@dan-uni/dan-any-plugin-detaolu 1.3.9 → 1.4.0

package/dist/index.js

@@ -2442,55 +2442,57 @@ const WIDTH_TABLE = new Map(Object.entries({
     Ｎ: 'N',
     Ｍ: 'M'
 }));
-const detaolu = (inp, config)=>{
+function detaolu_meta(config) {
     const TRIM_ENDING = config.TRIM_ENDING;
     const TRIM_SPACE = config.TRIM_SPACE;
     const TRIM_WIDTH = config.TRIM_WIDTH;
-    const FORCELIST = (config?.FORCELIST ?? DEFAULT_CONFIG.FORCELIST).map(([pattern, repl])=>[
+    const FORCELIST = config.FORCELIST.map(([pattern, repl])=>[
             new RegExp(pattern, 'giu'),
             repl
         ]);
     const FORCELIST_BREAK_ON_MATCH = !config.FORCELIST_CONTINUE_ON_MATCH;
-    let len = inp.length;
-    let text = '';
-    if (TRIM_ENDING) {
-        while(ENDING_CHARS.has(inp.charAt(len - 1)))len--;
-        if (0 === len) len = inp.length;
-    }
-    if (TRIM_WIDTH) for(let i = 0; i < len; i++){
-        const c = inp.charAt(i);
-        text += WIDTH_TABLE.get(c) || c;
-    }
-    else text = inp.slice(0, len);
-    if (TRIM_SPACE) text = text.replaceAll(/[ \u3000]+/g, ' ').replaceAll(/([\u3000-\u9FFF\uFF00-\uFFEF]) (?=[\u3000-\u9FFF\uFF00-\uFFEF])/g, '$1');
-    let taolu_matched = false;
-    for (const taolu of FORCELIST)if (taolu[0].test(text)) {
-        text = text.replace(taolu[0], taolu[1]);
-        taolu_matched = true;
-        if (FORCELIST_BREAK_ON_MATCH) break;
-    }
-    return [
-        taolu_matched,
-        text
-    ];
-};
-const whitelisted = (text, config)=>{
-    const WHITELIST = (config?.WHITELIST ?? DEFAULT_CONFIG.WHITELIST).map((x)=>new RegExp(x[0], 'iu'));
-    if (0 === WHITELIST.length) return false;
-    return WHITELIST.some((re)=>re.test(text));
-};
-const blacklisted = (text, config)=>{
-    const BLACKLIST = (config?.BLACKLIST ?? DEFAULT_CONFIG.BLACKLIST).map((x)=>x[0] ? new RegExp(x[1]) : x[1].toLowerCase());
-    if (0 === BLACKLIST.length) return null;
-    {
+    return (inp)=>{
+        let len = inp.length;
+        let text = '';
+        if (TRIM_ENDING) {
+            while(ENDING_CHARS.has(inp.charAt(len - 1)))len--;
+            if (0 === len) len = inp.length;
+        }
+        if (TRIM_WIDTH) for(let i = 0; i < len; i++){
+            const c = inp.charAt(i);
+            text += WIDTH_TABLE.get(c) || c;
+        }
+        else text = inp.slice(0, len);
+        if (TRIM_SPACE) text = text.replaceAll(/[ \u3000]+/g, ' ').replaceAll(/([\u3000-\u9FFF\uFF00-\uFFEF]) (?=[\u3000-\u9FFF\uFF00-\uFFEF])/g, '$1');
+        let taolu_matched = false;
+        for (const taolu of FORCELIST)if (taolu[0].test(text)) {
+            text = text.replace(taolu[0], taolu[1]);
+            taolu_matched = true;
+            if (FORCELIST_BREAK_ON_MATCH) break;
+        }
+        return [
+            taolu_matched,
+            text
+        ];
+    };
+}
+function whitelisted_meta(config) {
+    const WHITELIST = config.WHITELIST.map((x)=>new RegExp(x[0], 'iu'));
+    if (0 === WHITELIST.length) return ()=>false;
+    return (text)=>WHITELIST.some((re)=>re.test(text));
+}
+function blacklisted_meta(config) {
+    const BLACKLIST = config.BLACKLIST.map((x)=>x[0] ? new RegExp(x[1]) : x[1].toLowerCase());
+    if (0 === BLACKLIST.length) return ()=>null;
+    return (text)=>{
         const lower = text.toLowerCase();
         for (const pattern of BLACKLIST){
             const matched = 'string' == typeof pattern ? lower.includes(pattern) : pattern.test(text);
             if (matched) return 'string' == typeof pattern ? ` ${pattern}` : ` /${pattern.source}/`;
         }
         return null;
-    }
-};
+    };
+}
 function extract_special_danmu(text) {
     try {
         text = JSON.parse(text)[4];
@@ -2506,15 +2508,25 @@ function select_median_length(strs) {
     const mid = Math.floor(sorted.length / 2);
     return sorted[mid];
 }
+function u8array_to_arraybuffer(array) {
+    return array.buffer.slice(array.byteOffset, array.byteOffset + array.byteLength);
+}
 async function load_wasm(wasm_mod) {
-    await similarity_stub_init(wasm_mod ?? await lib_default().readFile(new URL('similarity-gen.wasm', import.meta.url)));
+    if (wasm_mod) return void await similarity_stub_init(wasm_mod);
+    const wasm_path = new URL('similarity-gen.wasm', import.meta.url).pathname;
+    const wasm_u8 = await lib_default().readFile(wasm_path);
+    await similarity_stub_init(u8array_to_arraybuffer(wasm_u8));
 }
 function make_ptr_idx(idx, is_next_chunk) {
     return is_next_chunk ? -1 - idx : idx;
 }
 async function merge(chunk, config = DEFAULT_CONFIG) {
+    const local_config = {
+        ...DEFAULT_CONFIG,
+        ...config
+    };
     await load_wasm();
-    begin_chunk(config);
+    begin_chunk(local_config);
     const ret = {
         clusters: [],
         stats: new Stats(),
@@ -2579,23 +2591,26 @@ async function merge(chunk, config = DEFAULT_CONFIG) {
             });
         }
     }
+    const detaolu = detaolu_meta(local_config);
+    const whitelisted = whitelisted_meta(local_config);
+    const blacklisted = blacklisted_meta(local_config);
     function obj_to_ir(objs, s, is_next_chunk) {
         return objs.map((obj, idx)=>{
-            if (!config.PROC_POOL1 && 1 === obj.pool) {
+            if (!local_config.PROC_POOL1 && 1 === obj.pool) {
                 if (s) {
                     s.ignored_type++;
                     apply_single_cluster(idx, obj, '已忽略字幕弹幕，可以在选项中修改');
                 }
                 return null;
             }
-            if (!config.PROC_TYPE7 && 4 === obj.mode) {
+            if (!local_config.PROC_TYPE7 && 4 === obj.mode) {
                 if (s) {
                     s.ignored_type++;
                     apply_single_cluster(idx, obj, '已忽略特殊弹幕，可以在选项中修改');
                 }
                 return null;
             }
-            if (!config.PROC_TYPE4 && 1 === obj.mode) {
+            if (!local_config.PROC_TYPE4 && 1 === obj.mode) {
                 if (s) {
                     s.ignored_type++;
                     apply_single_cluster(idx, obj, '已忽略底部弹幕，可以在选项中修改');
@@ -2604,7 +2619,7 @@ async function merge(chunk, config = DEFAULT_CONFIG) {
             }
             const disp_str = trim_dispstr(4 === obj.mode && '[' === obj.content[0] ? extract_special_danmu(obj.content) : obj.content);
             if (4 !== obj.mode) {
-                const matched = blacklisted(disp_str, config);
+                const matched = blacklisted(disp_str);
                 if (matched) {
                     if (s) {
                         s.deleted_blacklist++;
@@ -2619,17 +2634,17 @@ async function merge(chunk, config = DEFAULT_CONFIG) {
                     return null;
                 }
             }
-            if (whitelisted(disp_str, config)) {
+            if (whitelisted(disp_str)) {
                 if (s) {
                     s.ignored_whitelist++;
                     apply_single_cluster(idx, obj, '命中白名单');
                 }
                 return null;
             }
-            const [matched_taolu, detaolued] = detaolu(disp_str, config);
+            const [matched_taolu, detaolued] = detaolu(disp_str);
             if (matched_taolu) {
                 if (s) s.num_taolu_matched++;
-                if (config.FORCELIST_APPLY_SINGULAR) obj = {
+                if (local_config.FORCELIST_APPLY_SINGULAR) obj = {
                     ...obj,
                     content: detaolued
                 };
@@ -2644,7 +2659,7 @@ async function merge(chunk, config = DEFAULT_CONFIG) {
     }
     const danmus = obj_to_ir(chunk.objs, ret.stats, false);
     const nearby_danmus = new Queue();
-    const THRESHOLD_MS = (config?.THRESHOLD ?? DEFAULT_CONFIG.THRESHOLD) * 1000;
+    const THRESHOLD_MS = 1000 * local_config.THRESHOLD;
     for (const dm of danmus){
         while(true){
             const peeked = nearby_danmus.peek();

package/dist/index.umd.min.js

@@ -12008,7 +12008,7 @@
             },
             "./src/pakku.js/similarity-gen.wasm" (module1, __unused_rspack_exports, __webpack_require__) {
                 "use strict";
-                module1.exports = __webpack_require__.p + "static/wasm/54a7637a81.module.wasm";
+                module1.exports = __webpack_require__.p + "static/wasm/21072e5de5.module.wasm";
             },
             "?efad" () {},
             "?956c" () {},
@@ -38576,7 +38576,7 @@ and limitations under the License.
                 PlatformVideoSource: ()=>platform_PlatformVideoSource,
                 PlatformVideoSources: ()=>PlatformVideoSources
             });
-            var package_namespaceObject = JSON.parse('{"UU":"@dan-uni/dan-any","rE":"1.3.9","TB":"https://github.com/ani-uni/danuni/tree/master/packages/dan-any#readme"}');
+            var package_namespaceObject = JSON.parse('{"UU":"@dan-uni/dan-any","rE":"1.4.0","TB":"https://github.com/ani-uni/danuni/tree/master/packages/dan-any#readme"}');
             const color_pad = (s)=>s.length < 2 ? `0${s}` : s;
             const decimalToHex = (n)=>color_pad(n.toString(16));
             const isDarkColor = ({ r, g, b })=>0.299 * r + 0.587 * g + 0.114 * b < 0x30;
@@ -39132,6 +39132,9 @@ and limitations under the License.
                         platform: platform_PlatformVideoSource.Bilibili,
                         extra: {
                             bili: {
+                                dmid: args.id,
+                                attr: args.attr,
+                                mid: args.mid,
                                 command: args
                             }
                         }
@@ -39865,7 +39868,7 @@ and limitations under the License.
             function main(that) {
                 that.dans.forEach((d)=>{
                     if (d.platform !== platform_PlatformVideoSource.Bilibili) throw new Error('bili-dedupe: 仅支持B站(主站)的弹幕');
-                    if (!d.extra.bili?.dmid) throw new Error('bili-dedupe: 弹幕缺少bili extra dmid字段');
+                    if (!d.extra.bili?.dmid && !d.extra.bili?.command?.id) throw new Error('bili-dedupe: 弹幕缺少bili extra dmid字段');
                 });
                 const map = new Map();
                 that.dans.forEach((d)=>map.set(d.extra.bili.dmid, d));
@@ -40416,6 +40419,8 @@ and limitations under the License.
                         const ok = this.dans.every((d)=>d.senderID.endsWith(`@${platform_PlatformVideoSource.Bilibili}`));
                         if (!ok) throw new Error('存在其他来源的senderID，请关闭该功能再试！');
                     }
+                    let ds = this.dans.map((dan)=>dan.toBiliXML(options));
+                    if (options?.skipBiliCommand) ds = ds.filter((d)=>null !== d);
                     const builder = new json2xml({
                         ignoreAttributes: false
                     });
@@ -40436,7 +40441,7 @@ and limitations under the License.
                                 ...DanUniConvertTipTemplate,
                                 data: this.getShared('SOID')
                             },
-                            d: this.dans.map((dan)=>dan.toBiliXML(options))
+                            d: ds
                         }
                     });
                 }
@@ -41089,55 +41094,57 @@ and limitations under the License.
                 Ｎ: 'N',
                 Ｍ: 'M'
             }));
-            const detaolu = (inp, config)=>{
+            function detaolu_meta(config) {
                 const TRIM_ENDING = config.TRIM_ENDING;
                 const TRIM_SPACE = config.TRIM_SPACE;
                 const TRIM_WIDTH = config.TRIM_WIDTH;
-                const FORCELIST = (config?.FORCELIST ?? DEFAULT_CONFIG.FORCELIST).map(([pattern, repl])=>[
+                const FORCELIST = config.FORCELIST.map(([pattern, repl])=>[
                         new RegExp(pattern, 'giu'),
                         repl
                     ]);
                 const FORCELIST_BREAK_ON_MATCH = !config.FORCELIST_CONTINUE_ON_MATCH;
-                let len = inp.length;
-                let text = '';
-                if (TRIM_ENDING) {
-                    while(ENDING_CHARS.has(inp.charAt(len - 1)))len--;
-                    if (0 === len) len = inp.length;
-                }
-                if (TRIM_WIDTH) for(let i = 0; i < len; i++){
-                    const c = inp.charAt(i);
-                    text += WIDTH_TABLE.get(c) || c;
-                }
-                else text = inp.slice(0, len);
-                if (TRIM_SPACE) text = text.replaceAll(/[ \u3000]+/g, ' ').replaceAll(/([\u3000-\u9FFF\uFF00-\uFFEF]) (?=[\u3000-\u9FFF\uFF00-\uFFEF])/g, '$1');
-                let taolu_matched = false;
-                for (const taolu of FORCELIST)if (taolu[0].test(text)) {
-                    text = text.replace(taolu[0], taolu[1]);
-                    taolu_matched = true;
-                    if (FORCELIST_BREAK_ON_MATCH) break;
-                }
-                return [
-                    taolu_matched,
-                    text
-                ];
-            };
-            const whitelisted = (text, config)=>{
-                const WHITELIST = (config?.WHITELIST ?? DEFAULT_CONFIG.WHITELIST).map((x)=>new RegExp(x[0], 'iu'));
-                if (0 === WHITELIST.length) return false;
-                return WHITELIST.some((re)=>re.test(text));
-            };
-            const blacklisted = (text, config)=>{
-                const BLACKLIST = (config?.BLACKLIST ?? DEFAULT_CONFIG.BLACKLIST).map((x)=>x[0] ? new RegExp(x[1]) : x[1].toLowerCase());
-                if (0 === BLACKLIST.length) return null;
-                {
+                return (inp)=>{
+                    let len = inp.length;
+                    let text = '';
+                    if (TRIM_ENDING) {
+                        while(ENDING_CHARS.has(inp.charAt(len - 1)))len--;
+                        if (0 === len) len = inp.length;
+                    }
+                    if (TRIM_WIDTH) for(let i = 0; i < len; i++){
+                        const c = inp.charAt(i);
+                        text += WIDTH_TABLE.get(c) || c;
+                    }
+                    else text = inp.slice(0, len);
+                    if (TRIM_SPACE) text = text.replaceAll(/[ \u3000]+/g, ' ').replaceAll(/([\u3000-\u9FFF\uFF00-\uFFEF]) (?=[\u3000-\u9FFF\uFF00-\uFFEF])/g, '$1');
+                    let taolu_matched = false;
+                    for (const taolu of FORCELIST)if (taolu[0].test(text)) {
+                        text = text.replace(taolu[0], taolu[1]);
+                        taolu_matched = true;
+                        if (FORCELIST_BREAK_ON_MATCH) break;
+                    }
+                    return [
+                        taolu_matched,
+                        text
+                    ];
+                };
+            }
+            function whitelisted_meta(config) {
+                const WHITELIST = config.WHITELIST.map((x)=>new RegExp(x[0], 'iu'));
+                if (0 === WHITELIST.length) return ()=>false;
+                return (text)=>WHITELIST.some((re)=>re.test(text));
+            }
+            function blacklisted_meta(config) {
+                const BLACKLIST = config.BLACKLIST.map((x)=>x[0] ? new RegExp(x[1]) : x[1].toLowerCase());
+                if (0 === BLACKLIST.length) return ()=>null;
+                return (text)=>{
                     const lower = text.toLowerCase();
                     for (const pattern of BLACKLIST){
                         const matched = 'string' == typeof pattern ? lower.includes(pattern) : pattern.test(text);
                         if (matched) return 'string' == typeof pattern ? ` ${pattern}` : ` /${pattern.source}/`;
                     }
                     return null;
-                }
-            };
+                };
+            }
             function extract_special_danmu(text) {
                 try {
                     text = JSON.parse(text)[4];
@@ -41153,15 +41160,25 @@ and limitations under the License.
                 const mid = Math.floor(sorted.length / 2);
                 return sorted[mid];
             }
+            function u8array_to_arraybuffer(array) {
+                return array.buffer.slice(array.byteOffset, array.byteOffset + array.byteLength);
+            }
             async function load_wasm(wasm_mod) {
-                await similarity_stub_init(wasm_mod ?? await lib_default().readFile(new URL(__webpack_require__("./src/pakku.js/similarity-gen.wasm"), __webpack_require__.b)));
+                if (wasm_mod) return void await similarity_stub_init(wasm_mod);
+                const wasm_path = new URL(__webpack_require__("./src/pakku.js/similarity-gen.wasm"), __webpack_require__.b).pathname;
+                const wasm_u8 = await lib_default().readFile(wasm_path);
+                await similarity_stub_init(u8array_to_arraybuffer(wasm_u8));
             }
             function make_ptr_idx(idx, is_next_chunk) {
                 return is_next_chunk ? -1 - idx : idx;
             }
             async function pakku_js_merge(chunk, config = DEFAULT_CONFIG) {
+                const local_config = {
+                    ...DEFAULT_CONFIG,
+                    ...config
+                };
                 await load_wasm();
-                begin_chunk(config);
+                begin_chunk(local_config);
                 const ret = {
                     clusters: [],
                     stats: new Stats(),
@@ -41226,23 +41243,26 @@ and limitations under the License.
                         });
                     }
                 }
+                const detaolu = detaolu_meta(local_config);
+                const whitelisted = whitelisted_meta(local_config);
+                const blacklisted = blacklisted_meta(local_config);
                 function obj_to_ir(objs, s, is_next_chunk) {
                     return objs.map((obj, idx)=>{
-                        if (!config.PROC_POOL1 && 1 === obj.pool) {
+                        if (!local_config.PROC_POOL1 && 1 === obj.pool) {
                             if (s) {
                                 s.ignored_type++;
                                 apply_single_cluster(idx, obj, '已忽略字幕弹幕，可以在选项中修改');
                             }
                             return null;
                         }
-                        if (!config.PROC_TYPE7 && 4 === obj.mode) {
+                        if (!local_config.PROC_TYPE7 && 4 === obj.mode) {
                             if (s) {
                                 s.ignored_type++;
                                 apply_single_cluster(idx, obj, '已忽略特殊弹幕，可以在选项中修改');
                             }
                             return null;
                         }
-                        if (!config.PROC_TYPE4 && 1 === obj.mode) {
+                        if (!local_config.PROC_TYPE4 && 1 === obj.mode) {
                             if (s) {
                                 s.ignored_type++;
                                 apply_single_cluster(idx, obj, '已忽略底部弹幕，可以在选项中修改');
@@ -41251,7 +41271,7 @@ and limitations under the License.
                         }
                         const disp_str = trim_dispstr(4 === obj.mode && '[' === obj.content[0] ? extract_special_danmu(obj.content) : obj.content);
                         if (4 !== obj.mode) {
-                            const matched = blacklisted(disp_str, config);
+                            const matched = blacklisted(disp_str);
                             if (matched) {
                                 if (s) {
                                     s.deleted_blacklist++;
@@ -41266,17 +41286,17 @@ and limitations under the License.
                                 return null;
                             }
                         }
-                        if (whitelisted(disp_str, config)) {
+                        if (whitelisted(disp_str)) {
                             if (s) {
                                 s.ignored_whitelist++;
                                 apply_single_cluster(idx, obj, '命中白名单');
                             }
                             return null;
                         }
-                        const [matched_taolu, detaolued] = detaolu(disp_str, config);
+                        const [matched_taolu, detaolued] = detaolu(disp_str);
                         if (matched_taolu) {
                             if (s) s.num_taolu_matched++;
-                            if (config.FORCELIST_APPLY_SINGULAR) obj = {
+                            if (local_config.FORCELIST_APPLY_SINGULAR) obj = {
                                 ...obj,
                                 content: detaolued
                             };
@@ -41291,7 +41311,7 @@ and limitations under the License.
                 }
                 const danmus = obj_to_ir(chunk.objs, ret.stats, false);
                 const nearby_danmus = new Queue();
-                const THRESHOLD_MS = (config?.THRESHOLD ?? DEFAULT_CONFIG.THRESHOLD) * 1000;
+                const THRESHOLD_MS = 1000 * local_config.THRESHOLD;
                 for (const dm of danmus){
                     while(true){
                         const peeked = nearby_danmus.peek();

package/dist/pakku.js/similarity_stub.d.ts

@@ -1,6 +1,6 @@
 import type { Config } from '.';
 import type { int, Stats } from './types';
-export declare function init(wasm_module: ArrayBuffer | NonSharedBuffer): Promise<void>;
+export declare function init(wasm_module: ArrayBuffer): Promise<void>;
 export declare function begin_chunk(config: Config): void;
 export declare function begin_index_lock(): void;
 export declare function detect_similarity(str: string, mode: number, index_l: int, S: Stats): null | {

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@dan-uni/dan-any-plugin-detaolu",
-  "version": "1.3.9",
+  "version": "1.4.0",
   "description": "A filter, dedupe and anti-spam plugin of dan-any, a danmaku transformer lib, based on pakku.js.",
   "author": "rinne",
   "license": "GPL-3.0-or-later",

package/src/pakku.js/index.ts

@@ -109,6 +109,7 @@ export const DEFAULT_CONFIG = {
 }
 
 export type Config = Partial<typeof DEFAULT_CONFIG>
+type ResolvedConfig = typeof DEFAULT_CONFIG
 
 interface DanmuIr {
   obj: DanmuObject
@@ -224,80 +225,88 @@ const WIDTH_TABLE = new Map(
 /**
  * 反套路
  */
-const detaolu = (inp: string, config: Config): [boolean, string] => {
+function detaolu_meta(
+  config: ResolvedConfig,
+): (text: string) => [boolean, string] {
   const TRIM_ENDING = config.TRIM_ENDING
   const TRIM_SPACE = config.TRIM_SPACE
   const TRIM_WIDTH = config.TRIM_WIDTH
-  const FORCELIST = (config?.FORCELIST ?? DEFAULT_CONFIG.FORCELIST).map(
+  const FORCELIST = config.FORCELIST.map(
     ([pattern, repl]) => [new RegExp(pattern, 'giu'), repl] as [RegExp, string],
   )
   const FORCELIST_BREAK_ON_MATCH = !config.FORCELIST_CONTINUE_ON_MATCH
 
-  let len = inp.length
-  let text = ''
-
-  if (TRIM_ENDING) {
-    while (ENDING_CHARS.has(inp.charAt(len - 1)))
-      // assert str.charAt(-1)===''
-      len--
-    if (len === 0)
-      // all chars are ending chars, do nothing
-      len = inp.length
-  }
+  return (inp: string) => {
+    let len = inp.length
+    let text = ''
+
+    if (TRIM_ENDING) {
+      while (ENDING_CHARS.has(inp.charAt(len - 1)))
+        // assert str.charAt(-1)===''
+        len--
+      if (len === 0)
+        // all chars are ending chars, do nothing
+        len = inp.length
+    }
 
-  if (TRIM_WIDTH) {
-    for (let i = 0; i < len; i++) {
-      const c = inp.charAt(i)
-      text += WIDTH_TABLE.get(c) || c
+    if (TRIM_WIDTH) {
+      for (let i = 0; i < len; i++) {
+        const c = inp.charAt(i)
+        text += WIDTH_TABLE.get(c) || c
+      }
+    } else {
+      text = inp.slice(0, len)
     }
-  } else {
-    text = inp.slice(0, len)
-  }
 
-  if (TRIM_SPACE) {
-    // text = text
-    //   .replace(TRIM_EXTRA_SPACE_RE, ' ')
-    //   .replace(TRIM_CJK_SPACE_RE, '$1')
-    text = text
-      .replaceAll(/[ \u3000]+/g, ' ')
-      .replaceAll(
-        /([\u3000-\u9FFF\uFF00-\uFFEF]) (?=[\u3000-\u9FFF\uFF00-\uFFEF])/g,
-        '$1',
-      )
-  }
+    if (TRIM_SPACE) {
+      // text = text
+      //   .replace(TRIM_EXTRA_SPACE_RE, ' ')
+      //   .replace(TRIM_CJK_SPACE_RE, '$1')
+      text = text
+        .replaceAll(/[ \u3000]+/g, ' ')
+        .replaceAll(
+          /([\u3000-\u9FFF\uFF00-\uFFEF]) (?=[\u3000-\u9FFF\uFF00-\uFFEF])/g,
+          '$1',
+        )
+    }
 
-  let taolu_matched = false
-  for (const taolu of FORCELIST) {
-    if (taolu[0].test(text)) {
-      text = text.replace(taolu[0], taolu[1])
-      taolu_matched = true
-      if (FORCELIST_BREAK_ON_MATCH) break
+    let taolu_matched = false
+    for (const taolu of FORCELIST) {
+      if (taolu[0].test(text)) {
+        text = text.replace(taolu[0], taolu[1])
+        taolu_matched = true
+        if (FORCELIST_BREAK_ON_MATCH) break
+      }
     }
-  }
 
-  return [taolu_matched, text]
+    return [taolu_matched, text]
+  }
 }
 
 /**
  * 白名单处理
  */
-const whitelisted = (text: string, config: Config) => {
-  const WHITELIST = (config?.WHITELIST ?? DEFAULT_CONFIG.WHITELIST).map(
-    (x) => new RegExp(x[0], 'iu'),
-  )
-  if (WHITELIST.length === 0) return false
-  else return WHITELIST.some((re) => re.test(text))
+function whitelisted_meta(config: ResolvedConfig): (text: string) => boolean {
+  const WHITELIST = config.WHITELIST.map((x) => new RegExp(x[0], 'iu'))
+
+  if (WHITELIST.length === 0) return () => false
+
+  return (text: string) => WHITELIST.some((re) => re.test(text))
 }
 
 /**
  * 黑名单处理
  */
-const blacklisted = (text: string, config: Config) => {
-  const BLACKLIST = (config?.BLACKLIST ?? DEFAULT_CONFIG.BLACKLIST).map((x) =>
+function blacklisted_meta(
+  config: ResolvedConfig,
+): (text: string) => string | null {
+  const BLACKLIST = config.BLACKLIST.map((x) =>
     x[0] ? new RegExp(x[1]) : x[1].toLowerCase(),
   )
-  if (BLACKLIST.length === 0) return null
-  else {
+
+  if (BLACKLIST.length === 0) return () => null
+
+  return (text: string) => {
     const lower = text.toLowerCase()
     for (const pattern of BLACKLIST) {
       const matched =
@@ -339,11 +348,22 @@ function select_median_length(strs: string[]): string {
   return sorted[mid]
 }
 
+function u8array_to_arraybuffer(array: Uint8Array): ArrayBuffer {
+  return array.buffer.slice(
+    array.byteOffset,
+    array.byteOffset + array.byteLength,
+  ) as ArrayBuffer
+}
+
 async function load_wasm(wasm_mod?: ArrayBuffer) {
-  await sim_init(
-    wasm_mod ??
-      (await fs.readFile(new URL('similarity-gen.wasm', import.meta.url))),
-  )
+  if (wasm_mod) {
+    await sim_init(wasm_mod)
+    return
+  }
+
+  const wasm_path = new URL('similarity-gen.wasm', import.meta.url).pathname
+  const wasm_u8 = await fs.readFile(wasm_path)
+  await sim_init(u8array_to_arraybuffer(wasm_u8))
 }
 
 function make_ptr_idx(idx: int, is_next_chunk: boolean): int {
@@ -355,9 +375,11 @@ async function merge(
   // next_chunk: DanmuChunk<DanmuObject>,
   config: Config = DEFAULT_CONFIG,
 ): Promise<DanmuClusterOutput> {
+  const local_config: ResolvedConfig = { ...DEFAULT_CONFIG, ...config }
+
   await load_wasm()
 
-  begin_chunk(config)
+  begin_chunk(local_config)
 
   const ret: DanmuClusterOutput = {
     clusters: [],
@@ -419,6 +441,10 @@ async function merge(
     }
   }
 
+  const detaolu = detaolu_meta(local_config)
+  const whitelisted = whitelisted_meta(local_config)
+  const blacklisted = blacklisted_meta(local_config)
+
   function obj_to_ir(
     objs: DanmuObject[],
     s: Stats | null,
@@ -426,7 +452,7 @@ async function merge(
   ): DanmuIr[] {
     return objs
       .map((obj, idx) => {
-        if (!config.PROC_POOL1 && obj.pool === 1) {
+        if (!local_config.PROC_POOL1 && obj.pool === 1) {
           if (s) {
             s.ignored_type++
             apply_single_cluster(idx, obj, '已忽略字幕弹幕，可以在选项中修改')
@@ -434,7 +460,7 @@ async function merge(
           return null
         }
         // if (!config.PROC_TYPE7 && obj.mode === 7) {
-        if (!config.PROC_TYPE7 && obj.mode === 4) {
+        if (!local_config.PROC_TYPE7 && obj.mode === 4) {
           if (s) {
             s.ignored_type++
             apply_single_cluster(idx, obj, '已忽略特殊弹幕，可以在选项中修改')
@@ -442,7 +468,7 @@ async function merge(
           return null
         }
         // if (!config.PROC_TYPE4 && obj.mode === 4) {
-        if (!config.PROC_TYPE4 && obj.mode === 1) {
+        if (!local_config.PROC_TYPE4 && obj.mode === 1) {
           if (s) {
             s.ignored_type++
             apply_single_cluster(idx, obj, '已忽略底部弹幕，可以在选项中修改')
@@ -473,7 +499,7 @@ async function merge(
 
         // if (obj.mode !== 8 && obj.mode !== 9) {
         if (obj.mode !== 4) {
-          const matched = blacklisted(disp_str, config)
+          const matched = blacklisted(disp_str)
           if (matched) {
             if (s) {
               s.deleted_blacklist++
@@ -489,7 +515,7 @@ async function merge(
             return null
           }
         }
-        if (whitelisted(disp_str, config)) {
+        if (whitelisted(disp_str)) {
           if (s) {
             s.ignored_whitelist++
             apply_single_cluster(idx, obj, '命中白名单')
@@ -497,11 +523,11 @@ async function merge(
           return null
         }
 
-        const [matched_taolu, detaolued] = detaolu(disp_str, config)
+        const [matched_taolu, detaolued] = detaolu(disp_str)
 
         if (matched_taolu) {
           if (s) s.num_taolu_matched++
-          if (config.FORCELIST_APPLY_SINGULAR)
+          if (local_config.FORCELIST_APPLY_SINGULAR)
             obj = {
               ...obj,
               content: detaolued,
@@ -523,7 +549,7 @@ async function merge(
 
   const nearby_danmus: Queue<DanmuIr[]> = new Queue()
 
-  const THRESHOLD_MS = (config?.THRESHOLD ?? DEFAULT_CONFIG.THRESHOLD) * 1000
+  const THRESHOLD_MS = local_config.THRESHOLD * 1000
 
   for (const dm of danmus) {
     while (true) {

package/src/pakku.js/similarity-gen.js

@@ -1,5 +1,24 @@
-/* eslint-disable @eslint-community/eslint-comments/no-unlimited-disable */
-/* eslint-disable */
+/* eslint-disable import/no-default-export */
+/* eslint-disable unicorn/prefer-number-properties */
+/* eslint-disable object-shorthand */
+/* eslint-disable block-scoped-var */
+/* eslint-disable unicorn/catch-error-name */
+/* eslint-disable getter-return */
+/* eslint-disable no-setter-return */
+/* eslint-disable prefer-template */
+/* eslint-disable unicorn/throw-new-error */
+/* eslint-disable unicorn/new-for-builtins */
+/* eslint-disable unicorn/prefer-code-point */
+/* eslint-disable unicorn/no-negated-condition */
+/* eslint-disable no-console */
+/* eslint-disable unicorn/consistent-function-scoping */
+/* eslint-disable no-unused-expressions */
+/* eslint-disable vars-on-top */
+/* eslint-disable one-var */
+/* eslint-disable import/no-mutable-exports */
+/* eslint-disable no-var */
+/* eslint-disable prettier/prettier */
+
 /** @nocollapse */ var Module = function (moduleArg = {}) {
   var moduleRtn
 

package/src/pakku.js/similarity_stub.ts

@@ -9,7 +9,7 @@ let ptr_buf: number
 
 const MAX_STRING_LEN = 16005
 
-export async function init(wasm_module: ArrayBuffer | NonSharedBuffer) {
+export async function init(wasm_module: ArrayBuffer) {
   module = await generated_promise({ wasm: wasm_module })
   ptr_buf = module._malloc(MAX_STRING_LEN * 2 + 7)
   if (ptr_buf % 2)