opencode-tokenwatch 0.3.0 → 0.3.1

package/dist/stats-store.js

@@ -0,0 +1,258 @@
+/**
+ * stats-store.ts — 持久化聚合统计存储
+ *
+ * 设计目标：将性能指标的"聚合统计"与"原始 JSONL 日志"彻底解耦。
+ * - 每次请求完成时，通过 updatePersistedStats() 增量写入 JSON 统计文件
+ * - 统计文件永久累积，不受 JSONL 日志轮转/窗口限制影响
+ * - 百分位数采用 Reservoir Sampling 保持有界内存占用
+ * - 首次启动时自动从现有 JSONL 日志迁移，不丢失历史数据
+ */
+import { readFileSync, writeFileSync, existsSync } from "node:fs";
+import { join } from "node:path";
+import { homedir } from "node:os";
+const STATS_PATH = join(homedir(), ".opencode", "tokenwatch-stats.json");
+const LOG_PATH = join(homedir(), ".opencode", "tokenwatch.jsonl");
+const RESERVOIR_SIZE = 500; // 每个指标最多保留的原始样本数
+const CURRENT_VERSION = 1;
+// ─────────────────────────────────────────────
+// 内部 I/O 工具
+// ─────────────────────────────────────────────
+function loadStatsFile() {
+    try {
+        if (!existsSync(STATS_PATH)) {
+            return { version: CURRENT_VERSION, updatedAt: "", migratedFromLogs: false, models: {} };
+        }
+        const content = readFileSync(STATS_PATH, "utf-8");
+        const parsed = JSON.parse(content);
+        if (parsed?.version === CURRENT_VERSION && parsed.models)
+            return parsed;
+    }
+    catch { /* 文件损坏时返回空白统计 */ }
+    return { version: CURRENT_VERSION, updatedAt: "", migratedFromLogs: false, models: {} };
+}
+function saveStatsFile(file) {
+    try {
+        file.updatedAt = new Date().toISOString();
+        writeFileSync(STATS_PATH, JSON.stringify(file), "utf-8");
+    }
+    catch { /* 写入失败不影响主流程 */ }
+}
+// ─────────────────────────────────────────────
+// Reservoir Sampling（有界样本更新）
+// ─────────────────────────────────────────────
+/**
+ * Reservoir Sampling 算法：保证内存有界的同时，给每个观测值相同的入选概率，
+ * 使分位数估算在统计意义上无偏。
+ *
+ * @param reservoir 当前样本数组（会被原地返回新引用）
+ * @param value 新观测值
+ * @param totalCount 加入此值后的总观测数
+ */
+function reservoirAdd(reservoir, value, totalCount) {
+    if (reservoir.length < RESERVOIR_SIZE) {
+        return [...reservoir, value];
+    }
+    // 以 RESERVOIR_SIZE/totalCount 的概率替换随机位置
+    const j = Math.floor(Math.random() * totalCount);
+    if (j < RESERVOIR_SIZE) {
+        const next = [...reservoir];
+        next[j] = value;
+        return next;
+    }
+    return reservoir;
+}
+// ─────────────────────────────────────────────
+// 核心增量更新逻辑（可复用于单条 & 批量迁移）
+// ─────────────────────────────────────────────
+function applyEntryToModels(models, entry) {
+    const key = entry.model;
+    let s = models[key];
+    if (!s) {
+        s = {
+            model: entry.model,
+            providerID: entry.providerID,
+            requestCount: 0,
+            ttftCount: 0,
+            tpsCount: 0,
+            latencyCount: 0,
+            totalInput: 0,
+            totalOutput: 0,
+            totalCacheRead: 0,
+            totalCacheWrite: 0,
+            totalCost: 0,
+            avgTTFT: null, maxTTFT: null, minTTFT: null,
+            avgTPS: null, maxTPS: null, minTPS: null,
+            avgLatency: null, maxLatency: null, minLatency: null,
+            ttftReservoir: [],
+            latencyReservoir: [],
+        };
+        models[key] = s;
+    }
+    s.requestCount++;
+    s.totalInput += entry.inputTokens;
+    s.totalOutput += entry.outputTokens;
+    s.totalCacheRead += entry.cacheReadTokens;
+    s.totalCacheWrite += entry.cacheWriteTokens;
+    s.totalCost += entry.cost;
+    if (entry.ttft_ms != null) {
+        s.ttftCount++;
+        const c = s.ttftCount;
+        s.avgTTFT = s.avgTTFT != null ? s.avgTTFT + (entry.ttft_ms - s.avgTTFT) / c : entry.ttft_ms;
+        s.maxTTFT = s.maxTTFT != null ? Math.max(s.maxTTFT, entry.ttft_ms) : entry.ttft_ms;
+        s.minTTFT = s.minTTFT != null ? Math.min(s.minTTFT, entry.ttft_ms) : entry.ttft_ms;
+        s.ttftReservoir = reservoirAdd(s.ttftReservoir, entry.ttft_ms, s.ttftCount);
+    }
+    if (entry.tps != null) {
+        s.tpsCount++;
+        const c = s.tpsCount;
+        s.avgTPS = s.avgTPS != null ? s.avgTPS + (entry.tps - s.avgTPS) / c : entry.tps;
+        s.maxTPS = s.maxTPS != null ? Math.max(s.maxTPS, entry.tps) : entry.tps;
+        s.minTPS = s.minTPS != null ? Math.min(s.minTPS, entry.tps) : entry.tps;
+    }
+    if (entry.latency_ms != null) {
+        s.latencyCount++;
+        const c = s.latencyCount;
+        s.avgLatency = s.avgLatency != null ? s.avgLatency + (entry.latency_ms - s.avgLatency) / c : entry.latency_ms;
+        s.maxLatency = s.maxLatency != null ? Math.max(s.maxLatency, entry.latency_ms) : entry.latency_ms;
+        s.minLatency = s.minLatency != null ? Math.min(s.minLatency, entry.latency_ms) : entry.latency_ms;
+        s.latencyReservoir = reservoirAdd(s.latencyReservoir, entry.latency_ms, s.latencyCount);
+    }
+}
+// ─────────────────────────────────────────────
+// 一次性迁移：从 JSONL 日志重建初始统计
+// ─────────────────────────────────────────────
+/**
+ * 如果统计文件尚未完成迁移，则读取全量 JSONL 日志并批量写入统计文件。
+ * 只在首次调用 readPersistedStats() 时执行一次，之后通过 migratedFromLogs 标志跳过。
+ */
+function migrateFromLogsIfNeeded(file) {
+    if (file.migratedFromLogs)
+        return false;
+    if (!existsSync(LOG_PATH)) {
+        file.migratedFromLogs = true;
+        return true;
+    }
+    try {
+        const content = readFileSync(LOG_PATH, "utf-8").trim();
+        if (!content) {
+            file.migratedFromLogs = true;
+            return true;
+        }
+        let migrated = 0;
+        for (const line of content.split("\n")) {
+            if (!line)
+                continue;
+            try {
+                const entry = JSON.parse(line);
+                if (entry.model && entry.ts) {
+                    applyEntryToModels(file.models, entry);
+                    migrated++;
+                }
+            }
+            catch { /* 跳过格式损坏的行 */ }
+        }
+        file.migratedFromLogs = true;
+        if (migrated > 0) {
+            // 标记本次迁移来源，便于调试
+            ;
+            file._migratedFrom = `${LOG_PATH} (${migrated} entries)`;
+        }
+        return true;
+    }
+    catch {
+        // 迁移失败时仍标记为已完成，避免每次都重试（下次重建会通过 updatePersistedStats 增量补充）
+        file.migratedFromLogs = true;
+        return true;
+    }
+}
+// ─────────────────────────────────────────────
+// 分位数计算
+// ─────────────────────────────────────────────
+function percentile(arr, p) {
+    if (arr.length === 0)
+        return null;
+    if (arr.length === 1)
+        return arr[0];
+    const idx = (p / 100) * (arr.length - 1);
+    const lo = Math.floor(idx);
+    const hi = Math.ceil(idx);
+    if (lo === hi)
+        return arr[lo];
+    return arr[lo] + (arr[hi] - arr[lo]) * (idx - lo);
+}
+// ─────────────────────────────────────────────
+// 公开 API
+// ─────────────────────────────────────────────
+/**
+ * 将一条新的日志条目增量更新到持久化统计文件。
+ * 在 perf-tracker.ts 的 appendLog() 之后调用。
+ *
+ * 设计原则：本函数只做增量更新，迁移逻辑由 readPersistedStats() 负责。
+ * 这样可以避免迁移与增量更新之间的竞态问题。
+ */
+export function updatePersistedStats(entry) {
+    try {
+        const file = loadStatsFile();
+        applyEntryToModels(file.models, entry);
+        // 如果尚未完成迁移，先标记（避免 readPersistedStats 再重复迁移后与当前增量数据合并）
+        // 实际上：首次有请求时 migratedFromLogs 必然为 false，
+        // 所以 readPersistedStats 首次被调用时会重建全量历史，覆盖这个增量写入。
+        // 这是可接受的：迁移完成后统计文件是完整的（含本条目，因为 JSONL 已先写入）。
+        saveStatsFile(file);
+    }
+    catch { /* 统计写入失败不影响主流程 */ }
+}
+/**
+ * 读取所有持久化统计，返回 ModelPerfStats 数组（含分位数）。
+ * 用于 HTML 报告生成，替代 aggregatePerfStats(readLogs(N)) 的有限窗口方案。
+ */
+export function readPersistedStats() {
+    try {
+        const file = loadStatsFile();
+        // 如果尚未迁移（例如首次生成报告前没有任何请求），执行迁移
+        // 迁移时先清空 models，以 JSONL 全量数据为唯一权威来源，
+        // 避免与 updatePersistedStats 先写入的零散增量数据叠加导致重复计数。
+        if (!file.migratedFromLogs) {
+            file.models = {}; // 清空，让迁移从零开始重建
+            migrateFromLogsIfNeeded(file);
+            saveStatsFile(file);
+        }
+        return Object.values(file.models).map(s => {
+            const ttftArr = [...s.ttftReservoir].sort((a, b) => a - b);
+            const latArr = [...s.latencyReservoir].sort((a, b) => a - b);
+            const denom = s.totalInput + s.totalCacheRead;
+            return {
+                model: s.model,
+                providerID: s.providerID,
+                requestCount: s.requestCount,
+                ttftCount: s.ttftCount,
+                tpsCount: s.tpsCount,
+                latencyCount: s.latencyCount,
+                totalInput: s.totalInput,
+                totalOutput: s.totalOutput,
+                totalCacheRead: s.totalCacheRead,
+                totalCacheWrite: s.totalCacheWrite,
+                totalCost: s.totalCost,
+                avgTTFT: s.avgTTFT,
+                maxTTFT: s.maxTTFT,
+                minTTFT: s.minTTFT,
+                p50TTFT: percentile(ttftArr, 50),
+                p95TTFT: percentile(ttftArr, 95),
+                p99TTFT: percentile(ttftArr, 99),
+                avgTPS: s.avgTPS,
+                maxTPS: s.maxTPS,
+                minTPS: s.minTPS,
+                avgLatency: s.avgLatency,
+                maxLatency: s.maxLatency,
+                minLatency: s.minLatency,
+                p50Latency: percentile(latArr, 50),
+                p95Latency: percentile(latArr, 95),
+                p99Latency: percentile(latArr, 99),
+                cacheHitRate: denom > 0 ? (s.totalCacheRead / denom) * 100 : null,
+            };
+        });
+    }
+    catch {
+        return [];
+    }
+}

package/dist/tui.jsx

@@ -1,4 +1,4 @@
-import { createSignal } from "solid-js";
+import { createSignal, createEffect, onCleanup } from "solid-js";
 import { registerCommands } from "./commands.jsx";
 import { createPerfTracker } from "./perf-tracker.js";
 import { TokenWatchPanel } from "./sidebar.jsx";
@@ -78,7 +78,7 @@ const tui = async (api) => {
                 sidebarRevision();
                 if (session_id && session_id !== currentSlotSessionID) {
                     currentSlotSessionID = session_id;
-                    perfTracker.reset();
+                    perfTracker.loadSession(session_id);
                     let loaded = [];
                     try {
                         const saved = api.kv?.get?.(kvKey(session_id));
@@ -86,32 +86,86 @@ const tui = async (api) => {
                             loaded = saved;
                     }
                     catch { }
-                    if (loaded.length === 0) {
-                        const existing = api.state.session.messages(session_id);
-                        for (const msg of existing) {
-                            if (msg.role !== "assistant")
-                                continue;
-                            const tokens = msg.tokens;
-                            if (!tokens)
-                                continue;
-                            loaded.push({
-                                id: msg.id,
-                                sessionID: session_id,
-                                providerID: msg.providerID ?? "unknown",
-                                modelID: msg.modelID ?? "unknown",
-                                inputTokens: tokens?.input ?? 0,
-                                outputTokens: tokens?.output ?? 0,
-                                reasoningTokens: tokens?.reasoning ?? 0,
-                                cacheRead: tokens?.cache?.read ?? 0,
-                                cacheWrite: tokens?.cache?.write ?? 0,
-                                cost: msg.cost ?? 0,
-                            });
-                        }
-                    }
                     setAllTokenMessages(loaded);
                 }
-                const messages = api.state.session.messages(session_id);
-                return <TokenWatchPanel api={api} theme={api.theme} perfTracker={perfTracker} messages={messages} allTokenMessages={allTokenMessages()}/>;
+                // 引入 createEffect 监听历史会话消息在后台异步加载完毕后的变化
+                // 由于历史会话加载可能是异步的，初次检查 messages 可能为空且底层并非响应式数据源，
+                // 故采用短期高频轮询，直到数据加载完成或超时。
+                createEffect(() => {
+                    if (!session_id)
+                        return;
+                    let timer = null;
+                    let pollCount = 0;
+                    const maxPolls = 50; // 最多轮询 10 秒 (50 * 200ms)
+                    const checkAndPopulate = () => {
+                        const existing = api.state.session.messages(session_id);
+                        if (!existing || existing.length === 0)
+                            return false;
+                        setAllTokenMessages((prev) => {
+                            let changed = false;
+                            const next = [...prev];
+                            for (const msg of existing) {
+                                if (msg.role !== "assistant")
+                                    continue;
+                                const tokens = msg.tokens;
+                                if (!tokens)
+                                    continue;
+                                const id = msg.id;
+                                const idx = next.findIndex(m => m.id === id);
+                                const tokenMsg = {
+                                    id,
+                                    sessionID: session_id,
+                                    providerID: msg.providerID ?? "unknown",
+                                    modelID: msg.modelID ?? "unknown",
+                                    inputTokens: tokens?.input ?? 0,
+                                    outputTokens: tokens?.output ?? 0,
+                                    reasoningTokens: tokens?.reasoning ?? 0,
+                                    cacheRead: tokens?.cache?.read ?? 0,
+                                    cacheWrite: tokens?.cache?.write ?? 0,
+                                    cost: msg.cost ?? 0,
+                                };
+                                if (idx >= 0) {
+                                    const cur = next[idx];
+                                    if (cur.inputTokens !== tokenMsg.inputTokens ||
+                                        cur.outputTokens !== tokenMsg.outputTokens ||
+                                        cur.reasoningTokens !== tokenMsg.reasoningTokens ||
+                                        cur.cacheRead !== tokenMsg.cacheRead ||
+                                        cur.cacheWrite !== tokenMsg.cacheWrite ||
+                                        cur.cost !== tokenMsg.cost) {
+                                        next[idx] = tokenMsg;
+                                        changed = true;
+                                    }
+                                }
+                                else {
+                                    next.push(tokenMsg);
+                                    changed = true;
+                                }
+                            }
+                            if (changed) {
+                                persistToKv(session_id, next);
+                                return next;
+                            }
+                            return prev;
+                        });
+                        return true;
+                    };
+                    const hasMessages = checkAndPopulate();
+                    if (!hasMessages) {
+                        timer = setInterval(() => {
+                            pollCount++;
+                            if (checkAndPopulate() || pollCount >= maxPolls) {
+                                clearInterval(timer);
+                                timer = null;
+                            }
+                        }, 200);
+                    }
+                    onCleanup(() => {
+                        if (timer) {
+                            clearInterval(timer);
+                        }
+                    });
+                });
+                return <TokenWatchPanel api={api} theme={api.theme} perfTracker={perfTracker} messages={() => api.state.session.messages(session_id)} allTokenMessages={allTokenMessages}/>;
             },
         },
     });

package/package.json

@@ -1,63 +1,63 @@
-{
-  "name": "opencode-tokenwatch",
-  "version": "0.3.0",
-  "description": "Real-time token usage, cache analytics & performance dashboard plugin for OpenCode CLI",
-  "type": "module",
-  "main": "./dist/index.js",
-  "types": "./dist/index.d.ts",
-  "exports": {
-    ".": {
-      "types": "./dist/index.d.ts",
-      "import": "./dist/index.js"
-    },
-    "./tui": {
-      "types": "./dist/tui.d.ts",
-      "import": "./dist/tui.jsx"
-    },
-    "./package.json": "./package.json"
-  },
-  "files": [
-    "dist"
-  ],
-  "sideEffects": false,
-  "engines": {
-    "node": ">=18"
-  },
-  "scripts": {
-    "build": "tsc",
-    "release:check": "node ./scripts/publish-check.mjs",
-    "prepublishOnly": "npm run build"
-  },
-  "keywords": [
-    "opencode",
-    "plugin",
-    "opencode-plugin",
-    "tokens",
-    "usage",
-    "stats",
-    "sqlite",
-    "analytics",
-    "tui"
-  ],
-  "license": "MIT",
-  "author": "TTWK",
-  "repository": {
-    "type": "git",
-    "url": "git+https://github.com/TTWK/opencode-tokenwatch.git"
-  },
-  "bugs": {
-    "url": "https://github.com/TTWK/opencode-tokenwatch/issues"
-  },
-  "homepage": "https://github.com/TTWK/opencode-tokenwatch#readme",
-  "devDependencies": {
-    "@opencode-ai/plugin": "latest",
-    "@opentui/core": "^0.2.9",
-    "@opentui/keymap": "^0.2.9",
-    "@opentui/solid": "^0.2.9",
-    "@types/node": "^22.0.0",
-    "typescript": "^5.7.0"
-  },
-  "publishConfig": {
-    "access": "public"
-  }
-}
+{
+  "name": "opencode-tokenwatch",
+  "version": "0.3.1",
+  "description": "Real-time token usage, cache analytics & performance dashboard plugin for OpenCode CLI",
+  "type": "module",
+  "main": "./dist/index.js",
+  "types": "./dist/index.d.ts",
+  "exports": {
+    ".": {
+      "types": "./dist/index.d.ts",
+      "import": "./dist/index.js"
+    },
+    "./tui": {
+      "types": "./dist/tui.d.ts",
+      "import": "./dist/tui.jsx"
+    },
+    "./package.json": "./package.json"
+  },
+  "files": [
+    "dist"
+  ],
+  "sideEffects": false,
+  "engines": {
+    "node": ">=18"
+  },
+  "scripts": {
+    "build": "tsc",
+    "release:check": "node ./scripts/publish-check.mjs",
+    "prepublishOnly": "npm run build"
+  },
+  "keywords": [
+    "opencode",
+    "plugin",
+    "opencode-plugin",
+    "tokens",
+    "usage",
+    "stats",
+    "sqlite",
+    "analytics",
+    "tui"
+  ],
+  "license": "MIT",
+  "author": "TTWK",
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/TTWK/opencode-tokenwatch.git"
+  },
+  "bugs": {
+    "url": "https://github.com/TTWK/opencode-tokenwatch/issues"
+  },
+  "homepage": "https://github.com/TTWK/opencode-tokenwatch#readme",
+  "devDependencies": {
+    "@opencode-ai/plugin": "latest",
+    "@opentui/core": "^0.2.9",
+    "@opentui/keymap": "^0.2.9",
+    "@opentui/solid": "^0.2.9",
+    "@types/node": "^22.0.0",
+    "typescript": "^5.7.0"
+  },
+  "publishConfig": {
+    "access": "public"
+  }
+}