Fix tokenizer jieba tag flow and debug logging

core/debug-core.js

@@ -1,4 +1,4 @@
-import { EventCenter } from "./event-manager.js";
+import { EventCenter } from "./event-manager.js";
 
 const DEFAULT_MAX_LOGS = 200;
 
@@ -110,14 +110,14 @@ class LoggerCore {
         });
     }
 
-    info(moduleId, message) {
+    info(moduleId, ...args) {
-        this._log("info", moduleId, message, null);
+        const msg = args.map(a => (typeof a === 'string' ? a : safeStringify(a))).join(' ');
+        this._log('info', moduleId, msg, null);
     }
-
+    warn(moduleId, ...args) {
-    warn(moduleId, message) {
+        const msg = args.map(a => (typeof a === 'string' ? a : safeStringify(a))).join(' ');
-        this._log("warn", moduleId, message, null);
+        this._log('warn', moduleId, msg, null);
     }
-
     error(moduleId, message, err) {
         this._log("error", moduleId, message, err || null);
     }

libs/jieba-wasm/jieba_rs_wasm.js

@@ -0,0 +1,438 @@
+let wasm;
+
+let cachedUint8ArrayMemory0 = null;
+
+function getUint8ArrayMemory0() {
+    if (cachedUint8ArrayMemory0 === null || cachedUint8ArrayMemory0.byteLength === 0) {
+        cachedUint8ArrayMemory0 = new Uint8Array(wasm.memory.buffer);
+    }
+    return cachedUint8ArrayMemory0;
+}
+
+let cachedTextDecoder = (typeof TextDecoder !== 'undefined' ? new TextDecoder('utf-8', { ignoreBOM: true, fatal: true }) : { decode: () => { throw Error('TextDecoder not available') } } );
+
+if (typeof TextDecoder !== 'undefined') { cachedTextDecoder.decode(); };
+
+const MAX_SAFARI_DECODE_BYTES = 2146435072;
+let numBytesDecoded = 0;
+function decodeText(ptr, len) {
+    numBytesDecoded += len;
+    if (numBytesDecoded >= MAX_SAFARI_DECODE_BYTES) {
+        cachedTextDecoder = (typeof TextDecoder !== 'undefined' ? new TextDecoder('utf-8', { ignoreBOM: true, fatal: true }) : { decode: () => { throw Error('TextDecoder not available') } } );
+        cachedTextDecoder.decode();
+        numBytesDecoded = len;
+    }
+    return cachedTextDecoder.decode(getUint8ArrayMemory0().subarray(ptr, ptr + len));
+}
+
+function getStringFromWasm0(ptr, len) {
+    ptr = ptr >>> 0;
+    return decodeText(ptr, len);
+}
+
+function debugString(val) {
+    // primitive types
+    const type = typeof val;
+    if (type == 'number' || type == 'boolean' || val == null) {
+        return  `${val}`;
+    }
+    if (type == 'string') {
+        return `"${val}"`;
+    }
+    if (type == 'symbol') {
+        const description = val.description;
+        if (description == null) {
+            return 'Symbol';
+        } else {
+            return `Symbol(${description})`;
+        }
+    }
+    if (type == 'function') {
+        const name = val.name;
+        if (typeof name == 'string' && name.length > 0) {
+            return `Function(${name})`;
+        } else {
+            return 'Function';
+        }
+    }
+    // objects
+    if (Array.isArray(val)) {
+        const length = val.length;
+        let debug = '[';
+        if (length > 0) {
+            debug += debugString(val[0]);
+        }
+        for(let i = 1; i < length; i++) {
+            debug += ', ' + debugString(val[i]);
+        }
+        debug += ']';
+        return debug;
+    }
+    // Test for built-in
+    const builtInMatches = /\[object ([^\]]+)\]/.exec(toString.call(val));
+    let className;
+    if (builtInMatches && builtInMatches.length > 1) {
+        className = builtInMatches[1];
+    } else {
+        // Failed to match the standard '[object ClassName]'
+        return toString.call(val);
+    }
+    if (className == 'Object') {
+        // we're a user defined class or Object
+        // JSON.stringify avoids problems with cycles, and is generally much
+        // easier than looping through ownProperties of `val`.
+        try {
+            return 'Object(' + JSON.stringify(val) + ')';
+        } catch (_) {
+            return 'Object';
+        }
+    }
+    // errors
+    if (val instanceof Error) {
+        return `${val.name}: ${val.message}\n${val.stack}`;
+    }
+    // TODO we could test for more things here, like `Set`s and `Map`s.
+    return className;
+}
+
+let WASM_VECTOR_LEN = 0;
+
+const cachedTextEncoder = (typeof TextEncoder !== 'undefined' ? new TextEncoder('utf-8') : { encode: () => { throw Error('TextEncoder not available') } } );
+
+const encodeString = (typeof cachedTextEncoder.encodeInto === 'function'
+    ? function (arg, view) {
+    return cachedTextEncoder.encodeInto(arg, view);
+}
+    : function (arg, view) {
+    const buf = cachedTextEncoder.encode(arg);
+    view.set(buf);
+    return {
+        read: arg.length,
+        written: buf.length
+    };
+});
+
+function passStringToWasm0(arg, malloc, realloc) {
+
+    if (realloc === undefined) {
+        const buf = cachedTextEncoder.encode(arg);
+        const ptr = malloc(buf.length, 1) >>> 0;
+        getUint8ArrayMemory0().subarray(ptr, ptr + buf.length).set(buf);
+        WASM_VECTOR_LEN = buf.length;
+        return ptr;
+    }
+
+    let len = arg.length;
+    let ptr = malloc(len, 1) >>> 0;
+
+    const mem = getUint8ArrayMemory0();
+
+    let offset = 0;
+
+    for (; offset < len; offset++) {
+        const code = arg.charCodeAt(offset);
+        if (code > 0x7F) break;
+        mem[ptr + offset] = code;
+    }
+
+    if (offset !== len) {
+        if (offset !== 0) {
+            arg = arg.slice(offset);
+        }
+        ptr = realloc(ptr, len, len = offset + arg.length * 3, 1) >>> 0;
+        const view = getUint8ArrayMemory0().subarray(ptr + offset, ptr + len);
+        const ret = encodeString(arg, view);
+
+        offset += ret.written;
+        ptr = realloc(ptr, len, offset, 1) >>> 0;
+    }
+
+    WASM_VECTOR_LEN = offset;
+    return ptr;
+}
+
+let cachedDataViewMemory0 = null;
+
+function getDataViewMemory0() {
+    if (cachedDataViewMemory0 === null || cachedDataViewMemory0.buffer.detached === true || (cachedDataViewMemory0.buffer.detached === undefined && cachedDataViewMemory0.buffer !== wasm.memory.buffer)) {
+        cachedDataViewMemory0 = new DataView(wasm.memory.buffer);
+    }
+    return cachedDataViewMemory0;
+}
+
+function isLikeNone(x) {
+    return x === undefined || x === null;
+}
+
+function getArrayJsValueFromWasm0(ptr, len) {
+    ptr = ptr >>> 0;
+    const mem = getDataViewMemory0();
+    const result = [];
+    for (let i = ptr; i < ptr + 4 * len; i += 4) {
+        result.push(wasm.__wbindgen_export_2.get(mem.getUint32(i, true)));
+    }
+    wasm.__externref_drop_slice(ptr, len);
+    return result;
+}
+/**
+ * @param {string} text
+ * @param {boolean | null} [hmm]
+ * @returns {string[]}
+ */
+export function cut(text, hmm) {
+    const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.cut(ptr0, len0, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+    var v2 = getArrayJsValueFromWasm0(ret[0], ret[1]).slice();
+    wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
+    return v2;
+}
+
+/**
+ * @param {string} text
+ * @returns {string[]}
+ */
+export function cut_all(text) {
+    const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.cut_all(ptr0, len0);
+    var v2 = getArrayJsValueFromWasm0(ret[0], ret[1]).slice();
+    wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
+    return v2;
+}
+
+/**
+ * @param {string} text
+ * @param {boolean | null} [hmm]
+ * @returns {string[]}
+ */
+export function cut_for_search(text, hmm) {
+    const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.cut_for_search(ptr0, len0, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+    var v2 = getArrayJsValueFromWasm0(ret[0], ret[1]).slice();
+    wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
+    return v2;
+}
+
+function takeFromExternrefTable0(idx) {
+    const value = wasm.__wbindgen_export_2.get(idx);
+    wasm.__externref_table_dealloc(idx);
+    return value;
+}
+/**
+ * @param {string} text
+ * @param {string} mode
+ * @param {boolean | null} [hmm]
+ * @returns {Token[]}
+ */
+export function tokenize(text, mode, hmm) {
+    const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ptr1 = passStringToWasm0(mode, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len1 = WASM_VECTOR_LEN;
+    const ret = wasm.tokenize(ptr0, len0, ptr1, len1, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+    if (ret[3]) {
+        throw takeFromExternrefTable0(ret[2]);
+    }
+    var v3 = getArrayJsValueFromWasm0(ret[0], ret[1]).slice();
+    wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
+    return v3;
+}
+
+/**
+ * @param {string} word
+ * @param {number | null} [freq]
+ * @param {string | null} [tag]
+ * @returns {number}
+ */
+export function add_word(word, freq, tag) {
+    const ptr0 = passStringToWasm0(word, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    var ptr1 = isLikeNone(tag) ? 0 : passStringToWasm0(tag, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    var len1 = WASM_VECTOR_LEN;
+    const ret = wasm.add_word(ptr0, len0, isLikeNone(freq) ? 0x100000001 : (freq) >>> 0, ptr1, len1);
+    return ret >>> 0;
+}
+
+/**
+ * @param {string} sentence
+ * @param {boolean | null} [hmm]
+ * @returns {Tag[]}
+ */
+export function tag(sentence, hmm) {
+    const ptr0 = passStringToWasm0(sentence, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.tag(ptr0, len0, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+    var v2 = getArrayJsValueFromWasm0(ret[0], ret[1]).slice();
+    wasm.__wbindgen_free(ret[0], ret[1] * 4, 4);
+    return v2;
+}
+
+/**
+ * @param {string} dict
+ */
+export function with_dict(dict) {
+    const ptr0 = passStringToWasm0(dict, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    const ret = wasm.with_dict(ptr0, len0);
+    if (ret[1]) {
+        throw takeFromExternrefTable0(ret[0]);
+    }
+}
+
+const EXPECTED_RESPONSE_TYPES = new Set(['basic', 'cors', 'default']);
+
+async function __wbg_load(module, imports) {
+    if (typeof Response === 'function' && module instanceof Response) {
+        if (typeof WebAssembly.instantiateStreaming === 'function') {
+            try {
+                return await WebAssembly.instantiateStreaming(module, imports);
+
+            } catch (e) {
+                const validResponse = module.ok && EXPECTED_RESPONSE_TYPES.has(module.type);
+
+                if (validResponse && module.headers.get('Content-Type') !== 'application/wasm') {
+                    console.warn("`WebAssembly.instantiateStreaming` failed because your server does not serve Wasm with `application/wasm` MIME type. Falling back to `WebAssembly.instantiate` which is slower. Original error:\n", e);
+
+                } else {
+                    throw e;
+                }
+            }
+        }
+
+        const bytes = await module.arrayBuffer();
+        return await WebAssembly.instantiate(bytes, imports);
+
+    } else {
+        const instance = await WebAssembly.instantiate(module, imports);
+
+        if (instance instanceof WebAssembly.Instance) {
+            return { instance, module };
+
+        } else {
+            return instance;
+        }
+    }
+}
+
+function __wbg_get_imports() {
+    const imports = {};
+    imports.wbg = {};
+    imports.wbg.__wbg_Error_0497d5bdba9362e5 = function(arg0, arg1) {
+        const ret = Error(getStringFromWasm0(arg0, arg1));
+        return ret;
+    };
+    imports.wbg.__wbg_new_07b483f72211fd66 = function() {
+        const ret = new Object();
+        return ret;
+    };
+    imports.wbg.__wbg_set_3f1d0b984ed272ed = function(arg0, arg1, arg2) {
+        arg0[arg1] = arg2;
+    };
+    imports.wbg.__wbindgen_bigint_from_u64 = function(arg0) {
+        const ret = BigInt.asUintN(64, arg0);
+        return ret;
+    };
+    imports.wbg.__wbindgen_debug_string = function(arg0, arg1) {
+        const ret = debugString(arg1);
+        const ptr1 = passStringToWasm0(ret, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len1 = WASM_VECTOR_LEN;
+        getDataViewMemory0().setInt32(arg0 + 4 * 1, len1, true);
+        getDataViewMemory0().setInt32(arg0 + 4 * 0, ptr1, true);
+    };
+    imports.wbg.__wbindgen_init_externref_table = function() {
+        const table = wasm.__wbindgen_export_2;
+        const offset = table.grow(4);
+        table.set(0, undefined);
+        table.set(offset + 0, undefined);
+        table.set(offset + 1, null);
+        table.set(offset + 2, true);
+        table.set(offset + 3, false);
+        ;
+    };
+    imports.wbg.__wbindgen_number_new = function(arg0) {
+        const ret = arg0;
+        return ret;
+    };
+    imports.wbg.__wbindgen_string_new = function(arg0, arg1) {
+        const ret = getStringFromWasm0(arg0, arg1);
+        return ret;
+    };
+    imports.wbg.__wbindgen_throw = function(arg0, arg1) {
+        throw new Error(getStringFromWasm0(arg0, arg1));
+    };
+
+    return imports;
+}
+
+function __wbg_init_memory(imports, memory) {
+
+}
+
+function __wbg_finalize_init(instance, module) {
+    wasm = instance.exports;
+    __wbg_init.__wbindgen_wasm_module = module;
+    cachedDataViewMemory0 = null;
+    cachedUint8ArrayMemory0 = null;
+
+
+    wasm.__wbindgen_start();
+    return wasm;
+}
+
+function initSync(module) {
+    if (wasm !== undefined) return wasm;
+
+
+    if (typeof module !== 'undefined') {
+        if (Object.getPrototypeOf(module) === Object.prototype) {
+            ({module} = module)
+        } else {
+            console.warn('using deprecated parameters for `initSync()`; pass a single object instead')
+        }
+    }
+
+    const imports = __wbg_get_imports();
+
+    __wbg_init_memory(imports);
+
+    if (!(module instanceof WebAssembly.Module)) {
+        module = new WebAssembly.Module(module);
+    }
+
+    const instance = new WebAssembly.Instance(module, imports);
+
+    return __wbg_finalize_init(instance, module);
+}
+
+async function __wbg_init(module_or_path) {
+    if (wasm !== undefined) return wasm;
+
+
+    if (typeof module_or_path !== 'undefined') {
+        if (Object.getPrototypeOf(module_or_path) === Object.prototype) {
+            ({module_or_path} = module_or_path)
+        } else {
+            console.warn('using deprecated parameters for the initialization function; pass a single object instead')
+        }
+    }
+
+    if (typeof module_or_path === 'undefined') {
+        module_or_path = new URL('jieba_rs_wasm_bg.wasm', import.meta.url);
+    }
+    const imports = __wbg_get_imports();
+
+    if (typeof module_or_path === 'string' || (typeof Request === 'function' && module_or_path instanceof Request) || (typeof URL === 'function' && module_or_path instanceof URL)) {
+        module_or_path = fetch(module_or_path);
+    }
+
+    __wbg_init_memory(imports);
+
+    const { instance, module } = await __wbg_load(await module_or_path, imports);
+
+    return __wbg_finalize_init(instance, module);
+}
+
+export { initSync };
+export default __wbg_init;

libs/jieba-wasm/jieba_rs_wasm_bg.js

@@ -0,0 +1,372 @@
+let wasm;
+export function __wbg_set_wasm(val) {
+    wasm = val;
+}
+
+
+const heap = new Array(128).fill(undefined);
+
+heap.push(undefined, null, true, false);
+
+function getObject(idx) { return heap[idx]; }
+
+let heap_next = heap.length;
+
+function dropObject(idx) {
+    if (idx < 132) return;
+    heap[idx] = heap_next;
+    heap_next = idx;
+}
+
+function takeObject(idx) {
+    const ret = getObject(idx);
+    dropObject(idx);
+    return ret;
+}
+
+const lTextDecoder = typeof TextDecoder === 'undefined' ? (0, module.require)('util').TextDecoder : TextDecoder;
+
+let cachedTextDecoder = new lTextDecoder('utf-8', { ignoreBOM: true, fatal: true });
+
+cachedTextDecoder.decode();
+
+let cachedUint8ArrayMemory0 = null;
+
+function getUint8ArrayMemory0() {
+    if (cachedUint8ArrayMemory0 === null || cachedUint8ArrayMemory0.byteLength === 0) {
+        cachedUint8ArrayMemory0 = new Uint8Array(wasm.memory.buffer);
+    }
+    return cachedUint8ArrayMemory0;
+}
+
+function getStringFromWasm0(ptr, len) {
+    ptr = ptr >>> 0;
+    return cachedTextDecoder.decode(getUint8ArrayMemory0().subarray(ptr, ptr + len));
+}
+
+function addHeapObject(obj) {
+    if (heap_next === heap.length) heap.push(heap.length + 1);
+    const idx = heap_next;
+    heap_next = heap[idx];
+
+    heap[idx] = obj;
+    return idx;
+}
+
+function debugString(val) {
+    // primitive types
+    const type = typeof val;
+    if (type == 'number' || type == 'boolean' || val == null) {
+        return  `${val}`;
+    }
+    if (type == 'string') {
+        return `"${val}"`;
+    }
+    if (type == 'symbol') {
+        const description = val.description;
+        if (description == null) {
+            return 'Symbol';
+        } else {
+            return `Symbol(${description})`;
+        }
+    }
+    if (type == 'function') {
+        const name = val.name;
+        if (typeof name == 'string' && name.length > 0) {
+            return `Function(${name})`;
+        } else {
+            return 'Function';
+        }
+    }
+    // objects
+    if (Array.isArray(val)) {
+        const length = val.length;
+        let debug = '[';
+        if (length > 0) {
+            debug += debugString(val[0]);
+        }
+        for(let i = 1; i < length; i++) {
+            debug += ', ' + debugString(val[i]);
+        }
+        debug += ']';
+        return debug;
+    }
+    // Test for built-in
+    const builtInMatches = /\[object ([^\]]+)\]/.exec(toString.call(val));
+    let className;
+    if (builtInMatches.length > 1) {
+        className = builtInMatches[1];
+    } else {
+        // Failed to match the standard '[object ClassName]'
+        return toString.call(val);
+    }
+    if (className == 'Object') {
+        // we're a user defined class or Object
+        // JSON.stringify avoids problems with cycles, and is generally much
+        // easier than looping through ownProperties of `val`.
+        try {
+            return 'Object(' + JSON.stringify(val) + ')';
+        } catch (_) {
+            return 'Object';
+        }
+    }
+    // errors
+    if (val instanceof Error) {
+        return `${val.name}: ${val.message}\n${val.stack}`;
+    }
+    // TODO we could test for more things here, like `Set`s and `Map`s.
+    return className;
+}
+
+let WASM_VECTOR_LEN = 0;
+
+const lTextEncoder = typeof TextEncoder === 'undefined' ? (0, module.require)('util').TextEncoder : TextEncoder;
+
+let cachedTextEncoder = new lTextEncoder('utf-8');
+
+const encodeString = (typeof cachedTextEncoder.encodeInto === 'function'
+    ? function (arg, view) {
+    return cachedTextEncoder.encodeInto(arg, view);
+}
+    : function (arg, view) {
+    const buf = cachedTextEncoder.encode(arg);
+    view.set(buf);
+    return {
+        read: arg.length,
+        written: buf.length
+    };
+});
+
+function passStringToWasm0(arg, malloc, realloc) {
+
+    if (realloc === undefined) {
+        const buf = cachedTextEncoder.encode(arg);
+        const ptr = malloc(buf.length, 1) >>> 0;
+        getUint8ArrayMemory0().subarray(ptr, ptr + buf.length).set(buf);
+        WASM_VECTOR_LEN = buf.length;
+        return ptr;
+    }
+
+    let len = arg.length;
+    let ptr = malloc(len, 1) >>> 0;
+
+    const mem = getUint8ArrayMemory0();
+
+    let offset = 0;
+
+    for (; offset < len; offset++) {
+        const code = arg.charCodeAt(offset);
+        if (code > 0x7F) break;
+        mem[ptr + offset] = code;
+    }
+
+    if (offset !== len) {
+        if (offset !== 0) {
+            arg = arg.slice(offset);
+        }
+        ptr = realloc(ptr, len, len = offset + arg.length * 3, 1) >>> 0;
+        const view = getUint8ArrayMemory0().subarray(ptr + offset, ptr + len);
+        const ret = encodeString(arg, view);
+
+        offset += ret.written;
+        ptr = realloc(ptr, len, offset, 1) >>> 0;
+    }
+
+    WASM_VECTOR_LEN = offset;
+    return ptr;
+}
+
+let cachedDataViewMemory0 = null;
+
+function getDataViewMemory0() {
+    if (cachedDataViewMemory0 === null || cachedDataViewMemory0.buffer.detached === true || (cachedDataViewMemory0.buffer.detached === undefined && cachedDataViewMemory0.buffer !== wasm.memory.buffer)) {
+        cachedDataViewMemory0 = new DataView(wasm.memory.buffer);
+    }
+    return cachedDataViewMemory0;
+}
+
+function isLikeNone(x) {
+    return x === undefined || x === null;
+}
+
+function getArrayJsValueFromWasm0(ptr, len) {
+    ptr = ptr >>> 0;
+    const mem = getDataViewMemory0();
+    const result = [];
+    for (let i = ptr; i < ptr + 4 * len; i += 4) {
+        result.push(takeObject(mem.getUint32(i, true)));
+    }
+    return result;
+}
+/**
+ * @param {string} text
+ * @param {boolean | undefined} [hmm]
+ * @returns {any[]}
+ */
+export function cut(text, hmm) {
+    try {
+        const retptr = wasm.__wbindgen_add_to_stack_pointer(-16);
+        const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len0 = WASM_VECTOR_LEN;
+        wasm.cut(retptr, ptr0, len0, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+        var r0 = getDataViewMemory0().getInt32(retptr + 4 * 0, true);
+        var r1 = getDataViewMemory0().getInt32(retptr + 4 * 1, true);
+        var v2 = getArrayJsValueFromWasm0(r0, r1).slice();
+        wasm.__wbindgen_free(r0, r1 * 4, 4);
+        return v2;
+    } finally {
+        wasm.__wbindgen_add_to_stack_pointer(16);
+    }
+}
+
+/**
+ * @param {string} text
+ * @returns {any[]}
+ */
+export function cut_all(text) {
+    try {
+        const retptr = wasm.__wbindgen_add_to_stack_pointer(-16);
+        const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len0 = WASM_VECTOR_LEN;
+        wasm.cut_all(retptr, ptr0, len0);
+        var r0 = getDataViewMemory0().getInt32(retptr + 4 * 0, true);
+        var r1 = getDataViewMemory0().getInt32(retptr + 4 * 1, true);
+        var v2 = getArrayJsValueFromWasm0(r0, r1).slice();
+        wasm.__wbindgen_free(r0, r1 * 4, 4);
+        return v2;
+    } finally {
+        wasm.__wbindgen_add_to_stack_pointer(16);
+    }
+}
+
+/**
+ * @param {string} text
+ * @param {boolean | undefined} [hmm]
+ * @returns {any[]}
+ */
+export function cut_for_search(text, hmm) {
+    try {
+        const retptr = wasm.__wbindgen_add_to_stack_pointer(-16);
+        const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len0 = WASM_VECTOR_LEN;
+        wasm.cut_for_search(retptr, ptr0, len0, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+        var r0 = getDataViewMemory0().getInt32(retptr + 4 * 0, true);
+        var r1 = getDataViewMemory0().getInt32(retptr + 4 * 1, true);
+        var v2 = getArrayJsValueFromWasm0(r0, r1).slice();
+        wasm.__wbindgen_free(r0, r1 * 4, 4);
+        return v2;
+    } finally {
+        wasm.__wbindgen_add_to_stack_pointer(16);
+    }
+}
+
+/**
+ * @param {string} text
+ * @param {string} mode
+ * @param {boolean | undefined} [hmm]
+ * @returns {any[]}
+ */
+export function tokenize(text, mode, hmm) {
+    try {
+        const retptr = wasm.__wbindgen_add_to_stack_pointer(-16);
+        const ptr0 = passStringToWasm0(text, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len0 = WASM_VECTOR_LEN;
+        const ptr1 = passStringToWasm0(mode, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len1 = WASM_VECTOR_LEN;
+        wasm.tokenize(retptr, ptr0, len0, ptr1, len1, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+        var r0 = getDataViewMemory0().getInt32(retptr + 4 * 0, true);
+        var r1 = getDataViewMemory0().getInt32(retptr + 4 * 1, true);
+        var r2 = getDataViewMemory0().getInt32(retptr + 4 * 2, true);
+        var r3 = getDataViewMemory0().getInt32(retptr + 4 * 3, true);
+        if (r3) {
+            throw takeObject(r2);
+        }
+        var v3 = getArrayJsValueFromWasm0(r0, r1).slice();
+        wasm.__wbindgen_free(r0, r1 * 4, 4);
+        return v3;
+    } finally {
+        wasm.__wbindgen_add_to_stack_pointer(16);
+    }
+}
+
+/**
+ * @param {string} word
+ * @param {number | undefined} [freq]
+ * @param {string | undefined} [tag]
+ * @returns {number}
+ */
+export function add_word(word, freq, tag) {
+    const ptr0 = passStringToWasm0(word, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len0 = WASM_VECTOR_LEN;
+    var ptr1 = isLikeNone(tag) ? 0 : passStringToWasm0(tag, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    var len1 = WASM_VECTOR_LEN;
+    const ret = wasm.add_word(ptr0, len0, !isLikeNone(freq), isLikeNone(freq) ? 0 : freq, ptr1, len1);
+    return ret >>> 0;
+}
+
+/**
+ * @param {string} sentence
+ * @param {boolean | undefined} [hmm]
+ * @returns {any[]}
+ */
+export function tag(sentence, hmm) {
+    try {
+        const retptr = wasm.__wbindgen_add_to_stack_pointer(-16);
+        const ptr0 = passStringToWasm0(sentence, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+        const len0 = WASM_VECTOR_LEN;
+        wasm.tag(retptr, ptr0, len0, isLikeNone(hmm) ? 0xFFFFFF : hmm ? 1 : 0);
+        var r0 = getDataViewMemory0().getInt32(retptr + 4 * 0, true);
+        var r1 = getDataViewMemory0().getInt32(retptr + 4 * 1, true);
+        var v2 = getArrayJsValueFromWasm0(r0, r1).slice();
+        wasm.__wbindgen_free(r0, r1 * 4, 4);
+        return v2;
+    } finally {
+        wasm.__wbindgen_add_to_stack_pointer(16);
+    }
+}
+
+export function __wbindgen_object_drop_ref(arg0) {
+    takeObject(arg0);
+};
+
+export function __wbindgen_string_new(arg0, arg1) {
+    const ret = getStringFromWasm0(arg0, arg1);
+    return addHeapObject(ret);
+};
+
+export function __wbindgen_object_clone_ref(arg0) {
+    const ret = getObject(arg0);
+    return addHeapObject(ret);
+};
+
+export function __wbg_new_1e7c00339420672b() {
+    const ret = new Object();
+    return addHeapObject(ret);
+};
+
+export function __wbindgen_number_new(arg0) {
+    const ret = arg0;
+    return addHeapObject(ret);
+};
+
+export function __wbg_set_1754fb90457a8cce(arg0, arg1, arg2) {
+    getObject(arg0)[takeObject(arg1)] = takeObject(arg2);
+};
+
+export function __wbg_new_b44ab9ef6060dd36(arg0, arg1) {
+    const ret = new Error(getStringFromWasm0(arg0, arg1));
+    return addHeapObject(ret);
+};
+
+export function __wbindgen_debug_string(arg0, arg1) {
+    const ret = debugString(getObject(arg1));
+    const ptr1 = passStringToWasm0(ret, wasm.__wbindgen_malloc, wasm.__wbindgen_realloc);
+    const len1 = WASM_VECTOR_LEN;
+    getDataViewMemory0().setInt32(arg0 + 4 * 1, len1, true);
+    getDataViewMemory0().setInt32(arg0 + 4 * 0, ptr1, true);
+};
+
+export function __wbindgen_throw(arg0, arg1) {
+    throw new Error(getStringFromWasm0(arg0, arg1));
+};
+

libs/jieba-wasm/jieba_rs_wasm_bg.wasm.d.ts

@@ -0,0 +1,25 @@
+/* tslint:disable */
+/* eslint-disable */
+export const memory: WebAssembly.Memory;
+export const cut: (a: number, b: number, c: number) => [number, number];
+export const cut_all: (a: number, b: number) => [number, number];
+export const cut_for_search: (a: number, b: number, c: number) => [number, number];
+export const tokenize: (a: number, b: number, c: number, d: number, e: number) => [number, number, number, number];
+export const add_word: (a: number, b: number, c: number, d: number, e: number) => number;
+export const tag: (a: number, b: number, c: number) => [number, number];
+export const with_dict: (a: number, b: number) => [number, number];
+export const rust_zstd_wasm_shim_qsort: (a: number, b: number, c: number, d: number) => void;
+export const rust_zstd_wasm_shim_malloc: (a: number) => number;
+export const rust_zstd_wasm_shim_memcmp: (a: number, b: number, c: number) => number;
+export const rust_zstd_wasm_shim_calloc: (a: number, b: number) => number;
+export const rust_zstd_wasm_shim_free: (a: number) => void;
+export const rust_zstd_wasm_shim_memcpy: (a: number, b: number, c: number) => number;
+export const rust_zstd_wasm_shim_memmove: (a: number, b: number, c: number) => number;
+export const rust_zstd_wasm_shim_memset: (a: number, b: number, c: number) => number;
+export const __wbindgen_malloc: (a: number, b: number) => number;
+export const __wbindgen_realloc: (a: number, b: number, c: number, d: number) => number;
+export const __wbindgen_export_2: WebAssembly.Table;
+export const __externref_drop_slice: (a: number, b: number) => void;
+export const __wbindgen_free: (a: number, b: number, c: number) => void;
+export const __externref_table_dealloc: (a: number) => void;
+export const __wbindgen_start: () => void;

modules/story-summary/generate/generator.js

@@ -14,6 +14,12 @@ const MAX_CAUSED_BY = 2;
 // factUpdates 清洗
 // ═══════════════════════════════════════════════════════════════════════════
 
+function normalizeRelationPredicate(p) {
+    if (/^对.+的看法$/.test(p)) return p;
+    if (/^与.+的关系$/.test(p)) return p;
+    return null;
+}
+
 function sanitizeFacts(parsed) {
     if (!parsed) return;
 
@@ -22,23 +28,25 @@ function sanitizeFacts(parsed) {
 
     for (const item of updates) {
         const s = String(item?.s || '').trim();
-        const p = String(item?.p || '').trim();
+        const pRaw = String(item?.p || '').trim();
 
-        if (!s || !p) continue;
+        if (!s || !pRaw) continue;
 
         // 删除操作
         if (item.retracted === true) {
-            ok.push({ s, p, retracted: true });
+            ok.push({ s, p: pRaw, retracted: true });
             continue;
         }
 
         const o = String(item?.o || '').trim();
         if (!o) continue;
 
-        const fact = { s, p, o };
+        const relP = normalizeRelationPredicate(pRaw);
+        const isRel = !!relP;
+        const fact = { s, p: isRel ? relP : pRaw, o };
 
         // 关系类保留 trend
-        if (/^对.+的/.test(p) && item.trend) {
+        if (isRel && item.trend) {
             const validTrends = ['破裂', '厌恶', '反感', '陌生', '投缘', '亲密', '交融'];
             if (validTrends.includes(item.trend)) {
                 fact.trend = item.trend;

modules/story-summary/generate/llm.js

@@ -102,7 +102,8 @@ Acknowledged. Now reviewing the incremental summarization specifications:
 
 [Fact Tracking - SPO Triples]
 ├─ s: 主体（角色名/物品名）
-├─ p: 谓词（属性名/对X的看法）
+├─ p: 谓词（属性名）
+│   - 关系类只允许：对X的看法 / 与X的关系
 ├─ o: 值（当前状态）
 ├─ trend: 仅关系类填写
 ├─ retracted: 删除标记
@@ -191,7 +192,7 @@ Before generating, observe the USER and analyze carefully:
 ## factUpdates 规则
 - s+p 为键，相同键会覆盖旧值
 - 状态类：s=角色名, p=属性(生死/位置/状态等), o=值
-- 关系类：s=角色A, p="对B的看法", o=描述, trend=趋势
+- 关系类：s=角色A, p="对B的看法" 或 p="与B的关系"，trend 仅限关系类
 - 删除：设置 retracted: true（不需要填 o）
 - 只输出有变化的条目
 - 硬约束才记录，避免叙事化，确保少、硬、稳定

modules/story-summary/story-summary.js

@@ -847,29 +847,32 @@ async function sendFrameBaseData(store, totalFloors) {
 }
 
 function sendFrameFullData(store, totalFloors) {
-    const lastSummarized = store?.lastSummarizedMesId ?? -1;
     if (store?.json) {
-        const facts = store.json.facts || [];
-        const relationships = extractRelationshipsFromFacts(facts);
         postToFrame({
             type: "SUMMARY_FULL_DATA",
-            payload: {
+            payload: buildFramePayload(store),
-                keywords: store.json.keywords || [],
-                events: store.json.events || [],
-                characters: {
-                    main: store.json.characters?.main || [],
-                    relationships,
-                },
-                arcs: store.json.arcs || [],
-                facts,
-                lastSummarizedMesId: lastSummarized,
-            },
         });
     } else {
         postToFrame({ type: "SUMMARY_CLEARED", payload: { totalFloors } });
     }
 }
 
+function buildFramePayload(store) {
+    const json = store?.json || {};
+    const facts = json.facts || [];
+    return {
+        keywords: json.keywords || [],
+        events: json.events || [],
+        characters: {
+            main: json.characters?.main || [],
+            relationships: extractRelationshipsFromFacts(facts),
+        },
+        arcs: json.arcs || [],
+        facts,
+        lastSummarizedMesId: store?.lastSummarizedMesId ?? -1,
+    };
+}
+
 function openPanelForMessage(mesId) {
     createOverlay();
     showOverlay();
@@ -979,20 +982,11 @@ async function autoRunSummaryWithRetry(targetMesId, configForRun) {
             onStatus: (text) => postToFrame({ type: "SUMMARY_STATUS", statusText: text }),
             onError: (msg) => postToFrame({ type: "SUMMARY_ERROR", message: msg }),
             onComplete: async ({ merged, endMesId, newEventIds }) => {
-                postToFrame({
+                const store = getSummaryStore();
-                    type: "SUMMARY_FULL_DATA",
+                postToFrame({ type: "SUMMARY_FULL_DATA", payload: buildFramePayload(store) });
-                    payload: {
-                        keywords: merged.keywords || [],
-                        events: merged.events || [],
-                        characters: merged.characters || { main: [], relationships: [] },
-                        arcs: merged.arcs || [],
-                        world: merged.world || [],
-                        lastSummarizedMesId: endMesId,
-                    },
-                });
 
                 applyHideStateDebounced();
-                updateFrameStatsAfterSummary(endMesId, merged);
+                updateFrameStatsAfterSummary(endMesId, store.json || {});
 
                 // L2 自动增量向量化
                 await autoVectorizeNewEvents(newEventIds);
@@ -1279,20 +1273,11 @@ async function handleManualGenerate(mesId, config) {
         onStatus: (text) => postToFrame({ type: "SUMMARY_STATUS", statusText: text }),
         onError: (msg) => postToFrame({ type: "SUMMARY_ERROR", message: msg }),
         onComplete: async ({ merged, endMesId, newEventIds }) => {
-            postToFrame({
+            const store = getSummaryStore();
-                type: "SUMMARY_FULL_DATA",
+            postToFrame({ type: "SUMMARY_FULL_DATA", payload: buildFramePayload(store) });
-                payload: {
-                    keywords: merged.keywords || [],
-                    events: merged.events || [],
-                    characters: merged.characters || { main: [], relationships: [] },
-                    arcs: merged.arcs || [],
-                    world: merged.world || [],
-                    lastSummarizedMesId: endMesId,
-                },
-            });
 
             applyHideStateDebounced();
-            updateFrameStatsAfterSummary(endMesId, merged);
+            updateFrameStatsAfterSummary(endMesId, store.json || {});
 
             // L2 自动增量向量化
             await autoVectorizeNewEvents(newEventIds);

modules/story-summary/vector/recall.js

@@ -21,6 +21,10 @@ import {
     mergeAndSparsify,
 } from './state-recall.js';
 import { ensureEventTextIndex, searchEventsByText } from './text-search.js';
+import {
+    extractRareTerms,
+    extractNounsFromFactsO,
+} from './tokenizer.js';
 
 const MODULE_ID = 'recall';
 
@@ -50,6 +54,10 @@ const CONFIG = {
 
     RRF_K: 60,
     TEXT_SEARCH_LIMIT: 80,
+
+    // TEXT-only 质量控制
+    TEXT_SOFT_MIN_SIM: 0.50,
+    TEXT_TOTAL_MAX: 6,
 };
 
 // ═══════════════════════════════════════════════════════════════════════════
@@ -83,7 +91,7 @@ function fuseEventsByRRF(vectorRanked, textRanked, eventById, k = CONFIG.RRF_K)
 
     const upsert = (id) => {
         if (!map.has(id)) {
-            map.set(id, { id, rrf: 0, vRank: Infinity, tRank: Infinity, type: 'TEXT' });
+            map.set(id, { id, rrf: 0, vRank: Infinity, tRank: Infinity, type: 'TEXT', rawSim: 0, vector: null });
         }
         return map.get(id);
     };
@@ -96,6 +104,7 @@ function fuseEventsByRRF(vectorRanked, textRanked, eventById, k = CONFIG.RRF_K)
         o.rrf += 1 / (k + i + 1);
         o.type = o.tRank !== Infinity ? 'HYBRID' : 'VECTOR';
         o.vector = r.vector;
+        o.rawSim = r.rawSim || 0;
     });
 
     textRanked.forEach((r) => {
@@ -381,6 +390,70 @@ function normalizeEntityWeights(queryEntityWeights) {
     return normalized;
 }
 
+// ═══════════════════════════════════════════════════════════════════════════
+// 文本路 Query 构建（分层高信号词）
+// ═══════════════════════════════════════════════════════════════════════════
+
+async function buildTextSearchQuery(segments, queryEntities, facts, expandedTerms) {
+    const breakdown = {
+        entities: [],
+        rareTerms: [],
+        factsO: [],
+        expanded: [],
+    };
+
+    breakdown.entities = [...(queryEntities || [])];
+
+    const q2Segments = segments.slice(-2);
+    const q2Text = q2Segments.join(' ');
+
+    try {
+        breakdown.rareTerms = await extractRareTerms(q2Text, 15);
+    } catch (e) {
+        xbLog.warn(MODULE_ID, '稀有词提取失败', e);
+        breakdown.rareTerms = [];
+    }
+
+    const entitySet = new Set(breakdown.entities.map(e => e.toLowerCase()));
+    breakdown.rareTerms = breakdown.rareTerms.filter(t => !entitySet.has(t.toLowerCase()));
+
+    const relevantSubjects = new Set(queryEntities || []);
+    try {
+        breakdown.factsO = await extractNounsFromFactsO(facts, relevantSubjects, 5);
+    } catch (e) {
+        xbLog.warn(MODULE_ID, 'facts O 提取失败', e);
+        breakdown.factsO = [];
+    }
+
+    const existingSet = new Set([
+        ...breakdown.entities,
+        ...breakdown.rareTerms,
+    ].map(e => e.toLowerCase()));
+    breakdown.factsO = breakdown.factsO.filter(t => !existingSet.has(t.toLowerCase()));
+
+    const allExistingSet = new Set([
+        ...breakdown.entities,
+        ...breakdown.rareTerms,
+        ...breakdown.factsO,
+    ].map(e => e.toLowerCase()));
+
+    breakdown.expanded = (expandedTerms || [])
+        .filter(t => !allExistingSet.has(t.toLowerCase()))
+        .slice(0, 3);
+
+    const queryParts = [
+        ...breakdown.entities,
+        ...breakdown.entities,
+        ...breakdown.rareTerms,
+        ...breakdown.factsO,
+        ...breakdown.expanded,
+    ];
+
+    const query = queryParts.join(' ');
+
+    return { query, breakdown };
+}
+
 function stripFloorTag(s) {
     return String(s || '').replace(/\s*\(#\d+(?:-\d+)?\)\s*$/, '').trim();
 }
@@ -605,7 +678,7 @@ async function searchEvents(queryVector, queryTextForSearch, allEvents, vectorCo
 
     const scored = (allEvents || []).map((event, idx) => {
         const v = vectorMap.get(event.id);
-        const sim = v ? cosineSimilarity(queryVector, v) : 0;
+        const rawSim = v ? cosineSimilarity(queryVector, v) : 0;
 
         let bonus = 0;
 
@@ -635,14 +708,15 @@ async function searchEvents(queryVector, queryTextForSearch, allEvents, vectorCo
             _id: event.id,
             _idx: idx,
             event,
-            similarity: sim,
+            rawSim,
-            finalScore: sim + bonus,
+            finalScore: rawSim + bonus,
             vector: v,
             _entityBonus: entityBonus,
             _hasPresent: maxEntityWeight > 0,
         };
     });
 
+    const rawSimById = new Map(scored.map(s => [s._id, s.rawSim]));
     const entityBonusById = new Map(scored.map(s => [s._id, s._entityBonus]));
     const hasPresentById = new Map(scored.map(s => [s._id, s._hasPresent]));
 
@@ -665,14 +739,34 @@ async function searchEvents(queryVector, queryTextForSearch, allEvents, vectorCo
     const vectorRanked = candidates.map(s => ({
         event: s.event,
         similarity: s.finalScore,
+        rawSim: s.rawSim,
         vector: s.vector,
     }));
 
     const eventById = new Map(allEvents.map(e => [e.id, e]));
     const fused = fuseEventsByRRF(vectorRanked, textRanked, eventById);
 
-    const hasVector = vectorRanked.length > 0;
+    const textOnlyStats = {
-    const filtered = hasVector ? fused.filter(x => x.type !== 'TEXT') : fused;
+        total: 0,
+        passedSoftCheck: 0,
+        filtered: 0,
+        finalIncluded: 0,
+        truncatedByLimit: 0,
+    };
+
+    const filtered = fused.filter(x => {
+        if (x.type !== 'TEXT') return true;
+
+        textOnlyStats.total++;
+        const sim = x.rawSim || rawSimById.get(x.id) || 0;
+        if (sim >= CONFIG.TEXT_SOFT_MIN_SIM) {
+            textOnlyStats.passedSoftCheck++;
+            return true;
+        }
+
+        textOnlyStats.filtered++;
+        return false;
+    });
 
     const mmrInput = filtered.slice(0, CONFIG.CANDIDATE_EVENTS).map(x => ({
         ...x,
@@ -686,14 +780,27 @@ async function searchEvents(queryVector, queryTextForSearch, allEvents, vectorCo
         c => c.vector || null,
         c => c.rrf
     );
-    // 构造结果
+
-    const results = mmrOutput.map(x => ({
+    let textOnlyCount = 0;
+    const finalResults = mmrOutput.filter(x => {
+        if (x.type !== 'TEXT') return true;
+        if (textOnlyCount < CONFIG.TEXT_TOTAL_MAX) {
+            textOnlyCount++;
+            return true;
+        }
+        textOnlyStats.truncatedByLimit++;
+        return false;
+    });
+    textOnlyStats.finalIncluded = textOnlyCount;
+
+    const results = finalResults.map(x => ({
         event: x.event,
         similarity: x.rrf,
         _recallType: hasPresentById.get(x.event?.id) ? 'DIRECT' : 'SIMILAR',
         _recallReason: x.type,
         _rrfDetail: { vRank: x.vRank, tRank: x.tRank, rrf: x.rrf },
         _entityBonus: entityBonusById.get(x.event?.id) || 0,
+        _rawSim: rawSimById.get(x.event?.id) || 0,
     }));
 
     // 统计信息附加到第一条结果
@@ -704,8 +811,9 @@ async function searchEvents(queryVector, queryTextForSearch, allEvents, vectorCo
             textCount: textRanked.length,
             hybridCount: fused.filter(x => x.type === 'HYBRID').length,
             vectorOnlyCount: fused.filter(x => x.type === 'VECTOR').length,
-            textOnlyFiltered: fused.filter(x => x.type === 'TEXT').length,
+            textOnlyTotal: textOnlyStats.total,
         };
+        results[0]._textOnlyStats = textOnlyStats;
         results[0]._textGapInfo = textGapInfo;
     }
 
@@ -729,6 +837,7 @@ function formatRecallLog({
     l0Results = [],
     textGapInfo = null,
     expandedTerms = [],
+    textQueryBreakdown = null,
 }) {
     const lines = [
         '\u2554' + '\u2550'.repeat(62) + '\u2557',
@@ -775,6 +884,40 @@ function formatRecallLog({
         lines.push(`  扩散: ${expandedTerms.join('、')}`);
     }
 
+    lines.push('');
+    lines.push('\u250c' + '\u2500'.repeat(61) + '\u2510');
+    lines.push('\u2502 【文本路 Query 构成】                                        \u2502');
+    lines.push('\u2514' + '\u2500'.repeat(61) + '\u2518');
+
+    if (textQueryBreakdown) {
+        const bd = textQueryBreakdown;
+        if (bd.entities?.length) {
+            lines.push(`  强信号-实体 (${bd.entities.length}): ${bd.entities.slice(0, 8).join(' | ')}${bd.entities.length > 8 ? ' ...' : ''}`);
+        } else {
+            lines.push('  强信号-实体: (无)');
+        }
+
+        if (bd.rareTerms?.length) {
+            lines.push(`  强信号-稀有词 (${bd.rareTerms.length}): ${bd.rareTerms.slice(0, 10).join(' | ')}${bd.rareTerms.length > 10 ? ' ...' : ''}`);
+        } else {
+            lines.push('  强信号-稀有词: (无)');
+        }
+
+        if (bd.factsO?.length) {
+            lines.push(`  中信号-facts O (${bd.factsO.length}): ${bd.factsO.join(' | ')}`);
+        } else {
+            lines.push('  中信号-facts O: (无)');
+        }
+
+        if (bd.expanded?.length) {
+            lines.push(`  背景扩展 (${bd.expanded.length}): ${bd.expanded.join(' | ')}`);
+        } else {
+            lines.push('  背景扩展: (无)');
+        }
+    } else {
+        lines.push('  (降级模式，无分层信息)');
+    }
+
     lines.push('');
     lines.push('  实体归一化（用于加分）:');
     if (normalizedEntityWeights?.size) {
@@ -816,13 +959,37 @@ function formatRecallLog({
 
     // L2
     const rrfStats = eventResults[0]?._rrfStats || {};
+    const textOnlyStats = eventResults[0]?._textOnlyStats || {};
 
     lines.push('');
     lines.push('  L2 事件记忆 (RRF 混合检索):');
     lines.push(`    总事件: ${allEvents.length} 条 | 最终: ${eventResults.length} 条`);
     lines.push(`    向量路: ${rrfStats.vectorCount || 0} 条 | 文本路: ${rrfStats.textCount || 0} 条`);
-    lines.push(`    HYBRID: ${rrfStats.hybridCount || 0} 条 | 纯 VECTOR: ${rrfStats.vectorOnlyCount || 0} 条 | 纯 TEXT (已过滤): ${rrfStats.textOnlyFiltered || 0} 条`);
+    lines.push(`    HYBRID: ${rrfStats.hybridCount || 0} 条 | 纯 VECTOR: ${rrfStats.vectorOnlyCount || 0} 条`);
+
+    lines.push('');
+    lines.push('  TEXT-only 质量控制:');
+    lines.push(`    候选: ${textOnlyStats.total || 0} 条`);
+    lines.push(`    通过软校验 (sim>=${CONFIG.TEXT_SOFT_MIN_SIM}): ${textOnlyStats.passedSoftCheck || 0} 条`);
+    lines.push(`    语义过滤: ${textOnlyStats.filtered || 0} 条`);
+    lines.push(`    限额截断 (max=${CONFIG.TEXT_TOTAL_MAX}): ${textOnlyStats.truncatedByLimit || 0} 条`);
+    lines.push(`    最终入选: ${textOnlyStats.finalIncluded || 0} 条`);
+
+    const textOnlyEvents = eventResults.filter(e => e._recallReason === 'TEXT');
+    if (textOnlyEvents.length > 0) {
+        lines.push('');
+        lines.push('  TEXT-only 入选事件:');
+        textOnlyEvents.forEach((e, i) => {
+            const ev = e.event || {};
+            const id = ev.id || '?';
+            const title = (ev.title || '').slice(0, 25) || '(无标题)';
+            const sim = (e._rawSim || 0).toFixed(2);
+            const tRank = e._rrfDetail?.tRank ?? '?';
+            lines.push(`    ${i + 1}. [${id}] ${title.padEnd(25)} sim=${sim} tRank=${tRank}`);
+        });
+    }
     const entityBoostedEvents = eventResults.filter(e => e._entityBonus > 0).length;
+    lines.push('');
     lines.push(`    实体加分事件: ${entityBoostedEvents} 条`);
 
     if (textGapInfo) {
@@ -886,14 +1053,17 @@ export async function recallMemory(queryText, allEvents, vectorConfig, options =
     const expandedTerms = expandByFacts(queryEntities, facts, 2);
     const normalizedEntityWeights = normalizeEntityWeights(queryEntityWeights);
 
-    // 构建文本查询串：最后一条消息 + 实体 + 关键词
+    let queryTextForSearch = '';
-    const lastSeg = segments[segments.length - 1] || '';
+    let textQueryBreakdown = null;
-    const queryTextForSearch = [
+    try {
-        lastSeg,
+        const result = await buildTextSearchQuery(segments, queryEntities, facts, expandedTerms);
-        ...queryEntities,
+        queryTextForSearch = result.query;
-        ...expandedTerms,
+        textQueryBreakdown = result.breakdown;
-        ...(store?.json?.keywords || []).slice(0, 5).map(k => k.text),
+    } catch (e) {
-    ].join(' ');
+        xbLog.warn(MODULE_ID, '文本路 Query 构建失败，降级到简单拼接', e);
+        const lastSeg = segments[segments.length - 1] || '';
+        queryTextForSearch = [lastSeg, ...queryEntities, ...expandedTerms].join(' ');
+    }
 
     // L0 召回
     let l0Results = [];
@@ -954,6 +1124,7 @@ export async function recallMemory(queryText, allEvents, vectorConfig, options =
         l0Results,
         textGapInfo,
         expandedTerms,
+        textQueryBreakdown,
     });
 
     console.group('%c[Recall]', 'color: #7c3aed; font-weight: bold');

modules/story-summary/vector/tokenizer.js

@@ -0,0 +1,287 @@
+import { xbLog } from '../../../core/debug-core.js';
+import { extensionFolderPath } from '../../../core/constants.js';
+
+const MODULE_ID = 'tokenizer';
+
+// ═══════════════════════════════════════════════════════════════════════════
+// 词性过滤
+// ═══════════════════════════════════════════════════════════════════════════
+
+// 保留的词性（名词类 + 英文）
+const KEEP_POS_PREFIXES = ['n', 'eng'];
+
+function shouldKeepByPos(pos) {
+    return KEEP_POS_PREFIXES.some(prefix => pos.startsWith(prefix));
+}
+
+// ═══════════════════════════════════════════════════════════════════════════
+// 语言检测
+// ═══════════════════════════════════════════════════════════════════════════
+
+function shouldUseJieba(text) {
+    const zh = (text.match(/[\u4e00-\u9fff]/g) || []).length;
+    return zh >= 5;
+}
+
+function detectMainLanguage(text) {
+    const zh = (text.match(/[\u4e00-\u9fff]/g) || []).length;
+    const jp = (text.match(/[\u3040-\u309f\u30a0-\u30ff]/g) || []).length;
+    const en = (text.match(/[a-zA-Z]/g) || []).length;
+    const total = zh + jp + en || 1;
+
+    if (jp / total > 0.2) return 'jp';
+    if (en / total > 0.5) return 'en';
+    return 'zh';
+}
+
+// 替换原有的大停用词表
+const STOP_WORDS = new Set([
+    // 系统词
+    '用户', '角色', '玩家', '旁白', 'user', 'assistant', 'system',
+    // 时间泛词
+    '时候', '现在', '今天', '明天', '昨天', '早上', '晚上',
+    // 方位泛词
+    '这里', '那里', '上面', '下面', '里面', '外面',
+    // 泛化名词
+    '东西', '事情', '事儿', '地方', '样子', '意思', '感觉',
+    '一下', '一些', '一点', '一会', '一次',
+]);
+
+// 英文停用词（fallback 用）
+const EN_STOP_WORDS = new Set([
+    'the', 'a', 'an', 'is', 'are', 'was', 'were', 'be', 'been',
+    'have', 'has', 'had', 'do', 'does', 'did', 'will', 'would',
+    'could', 'should', 'may', 'might', 'must', 'can',
+    'to', 'of', 'in', 'on', 'at', 'for', 'with', 'by', 'from',
+    'and', 'or', 'but', 'if', 'that', 'this', 'it', 'its',
+    'i', 'you', 'he', 'she', 'we', 'they',
+    'my', 'your', 'his', 'her', 'our', 'their',
+    'what', 'which', 'who', 'whom', 'where', 'when', 'why', 'how',
+]);
+
+let jiebaModule = null;
+let jiebaReady = false;
+let jiebaLoading = false;
+
+async function ensureJieba() {
+    if (jiebaReady) return true;
+    if (jiebaLoading) {
+        for (let i = 0; i < 50; i++) {
+            await new Promise(r => setTimeout(r, 100));
+            if (jiebaReady) return true;
+        }
+        return false;
+    }
+
+    jiebaLoading = true;
+
+    try {
+        const jiebaPath = `/${extensionFolderPath}/libs/jieba-wasm/jieba_rs_wasm.js`;
+        // eslint-disable-next-line no-unsanitized/method
+        jiebaModule = await import(jiebaPath);
+
+        if (jiebaModule.default) {
+            await jiebaModule.default();
+        }
+
+        jiebaReady = true;
+        xbLog.info(MODULE_ID, 'jieba-wasm 加载成功');
+        const keys = Object.getOwnPropertyNames(jiebaModule || {});
+        const dkeys = Object.getOwnPropertyNames(jiebaModule?.default || {});
+        xbLog.info(MODULE_ID, `jieba keys: ${keys.join(',')}`);
+        xbLog.info(MODULE_ID, `jieba default keys: ${dkeys.join(',')}`);
+        xbLog.info(MODULE_ID, `jieba.tag: ${typeof jiebaModule?.tag}`);
+        return true;
+    } catch (e) {
+        xbLog.error(MODULE_ID, 'jieba-wasm 加载失败', e);
+        jiebaLoading = false;
+        return false;
+    }
+}
+
+function fallbackTokenize(text) {
+    const tokens = [];
+    const lang = detectMainLanguage(text);
+
+    // 英文
+    const enMatches = text.match(/[a-zA-Z]{2,20}/gi) || [];
+    tokens.push(...enMatches.filter(w => !EN_STOP_WORDS.has(w.toLowerCase())));
+
+    // 日语假名
+    if (lang === 'jp') {
+        const kanaMatches = text.match(/[\u3040-\u309f\u30a0-\u30ff]{2,10}/g) || [];
+        tokens.push(...kanaMatches);
+    }
+
+    // 中文/日语汉字
+    const zhMatches = text.match(/[\u4e00-\u9fff]{2,6}/g) || [];
+    tokens.push(...zhMatches);
+
+    // 数字+汉字组合
+    const numZhMatches = text.match(/\d+[\u4e00-\u9fff]{1,4}/g) || [];
+    tokens.push(...numZhMatches);
+
+    return tokens;
+}
+
+export async function extractNouns(text, options = {}) {
+    const { minLen = 2, maxCount = 0 } = options;
+    if (!text?.trim()) return [];
+
+    // 中文为主 → 用 jieba
+    if (shouldUseJieba(text)) {
+        const hasJieba = await ensureJieba();
+
+        if (hasJieba && jiebaModule?.tag) {
+            try {
+                const tagged = jiebaModule.tag(text, true);
+
+                const result = [];
+                const seen = new Set();
+
+                const list = Array.isArray(tagged) ? tagged : [];
+                for (const item of list) {
+                    let word = '';
+                    let pos = '';
+                    if (Array.isArray(item)) {
+                        [word, pos] = item;
+                    } else if (item && typeof item === 'object') {
+                        word = item.word || item.w || item.text || item.term || '';
+                        pos = item.tag || item.pos || item.p || '';
+                    }
+                    if (!word || !pos) continue;
+                    if (word.length < minLen) continue;
+                    if (!shouldKeepByPos(pos)) continue;
+                    if (STOP_WORDS.has(word)) continue;
+                    if (seen.has(word)) continue;
+
+                    seen.add(word);
+                    result.push(word);
+
+                    if (maxCount > 0 && result.length >= maxCount) break;
+                }
+
+                return result;
+            } catch (e) {
+                xbLog.warn(MODULE_ID, 'jieba tag 失败：' + (e && e.message ? e.message : String(e)));
+            }
+        }
+    }
+
+    // 非中文 / jieba 失败 → fallback
+    const tokens = fallbackTokenize(text);
+
+    const result = [];
+    const seen = new Set();
+
+    for (const t of tokens) {
+        if (t.length < minLen) continue;
+        if (STOP_WORDS.has(t)) continue;
+        if (seen.has(t)) continue;
+
+        seen.add(t);
+        result.push(t);
+
+        if (maxCount > 0 && result.length >= maxCount) break;
+    }
+
+    return result;
+}
+
+export async function extractRareTerms(text, maxCount = 15) {
+    if (!text?.trim()) return [];
+
+    // 中文为主 → 用 jieba
+    if (shouldUseJieba(text)) {
+        const hasJieba = await ensureJieba();
+
+        if (hasJieba && jiebaModule?.tag) {
+            try {
+                const tagged = jiebaModule.tag(text, true);
+
+                const candidates = [];
+                const seen = new Set();
+
+                const list = Array.isArray(tagged) ? tagged : [];
+                for (const item of list) {
+                    let word = '';
+                    let pos = '';
+                    if (Array.isArray(item)) {
+                        [word, pos] = item;
+                    } else if (item && typeof item === 'object') {
+                        word = item.word || item.w || item.text || item.term || '';
+                        pos = item.tag || item.pos || item.p || '';
+                    }
+                    if (!word || !pos) continue;
+                    if (word.length < 2) continue;
+                    if (!shouldKeepByPos(pos)) continue;
+                    if (STOP_WORDS.has(word)) continue;
+                    if (seen.has(word)) continue;
+
+                    seen.add(word);
+
+                    // 稀有度评分
+                    let score = 0;
+                    if (word.length >= 4) score += 3;
+                    else if (word.length >= 3) score += 1;
+                    if (/[a-zA-Z]/.test(word)) score += 2;
+                    if (/\d/.test(word)) score += 1;
+                    // 专名词性加分
+                    if (['nr', 'ns', 'nt', 'nz'].some(p => pos.startsWith(p))) score += 2;
+
+                    candidates.push({ term: word, score });
+                }
+
+                candidates.sort((a, b) => b.score - a.score);
+                return candidates.slice(0, maxCount).map(x => x.term);
+            } catch (e) {
+                xbLog.warn(MODULE_ID, 'jieba tag 失败：' + (e && e.message ? e.message : String(e)));
+            }
+        }
+    }
+
+    // 非中文 / jieba 失败 → fallback
+    const allNouns = await extractNouns(text, { minLen: 2, maxCount: 0 });
+
+    const scored = allNouns.map(t => {
+        let score = 0;
+        if (t.length >= 4) score += 3;
+        else if (t.length >= 3) score += 1;
+        if (/[a-zA-Z]/.test(t)) score += 2;
+        if (/\d/.test(t)) score += 1;
+        return { term: t, score };
+    });
+
+    scored.sort((a, b) => b.score - a.score);
+    return scored.slice(0, maxCount).map(x => x.term);
+}
+
+export async function extractNounsFromFactsO(facts, relevantSubjects, maxCount = 5) {
+    if (!facts?.length || !relevantSubjects?.size) return [];
+
+    const oTexts = [];
+
+    for (const f of facts) {
+        if (f.retracted) continue;
+
+        // 只取相关主体的 facts
+        const s = String(f.s || '').trim();
+        if (!relevantSubjects.has(s)) continue;
+
+        const o = String(f.o || '').trim();
+        if (!o) continue;
+
+        // 跳过太长的 O（可能是完整句子）
+        if (o.length > 30) continue;
+
+        oTexts.push(o);
+    }
+
+    if (!oTexts.length) return [];
+
+    const combined = oTexts.join(' ');
+    return await extractNouns(combined, { minLen: 2, maxCount });
+}
+
+export { ensureJieba };
+