@llangtop/pwiki-core 1.1.0 → 1.2.0

package/dist/lib/store.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"store.d.ts","sourceRoot":"","sources":["../../src/lib/store.ts"],"names":[],"mappings":"AAGA,OAAO,EACL,UAAU,EAAE,SAAS,EAAE,YAAY,EACnC,kBAAkB,EAAE,kBAAkB,EACtC,WAAW,EAAE,YAAY,EACzB,YAAY,EAAE,WAAW,GAC1B,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,QAAQ,EAAE,UAAU,EAAE,WAAW,EAAE,eAAe,EAAE,QAAQ,EAC5D,qBAAqB,EAAE,UAAU,GAClC,MAAM,kBAAkB,CAAC;AAE1B,OAAO,EACL,aAAa,EAAE,aAAa,EAC5B,YAAY,EAAE,YAAY,EAC1B,eAAe,EAAE,iBAAiB,EAAE,eAAe,EACnD,YAAY,GACb,MAAM,oBAAoB,CAAC;AAO5B,wBAAgB,KAAK;;;;;;EAWpB"}
+{"version":3,"file":"store.d.ts","sourceRoot":"","sources":["../../src/lib/store.ts"],"names":[],"mappings":"AAGA,OAAO,EACL,UAAU,EAAE,SAAS,EAAE,YAAY,EACnC,kBAAkB,EAAE,kBAAkB,EACtC,WAAW,EAAE,YAAY,EACzB,YAAY,EAAE,WAAW,GAC1B,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,QAAQ,EAAE,UAAU,EAAE,WAAW,EAAE,eAAe,EAAE,QAAQ,EAC5D,qBAAqB,EAAE,UAAU,EACjC,aAAa,EAAE,cAAc,GAC9B,MAAM,kBAAkB,CAAC;AAE1B,OAAO,EACL,aAAa,EAAE,aAAa,EAC5B,YAAY,EAAE,YAAY,EAC1B,eAAe,EAAE,iBAAiB,EAAE,eAAe,EACnD,YAAY,GACb,MAAM,oBAAoB,CAAC;AAO5B,wBAAgB,KAAK;;;;;;EAWpB"}

package/dist/lib/store.js

@@ -1,7 +1,7 @@
 // store.ts �?数据�?barrel
 // 聚合 store-config + store-index + store-vectors
 export { getSources, addSource, removeSource, getSemanticEnabled, setSemanticEnabled, readModelId, writeModelId, getWikiModel, configStats, } from "./store-config.js";
-export { getIndex, mergeIndex, removeEntry, updateEntryPath, getEntry, removeEntriesBySource, indexStats, } from "./store-index.js";
+export { getIndex, mergeIndex, removeEntry, updateEntryPath, getEntry, removeEntriesBySource, indexStats, readBm25Stats, writeBm25Stats, } from "./store-index.js";
 export { getEmbeddings, setEmbeddings, getChunkInfo, setChunkInfo, removeEmbedding, getEmbeddingModel, getEmbeddingDim, vectorsStats, } from "./store-vectors.js";
 import { configStats } from "./store-config.js";
 import { indexStats } from "./store-index.js";

package/dist/lib/store.js.map

@@ -1 +1 @@
-{"version":3,"file":"store.js","sourceRoot":"","sources":["../../src/lib/store.ts"],"names":[],"mappings":"AAAA,wBAAwB;AACxB,gDAAgD;AAEhD,OAAO,EACL,UAAU,EAAE,SAAS,EAAE,YAAY,EACnC,kBAAkB,EAAE,kBAAkB,EACtC,WAAW,EAAE,YAAY,EACzB,YAAY,EAAE,WAAW,GAC1B,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,QAAQ,EAAE,UAAU,EAAE,WAAW,EAAE,eAAe,EAAE,QAAQ,EAC5D,qBAAqB,EAAE,UAAU,GAClC,MAAM,kBAAkB,CAAC;AAE1B,OAAO,EACL,aAAa,EAAE,aAAa,EAC5B,YAAY,EAAE,YAAY,EAC1B,eAAe,EAAE,iBAAiB,EAAE,eAAe,EACnD,YAAY,GACb,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAClD,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AAEvD,MAAM,UAAU,KAAK;IACnB,MAAM,CAAC,GAAG,WAAW,EAAE,CAAC;IACxB,MAAM,CAAC,GAAG,UAAU,EAAE,CAAC;IACvB,MAAM,CAAC,GAAG,YAAY,EAAE,CAAC;IACzB,OAAO;QACL,OAAO,EAAE,CAAC,CAAC,OAAO;QAClB,KAAK,EAAE,CAAC,CAAC,KAAK;QACd,QAAQ,EAAE,CAAC,CAAC,QAAQ;QACpB,eAAe,EAAE,kBAAkB,EAAE;QACrC,UAAU,EAAE,CAAC,CAAC,UAAU;KACzB,CAAC;AACJ,CAAC"}
+{"version":3,"file":"store.js","sourceRoot":"","sources":["../../src/lib/store.ts"],"names":[],"mappings":"AAAA,wBAAwB;AACxB,gDAAgD;AAEhD,OAAO,EACL,UAAU,EAAE,SAAS,EAAE,YAAY,EACnC,kBAAkB,EAAE,kBAAkB,EACtC,WAAW,EAAE,YAAY,EACzB,YAAY,EAAE,WAAW,GAC1B,MAAM,mBAAmB,CAAC;AAE3B,OAAO,EACL,QAAQ,EAAE,UAAU,EAAE,WAAW,EAAE,eAAe,EAAE,QAAQ,EAC5D,qBAAqB,EAAE,UAAU,EACjC,aAAa,EAAE,cAAc,GAC9B,MAAM,kBAAkB,CAAC;AAE1B,OAAO,EACL,aAAa,EAAE,aAAa,EAC5B,YAAY,EAAE,YAAY,EAC1B,eAAe,EAAE,iBAAiB,EAAE,eAAe,EACnD,YAAY,GACb,MAAM,oBAAoB,CAAC;AAE5B,OAAO,EAAE,WAAW,EAAE,MAAM,mBAAmB,CAAC;AAChD,OAAO,EAAE,UAAU,EAAE,MAAM,kBAAkB,CAAC;AAC9C,OAAO,EAAE,YAAY,EAAE,MAAM,oBAAoB,CAAC;AAClD,OAAO,EAAE,kBAAkB,EAAE,MAAM,mBAAmB,CAAC;AAEvD,MAAM,UAAU,KAAK;IACnB,MAAM,CAAC,GAAG,WAAW,EAAE,CAAC;IACxB,MAAM,CAAC,GAAG,UAAU,EAAE,CAAC;IACvB,MAAM,CAAC,GAAG,YAAY,EAAE,CAAC;IACzB,OAAO;QACL,OAAO,EAAE,CAAC,CAAC,OAAO;QAClB,KAAK,EAAE,CAAC,CAAC,KAAK;QACd,QAAQ,EAAE,CAAC,CAAC,QAAQ;QACpB,eAAe,EAAE,kBAAkB,EAAE;QACrC,UAAU,EAAE,CAAC,CAAC,UAAU;KACzB,CAAC;AACJ,CAAC"}

package/dist/lib/tokenizer.d.ts

@@ -0,0 +1,10 @@
+/**
+ * 对一段文本进行分词
+ * @returns 小写化的 token 列表，不包含单字符 ASCII
+ *
+ * @example
+ *   tokenize("变压器故障排查，check transformer oil 温度")
+ *   // → ["变压","压器","器故","故障","障排","排查","check","transformer","oil","温度"]
+ */
+export declare function tokenize(text: string): string[];
+//# sourceMappingURL=tokenizer.d.ts.map

package/dist/lib/tokenizer.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"tokenizer.d.ts","sourceRoot":"","sources":["../../src/lib/tokenizer.ts"],"names":[],"mappings":"AA6BA;;;;;;;GAOG;AACH,wBAAgB,QAAQ,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAqD/C"}

package/dist/lib/tokenizer.js

@@ -0,0 +1,87 @@
+// tokenizer.ts — 中文 2-gram + ASCII 切词分词器
+//
+// 零外部依赖。状态机按字符 Unicode 属性分类：
+//   CJK 字符 → 收集连续序列 → 输出二元组 (2-gram)
+//   ASCII 字母/数字 → 按非字母数字切分，保留 ≥2 字符的词
+//   其他（标点、空白）→ 丢弃，作分隔符
+//
+// 查询和文档使用同一个 tokenize()，确保 BM25 token 对齐。
+/**
+ * 检查字符是否属于 CJK 统一表意文字范围
+ */
+function isCJK(cp) {
+    return (cp >= 0x4E00 && cp <= 0x9FFF) // CJK Unified Ideographs
+        || (cp >= 0x3400 && cp <= 0x4DBF) // CJK Extension A
+        || (cp >= 0x20000 && cp <= 0x2A6DF) // CJK Extension B
+        || (cp >= 0xF900 && cp <= 0xFAFF) // CJK Compatibility Ideographs
+        || (cp >= 0x2F800 && cp <= 0x2FA1F); // CJK Compatibility Supplement
+}
+/**
+ * 检查字符是否为 ASCII 字母或数字（参与分词）
+ */
+function isASCII(cp) {
+    return (cp >= 0x61 && cp <= 0x7A) // a-z
+        || (cp >= 0x41 && cp <= 0x5A) // A-Z
+        || (cp >= 0x30 && cp <= 0x39); // 0-9
+}
+/**
+ * 对一段文本进行分词
+ * @returns 小写化的 token 列表，不包含单字符 ASCII
+ *
+ * @example
+ *   tokenize("变压器故障排查，check transformer oil 温度")
+ *   // → ["变压","压器","器故","故障","障排","排查","check","transformer","oil","温度"]
+ */
+export function tokenize(text) {
+    const tokens = [];
+    const cjkBuf = []; // 连续 CJK 字符缓冲区
+    const asciiBuf = []; // 连续 ASCII 字符缓冲区
+    /**
+     * 将 CJK 缓冲区输出为 2-gram tokens
+     */
+    function flushCJK() {
+        if (cjkBuf.length < 2) {
+            // 单字符 CJK：保留
+            if (cjkBuf.length === 1)
+                tokens.push(cjkBuf[0]);
+        }
+        else {
+            // 滑动窗口生成所有相邻二元组
+            for (let i = 0; i < cjkBuf.length - 1; i++) {
+                tokens.push(cjkBuf[i] + cjkBuf[i + 1]);
+            }
+        }
+        cjkBuf.length = 0;
+    }
+    /**
+     * 将 ASCII 缓冲区输出为单词（≥2 字符）
+     */
+    function flushASCII() {
+        if (asciiBuf.length >= 2) {
+            tokens.push(asciiBuf.join("").toLowerCase());
+        }
+        asciiBuf.length = 0;
+    }
+    for (let i = 0; i < text.length; i++) {
+        const ch = text[i];
+        const cp = ch.codePointAt(0);
+        if (isCJK(cp)) {
+            flushASCII(); // 脚本切换：先输出之前的 ASCII 词
+            cjkBuf.push(ch);
+        }
+        else if (isASCII(cp)) {
+            flushCJK(); // 脚本切换：先输出之前的 CJK 词
+            asciiBuf.push(ch);
+        }
+        else {
+            // 标点/空白/Emoji → 分隔符，同时清空两个缓冲区
+            flushCJK();
+            flushASCII();
+        }
+    }
+    // 文本结束，输出残留
+    flushCJK();
+    flushASCII();
+    return tokens;
+}
+//# sourceMappingURL=tokenizer.js.map

package/dist/lib/tokenizer.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"tokenizer.js","sourceRoot":"","sources":["../../src/lib/tokenizer.ts"],"names":[],"mappings":"AAAA,yCAAyC;AACzC,EAAE;AACF,6BAA6B;AAC7B,qCAAqC;AACrC,sCAAsC;AACtC,uBAAuB;AACvB,EAAE;AACF,0CAA0C;AAE1C;;GAEG;AACH,SAAS,KAAK,CAAC,EAAU;IACvB,OAAO,CAAC,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM,CAAC,CAAG,yBAAyB;WAC1D,CAAC,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM,CAAC,CAAG,kBAAkB;WACnD,CAAC,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO,CAAC,CAAC,kBAAkB;WACnD,CAAC,EAAE,IAAI,MAAM,IAAI,EAAE,IAAI,MAAM,CAAC,CAAG,+BAA+B;WAChE,CAAC,EAAE,IAAI,OAAO,IAAI,EAAE,IAAI,OAAO,CAAC,CAAC,CAAA,+BAA+B;AACzE,CAAC;AAED;;GAEG;AACH,SAAS,OAAO,CAAC,EAAU;IACzB,OAAO,CAAC,EAAE,IAAI,IAAI,IAAI,EAAE,IAAI,IAAI,CAAC,CAAE,MAAM;WAClC,CAAC,EAAE,IAAI,IAAI,IAAI,EAAE,IAAI,IAAI,CAAC,CAAE,MAAM;WAClC,CAAC,EAAE,IAAI,IAAI,IAAI,EAAE,IAAI,IAAI,CAAC,CAAC,CAAC,MAAM;AAC3C,CAAC;AAED;;;;;;;GAOG;AACH,MAAM,UAAU,QAAQ,CAAC,IAAY;IACnC,MAAM,MAAM,GAAa,EAAE,CAAC;IAC5B,MAAM,MAAM,GAAa,EAAE,CAAC,CAAG,eAAe;IAC9C,MAAM,QAAQ,GAAa,EAAE,CAAC,CAAC,iBAAiB;IAEhD;;OAEG;IACH,SAAS,QAAQ;QACf,IAAI,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;YACtB,aAAa;YACb,IAAI,MAAM,CAAC,MAAM,KAAK,CAAC;gBAAE,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC;QAClD,CAAC;aAAM,CAAC;YACN,gBAAgB;YAChB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,MAAM,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;gBAC3C,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,MAAM,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;YACzC,CAAC;QACH,CAAC;QACD,MAAM,CAAC,MAAM,GAAG,CAAC,CAAC;IACpB,CAAC;IAED;;OAEG;IACH,SAAS,UAAU;QACjB,IAAI,QAAQ,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;YACzB,MAAM,CAAC,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,WAAW,EAAE,CAAC,CAAC;QAC/C,CAAC;QACD,QAAQ,CAAC,MAAM,GAAG,CAAC,CAAC;IACtB,CAAC;IAED,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QACrC,MAAM,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC;QACnB,MAAM,EAAE,GAAG,EAAE,CAAC,WAAW,CAAC,CAAC,CAAE,CAAC;QAE9B,IAAI,KAAK,CAAC,EAAE,CAAC,EAAE,CAAC;YACd,UAAU,EAAE,CAAC,CAAS,sBAAsB;YAC5C,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QAClB,CAAC;aAAM,IAAI,OAAO,CAAC,EAAE,CAAC,EAAE,CAAC;YACvB,QAAQ,EAAE,CAAC,CAAW,oBAAoB;YAC1C,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;QACpB,CAAC;aAAM,CAAC;YACN,8BAA8B;YAC9B,QAAQ,EAAE,CAAC;YACX,UAAU,EAAE,CAAC;QACf,CAAC;IACH,CAAC;IAED,YAAY;IACZ,QAAQ,EAAE,CAAC;IACX,UAAU,EAAE,CAAC;IAEb,OAAO,MAAM,CAAC;AAChB,CAAC"}

package/package.json

@@ -1,34 +1,34 @@
-{
-  "name": "@llangtop/pwiki-core",
-  "version": "1.1.0",
-  "description": "Wiki knowledge base engine — keyword/semantic/hybrid search with local ONNX embeddings",
-  "type": "module",
-  "main": "./dist/index.js",
-  "types": "./dist/index.d.ts",
-  "exports": {
-    ".": {
-      "types": "./dist/index.d.ts",
-      "import": "./dist/index.js"
-    }
-  },
-  "files": ["dist", "README.md"],
-  "scripts": {
-    "build": "tsc",
-    "start": "node dist/index.js"
-  },
-  "dependencies": {
-    "@huggingface/transformers": "^3.0.0",
-    "remark-parse": "^11.0.0",
-    "unified": "^11.0.0",
-    "unist-util-visit": "^5.0.0"
-  },
-  "devDependencies": {
-    "@types/mdast": "^4.0.0",
-    "@types/node": "^22.0.0",
-    "typescript": "^5.5.0"
-  },
-  "engines": {
-    "node": ">=18"
-  },
-  "license": "MIT"
-}
+{
+  "name": "@llangtop/pwiki-core",
+  "version": "1.2.0",
+  "description": "Wiki knowledge base engine — keyword/semantic/hybrid search with local ONNX embeddings",
+  "type": "module",
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js"
+    }
+  },
+  "files": ["dist", "README.md"],
+  "scripts": {
+    "build": "tsc",
+    "start": "node dist/index.js"
+  },
+  "dependencies": {
+    "@huggingface/transformers": "^3.0.0",
+    "remark-parse": "^11.0.0",
+    "unified": "^11.0.0",
+    "unist-util-visit": "^5.0.0"
+  },
+  "devDependencies": {
+    "@types/mdast": "^4.0.0",
+    "@types/node": "^22.0.0",
+    "typescript": "^5.5.0"
+  },
+  "engines": {
+    "node": ">=18"
+  },
+  "license": "MIT"
+}

package/dist/WikiEngine.d.ts

@@ -1,73 +0,0 @@
-import type { ModelInfo } from "./lib/model-registry.js";
-import type { SearchMode, SearchHit, FileEntry, FileLLMData, WikiStatus } from "./lib/types.js";
-export interface EngineConfig {
-    basePath?: string;
-    modelId?: string;
-}
-export declare class WikiEngine {
-    constructor(config?: EngineConfig);
-    get sources(): string[];
-    addSource(absPath: string): boolean;
-    removeSource(target: string): string | null;
-    loadSource(absPath: string): Promise<number>;
-    load(): Promise<{
-        files: number;
-        sources: number;
-    }>;
-    search(query: string, mode?: SearchMode): Promise<SearchHit[]>;
-    readEntry(pathOrRelPath: string): {
-        entry: FileEntry;
-        content: string;
-    } | null;
-    createEntry(sourceDir: string, relPath: string, title?: string, tags?: string[], content?: string): string;
-    renameEntry(relPath: string, newTitle: string): boolean;
-    moveEntry(relPath: string, newRelPath: string): boolean;
-    modifyEntry(sourceDir: string, relPath: string, content: string): boolean;
-    get semanticEnabled(): boolean;
-    enableSemantic(modelId?: string): Promise<{
-        ok: boolean;
-        msg: string;
-    }>;
-    disableSemantic(): void;
-    generateEmbeddings(sourceDir?: string): Promise<{
-        embedded: number;
-    }>;
-    downloadModel(modelId?: string): Promise<{
-        ok: boolean;
-        msg: string;
-    }>;
-    listModels(): ModelInfo[];
-    get llmInfo(): {
-        apiBase: string;
-        model: string;
-        hasKey: boolean;
-    };
-    compileStatus(sourceDir?: string): {
-        total: number;
-        compiled: number;
-        uncompiled: string[];
-    };
-    getCompilePrompt(relPath: string): {
-        system: string;
-        user: string;
-    } | null;
-    storeCompiled(relPath: string, data: FileLLMData): boolean;
-    compileFile(relPath: string, opts?: {
-        model?: string;
-        force?: boolean;
-    }): Promise<{
-        ok: boolean;
-        msg: string;
-    }>;
-    compileAll(sourceDir?: string, limit?: number, opts?: {
-        model?: string;
-        force?: boolean;
-    }): Promise<{
-        compiled: number;
-        skipped: number;
-        failed: number;
-        msgs: string[];
-    }>;
-    status(): WikiStatus;
-}
-//# sourceMappingURL=WikiEngine.d.ts.map

package/dist/lib/ast-chunker.d.ts

@@ -1,23 +0,0 @@
-/** AST 分块结果（兼容旧 extractChunks 接口 + 新字段） */
-export interface ChunkResult {
-    key: string;
-    heading: string;
-    level: number;
-    embedText: string;
-    rawText: string;
-    headingPath: string[];
-    chunkTypeHint: string;
-    wikilinks: string[];
-    startLine: number;
-    endLine: number;
-}
-/**
- * 使用 unified + remark-parse 解析 markdown，按标题切分为语义块。
- *
- * @param raw          原始 markdown 文本
- * @param relPath      文件相对路径（用于 key 生成）
- * @param defaultTitle 无标题时的默认标题
- * @returns 分块结果数组，AST 解析失败时返回空数组（调用方应降级 regex）
- */
-export declare function extractChunksAST(raw: string, relPath: string, defaultTitle: string, maxEmbedLen?: number): Promise<ChunkResult[]>;
-//# sourceMappingURL=ast-chunker.d.ts.map

package/dist/lib/content-cache.d.ts

@@ -1,13 +0,0 @@
-/** 存入缓存 */
-export declare function setContent(relPath: string, content: string): void;
-/** 读取缓存 */
-export declare function getContent(relPath: string): string | undefined;
-/** 检查是否已缓存 */
-export declare function hasContent(relPath: string): boolean;
-/** 按 sourceDir 清除缓存条目 */
-export declare function clearSource(sourceDir: string): void;
-/** 清除所有缓存 */
-export declare function clearAll(): void;
-/** 缓存大小 */
-export declare function cacheSize(): number;
-//# sourceMappingURL=content-cache.d.ts.map

package/dist/lib/embedder.d.ts

@@ -1,22 +0,0 @@
-export interface LocalModelInfo {
-    path: string;
-    variant: "fp32" | "int8" | "none";
-    onnxSize: number;
-    otherSize: number;
-}
-export declare function getLocalModelInfo(): LocalModelInfo | null;
-export declare function initialize(): Promise<boolean>;
-export declare function isAvailable(): boolean;
-export declare function downloadModel(modelId?: string): {
-    ok: boolean;
-    msg: string;
-};
-export declare function getInitError(): string | null;
-export declare function getModelName(): string;
-export declare function getModelRepo(): string;
-export declare function getModelSource(): string;
-export declare function getLoadedVariant(): string;
-export declare function embed(text: string): Promise<number[]>;
-export declare function embedBatch(texts: string[]): Promise<number[][]>;
-export declare function cosineSimilarity(a: number[], b: number[]): number;
-//# sourceMappingURL=embedder.d.ts.map

package/dist/lib/file-manifest.d.ts

@@ -1,36 +0,0 @@
-export interface FileManifestEntry {
-    md5: string;
-    fileSize: number;
-    astChunkCount: number;
-    astIndexedAt: string;
-    llmCompiled: boolean;
-    llmCompiledAt?: string;
-    compilingSince?: string;
-    hasSemanticVectors: boolean;
-    contentClass?: string;
-    deleted?: boolean;
-}
-export interface FileManifest {
-    version: 1;
-    files: Record<string, FileManifestEntry>;
-}
-export declare function getManifestStats(): {
-    total: number;
-    compiled: number;
-    withVectors: number;
-};
-export declare function getManifest(): FileManifest;
-export declare function getFileState(relPath: string): FileManifestEntry | undefined;
-export declare function updateFileState(relPath: string, patch: Partial<FileManifestEntry>): void;
-export declare function removeFileState(relPath: string): void;
-export declare function computeMD5(content: string): string;
-export declare function isFileChanged(relPath: string, currentMD5: string): boolean;
-export declare function isCompilationStale(relPath: string, currentMD5: string): boolean;
-export declare function ensureCompiledDir(): void;
-export declare function getCompiledFilePath(relPath: string): string;
-export declare function manifestStats(): {
-    total: number;
-    compiled: number;
-    stale: number;
-};
-//# sourceMappingURL=file-manifest.d.ts.map

package/dist/lib/indexer-compile.d.ts

@@ -1,18 +0,0 @@
-import type { CompiledChunk, RawChunk, FileSegment, FileLLMData } from "./types.js";
-import type { PreprocessedChunk } from "./preprocessor.js";
-/**
- * 鑾峰彇鎵€鏈夊凡绱㈠紩鏂囦欢鐨勫師濮嬪潡锛堜緵 AI 缂栬瘧锛? */
-export declare function getRawChunks(sourceDir?: string, uncompiledOnly?: boolean): Promise<RawChunk[]>;
-/**
- * 瀛樺偍鍧楃骇缂栬瘧缁撴灉骞堕噸寤?embedding (v5.1 閬楃暀)
- */
-export declare function storeCompiledChunks(compiled: CompiledChunk[]): Promise<number>;
-/**
- * 瀛樺偍鏂囦欢绾х紪璇戠粨鏋滃苟閲嶅缓 embedding (v5.2 鈫?v5.4 杩囨浮)
- * v5.4 TODO: 鏇挎崲涓?storeFileLLMVector 鈥?鍙寕 1 涓?###llm 鍚戦噺锛屼笉鍒?AST chunks
- */
-export declare function storeFileSegments(relPath: string, segments: FileSegment[], preprocessed: PreprocessedChunk[]): Promise<number>;
-/**
- * 瀛樺偍鏂囦欢绾?LLM 缂栬瘧缁撴灉锛氭寕杞?1 涓?###llm 鍚戦噺锛屼笉鍒犻櫎 AST chunks銆? */
-export declare function storeFileLLMVector(sourceDir: string, relPath: string, llmData: FileLLMData, llmModel?: string): Promise<boolean>;
-//# sourceMappingURL=indexer-compile.d.ts.map

package/dist/lib/indexer-embed.d.ts

@@ -1,21 +0,0 @@
-import type { FileEntry } from "./types.js";
-/**
- * 鎸夋爣棰樺皢鏂囦欢鍒嗗壊涓哄涓潡銆? * v5.3: 浼樺厛 AST 瑙ｆ瀽锛屽け璐ラ檷绾?regex銆? */
-export declare function extractChunks(filePath: string, relPath: string, defaultTitle: string, maxEmbedLen?: number): Promise<{
-    key: string;
-    heading: string;
-    level: number;
-    embedText: string;
-    rawText: string;
-}[]>;
-/**
- * 鎵归噺鐢熸垚 embedding 骞舵寔涔呭寲鍒?vectors.json
- */
-export declare function generateEmbeddings(sourceDir: string, entries: FileEntry[]): Promise<number>;
-/**
- * 涓哄崟涓枃浠剁敓鎴?鏇存柊 embedding
- */
-export declare function embedSingleFile(sourceDir: string, relPath: string, title: string): Promise<boolean>;
-/** 计算全部向量的均值（噪声基底），供语义搜索降噪 */
-export declare function recomputeCentroid(): void;
-//# sourceMappingURL=indexer-embed.d.ts.map

package/dist/lib/indexer-scan.d.ts

@@ -1,4 +0,0 @@
-import type { FileEntry } from "./types.js";
-/** 递归扫描目录，返回所有 .md 的 FileEntry */
-export declare function scanDir(sourceDir: string): Promise<FileEntry[]>;
-//# sourceMappingURL=indexer-scan.d.ts.map

package/dist/lib/model-registry.d.ts

@@ -1,18 +0,0 @@
-export interface ModelInfo {
-    id: string;
-    name: string;
-    hfRepo: string;
-    dim: number;
-    description: string;
-    languages: string[];
-    maxTokens: number;
-    int8Size: number;
-    fp32Size: number;
-}
-export declare const BUILTIN_MODELS: ModelInfo[];
-export declare function getBuiltinModels(): ModelInfo[];
-export declare function findModel(id: string): ModelInfo | undefined;
-export declare function getCurrentModel(): ModelInfo;
-export declare function selectModel(id: string): ModelInfo | null;
-export declare function getDefaultModelId(): string;
-//# sourceMappingURL=model-registry.d.ts.map

package/dist/lib/parser.d.ts

@@ -1,9 +0,0 @@
-import type { FileEntry } from "./types.js";
-/**
- * 解析单个 .md 文件为 FileEntry
- * @param root     数据源根目录
- * @param filePath 文件绝对路径
- * @param mtime    文件修改时间（可选；不传则通过 statSync 自动获取）
- */
-export declare function parseFileEntry(root: string, filePath: string, mtime?: string): FileEntry | null;
-//# sourceMappingURL=parser.d.ts.map

package/dist/lib/preprocessor.d.ts

@@ -1,36 +0,0 @@
-/** chunkType: markdown 特征推断 */
-export declare function inferChunkType(text: string, heading: string): string;
-/** contentClass: 文件路径推断 */
-export declare function inferContentClass(relPath: string): string;
-/** importance: 启发式打分 (0.1-1.0) */
-export declare function inferImportance(text: string, heading: string): number;
-/** temporalAnchor: 正则提取第一个 YYYY-MM-DD */
-export declare function inferTemporalAnchor(text: string): string | undefined;
-/** confidence: 基于文本长度的预设置信度 */
-export declare function inferConfidence(text: string): number;
-/** summary: 清洗后取前 30 字 */
-export declare function inferSummary(text: string): string;
-/** keywords: 英文标识符 + 中文高频词预提取 */
-export declare function inferKeywords(text: string): string[];
-/** 预处理器对文件全文的输出 */
-export interface PreprocessedChunk {
-    /** 程序切分的块（按标题简单切，仅作兜底） */
-    heading: string;
-    level: number;
-    text: string;
-    /** 程序提取的元数据 */
-    chunkType: string;
-    contentClass: string;
-    importance: number;
-    temporalAnchor?: string;
-    confidence: number;
-    summary: string;
-    keywords: string[];
-}
-/**
- * 对文件全文做预处理: AST 分块 + 提取每段元数据。
- * v5.3: 复用 ast-chunker 替代 regex 逐行扫描。
- * 这是兜底分块 — LLM 在文件级编译时可能输出不同的 segments。
- */
-export declare function preprocessFile(relPath: string, fullText: string, defaultTitle: string): Promise<PreprocessedChunk[]>;
-//# sourceMappingURL=preprocessor.d.ts.map

package/dist/lib/semantic-compiler.d.ts

@@ -1,44 +0,0 @@
-import type { RawChunk, CompiledChunk, FileSegment } from "./types.js";
-import type { PreprocessedChunk } from "./preprocessor.js";
-/** 推荐每批处理的 chunk 数（平衡上下文大小和效率） */
-export declare const BATCH_SIZE = 25;
-export declare const COMPILE_SYSTEM_PROMPT = "\u4F60\u662F\u4E00\u4E2A\"\u77E5\u8BC6\u8BED\u4E49\u7F16\u8BD1\u5668\"\u3002\n\n\u4F60\u7684\u4EFB\u52A1\u4E0D\u662F\u603B\u7ED3\u5185\u5BB9\u3002\n\n\u4F60\u7684\u4EFB\u52A1\u662F\uFF1A\n\u5C06\u4EBA\u7C7B\u968F\u624B\u8BB0\u5F55\u7684\u975E\u7ED3\u6784\u5316\u7B14\u8BB0\uFF0C\n\u8F6C\u6362\u4E3A\u9002\u5408\u673A\u5668\u8BED\u4E49\u7D22\u5F15\u3001\u6982\u5FF5\u68C0\u7D22\u3001\n\u77E5\u8BC6\u805A\u7C7B\u3001\u957F\u671F\u6F14\u5316\u7684\"\u8BA4\u77E5\u77E5\u8BC6\u5355\u5143\"\u3002\n\n\u6838\u5FC3\u539F\u5219\uFF1A\n1. \u4FDD\u7559\u539F\u59CB\u4FE1\u606F \u2014 \u4E0D\u5220\u6280\u672F\u7EC6\u8282\n2. \u4E0D\u6539\u53D8\u539F\u610F \u2014 \u53EA\u89C4\u8303\u5316\u8868\u8FBE\n3. \u8865\u5168\u9690\u5F0F\u8868\u8FBE \u2014 \u8865\u5145\u7701\u7565\u7684\u4E3B\u8BED\u3001\u5C55\u5F00\u7F29\u5199\n4. \u7EDF\u4E00\u672F\u8BED \u2014 \u5C06\u540C\u4E49\u8868\u8FBE\u5F52\u4E00\uFF08\u5982 \"\u72B6\u6001\u6C61\u67D3\" \u2194 \"stale closure\"\uFF09\n5. \u63D0\u53D6\u6838\u5FC3\u6982\u5FF5 \u2014 \u8BC6\u522B\u6280\u672F\u5173\u952E\u8BCD\n6. \u4FDD\u6301\u5355\u4E3B\u9898 \u2014 \u4E00\u4E2A chunk \u53EA\u63CF\u8FF0\u4E00\u4E2A\u8BA4\u77E5\u4E3B\u9898\n7. \u8F93\u51FA\u7ED3\u6784\u5316 JSON \u2014 \u4E25\u683C\u9075\u5FAA schema\n\n\u7981\u6B62\uFF1A\n1. \u8FC7\u5EA6\u603B\u7ED3\n2. \u5220\u9664\u539F\u6587\n3. \u6539\u5199\u903B\u8F91\n4. \u4E3B\u89C2\u63A8\u65AD\n5. \u5F15\u5165\u4E0D\u5B58\u5728\u7684\u4FE1\u606F\n\n\u4F60\u7684\u89D2\u8272\u662F\uFF1A\"\u8BED\u4E49\u6807\u51C6\u5316\u5668\"\uFF0C\u4E0D\u662F\"\u5185\u5BB9\u4F5C\u8005\"\u3002";
-/** 为一批 chunk 构建编译 prompt */
-export declare function buildCompilePrompt(chunks: RawChunk[]): string;
-/**
- * 从编译后的 ChunkInfo 构建最优 embedding 输入
- *
- * 蓝图推荐格式:
- *   [TOPIC] + [CONCEPTS] + [ALIASES] + [KEYWORDS] + [NORMALIZED] + [RAW]
- *
- * 原因: 增强隐式语义，让向量模型在检索时更稳定地匹配
- */
-export declare function buildEmbeddingText(topic: string, normalizedText: string, concepts: string[], aliases: string[], keywords: string[], contentClass: string, temporalAnchor: string | undefined, rawText: string): string;
-/** 尝试从 LLM 响应中提取 CompiledChunk 数组 */
-export declare function parseCompiledResult(text: string): CompiledChunk[] | null;
-/** 文件级编译 System Prompt（只要求 LLM 做 4 件事） */
-export declare const FILE_COMPILE_SYSTEM_PROMPT = "\u4F60\u662F\u4E00\u4E2A\"\u77E5\u8BC6\u8BED\u4E49\u7F16\u8BD1\u5668\"\u3002\n\n\u4F60\u7684\u4EFB\u52A1: \u5C06\u6574\u7BC7\u7B14\u8BB0\u8F6C\u6362\u4E3A\u7ED3\u6784\u5316\u7684\u8BED\u4E49\u77E5\u8BC6\u5355\u5143\u3002\n\n\u4F60\u9700\u8981\u505A\u7684 4 \u4EF6\u4E8B:\n1. \u81EA\u884C\u5224\u65AD\u8BED\u4E49\u8FB9\u754C \u2014 \u5C06\u6587\u4EF6\u5206\u6210\u82E5\u5E72\u8FDE\u7EED\u7684\u8BED\u4E49\u7247\u6BB5\uFF08segments\uFF09\n2. \u4E3A\u6BCF\u4E2A\u7247\u6BB5\u5199\u51FA topic\uFF08\u6838\u5FC3\u4E3B\u9898\uFF0C\u4E00\u53E5\u8BDD\uFF09\n3. \u4E3A\u6BCF\u4E2A\u7247\u6BB5\u5199\u51FA normalizedText\uFF08\u89C4\u8303\u5316\u6587\u672C\uFF1A\u8865\u5168\u7701\u7565\u3001\u7EDF\u4E00\u672F\u8BED\u3001\u4FDD\u7559\u6240\u6709\u6280\u672F\u7EC6\u8282\uFF09\n4. \u4E3A\u6BCF\u4E2A\u7247\u6BB5\u63D0\u53D6 concepts\uFF08\u6280\u672F\u6982\u5FF5\uFF09\u548C aliases\uFF08\u540C\u4E49\u8868\u8FBE\uFF0C\u683C\u5F0F \"\u4E2D\u6587 \u2194 English\"\uFF09\n\n\u6838\u5FC3\u539F\u5219:\n- \u4FDD\u7559\u539F\u59CB\u4FE1\u606F \u2014 \u4E0D\u5220\u6280\u672F\u7EC6\u8282\uFF08API \u540D\u3001\u53C2\u6570\u3001\u9519\u8BEF\u4FE1\u606F\u3001\u7F29\u5199\uFF09\n- \u4E0D\u6539\u53D8\u539F\u610F \u2014 \u53EA\u89C4\u8303\u5316\u8868\u8FBE\n- \u8BED\u4E49\u8FB9\u754C = \u540C\u4E00\u8BA4\u77E5\u4E3B\u9898\u7684\u81EA\u7136\u6BB5\u6216\u8FDE\u7EED\u6BB5\u843D\n- \u5982\u679C\u6574\u4E2A\u6587\u4EF6\u662F\u5355\u4E00\u4E3B\u9898\uFF0C\u53EA\u8F93\u51FA 1 \u4E2A segment\n\n\u7981\u6B62: \u8FC7\u5EA6\u603B\u7ED3\u3001\u5220\u9664\u539F\u6587\u3001\u6539\u5199\u903B\u8F91\u3001\u4E3B\u89C2\u63A8\u65AD\u3001\u5F15\u5165\u4E0D\u5B58\u5728\u7684\u4FE1\u606F\u3002";
-/**
- * 为文件级编译构建 prompt
- * @param relPath      文件路径
- * @param fullText     文件全文
- * @param preprocessed 预处理器输出（仅展示给 LLM 参考）
- */
-export declare function buildFileCompilePrompt(relPath: string, fullText: string, preprocessed: PreprocessedChunk[]): string;
-/** 从 LLM 响应中提取 FileSegment 数组 */
-export declare function parseFileSegments(text: string): FileSegment[] | null;
-/** v5.4 文件级 System Prompt（极简版） */
-export declare const FILE_LLM_SYSTEM_PROMPT = "\u4F60\u662F\u4E00\u4E2A\"\u77E5\u8BC6\u8BED\u4E49\u7F16\u8BD1\u5668\"\u3002\n\n\u4F60\u7684\u4EFB\u52A1: \u5C06\u6574\u7BC7\u7B14\u8BB0\u8F6C\u6362\u4E3A\u7ED3\u6784\u5316\u7684\u8BED\u4E49\u5143\u6570\u636E\uFF0C\u7528\u4E8E\u589E\u5F3A\u8BED\u4E49\u641C\u7D22\u3002\n\n\u4F60\u9700\u8981\u8F93\u51FA\u7684 4 \u4E2A\u5B57\u6BB5:\n1. topic \u2014 \u6838\u5FC3\u4E3B\u9898\uFF08\u4E00\u53E5\u8BDD\u6982\u62EC\u5168\u6587\uFF09\n2. normalizedText \u2014 \u89C4\u8303\u5316\u6587\u672C\uFF08\u8865\u5168\u7701\u7565\u4E3B\u8BED\u3001\u7EDF\u4E00\u672F\u8BED\u3001\u4FDD\u7559\u6240\u6709\u6280\u672F\u7EC6\u8282\uFF09\n3. concepts \u2014 \u6280\u672F\u6982\u5FF5\u5217\u8868\uFF083-8 \u4E2A\u6838\u5FC3\u6982\u5FF5\uFF09\n4. aliases \u2014 \u540C\u4E49\u8868\u8FBE\uFF08\u683C\u5F0F \"\u4E2D\u6587 \u2194 English\"\uFF0C2-5 \u7EC4\uFF09\n\n\u6838\u5FC3\u539F\u5219:\n- \u4FDD\u7559\u6240\u6709\u6280\u672F\u7EC6\u8282\uFF08API \u540D\u3001\u53C2\u6570\u3001\u9519\u8BEF\u4FE1\u606F\u3001\u7F29\u5199\u3001\u7248\u672C\u53F7\uFF09\n- \u4E0D\u6539\u53D8\u539F\u610F\uFF0C\u53EA\u89C4\u8303\u5316\u8868\u8FBE\n- concepts \u63D0\u53D6\u6280\u672F\u5173\u952E\u8BCD\uFF0C\u4E0D\u662F\u6458\u8981\n- aliases \u8986\u76D6\u4E2D\u82F1\u5BF9\u7167\u548C\u7F29\u5199\u5C55\u5F00\n\n\u7981\u6B62: \u8FC7\u5EA6\u603B\u7ED3\u3001\u5220\u9664\u539F\u6587\u3001\u6539\u5199\u903B\u8F91\u3001\u4E3B\u89C2\u63A8\u65AD\u3001\u5F15\u5165\u4E0D\u5B58\u5728\u7684\u4FE1\u606F\u3002";
-/**
- * v5.4 构建简化文件级编译 prompt
- */
-export declare function buildFileLLMPrompt(relPath: string, fullText: string): string;
-/**
- * v5.4 解析文件级 LLM 响应（单对象，非 segments 数组）
- */
-export declare function parseFileLLMResult(text: string): import("../lib/types.js").FileLLMData | null;
-/**
- * v5.4 构建文件级 LLM 向量的 embedding 文本
- */
-export declare function buildFileLLMEmbeddingText(data: import("../lib/types.js").FileLLMData, relPath?: string, maxEmbedLen?: number): string;
-//# sourceMappingURL=semantic-compiler.d.ts.map

package/dist/lib/semantic-search.d.ts

@@ -1,4 +0,0 @@
-import type { SearchHit } from "./types.js";
-export declare function semanticSearch(query: string): Promise<SearchHit[]>;
-export declare function hybridSearch(query: string): Promise<SearchHit[]>;
-//# sourceMappingURL=semantic-search.d.ts.map

package/dist/lib/store-config.d.ts

@@ -1,25 +0,0 @@
-interface WikiConfig {
-    sources: string[];
-    semanticEnabled: boolean;
-    currentModelId: string;
-    lastScan: string;
-}
-export declare function readConfig(): WikiConfig;
-export declare function writeConfig(c: WikiConfig): void;
-export declare function getSources(): string[];
-export declare function addSource(absPath: string): boolean;
-/** 移除数据�?�?返回被移除的路径，未找到返回 null */
-export declare function removeSource(target: string): string | null;
-export declare function getLastScan(): string;
-export declare function setLastScan(iso: string): void;
-export declare function getSemanticEnabled(): boolean;
-export declare function setSemanticEnabled(enabled: boolean): void;
-export declare function readModelId(): string;
-export declare function writeModelId(id: string): void;
-export declare function getWikiModel(): string;
-export declare function configStats(): {
-    sources: number;
-    lastScan: string;
-};
-export {};
-//# sourceMappingURL=store-config.d.ts.map

package/dist/lib/store-vectors.d.ts

@@ -1,17 +0,0 @@
-export declare function getEmbeddings(): Record<string, number[]>;
-export declare function setEmbeddings(entries: Record<string, number[]>, model?: string, dim?: number): void;
-export declare function getChunkInfo(): Record<string, import("./types.js").ChunkInfo>;
-export declare function setChunkInfo(chunkInfo: Record<string, import("./types.js").ChunkInfo>): void;
-export declare function removeEmbedding(relPath: string): void;
-export declare function getEmbeddingModel(): string | undefined;
-export declare function getEmbeddingDim(): number | undefined;
-export declare function getCentroid(): number[] | null;
-export declare function setCentroid(centroid: number[]): void;
-export declare function clearCentroid(): void;
-export declare function vectorsStats(): {
-    embeddings: number;
-    centroid: boolean;
-    model?: string;
-    dim?: number;
-};
-//# sourceMappingURL=store-vectors.d.ts.map