ",
+      enum_declaration: "enum ",
+      annotation_type_declaration: "@interface ",
+      method_declaration: "",
+      constructor_declaration: "",
+      record_declaration: "record "
+    },
+    commentTypes: /* @__PURE__ */ new Set(["line_comment", "block_comment"])
   }
 };
 function getLanguageSpec(language) {
@@ -466,11 +511,11 @@ var SemanticSplitter = class {
   code;
   language;
   constructor(config = {}) {
-    const maxChunkSize = config.maxChunkSize ?? 1e3;
+    const maxChunkSize = config.maxChunkSize ?? 2500;
     this.config = {
       maxChunkSize,
-      minChunkSize: config.minChunkSize ?? 50,
-      chunkOverlap: config.chunkOverlap ?? 100,
+      minChunkSize: config.minChunkSize ?? 100,
+      chunkOverlap: config.chunkOverlap ?? 200,
       // 物理字符硬上限：默认为 maxChunkSize * 4（假设 1 token ≈ 4 chars）
       maxRawChars: config.maxRawChars ?? maxChunkSize * 4
     };
@@ -619,22 +664,10 @@ ${displayCode}`,
     let nextContext = context;
     const spec = getLanguageSpec(this.language);
     if (spec?.hierarchy.has(node.type)) {
-      let name = null;
-      for (const child of node.namedChildren) {
-        if (child.type === "identifier" || child.type === "type_identifier" || child.type === "name") {
-          name = child.text;
-          break;
-        }
-      }
-      if (!name && node.firstNamedChild) {
-        const firstChild = node.firstNamedChild;
-        if (firstChild.text.length <= 100 && !firstChild.text.includes("\n")) {
-          name = firstChild.text;
-        }
-      }
+      const name = this.extractNodeName(node, spec);
       if (name) {
-        const typePrefix = this.getTypePrefix(node.type);
-        nextContext = [...context, `${typePrefix}${name}`];
+        const prefix = spec.prefixMap[node.type] ?? "";
+        nextContext = [...context, `${prefix}${name}`];
       }
     }
     if (nodeSize <= this.config.maxChunkSize) {
@@ -651,21 +684,33 @@ ${displayCode}`,
     return this.mergeAdjacentWindows(childWindows);
   }
   /**
-   * 获取节点类型的简短前缀
+   * 从节点中提取名称（数据驱动）
    */
-  getTypePrefix(nodeType) {
-    if (nodeType.includes("class")) return "class ";
-    if (nodeType.includes("interface")) return "interface ";
-    if (nodeType.includes("method")) return "method ";
-    if (nodeType.includes("function")) return "function ";
-    return "";
+  extractNodeName(node, spec) {
+    for (const child of node.namedChildren) {
+      if (spec.nameNodeTypes.has(child.type)) {
+        return child.text;
+      }
+    }
+    if (node.firstNamedChild) {
+      const firstChild = node.firstNamedChild;
+      if (firstChild.text.length <= 100 && !firstChild.text.includes("\n")) {
+        return firstChild.text;
+      }
+    }
+    return null;
   }
   /**
    * Gap-Aware 相邻窗口合并
    *
-   * 使用 NWS + Raw 双预算策略：
+   * 使用三重预算策略：
    * - NWS 预算：控制有效代码量
    * - Raw 预算：控制物理字符数，防止大量注释撑爆 Token
+   * - 语义边界惩罚：不同 contextPath 的窗口合并门槛更高
+   *
+   * 前向吸附策略：
+   * - 如果当前窗口以 comment 结尾，将 comment 推到下一个窗口
+   * - 保证 JSDoc/注释与其描述的代码在同一个 chunk
    */
   mergeAdjacentWindows(windows) {
     if (windows.length === 0) return [];
@@ -673,6 +718,11 @@ ${displayCode}`,
     let current = windows[0];
     for (let i = 1; i < windows.length; i++) {
       const next = windows[i];
+      this.forwardAbsorbComments(current, next);
+      if (current.nodes.length === 0) {
+        current = next;
+        continue;
+      }
       const currentStart = current.nodes[0].startIndex;
       const currentEnd = current.nodes[current.nodes.length - 1].endIndex;
       const nextStart = next.nodes[0].startIndex;
@@ -680,13 +730,18 @@ ${displayCode}`,
       const gapNws = this.adapter.nws(currentEnd, nextStart);
       const combinedNws = current.size + gapNws + next.size;
       const combinedRawLen = nextEnd - currentStart;
+      const sameContext = this.isSameContext(current.contextPath, next.contextPath);
+      const boundaryPenalty = sameContext ? 1 : 0.7;
       const isTiny = current.size < this.config.minChunkSize;
-      const fitsNwsBudget = combinedNws <= this.config.maxChunkSize || isTiny && combinedNws < this.config.maxChunkSize * 1.5;
-      const fitsRawBudget = combinedRawLen <= this.config.maxRawChars;
+      const effectiveBudget = this.config.maxChunkSize * boundaryPenalty;
+      const fitsNwsBudget = combinedNws <= effectiveBudget || isTiny && combinedNws < effectiveBudget * 1.5;
+      const fitsRawBudget = combinedRawLen <= this.config.maxRawChars * boundaryPenalty;
       if (fitsNwsBudget && fitsRawBudget) {
         current.nodes.push(...next.nodes);
         current.size = combinedNws;
-        current.contextPath = this.commonPrefix(current.contextPath, next.contextPath);
+        if (next.contextPath.length > current.contextPath.length) {
+          current.contextPath = next.contextPath;
+        }
       } else {
         merged.push(current);
         current = next;
@@ -695,6 +750,60 @@ ${displayCode}`,
     merged.push(current);
     return merged;
   }
+  /**
+   * 前向吸附：将 current 尾部的 comment 节点推到 next 头部
+   *
+   * 这确保 JSDoc/docstring/注释与其描述的函数/方法在同一个 chunk 中，
+   * 而不是被切到前一个 chunk 的末尾。
+   *
+   * 注意：此方法会直接修改 current 和 next
+   */
+  forwardAbsorbComments(current, next) {
+    const spec = getLanguageSpec(this.language);
+    const commentTypes = spec?.commentTypes ?? /* @__PURE__ */ new Set(["comment"]);
+    const absorbedNodes = [];
+    let absorbedNws = 0;
+    while (current.nodes.length > 0) {
+      const lastNode = current.nodes[current.nodes.length - 1];
+      if (commentTypes.has(lastNode.type)) {
+        current.nodes.pop();
+        const nodeNws = this.adapter.nws(lastNode.startIndex, lastNode.endIndex);
+        absorbedNodes.unshift(lastNode);
+        absorbedNws += nodeNws;
+        current.size -= nodeNws;
+      } else {
+        break;
+      }
+    }
+    if (absorbedNodes.length > 0) {
+      const gapNws = next.nodes.length > 0 ? this.adapter.nws(
+        absorbedNodes[absorbedNodes.length - 1].endIndex,
+        next.nodes[0].startIndex
+      ) : 0;
+      next.nodes.unshift(...absorbedNodes);
+      next.size += absorbedNws + gapNws;
+    }
+  }
+  /**
+   * 检查两个 contextPath 是否属于同一语义单元
+   *
+   * 规则：如果两者的公共前缀长度 >= 较短路径长度，认为是同一单元
+   * 例如：
+   * - ["file", "class A", "method foo"] 和 ["file", "class A", "method bar"] -> false（不同方法）
+   * - ["file", "class A"] 和 ["file", "class A", "method foo"] -> true（父子关系）
+   */
+  isSameContext(a, b) {
+    const minLen = Math.min(a.length, b.length);
+    let commonLen = 0;
+    for (let i = 0; i < minLen; i++) {
+      if (a[i] === b[i]) {
+        commonLen++;
+      } else {
+        break;
+      }
+    }
+    return commonLen >= minLen;
+  }
   /**
    * 将窗口转换为最终的 ProcessedChunk
    *
@@ -771,25 +880,6 @@ ${displayCode}`,
     }
     return Math.max(0, result);
   }
-  /**
-   * 计算两个路径数组的最长公共前缀（LCA）
-   *
-   * 用于合并窗口时更新 contextPath，避免 breadcrumb 误标
-   * 例如：["file", "class A", "method foo"] 和 ["file", "class A", "method bar"]
-   *       => ["file", "class A"]
-   */
-  commonPrefix(a, b) {
-    const result = [];
-    const len = Math.min(a.length, b.length);
-    for (let i = 0; i < len; i++) {
-      if (a[i] === b[i]) {
-        result.push(a[i]);
-      } else {
-        break;
-      }
-    }
-    return result;
-  }
 };
 function generateVectorText(code, contextPath) {
   const breadcrumb = contextPath.join(" > ");
@@ -1052,7 +1142,6 @@ async function scan(rootPath, options = {}) {
       setStoredEmbeddingDimensions(db, currentDimensions);
     }
     if (forceReindex) {
-      logger.info("\u5F3A\u5236\u91CD\u65B0\u7D22\u5F15...");
       clear(db);
       if (options.vectorIndex !== false) {
         const embeddingConfig = getEmbeddingConfig();
@@ -1065,15 +1154,12 @@ async function scan(rootPath, options = {}) {
     const scannedPaths = new Set(
       filePaths.map((p) => path3.relative(rootPath, p).replace(/\\/g, "/"))
     );
-    let processedCount = 0;
     const results = [];
     const batchSize = 100;
     for (let i = 0; i < filePaths.length; i += batchSize) {
       const batch = filePaths.slice(i, i + batchSize);
       const batchResults = await processFiles(rootPath, batch, knownFiles);
       results.push(...batchResults);
-      processedCount += batch.length;
-      options.onProgress?.(processedCount, filePaths.length);
     }
     const toAdd = [];
     const toUpdateMtime = [];
@@ -1124,15 +1210,19 @@ async function scan(rootPath, options = {}) {
       errors: results.filter((r) => r.status === "error").length
     };
     if (options.vectorIndex !== false) {
+      options.onProgress?.(45, 100, "\u6B63\u5728\u51C6\u5907\u5411\u91CF\u7D22\u5F15...");
       const embeddingConfig = getEmbeddingConfig();
       const indexer = await getIndexer(projectId, embeddingConfig.dimensions);
       const needsVectorIndex = results.filter(
         (r) => r.status === "added" || r.status === "modified"
       );
       const healingPathSet = new Set(getFilesNeedingVectorIndex(db));
-      const healingFiles = results.filter((r) => r.status === "unchanged" && healingPathSet.has(r.relPath)).map((r) => ({ ...r, status: "modified" }));
-      if (healingFiles.length > 0) {
-        logger.info({ count: healingFiles.length }, "\u81EA\u6108\uFF1A\u53D1\u73B0\u9700\u8981\u8865\u7D22\u5F15\u7684\u6587\u4EF6");
+      const healingFilePaths = results.filter((r) => r.status === "unchanged" && healingPathSet.has(r.relPath)).map((r) => r.absPath);
+      let healingFiles = [];
+      if (healingFilePaths.length > 0) {
+        logger.info({ count: healingFilePaths.length }, "\u81EA\u6108\uFF1A\u53D1\u73B0\u9700\u8981\u8865\u7D22\u5F15\u7684\u6587\u4EF6");
+        const processedHealingFiles = await processFiles(rootPath, healingFilePaths, /* @__PURE__ */ new Map());
+        healingFiles = processedHealingFiles.filter((r) => r.status === "added" || r.status === "modified").map((r) => ({ ...r, status: "modified" }));
       }
       const deletedResults = deletedPaths.map((path4) => ({
         absPath: "",
@@ -1147,7 +1237,11 @@ async function scan(rootPath, options = {}) {
       }));
       const allToIndex = [...needsVectorIndex, ...healingFiles, ...deletedResults];
       if (allToIndex.length > 0) {
-        const indexStats = await indexer.indexFiles(db, allToIndex);
+        options.onProgress?.(45, 100, `\u6B63\u5728\u751F\u6210\u5411\u91CF\u5D4C\u5165... (${allToIndex.length} \u4E2A\u6587\u4EF6)`);
+        const indexStats = await indexer.indexFiles(db, allToIndex, (completed, total) => {
+          const progress = 45 + Math.floor(completed / total * 54);
+          options.onProgress?.(progress, 100, `\u6B63\u5728\u751F\u6210\u5411\u91CF\u5D4C\u5165... (${completed}/${total} \u6279\u6B21)`);
+        });
         stats.vectorIndex = {
           indexed: indexStats.indexed,
           deleted: indexStats.deleted,
@@ -1155,6 +1249,7 @@ async function scan(rootPath, options = {}) {
         };
       }
     }
+    options.onProgress?.(100, 100, "\u7D22\u5F15\u5B8C\u6210");
     return stats;
   } finally {
     closeDb(db);

package/dist/{chunk-C7XDGBT5.js → chunk-JVKVSTQ3.js}

@@ -75,9 +75,7 @@ function createFormattedStream(filePath) {
         const { level: _l, time: _t, pid: _p, hostname: _h, name: _n, msg: _m, ...extra } = log;
         let line = `${time} [${level}] ${msg}`;
         if (Object.keys(extra).length > 0) {
-          const extraLines = JSON.stringify(extra, null, 2).split("\n").map((l, i) => i === 0 ? l : `    ${l}`).join("\n");
-          line += `
-    ${extraLines}`;
+          line += ` ${JSON.stringify(extra)}`;
         }
         writeStream.write(`${line}
 `, callback);

package/dist/{chunk-VW5RACJC.js → chunk-OXPWRE3G.js}

@@ -1,6 +1,6 @@
 import {
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 
 // src/search/fts.ts
 var tokenizerCache = /* @__PURE__ */ new WeakMap();

package/dist/{chunk-WWYSLCNZ.js → chunk-TJHS7BN7.js}

@@ -4,10 +4,10 @@ import {
   batchUpsertChunkFts,
   clearVectorIndexHash,
   isChunksFtsInitialized
-} from "./chunk-VW5RACJC.js";
+} from "./chunk-OXPWRE3G.js";
 import {
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 import {
   getEmbeddingConfig
 } from "./chunk-SKBAE26T.js";
@@ -259,15 +259,21 @@ var ProgressTracker = class {
   startTime;
   lastLogTime = 0;
   logIntervalMs = 2e3;
-  // 每 3 秒输出一次
-  constructor(total) {
+  // 每 2 秒输出一次
+  onProgress;
+  /** 是否跳过日志（单批次时跳过，避免与索引日志混淆） */
+  skipLogs;
+  constructor(total, onProgress) {
     this.total = total;
     this.startTime = Date.now();
+    this.onProgress = onProgress;
+    this.skipLogs = total <= 1;
   }
   /** 记录一个批次完成 */
   recordBatch(tokens) {
     this.completed++;
     this.totalTokens += tokens;
+    this.onProgress?.(this.completed, this.total);
     const now = Date.now();
     if (now - this.lastLogTime >= this.logIntervalMs) {
       this.logProgress();
@@ -276,6 +282,7 @@ var ProgressTracker = class {
   }
   /** 输出进度 */
   logProgress() {
+    if (this.skipLogs) return;
     const elapsed = (Date.now() - this.startTime) / 1e3;
     const percent = Math.round(this.completed / this.total * 100);
     const rate = this.completed / elapsed;
@@ -293,6 +300,7 @@ var ProgressTracker = class {
   }
   /** 完成时输出最终统计 */
   complete() {
+    if (this.skipLogs) return;
     const elapsed = (Date.now() - this.startTime) / 1e3;
     logger.info(
       {
@@ -447,8 +455,9 @@ var EmbeddingClient = class {
    * 批量获取 Embedding
    * @param texts 待处理的文本数组
    * @param batchSize 每批次发送的文本数量（默认 20）
+   * @param onProgress 可选的进度回调 (completed, total) => void
    */
-  async embedBatch(texts, batchSize = 20) {
+  async embedBatch(texts, batchSize = 20, onProgress) {
     if (texts.length === 0) {
       return [];
     }
@@ -456,7 +465,7 @@ var EmbeddingClient = class {
     for (let i = 0; i < texts.length; i += batchSize) {
       batches.push(texts.slice(i, i + batchSize));
     }
-    const progress = new ProgressTracker(batches.length);
+    const progress = new ProgressTracker(batches.length, onProgress);
     const batchResults = await Promise.all(
       batches.map(
         (batch, batchIndex) => this.processWithRateLimit(batch, batchIndex * batchSize, progress)
@@ -625,8 +634,9 @@ var Indexer = class {
    *
    * @param db SQLite 数据库实例
    * @param results 文件处理结果
+   * @param onProgress 可选的进度回调 (indexed, total) => void
    */
-  async indexFiles(db, results) {
+  async indexFiles(db, results, onProgress) {
     if (!this.vectorStore) {
       await this.init();
     }
@@ -670,7 +680,7 @@ var Indexer = class {
       stats.deleted = toDelete.length;
     }
     if (toIndex.length > 0) {
-      const indexResult = await this.batchIndex(db, toIndex);
+      const indexResult = await this.batchIndex(db, toIndex, onProgress);
       stats.indexed = indexResult.success;
       stats.errors = indexResult.errors;
     }
@@ -694,7 +704,7 @@ var Indexer = class {
    * 3. FTS 写入批量化：N 次删除+插入 → 1 次批量删除 + 1 次批量插入
    * 4. 日志汇总化：逐文件日志 → 汇总日志
    */
-  async batchIndex(db, files) {
+  async batchIndex(db, files, onProgress) {
     if (files.length === 0) {
       return { success: 0, errors: 0 };
     }
@@ -715,7 +725,7 @@ var Indexer = class {
     logger.info({ count: allTexts.length, files: files.length }, "\u5F00\u59CB\u6279\u91CF Embedding");
     let embeddings;
     try {
-      const results = await this.embeddingClient.embedBatch(allTexts);
+      const results = await this.embeddingClient.embedBatch(allTexts, 20, onProgress);
       embeddings = results.map((r) => r.embedding);
     } catch (err) {
       const error = err;

package/dist/{codebaseRetrieval-AV4GK6FT.js → codebaseRetrieval-VZ2E5JYE.js}

@@ -1,9 +1,9 @@
 import {
   codebaseRetrievalSchema,
   handleCodebaseRetrieval
-} from "./chunk-O3HDM3CF.js";
-import "./chunk-VW5RACJC.js";
-import "./chunk-C7XDGBT5.js";
+} from "./chunk-2SIQBQ2M.js";
+import "./chunk-OXPWRE3G.js";
+import "./chunk-JVKVSTQ3.js";
 import "./chunk-SKBAE26T.js";
 export {
   codebaseRetrievalSchema,

package/dist/index.js

@@ -1,22 +1,30 @@
 #!/usr/bin/env node
 import {
   scan
-} from "./chunk-QWQ64TBE.js";
-import "./chunk-WWYSLCNZ.js";
+} from "./chunk-AA3ILFHL.js";
+import "./chunk-TJHS7BN7.js";
 import {
   generateProjectId
-} from "./chunk-VW5RACJC.js";
+} from "./chunk-OXPWRE3G.js";
 import {
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 import "./chunk-SKBAE26T.js";
 
 // src/index.ts
 import { promises as fs } from "fs";
 import os from "os";
 import path from "path";
+import { fileURLToPath } from "url";
 import cac from "cac";
+var __dirname = path.dirname(fileURLToPath(import.meta.url));
+var pkgPath = path.resolve(__dirname, "../package.json");
+var pkg = JSON.parse(await fs.readFile(pkgPath, "utf-8"));
 var cli = cac("contextweaver");
+if (process.argv.includes("-v") || process.argv.includes("--version")) {
+  console.log(pkg.version);
+  process.exit(0);
+}
 cli.command("init", "\u521D\u59CB\u5316 ContextWeaver \u914D\u7F6E").action(async () => {
   const configDir = path.join(os.homedir(), ".contextweaver");
   const envFile = path.join(configDir, ".env");
@@ -80,32 +88,33 @@ cli.command("index [path]", "\u626B\u63CF\u4EE3\u7801\u5E93\u5E76\u5EFA\u7ACB\u7
   }
   const startTime = Date.now();
   try {
+    let lastLoggedPercent = 0;
     const stats = await scan(rootPath, {
       force: options.force,
-      onProgress: (current, total) => {
-        const percent = (current / total * 100).toFixed(1);
-        logger.info({ current, total, percent: `${percent}%` }, "\u626B\u63CF\u8FDB\u5EA6");
+      onProgress: (current, total, message) => {
+        if (total !== void 0) {
+          const percent = Math.floor(current / total * 100);
+          if (percent >= lastLoggedPercent + 30 && percent < 100) {
+            logger.info(`\u7D22\u5F15\u8FDB\u5EA6: ${percent}% - ${message || ""}`);
+            lastLoggedPercent = Math.floor(percent / 30) * 30;
+          }
+        }
       }
     });
     process.stdout.write("\n");
     const duration = ((Date.now() - startTime) / 1e3).toFixed(2);
-    logger.info("\u626B\u63CF\u5B8C\u6210\uFF01");
-    logger.info(`\u8017\u65F6: ${duration}s`);
-    logger.info(`\u6587\u4EF6\u603B\u6570: ${stats.totalFiles}`);
-    logger.info(`\u65B0\u589E: ${stats.added}`);
-    logger.info(`\u4FEE\u6539: ${stats.modified}`);
-    logger.info(`\u672A\u53D8: ${stats.unchanged}`);
-    logger.info(`\u5220\u9664: ${stats.deleted}`);
-    logger.info(`\u8DF3\u8FC7: ${stats.skipped}`);
-    logger.info(`\u9519\u8BEF: ${stats.errors}`);
+    logger.info(`\u7D22\u5F15\u5B8C\u6210 (${duration}s)`);
+    logger.info(
+      `\u603B\u6570:${stats.totalFiles} \u65B0\u589E:${stats.added} \u4FEE\u6539:${stats.modified} \u672A\u53D8:${stats.unchanged} \u5220\u9664:${stats.deleted} \u8DF3\u8FC7:${stats.skipped} \u9519\u8BEF:${stats.errors}`
+    );
   } catch (err) {
     const error = err;
-    logger.error({ err, stack: error.stack }, `\u626B\u63CF\u5931\u8D25: ${error.message}`);
+    logger.error({ err, stack: error.stack }, `\u7D22\u5F15\u5931\u8D25: ${error.message}`);
     process.exit(1);
   }
 });
 cli.command("mcp", "\u542F\u52A8 MCP \u670D\u52A1\u5668").action(async () => {
-  const { startMcpServer } = await import("./server-7PYHHTOM.js");
+  const { startMcpServer } = await import("./server-SOOY6RCA.js");
   try {
     await startMcpServer();
   } catch (err) {
@@ -127,7 +136,7 @@ cli.command("search", "\u672C\u5730\u68C0\u7D22\uFF08\u53C2\u6570\u5BF9\u9F50 MC
     }
     const technicalTerms = (options.technicalTerms || "").split(",").map((t) => t.trim()).filter(Boolean);
     const useZen = options.zen !== false;
-    const { handleCodebaseRetrieval } = await import("./codebaseRetrieval-AV4GK6FT.js");
+    const { handleCodebaseRetrieval } = await import("./codebaseRetrieval-VZ2E5JYE.js");
     const response = await handleCodebaseRetrieval(
       {
         repo_path: repoPath,

package/dist/{lock-PX2BX2YN.js → lock-RC33CJZA.js}

@@ -1,6 +1,6 @@
 import {
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 import "./chunk-SKBAE26T.js";
 
 // src/utils/lock.ts

package/dist/scanner-NM7WPJJE.js

@@ -0,0 +1,10 @@
+import {
+  scan
+} from "./chunk-AA3ILFHL.js";
+import "./chunk-TJHS7BN7.js";
+import "./chunk-OXPWRE3G.js";
+import "./chunk-JVKVSTQ3.js";
+import "./chunk-SKBAE26T.js";
+export {
+  scan
+};

package/dist/{server-7PYHHTOM.js → server-SOOY6RCA.js}

@@ -1,11 +1,11 @@
 import {
   codebaseRetrievalSchema,
   handleCodebaseRetrieval
-} from "./chunk-O3HDM3CF.js";
-import "./chunk-VW5RACJC.js";
+} from "./chunk-2SIQBQ2M.js";
+import "./chunk-OXPWRE3G.js";
 import {
   logger
-} from "./chunk-C7XDGBT5.js";
+} from "./chunk-JVKVSTQ3.js";
 import "./chunk-SKBAE26T.js";
 
 // src/mcp/server.ts
@@ -94,14 +94,31 @@ async function startMcpServer() {
     logger.debug("\u6536\u5230 list_tools \u8BF7\u6C42");
     return { tools: TOOLS };
   });
-  server.setRequestHandler(CallToolRequestSchema, async (request) => {
+  server.setRequestHandler(CallToolRequestSchema, async (request, extra) => {
     const { name, arguments: args } = request.params;
     logger.info({ tool: name }, "\u6536\u5230 call_tool \u8BF7\u6C42");
+    const rawToken = extra._meta?.progressToken;
+    const progressToken = typeof rawToken === "string" || typeof rawToken === "number" ? rawToken : void 0;
+    const onProgress = progressToken ? async (current, total, message) => {
+      try {
+        await extra.sendNotification({
+          method: "notifications/progress",
+          params: {
+            progressToken,
+            progress: current,
+            total,
+            message
+          }
+        });
+      } catch (err) {
+        logger.debug({ error: err.message }, "\u53D1\u9001\u8FDB\u5EA6\u901A\u77E5\u5931\u8D25");
+      }
+    } : void 0;
     try {
       switch (name) {
         case "codebase-retrieval": {
           const parsed = codebaseRetrievalSchema.parse(args);
-          return await handleCodebaseRetrieval(parsed);
+          return await handleCodebaseRetrieval(parsed, void 0, onProgress);
         }
         default:
           throw new Error(`Unknown tool: ${name}`);
@@ -121,8 +138,8 @@ async function startMcpServer() {
     }
   });
   const transport = new StdioServerTransport();
-  await server.connect(transport);
   logger.info("MCP \u670D\u52A1\u5668\u5DF2\u542F\u52A8\uFF0C\u7B49\u5F85\u8FDE\u63A5...");
+  await server.connect(transport);
 }
 export {
   startMcpServer

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@hsingjui/contextweaver",
-  "version": "0.0.4",
+  "version": "0.0.6",
   "description": "A context weaving tool for LLMs",
   "license": "MIT",
   "author": "hsingjui",
@@ -29,9 +29,10 @@
     "fmt": "biome check --write ./src"
   },
   "dependencies": {
-    "@lancedb/lancedb": "^0.19.1",
+    "@keqingmoe/tree-sitter": "^0.26.2",
+    "@lancedb/lancedb": "^0.22.0",
     "@modelcontextprotocol/sdk": "^1.25.1",
-    "better-sqlite3": "^11.10.0",
+    "better-sqlite3": "^12.2.0",
     "cac": "^6.7.14",
     "chardet": "^2.1.1",
     "dotenv": "^17.2.3",
@@ -40,28 +41,27 @@
     "ignore": "^7.0.5",
     "p-limit": "^7.2.0",
     "pino": "^10.1.0",
-    "tree-sitter": "0.20.6",
-    "tree-sitter-go": "0.20.0",
-    "tree-sitter-java": "0.20.2",
-    "tree-sitter-javascript": "0.20.4",
-    "tree-sitter-python": "0.20.4",
-    "tree-sitter-rust": "0.20.4",
-    "tree-sitter-typescript": "0.20.5",
+    "tree-sitter-go": "0.23.4",
+    "tree-sitter-java": "0.23.5",
+    "tree-sitter-javascript": "0.23.1",
+    "tree-sitter-python": "0.25.0",
+    "tree-sitter-rust": "0.21.0",
+    "tree-sitter-typescript": "0.23.2",
     "zod": "^4.2.1"
   },
   "devDependencies": {
     "@biomejs/biome": "2.3.10",
     "@types/better-sqlite3": "^7.6.13",
-    "@types/node": "^25.0.3",
+    "@types/node": "^24.0.0",
     "knip": "^5.78.0",
     "tsup": "^8.5.1",
     "typescript": "^5.9.3"
   },
   "pnpm": {
     "onlyBuiltDependencies": [
+      "@keqingmoe/tree-sitter",
       "better-sqlite3",
       "esbuild",
-      "tree-sitter",
       "tree-sitter-go",
       "tree-sitter-java",
       "tree-sitter-javascript",

package/dist/scanner-HYP3L57R.js

@@ -1,10 +0,0 @@
-import {
-  scan
-} from "./chunk-QWQ64TBE.js";
-import "./chunk-WWYSLCNZ.js";
-import "./chunk-VW5RACJC.js";
-import "./chunk-C7XDGBT5.js";
-import "./chunk-SKBAE26T.js";
-export {
-  scan
-};