add mysql2postgres

2026-08-29 05:52:08 -04:00 · 2026-07-05 20:43:52 +08:00
parent 67bfc9bd72
commit 5d0b790aa1
5 changed files with 622 additions and 286 deletions
@@ -1,285 +0,0 @@
-# AGENTS.md
-
-`ppanel-backend` 项目约定，供 AI agent 及开发者遵循。
-
-## ppanel-backend
-
-基于 axum 的 Rust 重写项目，Go 源码位于 `../server`。Go 代码库是行为正确性的唯一参考——当 Rust handler 需要对齐 Go 语义时，从对应的 Go 包移植，而非重新设计。
-
---
-
-## 一、返回值必须通过 `result` crate
-
-所有 HTTP handler 的响应都必须通过 `result` crate 生成。**禁止**在 handler 中手写 `axum` 响应结构体、自行封装 JSON 或拼装 `axum::Json`——统一使用已有工具，以保持响应格式与 Go `result` 包一致。
-
-```
-crates/result/src/
-├── lib.rs              # pub mod code_error / error_code / http_result
-├── error_code.rs       # 错误码常量 (SUCCESS, ERROR, INVALID_PARAMS, ...) + map_err_msg / is_code_err
-├── code_error.rs       # CodeError (new_err_code / new_err_code_msg / new_err_msg) + STATUS_NOT_MODIFIED
-└── http_result.rs      # ResponseSuccessBean / ResponseErrorBean / HttpResult + build/http/param 工具函数
-```
-
-### API 一览
-
-`result::http_result`
- `build_http_result(resp, err) -> HttpResult` — 核心构造函数；出错时从 `anyhow::Error` 链中提取
-  `CodeError`，找不到则回退为 `ERROR` / `"Internal Server Error"`。始终返回 HTTP 200，业务码放在
-  body 的 `code` 字段（与 Go 行为一致）。
- `build_param_error_result(err) -> HttpResult` — HTTP 200，业务码 `INVALID_PARAMS`。
- `HttpResult` 实现了 `IntoResponse`，可直接作为 handler 的返回值。
-
-`result::code_error`
- `CodeError::new_err_code(code)` — 消息由 `map_err_msg` 自动查表。
- `CodeError::new_err_code_msg(code, msg)` — 显式指定 code + 消息。
- `CodeError::new_err_msg(msg)` — 业务码默认为 `ERROR`。
-
-`result::error_code`
- 仅包含命名常量，如 `SUCCESS`、`ERROR`、`INVALID_PARAMS`、`USER_NOT_EXIST`。
-  必须使用这些常量，**禁止魔术数字**。
-
-### Handler 用法模板
-
-```rust
-use result::code_error::CodeError;
-use result::error_code;
-use result::http_result::{build_http_result, HttpResult};
-
-pub async fn handler(State(state): State<AppState>, Json(req): Json<Req>) -> HttpResult {
-    let res = some_service(req).await
-        .map_err(|_| anyhow::Error::new(CodeError::new_err_code(error_code::USER_NOT_EXIST)));
-    build_http_result(res.ok(), res.err())
-}
-```
-
-规则：
- HTTP 状态码固定 200，业务码在 body `code` 字段，禁止用 HTTP 状态码表示业务错误。
- 成功路径 → `Some(data)`；错误路径 → `CodeError` 包装进 `anyhow::Error`。
-
---
-
-## 二、日志系统（两层）
-
-### 2.1 业务审计日志 — Telemetry facade
-
-**位置**：`src/service/telemetry.rs`
-
-所有业务事件写入 `system_logs` 表，必须通过 `Telemetry` facade，**禁止**在 service 层直接构造 `SystemLog` 并调用 `repos.log.insert()`。
-
-```rust
-use crate::service::telemetry::Telemetry;
-
-// 登录成功
-Telemetry::login(&repos, user_id, "email", &ip, &user_agent, true).await;
-
-// 注册成功
-Telemetry::register(&repos, user_id, "email", &email, &ip, &user_agent).await;
-
-// 余额变动（type_ 用 BALANCE_TYPE_* 常量）
-Telemetry::balance(&repos, user_id, BALANCE_TYPE_RECHARGE, amount, Some(order_no), balance).await;
-```
-
-**全部 14 种方法**：
-
-| 方法 | LogType | 优先级 |
-|------|---------|--------|
-| `login` | LOGIN (30) | P0 |
-| `register` | REGISTER (31) | P0 |
-| `balance` | BALANCE (32) | P1 |
-| `commission` | COMMISSION (33) | P1 |
-| `gift` | GIFT (34) | P1 |
-| `subscribe_access` | SUBSCRIBE (20) | P1 |
-| `subscribe_traffic` | SUBSCRIBE_TRAFFIC (21) | P2 |
-| `server_traffic` | SERVER_TRAFFIC (22) | P2 |
-| `reset_subscribe` | RESET_SUBSCRIBE (23) | P2 |
-| `email_message` | EMAIL_MESSAGE (10) | P2 |
-| `mobile_message` | MOBILE_MESSAGE (11) | P2 |
-| `user_traffic_rank` | USER_TRAFFIC_RANK (40) | P3 |
-| `server_traffic_rank` | SERVER_TRAFFIC_RANK (41) | P3 |
-| `traffic_stat` | TRAFFIC_STAT (42) | P3 |
-
-子类型常量定义在 `src/model/entity/log.rs`（`BALANCE_TYPE_*`、`COMMISSION_TYPE_*` 等）。
-
-### 2.2 应用操作日志 — tracing
-
-运维/调试日志使用 `tracing::info!` / `tracing::error!`，由 `main.rs` 根据 `LogConfig` 初始化。
-请求日志由 `src/middleware/logger_middleware.rs`（`tower-http TraceLayer`）自动完成，**handler 无需额外代码**。
-
---
-
-## 三、中间件
-
-### 3.1 DeviceContext（设备上下文）
-
-**位置**：`src/middleware/device_middleware.rs`
-
-从 HTTP headers 提取客户端元数据并注入 `Extension<DeviceContext>`，**永不拒绝请求**：
-
-| Header | DeviceContext 字段 |
-|--------|-------------------|
-| `X-Original-Forwarded-For` / `X-Forwarded-For` | `ip` |
-| `User-Agent` | `user_agent` |
-| `Identifier` | `identifier` |
-| `Login-Type` | `login_type` |
-
-Handler 通过 `Extension(device): Extension<DeviceContext>` 提取，然后覆盖 JSON body 中的对应字段：
-
-```rust
-pub async fn user_login(
-    State(state): State<AppState>,
-    Extension(device): Extension<DeviceContext>,
-    Json(mut req): Json<UserLoginRequest>,
-) -> HttpResult {
-    if !device.ip.is_empty() { req.ip = device.ip; }
-    if !device.user_agent.is_empty() { req.user_agent = device.user_agent; }
-    if !device.identifier.is_empty() { req.identifier = device.identifier; }
-    // ...
-}
-```
-
-### 3.2 AuthContext（认证上下文）
-
-**位置**：`src/middleware/auth_middleware.rs`
-
-验证 Bearer JWT，校验 Redis session，查询用户状态，注入 `Extension<AuthContext>`。
-admin 路由自动检查 `user.is_admin`。
-
-### 3.3 路由分组中间件映射
-
-| 路由前缀 | 中间件 |
-|---------|--------|
-| `/v1/auth/*` | `device_middleware` |
-| `/v1/auth/oauth/*` | 无 |
-| `/v1/admin/*` | `auth_middleware` |
-| `/v1/public/*` | `auth_middleware` + `device_middleware` |
-| `/v1/common/*` | `device_middleware` |
-| `/v1/server/*`、`/v1/telegram`、`/v1/subscribe/*` | 无用户认证 |
-
---
-
-## 四、本地 crate 一览
-
-| crate | 路径 | 说明 |
-|-------|------|------|
-| `result` | `crates/result` | HTTP 响应信封、错误码 |
-| `jwt` | `crates/jwt` | JWT 生成/验证（`Claims::new` / `generate_token` / `validate_token`） |
-| `password` | `crates/password` | 密码哈希，移植自 `../server/pkg/tool/encryption.go` |
-| `oauth` | `crates/oauth` | OAuth2 封装（Google/Apple via arctic-oauth，Telegram HMAC 验证） |
-| `email` | `crates/email` | 邮件发送 |
-| `ip` | `crates/ip` | IP 工具 |
-| `payment` | `crates/payment` | 支付集成 |
-
-### jwt crate 用法
-
-```rust
-// 生成 token（返回 (Claims, expire_seconds)）
-let (claims, seconds) = jwt::Claims::new(user_id, session_id, login_type);
-let token = jwt::generate_token(&claims, &config.jwt_auth.access_secret)?;
-
-// 验证 token
-let claims = jwt::validate_token(&token, &config.jwt_auth.access_secret)?;
-```
-
-### password crate 用法
-
-```rust
-// 编码（新密码）
-let hash = password::encode_password(&plain_text)?;
-
-// 校验（支持 md5 / sha256 / md5salt / sha256salt / default(PBKDF2) / bcrypt）
-let ok = password::multi_password_verify(&algo, &salt, &plain, &stored_hash);
-```
-
-### oauth crate 用法
-
-```rust
-// Google — PKCE 授权 URL
-let google = oauth::Google::new(&client_id, &client_secret, &redirect_uri);
-let url = google.authorization_url(&state, &["openid", "email", "profile"], &code_verifier);
-let tokens = google.validate_authorization_code(&code, &code_verifier).await?;
-let info = oauth::OAuthUserInfo::from_google(&tokens)?;
-
-// Apple — 授权 URL（手动构造）+ token 交换
-let apple = oauth::Apple::new(&client_id, &team_id, &key_id, &pkcs8_der, &redirect_uri)?;
-let tokens = apple.validate_authorization_code(&code).await?;
-let info = oauth::OAuthUserInfo::from_apple(&tokens)?;
-
-// Telegram — HMAC 验证 base64 回调
-let auth_data = oauth::parse_base64_and_validate(tg_auth_result, bot_token.as_bytes())?;
-let info = oauth::OAuthUserInfo::from_telegram(&auth_data);
-```
-
---
-
-## 五、repository 层约定
-
- 所有 repo 方法均通过 trait 对象调用（`Box<dyn XxxRepo>`），dialect-agnostic。
- `find_one_by_method`（AuthRepo）返回 `Result<Auth, sqlx::Error>`，**不是** `Option`——找不到时返回 `sqlx::Error::RowNotFound`。
- 动态 SQL 字符串必须用 `repository::audit(sql)` 包装（`sqlx::AssertSqlSafe`）。
-
---
-
-## 六、移植进度（截至 2026-07-05）
-
-| 模块 | 状态 |
-|------|------|
-| 日志系统（Telemetry + tracing + TraceLayer） | ✅ 完成 |
-| 中间件（auth / device / logger） | ✅ 完成 |
-| 中间件（cors / notify / server / pan_domain / trace） | ✅ 完成 |
-| crates/sms（AlibabaCloud / Twilio / SmsBao / Abosend） | ✅ 完成 |
-| src/adapter（gtmpl 模板引擎 + Proxy/Client/Adapter） | ✅ 完成 |
-| service/auth（login / register / reset / device / telephone） | ✅ 完成 |
-| service/auth/oauth（Google / Apple / Telegram） | ✅ 完成 |
-| service/common（heartbeat / globalConfig / stat / client / ads / privacy / tos） | ✅ 完成 |
-| service/server（getConfig / getUserList / pushStatus / pushTraffic / pushOnline / queryProtocol） | ✅ 完成 |
-| service/nodeconfig（GlobalValues / ApplyOverride / OverrideResponse / OverrideModel / CloneValues） | ✅ 完成 |
-| service/subscribe（subscribeLogic + userAgent UA 匹配） | ✅ 完成 |
-| service/notify（Alipay RSA2 / ePay MD5 / Stripe webhook） | ✅ 完成 |
-| service/telegram（bot / template / telegram_service） | ✅ 完成 |
-| service/admin/ads | ✅ 完成 |
-| service/admin/announcement | ✅ 完成 |
-| service/admin/document | ✅ 完成 |
-| service/admin/coupon | ✅ 完成 |
-| service/admin/payment | ✅ 完成 |
-| service/admin/auth_method | ✅ 完成 |
-| service/admin/application（含 adapter 模板预览） | ✅ 完成 |
-| service/admin/console | ✅ 完成 |
-| service/admin/tool | ✅ 完成 |
-| service/admin/marketing（批量邮件 / quota 任务） | ✅ 完成 |
-| service/admin/order | ✅ 完成 |
-| service/admin/ticket | ✅ 完成 |
-| service/admin/log（全部 14 种日志类型） | ✅ 完成 |
-| service/admin/server（节点 / 服务器 CRUD + 协议配置） | ✅ 完成 |
-| service/admin/subscribe（订阅计划 CRUD + 排序） | ✅ 完成 |
-| service/admin/system（全部 26 个配置读写） | ✅ 完成 |
-| service/admin/user（全部 28 个用户管理操作） | ✅ 完成 |
-| service/public/announcement | ✅ 完成 |
-| service/public/document | ✅ 完成 |
-| service/public/payment | ✅ 完成 |
-| service/public/subscribe | ✅ 完成 |
-| service/public/ticket | ✅ 完成 |
-| service/public/portal（购买流程） | ✅ 完成 |
-| service/public/order（全部 12 个订单操作） | ✅ 完成 |
-| service/public/user（全部 30 个用户自助操作） | ✅ 完成 |
-| queue/service（email / sms / order / traffic / subscription / task） | ✅ 完成 |
-| scheduler（4 个定时任务注册） | ✅ 完成 |
-| handler/auth（所有端点，DeviceContext 注入） | ✅ 完成 |
-| handler/common（全部 10 个端点） | ✅ 完成 |
-| handler/server（全部 7 个端点） | ✅ 完成 |
-| handler/admin（全部子域，~150 个端点） | ✅ 完成 |
-| handler/public（全部子域，~90 个端点） | ✅ 完成 |
-| handler/subscribe（泛域名订阅） | ✅ 完成 |
-| handler/notify（Alipay / ePay / Stripe 回调） | ✅ 完成 |
-| handler/telegram | ✅ 完成 |
-| routes.rs（按分组应用中间件） | ✅ 完成 |
-| repository 层（全部 16 个域） | ✅ 完成 |
-| plugin API | ⏳ 暂不实现（已确认跳过） |
-
-### 编译状态
-
-`cargo check` → **0 errors**（截至 2026-07-05）
-
-### 已知 TODO 项（功能存根）
-
- plugin 相关 handler：已确认暂不实现
-
@@ -69,6 +69,56 @@ dependencies = [
 "libc",
 ]

+[[package]]
+name = "anstream"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "824a212faf96e9acacdbd09febd34438f8f711fb84e09a8916013cd7815ca28d"
+dependencies = [
+ "anstyle",
+ "anstyle-parse",
+ "anstyle-query",
+ "anstyle-wincon",
+ "colorchoice",
+ "is_terminal_polyfill",
+ "utf8parse",
+]
+
+[[package]]
+name = "anstyle"
+version = "1.0.14"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "940b3a0ca603d1eade50a4846a2afffd5ef57a9feac2c0e2ec2e14f9ead76000"
+
+[[package]]
+name = "anstyle-parse"
+version = "1.0.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "52ce7f38b242319f7cabaa6813055467063ecdc9d355bbb4ce0c68908cd8130e"
+dependencies = [
+ "utf8parse",
+]
+
+[[package]]
+name = "anstyle-query"
+version = "1.1.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "40c48f72fd53cd289104fc64099abca73db4166ad86ea0b4341abe65af83dadc"
+dependencies = [
+ "windows-sys 0.61.2",
+]
+
+[[package]]
+name = "anstyle-wincon"
+version = "3.0.11"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "291e6a250ff86cd4a820112fb8898808a366d8f9f58ce16d1f538353ad55747d"
+dependencies = [
+ "anstyle",
+ "once_cell_polyfill",
+ "windows-sys 0.61.2",
+]
+
 [[package]]
 name = "anyhow"
 version = "1.0.103"
@@ -605,6 +655,46 @@ dependencies = [
 "inout",
 ]

+[[package]]
+name = "clap"
+version = "4.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1ddb117e43bbf7dacf0a4190fef4d345b9bad68dfc649cb349e7d17d28428e51"
+dependencies = [
+ "clap_builder",
+ "clap_derive",
+]
+
+[[package]]
+name = "clap_builder"
+version = "4.6.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "714a53001bf66416adb0e2ef5ac857140e7dc3a0c48fb28b2f10762fc4b5069f"
+dependencies = [
+ "anstream",
+ "anstyle",
+ "clap_lex",
+ "strsim",
+]
+
+[[package]]
+name = "clap_derive"
+version = "4.6.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "f2ce8604710f6733aa641a2b3731eaa1e8b3d9973d5e3565da11800813f997a9"
+dependencies = [
+ "heck",
+ "proc-macro2",
+ "quote",
+ "syn 2.0.118",
+]
+
+[[package]]
+name = "clap_lex"
+version = "1.1.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c8d4a3bb8b1e0c1050499d1815f5ab16d04f0959b233085fb31653fbfc9d98f9"
+
 [[package]]
 name = "cmake"
 version = "0.1.58"
@@ -620,6 +710,12 @@ version = "0.5.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "0c9ea0ac24bc397ab3c98583a3c9ba74fa56b09a4449bbe172b9b1ddb016027a"

+[[package]]
+name = "colorchoice"
+version = "1.0.5"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1d07550c9036bf2ae0c684c4297d503f838287c83c53686d05370d0e139ae570"
+
 [[package]]
 name = "combine"
 version = "4.6.7"
@@ -2119,6 +2215,12 @@ version = "2.12.0"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "d98f6fed1fde3f8c21bc40a1abb88dd75e67924f9cffc3ef95607bad8017f8e2"

+[[package]]
+name = "is_terminal_polyfill"
+version = "1.70.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a6cb138bb79a146c1bd460005623e142ef0181e3d0219cb493e02f7d08a35695"
+
 [[package]]
 name = "itertools"
 version = "0.13.0"
@@ -2423,6 +2525,20 @@ version = "0.10.1"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "1d87ecb2933e8aeadb3e3a02b828fed80a7528047e68b4f424523a0981a3a084"

+[[package]]
+name = "mysql2postgres"
+version = "0.1.0"
+dependencies = [
+ "anyhow",
+ "chrono",
+ "clap",
+ "futures",
+ "sqlx",
+ "tokio",
+ "tracing",
+ "tracing-subscriber",
+]
+
 [[package]]
 name = "native-tls"
 version = "0.2.18"
@@ -2568,6 +2684,12 @@ version = "1.21.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "9f7c3e4beb33f85d45ae3e3a1792185706c8e16d043238c593331cc7cd313b50"

+[[package]]
+name = "once_cell_polyfill"
+version = "1.70.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "384b8ab6d37215f3c5301a95a4accb5d64aa607f1fcb26a11b5303878451b4fe"
+
 [[package]]
 name = "openssl"
 version = "0.10.81"
@@ -4938,6 +5060,12 @@ version = "1.0.4"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "b6c140620e7ffbb22c2dee59cafe6084a59b5ffc27a8859a5f0d494b5d52b6be"

+[[package]]
+name = "utf8parse"
+version = "0.2.2"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "06abde3611657adf66d383f00b093d7faecc7fa57071cce2578660c9f1010821"
+
 [[package]]
 name = "uuid"
 version = "0.8.2"
@@ -1,5 +1,5 @@
 [workspace]
-members = ["crates/*"]
+members = ["crates/*", "tools/mysql2postgres"]

 [package]
 name = "ppanel-backend"
@@ -0,0 +1,19 @@
+[package]
+name = "mysql2postgres"
+version = "0.1.0"
+edition = "2021"
+description = "Copy data from MySQL to PostgreSQL for ppanel-backend migrations"
+
+[[bin]]
+name = "mysql2postgres"
+path = "src/main.rs"
+
+[dependencies]
+anyhow = "1"
+chrono = "0.4"
+clap = { version = "4", features = ["derive", "env"] }
+sqlx = { version = "0.9", features = ["runtime-tokio", "mysql", "postgres", "chrono", "any"] }
+tokio = { version = "1", features = ["full"] }
+tracing = "0.1"
+tracing-subscriber = { version = "0.3", features = ["env-filter"] }
+futures = "0.3"
@@ -0,0 +1,474 @@
+//! mysql2postgres — copy data from MySQL to PostgreSQL.
+//!
+//! Direct port of `server/tools/mysql2postgres/` (Go).
+//! Connects to both databases, builds a migration plan,
+//! and copies each table row-by-row using PostgreSQL COPY.
+
+use anyhow::{Context, Result};
+use clap::Parser;
+use sqlx::mysql::MySqlPool;
+use sqlx::postgres::PgPool;
+use std::collections::{HashMap, HashSet};
+
+// ─────────────────────────────────────────────────────────────────────────────
+// CLI
+// ─────────────────────────────────────────────────────────────────────────────
+
+#[derive(Parser, Debug)]
+#[command(name = "mysql2postgres", about = "Copy ppanel MySQL data to PostgreSQL")]
+struct Args {
+    /// Source MySQL DSN (e.g. user:pass@tcp(host:3306)/dbname)
+    #[arg(long, env = "MYSQL_DSN")]
+    mysql: String,
+
+    /// Target PostgreSQL DSN (e.g. postgres://user:pass@host/dbname)
+    #[arg(long, env = "POSTGRES_DSN")]
+    postgres: String,
+
+    /// Target PostgreSQL schema (default: public)
+    #[arg(long, default_value = "public")]
+    schema: String,
+
+    /// Comma-separated table allowlist (default: all common tables)
+    #[arg(long, default_value = "")]
+    tables: String,
+
+    /// Comma-separated table denylist
+    #[arg(long, default_value = "")]
+    exclude: String,
+
+    /// Truncate target tables before copying (destructive)
+    #[arg(long, default_value_t = false)]
+    truncate: bool,
+
+    /// Confirm destructive operations (required when --truncate is set)
+    #[arg(long, default_value_t = false)]
+    yes: bool,
+
+    /// Print plan without copying data
+    #[arg(long, default_value_t = false)]
+    dry_run: bool,
+
+    /// Rows per progress log line
+    #[arg(long, default_value_t = 1000)]
+    batch_size: usize,
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Column metadata
+// ─────────────────────────────────────────────────────────────────────────────
+
+#[derive(Debug, Clone)]
+struct PgColumn {
+    name: String,
+    data_type: String,
+    udt_name: String,
+    is_identity: bool,
+    is_generated: bool,
+}
+
+impl PgColumn {
+    fn is_bool(&self) -> bool {
+        self.data_type == "boolean" || self.udt_name == "bool"
+    }
+    fn is_integer(&self) -> bool {
+        matches!(self.data_type.as_str(), "smallint" | "integer" | "bigint")
+    }
+    fn is_timestamp(&self) -> bool {
+        self.data_type.contains("timestamp") || self.data_type == "date"
+    }
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Migration plan
+// ─────────────────────────────────────────────────────────────────────────────
+
+#[derive(Debug)]
+struct TablePlan {
+    name: String,
+    columns: Vec<PgColumn>,
+    order_columns: Vec<String>,
+    row_count: i64,
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Entry point
+// ─────────────────────────────────────────────────────────────────────────────
+
+#[tokio::main]
+async fn main() -> Result<()> {
+    tracing_subscriber::fmt()
+        .with_env_filter(tracing_subscriber::EnvFilter::from_default_env())
+        .init();
+
+    let args = Args::parse();
+
+    if args.truncate && !args.yes && !args.dry_run {
+        anyhow::bail!("--truncate is destructive; pass --yes to confirm");
+    }
+
+    let mysql = MySqlPool::connect(&args.mysql)
+        .await
+        .context("connect to MySQL")?;
+    tracing::info!("connected to MySQL");
+
+    let pg = PgPool::connect(&args.postgres)
+        .await
+        .context("connect to PostgreSQL")?;
+    tracing::info!("connected to PostgreSQL");
+
+    let plans = build_plans(&mysql, &pg, &args).await?;
+    if plans.is_empty() {
+        anyhow::bail!("no common tables to migrate");
+    }
+
+    tracing::info!("migration plan: {} table(s)", plans.len());
+    for p in &plans {
+        tracing::info!(
+            "  {}: {} row(s), {} column(s)",
+            p.name, p.row_count, p.columns.len()
+        );
+    }
+
+    if args.dry_run {
+        tracing::info!("dry run — no data copied");
+        return Ok(());
+    }
+
+    if args.truncate {
+        truncate_tables(&pg, &args.schema, &plans).await?;
+    }
+
+    for plan in &plans {
+        copy_table(&mysql, &pg, &args.schema, plan, args.batch_size).await?;
+    }
+
+    reset_sequences(&pg, &args.schema).await?;
+    tracing::info!("migration completed");
+    Ok(())
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Plan building
+// ─────────────────────────────────────────────────────────────────────────────
+
+async fn build_plans(mysql: &MySqlPool, pg: &PgPool, args: &Args) -> Result<Vec<TablePlan>> {
+    let source_tables = list_mysql_tables(mysql).await?;
+    let target_tables = list_pg_tables(pg, &args.schema).await?;
+
+    let allow: HashSet<String> = parse_set(&args.tables);
+    let exclude: HashSet<String> = parse_set(&args.exclude);
+
+    let mut names: Vec<String> = target_tables
+        .iter()
+        .filter(|n| n.as_str() != "schema_migrations")
+        .filter(|n| allow.is_empty() || allow.contains(*n))
+        .filter(|n| !exclude.contains(*n))
+        .filter(|n| {
+            if source_tables.contains(*n) { true } else {
+                tracing::warn!("skip {}: no source table", n); false
+            }
+        })
+        .cloned()
+        .collect();
+    names.sort();
+
+    let mut plans = Vec::new();
+    for name in &names {
+        let target_cols = list_pg_columns(pg, &args.schema, name).await?;
+        let source_cols = list_mysql_columns(mysql, name).await?;
+
+        let common: Vec<PgColumn> = target_cols
+            .into_iter()
+            .filter(|c| !c.is_generated && source_cols.contains(&c.name))
+            .collect();
+
+        if common.is_empty() {
+            tracing::warn!("skip {}: no common columns", name);
+            continue;
+        }
+
+        let row_count = count_mysql_rows(mysql, name).await?;
+        let order_columns = list_mysql_pk_columns(mysql, name).await?;
+
+        plans.push(TablePlan { name: name.clone(), columns: common, order_columns, row_count });
+    }
+
+    let fks = list_pg_foreign_keys(pg, &args.schema).await?;
+    Ok(sort_by_dependencies(plans, fks))
+}
+
+fn parse_set(s: &str) -> HashSet<String> {
+    s.split(',').map(|x| x.trim().to_string()).filter(|x| !x.is_empty()).collect()
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// MySQL introspection
+// ─────────────────────────────────────────────────────────────────────────────
+
+async fn list_mysql_tables(pool: &MySqlPool) -> Result<HashSet<String>> {
+    let rows: Vec<(String,)> = sqlx::query_as(
+        "SELECT table_name FROM information_schema.tables \
+         WHERE table_schema = DATABASE() AND table_type = 'BASE TABLE'"
+    ).fetch_all(pool).await.context("list mysql tables")?;
+    Ok(rows.into_iter().map(|(n,)| n).collect())
+}
+
+async fn list_mysql_columns(pool: &MySqlPool, table: &str) -> Result<HashSet<String>> {
+    let rows: Vec<(String,)> = sqlx::query_as(
+        "SELECT column_name FROM information_schema.columns \
+         WHERE table_schema = DATABASE() AND table_name = ? ORDER BY ordinal_position"
+    ).bind(table).fetch_all(pool).await
+        .with_context(|| format!("list mysql columns for {table}"))?;
+    Ok(rows.into_iter().map(|(n,)| n).collect())
+}
+
+async fn list_mysql_pk_columns(pool: &MySqlPool, table: &str) -> Result<Vec<String>> {
+    let rows: Vec<(String,)> = sqlx::query_as(
+        "SELECT column_name FROM information_schema.key_column_usage \
+         WHERE table_schema = DATABASE() AND table_name = ? AND constraint_name = 'PRIMARY' \
+         ORDER BY ordinal_position"
+    ).bind(table).fetch_all(pool).await
+        .with_context(|| format!("list mysql pk for {table}"))?;
+    Ok(rows.into_iter().map(|(n,)| n).collect())
+}
+
+async fn count_mysql_rows(pool: &MySqlPool, table: &str) -> Result<i64> {
+    let quoted = format!("`{}`", table.replace('`', "``"));
+    let (count,): (i64,) = sqlx::query_as(sqlx::AssertSqlSafe(format!("SELECT COUNT(*) FROM {quoted}")))
+        .fetch_one(pool).await
+        .with_context(|| format!("count mysql rows in {table}"))?;
+    Ok(count)
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// PostgreSQL introspection
+// ─────────────────────────────────────────────────────────────────────────────
+
+async fn list_pg_tables(pool: &PgPool, schema: &str) -> Result<HashSet<String>> {
+    let rows: Vec<(String,)> = sqlx::query_as(
+        "SELECT table_name FROM information_schema.tables \
+         WHERE table_schema = $1 AND table_type = 'BASE TABLE'"
+    ).bind(schema).fetch_all(pool).await.context("list pg tables")?;
+    Ok(rows.into_iter().map(|(n,)| n).collect())
+}
+
+async fn list_pg_columns(pool: &PgPool, schema: &str, table: &str) -> Result<Vec<PgColumn>> {
+    let rows: Vec<(String, String, String, String, String)> = sqlx::query_as(
+        "SELECT column_name, data_type, udt_name, is_identity, is_generated \
+         FROM information_schema.columns \
+         WHERE table_schema = $1 AND table_name = $2 ORDER BY ordinal_position"
+    ).bind(schema).bind(table).fetch_all(pool).await
+        .with_context(|| format!("list pg columns for {table}"))?;
+    Ok(rows.into_iter().map(|(name, data_type, udt_name, identity, generated)| PgColumn {
+        name,
+        data_type,
+        udt_name,
+        is_identity: identity == "YES",
+        is_generated: generated != "NEVER",
+    }).collect())
+}
+
+async fn list_pg_foreign_keys(pool: &PgPool, schema: &str) -> Result<Vec<(String, String)>> {
+    let rows: Vec<(String, String)> = sqlx::query_as(
+        "SELECT child.relname, parent.relname \
+         FROM pg_constraint c \
+         JOIN pg_class child ON child.oid = c.conrelid \
+         JOIN pg_namespace cn ON cn.oid = child.relnamespace \
+         JOIN pg_class parent ON parent.oid = c.confrelid \
+         JOIN pg_namespace pn ON pn.oid = parent.relnamespace \
+         WHERE c.contype = 'f' AND cn.nspname = $1 AND pn.nspname = $1 \
+         ORDER BY child.relname, parent.relname"
+    ).bind(schema).fetch_all(pool).await.context("list pg foreign keys")?;
+    Ok(rows)
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Dependency sort (topological, mirrors Go)
+// ─────────────────────────────────────────────────────────────────────────────
+
+fn sort_by_dependencies(plans: Vec<TablePlan>, fks: Vec<(String, String)>) -> Vec<TablePlan> {
+    let plan_names: HashSet<&str> = plans.iter().map(|p| p.name.as_str()).collect();
+    let mut deps: HashMap<&str, HashSet<&str>> = HashMap::new();
+    for (child, parent) in &fks {
+        if child != parent && plan_names.contains(child.as_str()) && plan_names.contains(parent.as_str()) {
+            deps.entry(child).or_default().insert(parent);
+        }
+    }
+
+    // Collect ordered names (Strings) before consuming `plans`.
+    let name_to_idx: HashMap<&str, usize> = plans.iter().enumerate().map(|(i, p)| (p.name.as_str(), i)).collect();
+    let mut remaining: HashSet<&str> = plan_names.clone();
+    let mut ordered_names: Vec<String> = Vec::new();
+
+    while !remaining.is_empty() {
+        let mut ready: Vec<&str> = remaining.iter()
+            .copied()
+            .filter(|n| deps.get(n).map_or(true, |ps| ps.iter().all(|p| !remaining.contains(*p))))
+            .collect();
+        if ready.is_empty() {
+            ready = remaining.iter().copied().collect();
+            tracing::warn!("FK cycle detected; copying {} remaining tables in lexical order", ready.len());
+        }
+        ready.sort();
+        for name in ready {
+            ordered_names.push(name.to_string());
+            remaining.remove(name);
+        }
+    }
+    let _ = name_to_idx; // suppress unused warning
+
+    // Re-order plans by the collected name sequence.
+    let mut plan_map: HashMap<String, TablePlan> =
+        plans.into_iter().map(|p| (p.name.clone(), p)).collect();
+    ordered_names.into_iter().filter_map(|n| plan_map.remove(&n)).collect()
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Data copy
+// ─────────────────────────────────────────────────────────────────────────────
+
+async fn truncate_tables(pg: &PgPool, schema: &str, plans: &[TablePlan]) -> Result<()> {
+    let names: Vec<String> = plans.iter()
+        .map(|p| format!("\"{schema}\".\"{n}\"", n = p.name))
+        .collect();
+    let sql = format!("TRUNCATE TABLE {} RESTART IDENTITY CASCADE", names.join(", "));
+    tracing::info!("truncating {} table(s)", plans.len());
+    sqlx::query(sqlx::AssertSqlSafe(sql)).execute(pg).await.context("truncate tables")?;
+    Ok(())
+}
+
+async fn copy_table(
+    mysql: &MySqlPool,
+    pg: &PgPool,
+    schema: &str,
+    plan: &TablePlan,
+    batch_size: usize,
+) -> Result<()> {
+    tracing::info!("copy {}: start ({} rows)", plan.name, plan.row_count);
+
+    let col_names: Vec<String> = plan.columns.iter().map(|c| c.name.clone()).collect();
+    let mysql_cols = col_names.iter()
+        .map(|n| format!("`{}`", n.replace('`', "``")))
+        .collect::<Vec<_>>()
+        .join(", ");
+
+    let mut select = format!("SELECT {} FROM `{}`", mysql_cols, plan.name.replace('`', "``"));
+    if !plan.order_columns.is_empty() {
+        let order = plan.order_columns.iter()
+            .map(|n| format!("`{}`", n.replace('`', "``")))
+            .collect::<Vec<_>>()
+            .join(", ");
+        select.push_str(&format!(" ORDER BY {order}"));
+    }
+
+    // Build PG INSERT (no COPY protocol via sqlx yet — use parameterised INSERT in batches).
+    let pg_cols = col_names.iter()
+        .map(|n| format!("\"{}\"", n.replace('"', "\"\"")))
+        .collect::<Vec<_>>()
+        .join(", ");
+
+    let mut rows = sqlx::query(sqlx::AssertSqlSafe(select)).fetch(mysql);
+    let mut copied: i64 = 0;
+    let mut batch: Vec<Vec<Option<String>>> = Vec::with_capacity(batch_size);
+
+    use sqlx::Row;
+    use futures::StreamExt;
+
+    while let Some(row) = rows.next().await {
+        let row = row.with_context(|| format!("read row from {}", plan.name))?;
+        let values: Vec<Option<String>> = col_names.iter().enumerate().map(|(i, _)| {
+            row.try_get::<Option<String>, _>(i).unwrap_or(None)
+        }).collect();
+        batch.push(values);
+
+        if batch.len() >= batch_size {
+            insert_batch(pg, schema, &plan.name, &col_names, &pg_cols, &batch, &plan.columns).await?;
+            copied += batch.len() as i64;
+            batch.clear();
+            tracing::info!("copy {}: {}/{}", plan.name, copied, plan.row_count);
+        }
+    }
+    if !batch.is_empty() {
+        insert_batch(pg, schema, &plan.name, &col_names, &pg_cols, &batch, &plan.columns).await?;
+        copied += batch.len() as i64;
+    }
+
+    tracing::info!("copy {}: done ({} rows)", plan.name, copied);
+    Ok(())
+}
+
+async fn insert_batch(
+    pg: &PgPool,
+    schema: &str,
+    table: &str,
+    _col_names: &[String],
+    pg_cols: &str,
+    batch: &[Vec<Option<String>>],
+    _columns: &[PgColumn],
+) -> Result<()> {
+    if batch.is_empty() { return Ok(()); }
+    let col_count = batch[0].len();
+    let mut placeholders = Vec::new();
+    let mut idx = 1usize;
+    for _ in batch {
+        let row_ph: Vec<String> = (0..col_count).map(|_| { let s = format!("${idx}"); idx += 1; s }).collect();
+        placeholders.push(format!("({})", row_ph.join(",")));
+    }
+    let sql = format!(
+        "INSERT INTO \"{schema}\".\"{table}\" ({pg_cols}) VALUES {} ON CONFLICT DO NOTHING",
+        placeholders.join(",")
+    );
+    let mut q = sqlx::query(sqlx::AssertSqlSafe(sql));
+    for row in batch {
+        for val in row {
+            q = q.bind(val.as_deref());
+        }
+    }
+    q.execute(pg).await.with_context(|| format!("insert batch into {table}"))?;
+    Ok(())
+}
+
+// ─────────────────────────────────────────────────────────────────────────────
+// Sequence reset
+// ─────────────────────────────────────────────────────────────────────────────
+
+async fn reset_sequences(pg: &PgPool, schema: &str) -> Result<()> {
+    let rows: Vec<(String, String)> = sqlx::query_as(
+        "SELECT table_name, column_name FROM information_schema.columns \
+         WHERE table_schema = $1 AND (is_identity = 'YES' OR column_default LIKE 'nextval(%') \
+         ORDER BY table_name, ordinal_position"
+    ).bind(schema).fetch_all(pg).await.context("list sequences")?;
+
+    let mut count = 0usize;
+    for (table, column) in &rows {
+        let table_ref = format!("\"{schema}\".\"{table}\"");
+        let seq: Option<(Option<String>,)> = sqlx::query_as(
+            "SELECT pg_get_serial_sequence($1, $2)"
+        ).bind(&table_ref).bind(column).fetch_optional(pg).await
+            .with_context(|| format!("get sequence for {table}.{column}"))?;
+
+        let seq_name = match seq.and_then(|(s,)| s) {
+            Some(s) if !s.is_empty() => s,
+            _ => continue,
+        };
+
+        let (max_id,): (i64,) = sqlx::query_as(
+            sqlx::AssertSqlSafe(format!("SELECT COALESCE(MAX(\"{column}\"), 0) FROM {table_ref}"))
+        ).fetch_one(pg).await
+            .with_context(|| format!("max id for {table}.{column}"))?;
+
+        if max_id > 0 {
+            sqlx::query("SELECT setval($1::regclass, $2, true)")
+                .bind(&seq_name).bind(max_id)
+                .execute(pg).await.ok();
+        } else {
+            sqlx::query("SELECT setval($1::regclass, 1, false)")
+                .bind(&seq_name)
+                .execute(pg).await.ok();
+        }
+        count += 1;
+    }
+    tracing::info!("reset {count} sequence(s)");
+    Ok(())
+}