增加GROUP_MINUTE环境变量，合并去重分钟数，表示按n分钟分组后对弹幕合并去重，默认为1，最大值为30，0表示不去重 (#38)

* 增加GROUP_MINUTE环境变量，合并去重分钟数，表示按n分钟分组后对弹幕合并去重，默认为1，最大值为30，0表示不去重

* 将allowedPlatforms和vodAllowedPlatforms提取成全局变量

README.md

@@ -30,7 +30,7 @@ LogVar 弹幕 API 服务器
 ## 功能
 - **API 接口**：
   - `GET /api/v2/search/anime?keyword=${queryTitle}`：根据关键字搜索动漫。
-  - `POST /api/v2/match`：根据关键字匹配动漫，用于自动匹配。
+  - `POST /api/v2/match`：根据关键字匹配动漫，用于自动匹配。（目前已支持在match接口中通过@语法动态指定平台优先级，如`赴山海 S01E28 @qiyi`）
   - `GET /api/v2/search/episodes`：根据关键词搜索所有匹配的剧集信息。
   - `GET /api/v2/bangumi/:animeId`：获取指定动漫的详细信息。
   - `GET /api/v2/comment/:commentId?withRelated=true&chConvert=1`：获取指定弹幕评论，支持返回相关评论和字符转换。
@@ -227,6 +227,7 @@ Settings > Functions > Advanced Setting > Function Region 切换为 Hong Kong，
 | PLATFORM_ORDER    | 【可选】自动匹配优选平台，按顺序优先返回指定平台弹幕，默认为空，即返回第一个满足条件的平台，示例：`bilibili1,qq`，表示如果有b站的播放源，则优先返回b站的弹幕，否则就返回腾讯的弹幕，两者都没有，则返回第一个满足条件的平台；当前可选择的平台字段有 `qiyi, bilibili1, imgo, youku, qq, renren, hanjutv`  |
 | EPISODE_TITLE_FILTER    | 【可选】剧集标题正则过滤，按正则关键字对剧集或综艺的集标题进行过滤，适用于过滤一些预告或综艺非正式集，默认值如下 |
 | BLOCKED_WORDS    | 【可选】弹幕屏蔽词列表，默认为空，示例如下       |
+| GROUP_MINUTE    | 【可选】合并去重分钟数，表示按n分钟分组后对弹幕合并去重，默认为1，最大值为30，0表示不去重       |
 
 ```regex
 # EPISODE_TITLE_FILTER 默认值

danmu_api/worker.js

@@ -1,6 +1,6 @@
 // 全局状态（Cloudflare 和 Vercel 都可能重用实例）
 // ⚠️ 不是持久化存储，每次冷启动会丢失
-const VERSION = "1.2.2";
+const VERSION = "1.2.3";
 let animes = [];
 let episodeIds = [];
 let episodeNum = 10001; // 全局变量，用于自增 ID
@@ -9,7 +9,8 @@ let episodeNum = 10001; // 全局变量，用于自增 ID
 const logBuffer = [];
 const MAX_LOGS = 500;
 const MAX_ANIMES = 100;
-const allowedPlatforms = ["qiyi", "bilibili1", "imgo", "youku", "qq"];
+const vodAllowedPlatforms = ["qiyi", "bilibili1", "imgo", "youku", "qq"];
+const allowedPlatforms = ["qiyi", "bilibili1", "imgo", "youku", "qq", "renren", "hanjutv"];
 
 // =====================
 // 环境变量处理
@@ -112,14 +113,11 @@ function resolvePlatformOrder(env) {
     platformOrder = process.env.PLATFORM_ORDER;  // Vercel / Node
   }
 
-  // 解析并校验 platformOrder
-  const _allowedPlatforms = ["qiyi", "bilibili1", "imgo", "youku", "qq", "renren", "hanjutv"];
-
   // 转换为数组并去除空格，过滤无效项
   const orderArr = platformOrder
     .split(',')
     .map(s => s.trim())  // 去除空格
-    .filter(s => _allowedPlatforms.includes(s));  // 只保留有效来源
+    .filter(s => allowedPlatforms.includes(s));  // 只保留有效来源
 
   // 如果没有有效的来源，使用默认顺序
   if (orderArr.length === 0) {
@@ -188,6 +186,22 @@ function resolveBlockedWords(env) {
   return DEFAULT_BLOCKED_WORDS;
 }
 
+// 分钟内合并去重（默认 1，最大值30，0表示不去重）
+const DEFAULT_GROUP_MINUTE = 1;
+let groupMinute = DEFAULT_GROUP_MINUTE;
+
+function resolveGroupMinute(env) {
+  if (env && env.GROUP_MINUTE) {
+    const n = parseInt(env.GROUP_MINUTE, 10);
+    if (!Number.isNaN(n) && n > 0) return Math.min(n, 30);
+  }
+  if (typeof process !== "undefined" && process.env?.GROUP_MINUTE) {
+    const n = parseInt(process.env.GROUP_MINUTE, 10);
+    if (!Number.isNaN(n) && n > 0) return Math.min(n, 30);
+  }
+  return Math.min(DEFAULT_GROUP_MINUTE, 30);
+}
+
 // =====================
 // 数据结构处理函数
 // =====================
@@ -574,6 +588,69 @@ function printFirst200Chars(data) {
   log("log", dataToPrint.slice(0, 200));  // 打印前200个字符
 }
 
+function groupDanmusByMinute(filteredDanmus, n) {
+  // 如果 n 为 0，直接返回原始数据
+  if (n === 0) {
+    return filteredDanmus.map(danmu => ({
+      ...danmu,
+      t: danmu.t !== undefined ? danmu.t : parseFloat(danmu.p.split(',')[0])
+    }));
+  }
+
+  // 按 n 分钟分组
+  const groupedByMinute = filteredDanmus.reduce((acc, danmu) => {
+    // 获取时间：优先使用 t 字段，如果没有则使用 p 的第一个值
+    const time = danmu.t !== undefined ? danmu.t : parseFloat(danmu.p.split(',')[0]);
+    // 计算分组（每 n 分钟一组，向下取整）
+    const group = Math.floor(time / (n * 60));
+
+    // 初始化分组
+    if (!acc[group]) {
+      acc[group] = [];
+    }
+
+    // 添加到对应分组
+    acc[group].push({ ...danmu, t: time });
+    return acc;
+  }, {});
+
+  // 处理每组的弹幕
+  const result = Object.keys(groupedByMinute).map(group => {
+    const danmus = groupedByMinute[group];
+
+    // 按消息内容分组
+    const groupedByMessage = danmus.reduce((acc, danmu) => {
+      const message = danmu.m.split(' X')[0]; // 提取原始消息（去除 Xn 后缀）
+      if (!acc[message]) {
+        acc[message] = {
+          count: 0,
+          earliestT: danmu.t,
+          cid: danmu.cid,
+          p: danmu.p
+        };
+      }
+      acc[message].count += 1;
+      // 更新最早时间
+      acc[message].earliestT = Math.min(acc[message].earliestT, danmu.t);
+      return acc;
+    }, {});
+
+    // 转换为结果格式
+    return Object.keys(groupedByMessage).map(message => {
+      const data = groupedByMessage[message];
+      return {
+        cid: data.cid,
+        p: data.p,
+        m: data.count > 1 ? `${message} X${data.count}` : message,
+        t: data.earliestT
+      };
+    });
+  });
+
+  // 展平结果并按时间排序
+  return result.flat().sort((a, b) => a.t - b.t);
+}
+
 function convertToDanmakuJson(contents, platform) {
   let danmus = [];
   let cidCounter = 1;
@@ -633,12 +710,21 @@ function convertToDanmakuJson(contents, platform) {
       }
       // 处理 XML 解析后的格式
       const pValues = item.p.split(",");
-      attributes = [
-        parseFloat(pValues[0]).toFixed(2),
-        pValues[1] || 0,
-        pValues[3] || 16777215,
-        `[${platform}]`
-      ].join(",");
+      if (pValues.length === 4) {
+        attributes = [
+          parseFloat(pValues[0]).toFixed(2),
+          pValues[1] || 0,
+          pValues[2] || 16777215,
+          `[${platform}]`
+        ].join(",");
+      } else {
+        attributes = [
+          parseFloat(pValues[0]).toFixed(2),
+          pValues[1] || 0,
+          pValues[3] || 16777215,
+          `[${platform}]`
+        ].join(",");
+      }
       m = item.m;
     }
 
@@ -670,11 +756,16 @@ function convertToDanmakuJson(contents, platform) {
     return !regexArray.some(regex => regex.test(item.m)); // 针对 `m` 字段进行匹配
   });
 
+  // 按n分钟内去重
+  log("log", "去重分钟数:", groupMinute);
+  const groupedDanmus = groupDanmusByMinute(filteredDanmus, groupMinute);
+
   log("log", "danmus_original:", danmus.length);
-  log("log", "danmus:", filteredDanmus.length);
+  log("log", "danmus_filter:", filteredDanmus.length);
+  log("log", "danmus_group:", groupedDanmus.length);
   // 输出前五条弹幕
-  log("log", "Top 5 danmus:", JSON.stringify(filteredDanmus.slice(0, 5), null, 2));
-  return filteredDanmus;
+  log("log", "Top 5 danmus:", JSON.stringify(groupedDanmus.slice(0, 5), null, 2));
+  return groupedDanmus;
 }
 
 function buildQueryString(params) {
@@ -1036,9 +1127,6 @@ function normalizePlatformName(inputPlatform) {
 
   const input = inputPlatform.trim();
 
-  // 支持的平台名称：["qiyi", "bilibili1", "imgo", "youku", "qq", "renren", "hanjutv"]
-  const allowedPlatforms = ["qiyi", "bilibili1", "imgo", "youku", "qq", "renren", "hanjutv"];
-
   // 直接返回输入的平台名称（如果有效）
   if (allowedPlatforms.includes(input)) {
     return input;
@@ -1055,7 +1143,6 @@ function createDynamicPlatformOrder(preferredPlatform) {
   }
 
   // 验证平台是否有效
-  const allowedPlatforms = ["qiyi", "bilibili1", "imgo", "youku", "qq", "renren", "hanjutv"];
   if (!allowedPlatforms.includes(preferredPlatform)) {
     log("warn", `Invalid platform: ${preferredPlatform}, using default order`);
     return [...platformOrderArr];
@@ -2648,35 +2735,17 @@ function parseRRSPPFields(pField) {
   return { timestamp, mode, size, color, userId, contentId };
 }
 
-function formatComments(items) {
-  const unique = {};
-  for(const it of items){
-    const text = String(it.d||"");
-    const meta = parseRRSPPFields(it.p);
-    if(!unique[meta.contentId]) unique[meta.contentId] = { content: text, ...meta };
-  }
-
-  const grouped = {};
-  for(const c of Object.values(unique)){
-    if(!grouped[c.content]) grouped[c.content] = [];
-    grouped[c.content].push(c);
-  }
-
-  const processed = [];
-  for(const [content, group] of Object.entries(grouped)){
-    if(group.length===1) processed.push(group[0]);
-    else{
-      const first = group.reduce((a,b)=>a.timestamp<b.timestamp?a:b);
-      processed.push({...first, content:`${first.content} X${group.length}`});
-    }
-  }
-
-  return processed.map(c=>({
-    cid: Number(c.contentId),
-    p: `${c.timestamp.toFixed(2)},${c.mode},${c.color},[renren]`,
-    m: c.content,
-    t: c.timestamp
-  }));
+function formatRenrenComments(items) {
+  return items.map(item => {
+    const text = String(item.d || "");
+    const meta = parseRRSPPFields(item.p);
+    return {
+      cid: Number(meta.contentId),
+      p: `${meta.timestamp.toFixed(2)},${meta.mode},${meta.color},[renren]`,
+      m: text,
+      t: meta.timestamp
+    };
+  });
 }
 
 async function getRenRenComments(episodeId, progressCallback=null){
@@ -2685,12 +2754,10 @@ async function getRenRenComments(episodeId, progressCallback=null){
   const raw = await fetchEpisodeDanmu(episodeId);
   if(progressCallback) await progressCallback(85,`原始弹幕 ${raw.length} 条，正在规范化`);
   log("log", `原始弹幕 ${raw.length} 条，正在规范化`);
-  const formatted = formatComments(raw);
+  const formatted = formatRenrenComments(raw);
   if(progressCallback) await progressCallback(100,`弹幕处理完成，共 ${formatted.length} 条`);
   log("log", `弹幕处理完成，共 ${formatted.length} 条`);
-  // 输出前五条弹幕
-  log("log", "Top 5 danmus:", JSON.stringify(formatted.slice(0, 5), null, 2));
-  return formatted;
+  return convertToDanmakuJson(formatted, "renren");
 }
 
 // ---------------------
@@ -2868,9 +2935,7 @@ async function getHanjutvComments(pid, progressCallback=null){
   const formatted = formatHanjutvComments(raw);
   if(progressCallback) await progressCallback(100,`弹幕处理完成，共 ${formatted.length} 条`);
   log("log", `弹幕处理完成，共 ${formatted.length} 条`);
-  // 输出前五条弹幕
-  log("log", "Top 5 danmus:", JSON.stringify(formatted.slice(0, 5), null, 2));
-  return formatted;
+  return convertToDanmakuJson(formatted, "hanjutv");
 }
 
 // =====================
@@ -3013,7 +3078,7 @@ async function handleVodAnimes(animesVod, curAnimes) {
 
     const vodPlayUrlList = anime.vod_play_url.split("$$$");
     const validIndices = vodPlayFromList
-        .map((item, index) => allowedPlatforms.includes(item) ? index : -1)
+        .map((item, index) => vodAllowedPlatforms.includes(item) ? index : -1)
         .filter(index => index !== -1);
 
     let links = [];
@@ -3064,7 +3129,7 @@ async function handle360Animes(animes360, curAnimes) {
     let links = [];
     if (anime.cat_name === "电影") {
       for (const key of Object.keys(anime.playlinks)) {
-        if (allowedPlatforms.includes(key)) {
+        if (vodAllowedPlatforms.includes(key)) {
           links.push({
             "name": key,
             "url": anime.playlinks[key],
@@ -3073,7 +3138,7 @@ async function handle360Animes(animes360, curAnimes) {
         }
       }
     } else if (anime.cat_name === "电视剧" || anime.cat_name === "动漫") {
-      if (allowedPlatforms.includes(anime.seriesSite)) {
+      if (vodAllowedPlatforms.includes(anime.seriesSite)) {
         for (let i = 0; i < anime.seriesPlaylinks.length; i++) {
           const item = anime.seriesPlaylinks[i];
           links.push({
@@ -3086,7 +3151,7 @@ async function handle360Animes(animes360, curAnimes) {
     } else if (anime.cat_name === "综艺") {
       const zongyiLinks = await Promise.all(
           Object.keys(anime.playlinks_year).map(async (site) => {
-            if (allowedPlatforms.includes(site)) {
+            if (vodAllowedPlatforms.includes(site)) {
               const yearLinks = await Promise.all(
                   anime.playlinks_year[site].map(async (year) => {
                     return await get360Zongyi(anime.id, site, year);
@@ -3678,7 +3743,7 @@ async function getComment(path) {
   }
 
   // 如果弹幕为空，则请求第三方弹幕服务器作为兜底
-  if (danmus.length === 0) {
+  if (danmus.length === 0 && urlPattern.test(url)) {
     danmus = await fetchOtherServer(url);
   }
 
@@ -3695,6 +3760,7 @@ async function handleRequest(req, env) {
   platformOrderArr = resolvePlatformOrder(env);
   episodeTitleFilter = resolveEpisodeTitleFilter(env);
   blockedWords = resolveBlockedWords(env);
+  groupMinute = resolveGroupMinute(env);
 
   const url = new URL(req.url);
   let path = url.pathname;