update

2026-08-16 21:48:49 +08:00
parent a82f855743
commit 422a4b9d71
4 changed files with 57 additions and 33 deletions
@@ -1,7 +1,7 @@
 # Mini AI Assistant 配置（可由设置页面修改）
-BASE_URL=https://opencode.ai/zen/v1
-API_KEY=sk-ihICiTIZTPdQ0FlhA7ru1kFMpVnxcvHNaK0oxem0ecnz7CpPx0xRoBMCPQhPDmmf
-MODEL=deepseek-v4-flash-free
+BASE_URL=https://ark.cn-beijing.volces.com/api/coding/v3
+API_KEY=ark-ef9b970f-5301-4691-a95e-3f01597ca3f2-b430e
+MODEL=deepseek-v4-flash
 WORKSPACE=D:\workbench\miniai
 CMD_APPROVE_LEVEL=medium
 PORT=3100
@@ -9,3 +9,5 @@ LOG_LEVEL=debug
 TEMPERATURE=0.3
 WHITELIST=d:\
 BLACKLIST=
+DUCKDB_EXE=D:\workbench\miniai\bin\duckdb.exe
+DUCKDB_FILE=D:\workbench\miniai\data\duck.db
@@ -26,3 +26,7 @@ WHITELIST=d:\

 # 路径黑名单（优先级最高，逗号分隔；留空表示无）
 BLACKLIST=
+
+# DuckDB：内置 CLI 可执行文件路径与本地数据库文件（无需原生 .node 依赖）
+DUCKDB_EXE=bin/duckdb.exe
+DUCKDB_FILE=data/duck.db
@@ -13,6 +13,10 @@ _be.*
 # 会话记忆数据
 conversations/

+# 内置 duckdb 可执行文件与运行时数据库
+bin/
+data/
+
 # 环境变量（含密钥，勿提交；参考 .env.example）
 .env
 .env.local
@@ -39,10 +39,12 @@ const config = {
  TEMPERATURE: parseFloat(process.env.TEMPERATURE || ENV.TEMPERATURE || "0.3"),
  WHITELIST: parseList(process.env.WHITELIST || ENV.WHITELIST).length ? parseList(process.env.WHITELIST || ENV.WHITELIST) : DEFAULT_WHITELIST,
  BLACKLIST: parseList(process.env.BLACKLIST || ENV.BLACKLIST),
+  DUCKDB_EXE: process.env.DUCKDB_EXE || ENV.DUCKDB_EXE || resolve("bin", "duckdb.exe"),
+  DUCKDB_FILE: process.env.DUCKDB_FILE || ENV.DUCKDB_FILE || resolve("data", "duck.db"),
 };

 // 持久化配置到 .env（WORKSPACE 为内部默认值，不在设置页暴露）
-const ENV_KEYS = ["BASE_URL", "API_KEY", "MODEL", "WORKSPACE", "CMD_APPROVE_LEVEL", "PORT", "LOG_LEVEL", "TEMPERATURE", "WHITELIST", "BLACKLIST"] as const;
+const ENV_KEYS = ["BASE_URL", "API_KEY", "MODEL", "WORKSPACE", "CMD_APPROVE_LEVEL", "PORT", "LOG_LEVEL", "TEMPERATURE", "WHITELIST", "BLACKLIST", "DUCKDB_EXE", "DUCKDB_FILE"] as const;
 function saveEnv() {
  const lines = ENV_KEYS.map((k) => `${k}=${String((config as any)[k])}`);
  const header = "# Mini AI Assistant 配置（可由设置页面修改）\n";
@@ -88,9 +90,9 @@ function clip(s: any, n = 600): string {

 // ---------- 依赖 ----------
 import { readFile, writeFile, readdir, rename, unlink, stat, mkdir } from "fs/promises";
-import { resolve, normalize, extname, sep } from "path";
+import { resolve, normalize, extname, sep, dirname } from "path";
 import { read as xlsxRead, write as xlsxWrite, utils } from "xlsx";
-import duckdb from "duckdb";
+import { spawn } from "child_process";

 // ---------- 路径安全（白名单 + 黑名单，黑名单优先） ----------
 // 判断 target 是否等于 base 或位于 base 的子目录下（跨平台）
@@ -125,33 +127,38 @@ function isDangerousCommand(cmd: string): boolean {
  return DANGEROUS_PATTERNS.some((p) => lower.includes(p));
 }

-// ---------- DuckDB（Excel 数据处理） ----------
-let duckDb: any = null;
-function getDuck(): Promise<any> {
+// ---------- DuckDB（内置 duckdb.exe 子进程 + 本地数据库文件，无原生 .node 依赖） ----------
+function duckRun(sql: string): Promise<string> {
  return new Promise((resolve, reject) => {
-    if (duckDb) return resolve(duckDb);
-    duckDb = new duckdb.Database(":memory:", (err: any) => {
-      if (err) return reject(err);
-      resolve(duckDb);
+    mkdirSync(dirname(config.DUCKDB_FILE), { recursive: true });
+    const child = spawn(config.DUCKDB_EXE, [config.DUCKDB_FILE, "-json"], { stdio: ["pipe", "pipe", "pipe"] });
+    let out = "", err = "";
+    child.stdout.on("data", (d) => (out += d.toString("utf8")));
+    child.stderr.on("data", (d) => (err += d.toString("utf8")));
+    child.on("error", (e) => reject(new Error("无法启动 duckdb: " + e.message)));
+    child.on("close", (code) => {
+      if (code === 0) resolve(out);
+      else reject(new Error(err.trim() || `duckdb 退出码 ${code}`));
    });
+    child.stdin.write(sql, "utf8");
+    child.stdin.end();
  });
 }
-function duckExec(db: any, sql: string): Promise<void> {
-  return new Promise((resolve, reject) => db.exec(sql, (e: any) => (e ? reject(e) : resolve())));
+async function duckExec(sql: string): Promise<void> {
+  await duckRun(sql);
 }
-function duckAll(db: any, sql: string): Promise<any[]> {
-  return new Promise((resolve, reject) => db.all(sql, (e: any, rows: any) => (e ? reject(e) : resolve(rows || []))));
+async function duckAll(sql: string): Promise<any[]> {
+  const out = await duckRun(sql);
+  const t = out.trim();
+  if (!t) return [];
+  try { return JSON.parse(t); } catch { return []; }
 }
-function sqlVal(v: any): string {
-  if (v === null || v === undefined || v === "") return "NULL";
-  if (typeof v === "number") return Number.isFinite(v) ? String(v) : "NULL";
-  if (typeof v === "boolean") return v ? "TRUE" : "FALSE";
-  if (v instanceof Date) return `'${v.toISOString().replace(/'/g, "''")}'`;
-  return `'${String(v).replace(/'/g, "''")}'`;
-}
-function safeCol(c: string): string {
-  const s = String(c).trim() || "col";
-  return s.replace(/[^\w一-龥]/g, "_");
+function csvCell(v: any): string {
+  if (v === null || v === undefined) return "";
+  if (typeof v === "number") return Number.isFinite(v) ? String(v) : "";
+  if (v instanceof Date) return '"' + v.toISOString().replace(/"/g, '""') + '"';
+  const s = String(v);
+  return /[",\n\r]/.test(s) ? '"' + s.replace(/"/g, '""') + '"' : s;
 }

 // ---------- 工具执行 ----------
@@ -248,7 +255,6 @@ async function execTool(name: string, args: any): Promise<{ ok: boolean; result:
        if (!ws) throw new Error("工作表不存在: " + sheetName);
        const rows: any[] = utils.sheet_to_json(ws, { defval: null });
        const table = args.table || "excel_data";
-        const db = await getDuck();
        if (!rows.length) return { ok: true, result: `工作表 "${sheetName}" 无数据行` };
        const rawCols = Object.keys(rows[0]);
        const seen = new Set<string>();
@@ -272,11 +278,20 @@ async function execTool(name: string, args: any): Promise<{ ok: boolean; result:
          types[c] = hasVal && allNum ? "DOUBLE" : "VARCHAR";
        }
        const colDefs = cols.map((c) => `"${c}" ${types[c]}`).join(", ");
-        await duckExec(db, `CREATE OR REPLACE TABLE "${table}" (${colDefs})`);
        const MAX = 50000;
        const slice = rows.slice(0, MAX);
-        const values = slice.map((r) => `(${cols.map((c) => sqlVal(r[c])).join(", ")})`).join(", ");
-        if (values) await duckExec(db, `INSERT INTO "${table}" VALUES ${values}`);
+        // 写 UTF-8 CSV → duckdb.exe COPY 导入
+        const csvPath = resolve(config.DUCKDB_FILE, "..", `load_${Date.now()}.csv`);
+        mkdirSync(dirname(csvPath), { recursive: true });
+        const csvLines = [cols.join(",")];
+        for (const r of slice) csvLines.push(cols.map((c) => csvCell(r[c])).join(","));
+        writeFileSync(csvPath, csvLines.join("\n"), "utf-8");
+        const csvSql = csvPath.replace(/\\/g, "/");
+        try {
+          await duckExec(`CREATE OR REPLACE TABLE "${table}" (${colDefs}); COPY "${table}" FROM '${csvSql}' (FORMAT csv, HEADER true, DELIMITER ',');`);
+        } finally {
+          try { unlinkSync(csvPath); } catch {}
+        }
        return {
          ok: true,
          result: {
@@ -290,9 +305,8 @@ async function execTool(name: string, args: any): Promise<{ ok: boolean; result:
      case "query_duckdb": {
        const sql = String(args.sql || "").trim();
        if (!sql) throw new Error("缺少 SQL 语句");
-        const db = await getDuck();
        try {
-          const rows = await duckAll(db, sql);
+          const rows = await duckAll(sql);
          const cols = rows.length ? Object.keys(rows[0]) : [];
          return { ok: true, result: { rowCount: rows.length, columns: cols, data: rows.slice(0, 200) } };
        } catch (e: any) {