koishi-plugin-chat-analyse 0.6.4 → 0.7.0

package/lib/Analyse.d.ts

@@ -5,18 +5,20 @@ export interface WordCloudData {
     time: Date;
     words: [string, number][];
 }
+/**
+ * @class Analyse
+ * @description 提供文本分析功能，如生成词云。
+ */
 export declare class Analyse {
     private ctx;
     private config;
     private renderer;
-    private nlp;
-    private isNlpReady;
+    private readonly jieba;
     constructor(ctx: Context, config: Config);
     /**
-     * @private
-     * @method initializeNlp
-     * @description 异步加载并训练 NLP 模型。
+     * @public @method registerCommands
+     * @description 在主命令下注册子命令。
+     * @param cmd - 主命令实例。
      */
-    private initializeNlp;
     registerCommands(cmd: Command): void;
 }

package/lib/index.d.ts

@@ -19,7 +19,6 @@ export interface Config {
     atRetentionDays: number;
     rankRetentionDays: number;
     enableWordCloud: boolean;
-    enableVocabulary: boolean;
 }
 /** @description 插件的配置项定义 */
 export declare const Config: Schema<Config>;

package/lib/index.js

@@ -553,7 +553,7 @@ var Stat = class {
       };
     }, "createHandler");
     if (this.config.enableCmdStat) {
-      cmd.subcommand("cmdstat", "命令统计").option("user", "-u <user:string> 指定用户").option("guild", "-g <guildId:string> 指定群组").option("separate", "-h 分离展示").option("all", "-a 全局").action(createHandler(async (scope, options) => {
+      cmd.subcommand("cmdstat", "命令统计").usage("查询命令统计，可指定查询范围，默认当前群组。").option("user", "-u <user:string> 指定用户").option("guild", "-g <guildId:string> 指定群组").option("separate", "-h 分离展示").option("all", "-a 全局").action(createHandler(async (scope, options) => {
         const stats = await this.ctx.database.select("analyse_cmd").where({ uid: { $in: scope.uids } }).groupBy("command", { count: /* @__PURE__ */ __name((row) => import_koishi3.$.sum(row.count), "count"), lastUsed: /* @__PURE__ */ __name((row) => import_koishi3.$.max(row.timestamp), "lastUsed") }).orderBy("count", "desc").execute();
         if (stats.length === 0) return "暂无统计数据";
         let processedStats;
@@ -581,7 +581,7 @@ var Stat = class {
       }));
     }
     if (this.config.enableMsgStat) {
-      cmd.subcommand("msgstat", "发言统计").option("user", "-u <user:string> 指定用户").option("guild", "-g <guildId:string> 指定群组").option("type", "-t <type:string> 指定类型").option("all", "-a 全局").action(createHandler(async (scope, options) => {
+      cmd.subcommand("msgstat", "发言统计").usage("查询发言统计，可指定查询范围，默认当前群组。").option("user", "-u <user:string> 指定用户").option("guild", "-g <guildId:string> 指定群组").option("type", "-t <type:string> 指定类型").option("all", "-a 全局").action(createHandler(async (scope, options) => {
         const { type } = options;
         const query = { uid: { $in: scope.uids } };
         if (type) query.type = type;
@@ -597,7 +597,7 @@ var Stat = class {
       }));
     }
     if (this.config.enableRankStat) {
-      cmd.subcommand("rankstat", "发言排行").option("guild", "-g <guildId:string> 指定群组").option("type", "-t <type:string> 指定类型").option("hours", "-h <hours:number> 指定时长", { fallback: 24 }).option("all", "-a 全局").action(createHandler(async (scope, options) => {
+      cmd.subcommand("rankstat", "发言排行").usage("查询发言排行，可指定查询范围，默认当前群组。").option("guild", "-g <guildId:string> 指定群组").option("type", "-t <type:string> 指定类型").option("hours", "-h <hours:number> 指定时长", { fallback: 24 }).option("all", "-a 全局").action(createHandler(async (scope, options) => {
         const { hours, type } = options;
         const since = new Date(Date.now() - hours * import_koishi3.Time.hour);
         const query = { uid: { $in: scope.uids }, timestamp: { $gte: since } };
@@ -614,7 +614,7 @@ var Stat = class {
       }));
     }
     if (this.config.enableActivity) {
-      cmd.subcommand("activity", "活跃统计").option("user", "-u <user:string> 指定用户").option("guild", "-g <guildId:string> 指定群组").option("hours", "-h <hours:number> 指定偏移时长").option("all", "-a 全局").option("days", "-d 切换至天数").action(createHandler(async (scope, options) => {
+      cmd.subcommand("activity", "活跃统计").usage("查询活跃统计，可指定查询范围，默认当前群组。").option("user", "-u <user:string> 指定用户").option("guild", "-g <guildId:string> 指定群组").option("hours", "-h <hours:number> 指定偏移时长").option("all", "-a 全局").option("days", "-d 切换至天数").action(createHandler(async (scope, options) => {
         const { days, hours } = options;
         if (days) {
           const timeRangeInDays = 24;
@@ -798,7 +798,7 @@ var Data = class {
         return "数据恢复失败";
       }
     });
-    cmd.subcommand(".clear", "清除数据", { authority: 4 }).option("table", "-t <table:string> 指定表名").option("guild", "-g <guildId:string> 指定群组").option("user", "-u <user:string> 指定用户").option("days", "-d <days:number> 指定天数").option("command", "-c <command:string> 指定命令").option("all", "-a 清除全部").usage(`根据指定条件清理统计数据，可以组合多个选项以精确控制清除范围。`).action(async ({ options }) => {
+    cmd.subcommand(".clear", "清除数据", { authority: 4 }).usage(`清除指定统计数据，可精确控制清除范围。`).option("table", "-t <table:string> 指定表名").option("guild", "-g <guildId:string> 指定群组").option("user", "-u <user:string> 指定用户").option("days", "-d <days:number> 指定天数").option("command", "-c <command:string> 指定命令").option("all", "-a 清除全部").action(async ({ options }) => {
       if (Object.keys(options).length === 0) return "请指定清除条件";
       if (options.table && !ALL_TABLES.includes(options.table)) return `表名 ${options.table} 无效`;
       try {
@@ -871,107 +871,64 @@ ${commandOutput}`;
 
 // src/Analyse.ts
 var import_koishi6 = require("koishi");
-var import_basic = require("@nlpjs/basic");
-var import_lang_zh = require("@nlpjs/lang-zh");
+var import_jieba = require("@node-rs/jieba");
+var import_dict = require("@node-rs/jieba/dict");
 var Analyse = class {
   constructor(ctx, config) {
     this.ctx = ctx;
     this.config = config;
     this.renderer = new Renderer(ctx);
-    this.nlp = new import_basic.Nlp({ languages: ["zh"], nlu: { log: false } });
-    this.nlp.settings.autoSave = false;
-    this.nlp.container.register("extract-lang-zh", new import_lang_zh.LangZh());
-    this.initializeNlp().catch((err) => {
-      this.ctx.logger.error("NLP 语言模型加载失败:", err);
-    });
+    if (config.enableWordCloud) this.jieba = import_jieba.Jieba.withDict(import_dict.dict);
   }
   static {
     __name(this, "Analyse");
   }
   renderer;
-  nlp;
-  isNlpReady = false;
+  jieba = null;
   /**
-   * @private
-   * @method initializeNlp
-   * @description 异步加载并训练 NLP 模型。
+   * @public @method registerCommands
+   * @description 在主命令下注册子命令。
+   * @param cmd - 主命令实例。
    */
-  async initializeNlp() {
-    await this.nlp.train();
-    this.isNlpReady = true;
-  }
   registerCommands(cmd) {
     if (this.config.enableWordCloud) {
-      cmd.subcommand(".wordcloud", "生成词云").usage("基于指定范围内的聊天记录生成词云图。").option("guild", "-g <guildId:string> 指定群组").option("user", "-u <user:string> 指定用户").option("all", "-a 全局").option("hours", "-h <hours:number> 指定时长", { fallback: 24 }).action(async ({ session, options }) => {
-        if (!this.isNlpReady) return "文本分析尚未就绪，请稍后再试";
+      cmd.subcommand("wordcloud", "生成词云").usage("基于聊天记录生成词云图，可指定范围，默认当前群组。").option("guild", "-g <guildId:string> 指定群组").option("user", "-u <user:string> 指定用户").option("hours", "-h <hours:number> 指定时长", { fallback: 24 }).option("all", "-a 全局").action(async ({ session, options }) => {
+        if (!this.jieba) return "Jieba 分词服务未就绪";
         const scope = await parseQueryScope(this.ctx, session, options);
         if (scope.error) return scope.error;
+        scope.uids ??= (await this.ctx.database.get("analyse_user", {}, ["uid"])).map((u) => u.uid);
+        if (!scope.uids?.length) return "暂无用户数据";
         const since = new Date(Date.now() - options.hours * import_koishi6.Time.hour);
-        const records = await this.ctx.database.select("analyse_cache").where({ uid: { $in: scope.uids }, timestamp: { $gte: since } }).project(["content"]).execute();
-        if (records.length === 0) return "暂无统计数据";
+        const records = await this.ctx.database.get("analyse_cache", { uid: { $in: scope.uids }, timestamp: { $gte: since } }, ["content"]);
+        if (!records.length) return "暂无统计数据";
         const allText = records.map((r) => r.content).join(" ");
-        const result = await this.nlp.process("zh", allText);
-        const words = (result.stems || []).filter((stem) => stem.length > 1);
-        const wordCounts = words.reduce((map, word) => {
-          map.set(word, (map.get(word) || 0) + 1);
-          return map;
-        }, /* @__PURE__ */ new Map());
-        if (wordCounts.size === 0) return "暂无有效词语";
+        const exclusionSet = /* @__PURE__ */ new Set([
+          "[face]",
+          "[file]",
+          "[forward]",
+          "[img]",
+          "[audio]",
+          "[video]",
+          "[json]",
+          "[rps]",
+          "[markdown]",
+          "[dice]"
+        ]);
+        const words = this.jieba.cut(allText).filter((w) => {
+          if (w.length <= 1) return false;
+          if (/^\d+$/.test(w)) return false;
+          if (exclusionSet.has(w)) return false;
+          if (/^\[at:.*?\]$/.test(w)) return false;
+          return true;
+        });
+        if (!words.length) return "暂无有效词语";
+        const wordCounts = words.reduce((map, word) => map.set(word, (map.get(word) || 0) + 1), /* @__PURE__ */ new Map());
         const wordList = Array.from(wordCounts.entries()).sort((a, b) => b[1] - a[1]).slice(0, 150);
         const title = await generateTitle(this.ctx, scope.scopeDesc, { main: "词云" });
-        const renderResult = await this.renderer.renderWordCloud({ title, time: /* @__PURE__ */ new Date(), words: wordList });
-        if (typeof renderResult === "string") return renderResult;
-        if (Array.isArray(renderResult) && renderResult.length > 0) {
-          for (const buffer of renderResult) await session.sendQueued(import_koishi6.h.image(buffer, "image/png"));
-        }
-      });
-    }
-    if (this.config.enableVocabulary) {
-      cmd.subcommand(".vocabulary", "词汇排行").usage("根据不重复词汇量占总词汇量的比例进行排行。").option("guild", "-g <guildId:string> 指定群组").option("all", "-a 全局").option("hours", "-h <hours:number> 指定时长", { fallback: 24 }).action(async ({ session, options }) => {
-        if (!this.isNlpReady) return "文本分析尚未就绪，请稍后再试";
-        const scope = await parseQueryScope(this.ctx, session, options);
-        if (scope.error) return scope.error;
-        const users = await this.ctx.database.get("analyse_user", { uid: { $in: scope.uids } }, ["uid", "userName"]);
-        const userNameMap = new Map(users.map((u) => [u.uid, u.userName]));
-        const since = new Date(Date.now() - options.hours * import_koishi6.Time.hour);
-        const allRecords = await this.ctx.database.get("analyse_cache", { uid: { $in: scope.uids }, timestamp: { $gte: since } }, ["uid", "content"]);
-        if (allRecords.length === 0) return "暂无统计数据";
-        const messagesByUid = /* @__PURE__ */ new Map();
-        for (const record of allRecords) {
-          if (!messagesByUid.has(record.uid)) messagesByUid.set(record.uid, []);
-          messagesByUid.get(record.uid).push(record.content);
-        }
-        const richnessData = [];
-        for (const [uid, messages] of messagesByUid.entries()) {
-          const allText = messages.join(" ");
-          const result = await this.nlp.process("zh", allText);
-          const words = (result.stems || []).filter((stem) => stem.length > 1);
-          if (words.length < 50) continue;
-          const uniqueWords = new Set(words);
-          const richness = uniqueWords.size / words.length;
-          richnessData.push({
-            name: userNameMap.get(uid) || `UID ${uid}`,
-            total: words.length,
-            unique: uniqueWords.size,
-            richness
-          });
-        }
-        if (richnessData.length === 0) return "暂无有效词语";
-        richnessData.sort((a, b) => b.richness - a.richness);
-        const list = richnessData.map((item) => [
-          item.name,
-          item.unique,
-          item.total,
-          `${(item.richness * 100).toFixed(2)}%`
-        ]);
-        const title = await generateTitle(this.ctx, scope.scopeDesc, { main: "词汇排行" });
-        const renderResult = await this.renderer.renderList(
-          { title, time: /* @__PURE__ */ new Date(), total: richnessData.length, list },
-          ["用户", "不重复词数", "总词数", "丰富度"]
-        );
-        if (typeof renderResult === "string") return renderResult;
-        if (Array.isArray(renderResult) && renderResult.length > 0) {
-          for (const buffer of renderResult) await session.sendQueued(import_koishi6.h.image(buffer, "image/png"));
+        const result = await this.renderer.renderWordCloud({ title, time: /* @__PURE__ */ new Date(), words: wordList });
+        if (typeof result === "string") return result;
+        if (Array.isArray(result) && result.length > 0) {
+          for (const buffer of result) await session.sendQueued(import_koishi6.h.image(buffer, "image/png"));
         }
       });
     }
@@ -1009,8 +966,7 @@ var Config3 = import_koishi7.Schema.intersect([
   }).description("基础分析配置"),
   import_koishi7.Schema.object({
     enableOriRecord: import_koishi7.Schema.boolean().default(true).description("启用原始记录"),
-    enableWordCloud: import_koishi7.Schema.boolean().default(true).description("启用词云生成"),
-    enableVocabulary: import_koishi7.Schema.boolean().default(true).description("启用词汇排行")
+    enableWordCloud: import_koishi7.Schema.boolean().default(true).description("启用词云生成")
   }).description("高级分析配置")
 ]);
 async function parseQueryScope(ctx, session, options) {
@@ -1057,7 +1013,7 @@ function apply(ctx, config) {
   new Stat(ctx, config).registerCommands(analyse);
   if (config.enableWhoAt) new WhoAt(ctx, config).registerCommand(analyse);
   if (config.enableDataIO) new Data(ctx).registerCommands(analyse);
-  if (config.enableOriRecord && (config.enableWordCloud || config.enableVocabulary)) new Analyse(ctx, config).registerCommands(analyse);
+  if (config.enableOriRecord && config.enableWordCloud) new Analyse(ctx, config).registerCommands(analyse);
 }
 __name(apply, "apply");
 // Annotate the CommonJS export names for ESM import in node:

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "koishi-plugin-chat-analyse",
   "description": "聊天记录分析",
-  "version": "0.6.4",
+  "version": "0.7.0",
   "contributors": [
     "Yis_Rime <yis_rime@outlook.com>"
   ],
@@ -39,7 +39,6 @@
     "koishi": "4.18.8"
   },
   "dependencies": {
-    "@nlpjs/basic": "^4.26.1",
-    "@nlpjs/lang-zh": "^4.26.1"
+    "@node-rs/jieba": "^2.0.1"
   }
 }