exwiw 0.1.6 → 0.1.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: c3fe2fb0c2cff1c962899799f4cfe22981a7b730759d456dc01ab8d19254dbec
-  data.tar.gz: 7bf651780e74074b50e9b0345a7cdcb9ec0281cd3d60ff04764cf03c1804b32e
+  metadata.gz: 7c5e29a492af74dfbfa0e778fcb527a218d3a33507024646b2fe88b495581c2f
+  data.tar.gz: bd66516a56f40e4147e76e3fc662c98cd3c0261eb54a310ef7af49bcc5373cf0
 SHA512:
-  metadata.gz: a8ee7520812cb8a655cd4f9b8dffa3afdd622311c6b39c5b7942aae5743c3642cb44a7b6b2fdd56ffdf967a8225c220866691a8a001bf75606c70871d4d731ca
-  data.tar.gz: 10182f0c69f0d773e1a515f932148c7b8795a51ba64e46141d0630459c815594e470ac62e4d082a695426ec254430dcac6c3cdbaff74b47e34c49f0dfa9a901a
+  metadata.gz: 1b63d52ce0abd624695b73d782c64a5d4dd861e701f7c5989e977dd506d0178f4e2d394e9ae57e1106bf83898b622bd93681c60bdfbbde7dcd72bf1796cd4847
+  data.tar.gz: 36ea50859424c45eb3bd83ffc84fb148eb4080345e7dfc012ff81b32003fa27e83f43324c4e7fa296c4d35ee8de6f89754a9e9f1449bd69d3d1b4066015e51bc

data/CHANGELOG.md

@@ -2,6 +2,26 @@
 
 ## [Unreleased]
 
+## [0.1.8] - 2026-05-16
+
+### Added
+
+- Emit a leading `insert-000-schema.{sql,js}` file alongside the per-table `insert-*` files so the generated dump can be applied to an empty database in one go. ([#14](https://github.com/heyinc/exwiw/pull/14))
+  - SQL adapters (`mysql2`, `postgresql`, `sqlite3`) write idempotent `CREATE TABLE IF NOT EXISTS` (and `CREATE INDEX IF NOT EXISTS` where the engine supports it) by shelling out to `mysqldump` / `pg_dump` / reading `sqlite_master`. PostgreSQL `ALTER TABLE ... ADD CONSTRAINT` is wrapped in a `DO $$ EXCEPTION WHEN duplicate_object` block.
+  - MongoDB adapter writes `insert-000-schema.js` containing `db.createCollection(...)` (wrapped in `try/catch` for `NamespaceExists`) and `db.<col>.createIndex(...)` calls for every top-level collection. Apply with `mongosh < dump/insert-000-schema.js`.
+
+### Fixed
+
+- PostgreSQL adapter now appends a `setval` for each table's sequence at the end of the `insert-*.sql` file, transcribing the source DB's `last_value` so `nextval` after restore does not collide with imported IDs. ([#19](https://github.com/heyinc/exwiw/pull/19))
+
+## [0.1.7] - 2026-05-14
+
+### Added
+
+- Add embedded document support to the MongoDB adapter via `embedded_in: { collection_name, path }`. Embedded configs are not dumped as their own jsonl; their `replace_with` rules apply to subdocuments (Array or Hash, with multi-level nesting) inside the parent collection.
+
+## [0.1.6] - 2026-03-14
+
 ### Added
 
 - Add `bulk_insert_chunk_size` table config to split the generated `INSERT` statement into chunks of the specified size. ([#8](https://github.com/riseshia/exwiw/pull/8))

data/README.md

@@ -64,12 +64,15 @@ exwiw \
 
 This command will generate sql files in the `dump` directory.
 
+- `dump/insert-000-schema.sql` — idempotent `CREATE TABLE IF NOT EXISTS ...` for every table in scope. Apply this first to provision an empty database.
 - `dump/insert-{idx}-{table_name}.sql`
 - `dump/delete-{idx}-{table_name}.sql`
 
 idx means the order of the dump. bigger idx might depend on smaller idx,
 so you should import the dump in order.
 
+`insert-000-schema.sql` is generated by shelling out to the database client tools (`mysqldump` for `mysql2`, `pg_dump` for `postgresql`, and the sqlite3 driver for `sqlite3`), so the corresponding client must be available on PATH when running exwiw. The output is post-processed to make it idempotent: `CREATE TABLE IF NOT EXISTS`, `CREATE INDEX IF NOT EXISTS` (where the engine supports it), and PostgreSQL's `ALTER TABLE ... ADD CONSTRAINT` statements are wrapped in `DO $$ ... EXCEPTION WHEN duplicate_object`.
+
 you need to delete the records before importing the dump,
 `delete-{idx}-{table_name}.sql` will help you to do that.
 This sql will delete "all" related records to the extract targets.
@@ -182,10 +185,52 @@ The MongoDB adapter is experimental. To use it:
   ```bash
   mongoimport --db app_dev --collection users --file dump/insert-002-users.jsonl
   ```
+- The leading `dump/insert-000-schema.js` contains `db.createCollection(...)` and `db.<col>.createIndex(...)` calls for every top-level collection (indexes are introspected from the source via `listIndexes`; the auto-created `_id_` index is skipped). Apply it with mongosh **before** running `mongoimport`:
+  ```bash
+  mongosh "mongodb://localhost/app_dev" dump/insert-000-schema.js
+  ```
 - Unlike SQL adapters, the MongoDB adapter does not emit `delete-*.jsonl` files (drop the database / collection yourself before importing if needed).
 - `raw_sql` is not supported (the `MongodbField` schema does not declare it; any `raw_sql` keys in scenario JSON are silently dropped on load). Use `replace_with` for masking.
 - The MongoDB adapter does not support the collection-level `filter` field (it raises `NotImplementedError` if set, since the SQL-string filter cannot be applied to MongoDB).
 
+#### Embedded documents
+
+MongoDB models often store one-to-many relationships as embedded subdocument arrays (e.g. `users` documents with a `posts: [...]` field). To mask fields inside embedded subdocuments, declare a separate config with `embedded_in`:
+
+```jsonc
+// scenario/users.json — top-level collection
+{
+  "name": "users",
+  "primary_key": "_id",
+  "belongs_tos": [{ "table_name": "shops", "foreign_key": "shop_id" }],
+  "fields": [
+    { "name": "_id" },
+    { "name": "name", "replace_with": "masked{_id}" },
+    { "name": "shop_id" }
+  ]
+}
+
+// scenario/posts.json — embedded under users.posts
+{
+  "name": "posts",
+  "primary_key": "_id",
+  "embedded_in": { "collection_name": "users", "path": "posts" },
+  "belongs_tos": [],
+  "fields": [
+    { "name": "_id" },
+    { "name": "title", "replace_with": "masked-{_id}" }
+  ]
+}
+```
+
+At runtime:
+
+- `posts` is **not** dumped as its own jsonl file. Its `replace_with` rules are applied to the subdocuments inside the parent `users` document at the path `posts`.
+- `path` accepts dot-separated paths for nested fields (e.g. `"profile.contacts"`).
+- Both arrays of subdocuments and a single Hash subdocument at `path` are supported. Multiple levels of nesting work via embedded chains.
+- Cross-collection references from inside an embedded subdocument (`belongs_tos` on an embedded config) are not supported and raise `ArgumentError` on load.
+- Specifying an embedded config as `--target-table` raises `NotImplementedError`; pass the top-level collection name instead.
+
 ## How it works
 
 - Load the table information from the specified config file.

data/docs/plans/2026-05-15-insert-000-schema-file.md

@@ -0,0 +1,151 @@
+# Plan: `insert-000-schema.{sql,js}` を dump 出力に追加する
+
+## Context
+
+現在 `exwiw` の dump 出力 (`dump/`) には `insert-NNN-{table}.sql` と `delete-NNN-{table}.sql` のみが書かれ、`CREATE TABLE` などのスキーマ定義は別管理になっている。そのため取得した dump を別環境にインポートしようとすると、利用者側で別途スキーマを用意する必要があり、空 DB に流すと失敗する／idempotent に流せない。
+
+ゴール: 既存の per-table insert ファイルの前段として、
+
+- **SQL adapters** (`mysql2` / `postgresql` / `sqlite3`): `insert-000-schema.sql` に `CREATE TABLE IF NOT EXISTS ...` などをまとめて出力する
+- **MongoDB adapter**: `insert-000-schema.js` に `db.createCollection(...)` と `db.<col>.createIndex(...)` をまとめて出力する
+
+これにより、空 DB に対しても `insert-000-schema.*` → `insert-001-...` → ... の順で適用するだけで dump が完結する。
+
+## Design
+
+### 採用した方針 (ユーザ確認済み)
+- DDL の取得元は **外部コマンドのシェルアウト** (SQL 系は `mysqldump` / `pg_dump` / `sqlite3 .schema`)。MongoDB は CLI に相当するものがないので、既に require 済みの `mongo` Ruby ドライバで `listCollections` / `listIndexes` を使う (mongosh を追加要件にしない)。
+- 出力は **1 ファイルに統合**: `insert-000-schema.{sql,js}`
+- スコープは **DB 上の完全定義をそのまま転写**。config の `columns` / `fields` で絞り込まない。
+- Mongo は **mongosh で流せる .js** を出力。
+
+### アーキテクチャ
+1. `Adapter::Base` に新規メソッドを追加 (デフォルトは no-op):
+   - `dump_schema(ordered_tables, output_path, logger)` — `output_path` (ファイル絶対パス) にスキーマ DDL を書く
+   - `schema_output_extension` — デフォルト `'sql'`、Mongo は override で `'js'`
+2. `Runner#run` の `mkdir_p` 直後、per-table ループの前で 1 回だけ呼び出す:
+   ```ruby
+   ordered_tables = ordered_table_names.map { |n| table_by_name.fetch(n) }
+   schema_path = File.join(@output_dir, "insert-000-schema.#{adapter.schema_output_extension}")
+   adapter.dump_schema(ordered_tables, schema_path, @logger)
+   ```
+   - `ordered_table_names` は依存順に並んでいる (`DetermineTableProcessingOrder`) ので、FK 制約付き DDL もそのまま流せる。
+3. 各 adapter の `dump_schema` 実装:
+
+   **Sqlite3Adapter** — 既に require 済みの `sqlite3` gem の connection を使い `SELECT type, sql FROM sqlite_master WHERE sql IS NOT NULL ORDER BY CASE type WHEN 'table' THEN 0 WHEN 'index' THEN 1 ELSE 2 END` を流す。`sqlite_sequence` 等の自動テーブル (`sql` が NULL) は除外される。各行の `sql` を後処理して:
+   - `CREATE TABLE ` → `CREATE TABLE IF NOT EXISTS `
+   - `CREATE INDEX ` → `CREATE INDEX IF NOT EXISTS ` (SQLite はサポート)
+   - `CREATE UNIQUE INDEX ` → 同上
+   - `CREATE TRIGGER ` / `CREATE VIEW ` も同様に IF NOT EXISTS を付与
+
+   設計上の選択: 「シェルアウト」をユーザは選んだが、SQLite については既存接続を再利用する方が安全 (外部 `sqlite3` バイナリの有無に依存しない)。シェルアウトに統一したい場合はオプションBへ。
+
+   **Mysql2Adapter** — `mysqldump` をシェルアウト:
+   ```
+   MYSQL_PWD=$password mysqldump \
+     --host={host} --port={port} --user={user} \
+     --no-data --skip-add-drop-table --skip-comments \
+     --skip-extended-insert --skip-set-charset \
+     --compact \
+     {database}
+   ```
+   - パスワードはコマンドライン引数ではなく `MYSQL_PWD` env 経由 (プロセス一覧に出さない)
+   - stdout を後処理して `CREATE TABLE ` → `CREATE TABLE IF NOT EXISTS `。MySQL は `CREATE INDEX IF NOT EXISTS` を **サポートしない** が、mysqldump はインデックスを `CREATE TABLE` 内にインラインで含めるため通常問題なし
+   - `/*!40101 ...*/` のような実行時 SET 文はそのまま残す (mysqldump が import 互換のために出すもの)
+
+   **PostgresqlAdapter** — `pg_dump` をシェルアウト:
+   ```
+   PGPASSWORD=$password pg_dump \
+     --host={host} --port={port} --username={user} \
+     --schema-only --no-owner --no-acl --no-comments \
+     {database}
+   ```
+   - stdout を後処理して `CREATE TABLE ` → `CREATE TABLE IF NOT EXISTS `、`CREATE INDEX ` → `CREATE INDEX IF NOT EXISTS `、`CREATE UNIQUE INDEX ` → `CREATE UNIQUE INDEX ... IF NOT EXISTS` (PG は両方サポート)
+   - `ALTER TABLE ... ADD CONSTRAINT` は重複適用で失敗する。**`information_schema.table_constraints` で存在チェックする `DO $$` ブロックにラップする** か、よりシンプルには `ALTER TABLE ONLY ... ADD CONSTRAINT` 行を検出して `IF NOT EXISTS` 版 (PG 9.6+ の `ALTER TABLE ADD CONSTRAINT` には IF NOT EXISTS はないので) **DO ブロックでラップ** する。実装方針: 各 `ALTER TABLE ... ADD CONSTRAINT "name" ...;` を:
+     ```sql
+     DO $$ BEGIN
+       ALTER TABLE ... ADD CONSTRAINT "name" ...;
+     EXCEPTION WHEN duplicate_object THEN NULL; END $$;
+     ```
+     でラップ。
+   - `SET ...` / `SELECT pg_catalog.set_config(...)` 行はそのまま残す。
+   - `CREATE SCHEMA public` は素の pg_dump では出ないが、出力に含まれた場合は `CREATE SCHEMA IF NOT EXISTS` に書き換える。
+
+   **MongodbAdapter** — 既に require 済みの `mongo` ドライバを使い、`tables.reject(&:embedded?)` をループ:
+   ```ruby
+   db.list_collections.each { |c| existing_collections << c['name'] }  # for skip-creation idempotency hint
+   ordered_tables.reject(&:embedded?).each do |config|
+     indexes = db[config.name].indexes.to_a.reject { |idx| idx['name'] == '_id_' }
+     # emit JS lines
+   end
+   ```
+   出力 JS のサンプル:
+   ```js
+   // Auto-generated by exwiw. Apply with: mongosh < insert-000-schema.js
+   try { db.createCollection("users"); } catch (e) { if (e.code !== 48) throw e; } // 48 = NamespaceExists
+   db.users.createIndex({"shop_id":1}, {"name":"index_users_on_shop_id"});
+   try { db.createCollection("shops"); } catch (e) { if (e.code !== 48) throw e; }
+   ```
+   - `_id_` index は MongoDB が自動作成するので除外
+   - `createIndex` はキー仕様とオプションが一致していれば idempotent (重複作成は no-op)
+   - index doc には `v` / `ns` など driver 由来のメタが入るので、`key` と `name` 以外で出力に含めるのは `unique` / `sparse` / `partialFilterExpression` / `expireAfterSeconds` / `collation` のみ (allowlist) にする
+   - `MongodbAdapter#schema_output_extension` で `'js'` を返す
+   - 埋め込み (embedded_in) は親 collection に内包されるので index 不要 → スキップ
+
+### 後処理ユーティリティ
+新規ファイル `lib/exwiw/ddl_postprocessor.rb` を作って各 SQL adapter から呼ぶ:
+- `.add_if_not_exists_to_create_table(sql)` — 行頭の `CREATE TABLE ` → `CREATE TABLE IF NOT EXISTS ` (既に IF NOT EXISTS が付いている場合はスキップ)
+- `.add_if_not_exists_to_create_index(sql)` — `CREATE [UNIQUE] INDEX ` 系
+- `.wrap_add_constraint_in_do_block(sql)` — PG 専用
+
+### CLI 側の挙動
+特に CLI フラグの追加はしない。常に `insert-000-schema.*` を出力する (ユーザの選択にあった通り)。ただし、外部コマンドが PATH に無い場合は明確なエラーメッセージで停止する:
+```
+Error: `pg_dump` not found in PATH. exwiw needs pg_dump to generate insert-000-schema.sql for the postgresql adapter.
+```
+
+## Files to modify / add
+
+| パス | 変更 |
+|---|---|
+| `lib/exwiw/adapter.rb` | `Base` に `dump_schema(tables, path, logger)` と `schema_output_extension` を追加 |
+| `lib/exwiw/runner.rb` | `mkdir_p` 直後で `adapter.dump_schema(...)` を呼ぶ |
+| `lib/exwiw/adapter/sqlite3_adapter.rb` | `dump_schema` 実装 (既存 connection を使い `sqlite_master` 経由) |
+| `lib/exwiw/adapter/mysql2_adapter.rb` | `dump_schema` 実装 (`mysqldump` シェルアウト) |
+| `lib/exwiw/adapter/postgresql_adapter.rb` | `dump_schema` 実装 (`pg_dump` シェルアウト, ADD CONSTRAINT DO ブロック化) |
+| `lib/exwiw/adapter/mongodb_adapter.rb` | `dump_schema` 実装、`schema_output_extension` を override |
+| `lib/exwiw/ddl_postprocessor.rb` (新規) | `IF NOT EXISTS` 書き換え / DO ブロックラップ |
+| `lib/exwiw.rb` | 新規ファイルの require |
+| `README.md` | `dump/` の出力に `insert-000-schema.{sql,js}` を追記、import 手順を更新 |
+| `spec/adapter/sqlite3_adapter_spec.rb` | `dump_schema` 統合テスト (`scenario/initdb/init.sqlite3` に対して実行し、出力が `CREATE TABLE IF NOT EXISTS` を含むことを assert) |
+| `spec/adapter/mongodb_adapter_spec.rb` | `dump_schema` テスト (db スタブで `listIndexes` を返し、出力 JS を assert) |
+| `spec/runner_spec.rb` | `insert-000-schema.sql` が `output_dir` に書かれることを assert (Sqlite3 経由で実際に流れることを確認) |
+
+## 再利用する既存コード
+- `lib/exwiw/connection_config.rb` (host / port / user / password / database_name) — シェルアウトの引数組み立てに使う
+- `lib/exwiw/determine_table_processing_order.rb` — schema dump も依存順で並べるためそのまま使う
+- `MongodbCollectionConfig#embedded?` (lib/exwiw/mongodb_collection_config.rb:33) — 埋め込み collection をスキップ
+- `MongodbAdapter#db` (lib/exwiw/adapter/mongodb_adapter.rb:197) — Mongo クライアントを取り出す既存 lazy getter
+
+## Verification
+
+### ユニット / 統合テスト
+1. `bundle exec rspec spec/runner_spec.rb spec/adapter/sqlite3_adapter_spec.rb` — sqlite3 経路で `insert-000-schema.sql` が生成され、内容に `CREATE TABLE IF NOT EXISTS "shops"` が含まれることを確認。
+2. `bundle exec rspec spec/adapter/mongodb_adapter_spec.rb` — mongo クライアントをスタブして JS 出力に `db.createCollection("users")` と該当 collection の `createIndex(...)` が含まれることを確認。
+
+### E2E (scenario スクリプト経由)
+3. `scenario/test_with_sqlite3.sh` を実行し、`dump/insert-000-schema.sql` が生成されることと、空 DB に対して `sqlite3 empty.db < dump/insert-000-schema.sql && for f in dump/insert-*.sql; do sqlite3 empty.db < $f; done` が成功することを確認する。
+4. `scenario/test_with_mysql2.sh`, `scenario/test_with_postgresql.sh` も同様に、`mysql empty_db < dump/insert-000-schema.sql` / `psql empty_db -f dump/insert-000-schema.sql` が成功 → 続けて insert ファイル群が流せることを確認。**`mysqldump` / `pg_dump` を docker compose のコンテナ内 (`compose.yml` で起動する DB コンテナ) で実行する必要がある場合は、scenario スクリプトを更新する。**
+5. `scenario/test_with_mongodb.sh` を実行し、`dump/insert-000-schema.js` が出力されることと、空 DB に対して `mongosh "mongodb://localhost/empty_db" < dump/insert-000-schema.js` が成功すること、続いて `mongoimport` で各 jsonl が流せることを確認。
+6. **idempotency 確認**: 同じ schema ファイルを 2 回流してもエラーにならないこと (`IF NOT EXISTS` / `DO $$ EXCEPTION WHEN duplicate_object` / `try/catch on createCollection` が効いている)。
+
+### 手動確認のチェックポイント
+- `mysqldump` / `pg_dump` が PATH にない環境で実行した場合、わかりやすいエラーで止まる
+- `DATABASE_PASSWORD` env が無い場合に外部コマンドが認証エラーで落ちないこと (シェルアウト時にも `MYSQL_PWD` / `PGPASSWORD` を渡す)
+- `CHANGELOG.md` への追記
+
+## 留意点 / 既知のリスク
+- **MySQL の `CREATE INDEX IF NOT EXISTS` 非対応**: mysqldump はインデックスをテーブル定義内にインラインで吐くので通常問題にならないが、`--no-create-info --no-data --routines --triggers` などのオプションで吐き分ける場合は別途対応が必要。今回はデフォルト挙動のみサポート。
+- **PG の COMMENT ON / GRANT / SEQUENCE OWNED BY**: `--no-owner --no-acl --no-comments` で削減できる。SEQUENCE 自体の `CREATE SEQUENCE` は `IF NOT EXISTS` 書き換え対象に含める。
+- **mongosh 依存**: 出力 JS を流すこと自体には mongosh が必要 (README に明記)。exwiw 本体はあくまで Ruby driver のみで生成するので、exwiw 実行ホストには mongosh は不要。
+- **巨大 schema**: pg_dump / mysqldump の出力をメモリに乗せて後処理するので、超巨大スキーマだとメモリ使用量が増える。実用上は問題にならない見込み。

data/docs/plans/2026-05-16-mongodb-from-clean-scenario.md

@@ -0,0 +1,76 @@
+# Plan: MongoDB の `insert-000-schema.js` を scenario で end-to-end 検証する
+
+## Context
+
+`lib/exwiw/adapter/mongodb_adapter.rb#dump_schema` は `insert-000-schema.js` に
+`createCollection` / `createIndex` を書き出す実装を既に持っているが、scenario 側で
+これを apply するパスが無く、CI でも検証できていなかった。具体的なギャップ:
+
+1. `scenario/setup_with_mongodb.rb` は seed を `insert_many` で流すだけで、index を一切作っていない
+2. その結果 `tmp/mongodb/insert-000-schema.js` は `createCollection` 行のみで `createIndex` が 0 行
+3. `scenario/import_with_mongodb.rb` は `insert-*.jsonl` だけを glob して処理しており、`insert-000-schema.js` を一切実行しない
+
+sqlite3 / mysql2 / postgresql で導入済みの「from clean DB から立ち上げる」流れと
+MongoDB の `insert-000-schema.js` が連動していない状態だった (issue #16)。
+
+## ゴール
+
+- 空の target DB に対して `mongosh insert-000-schema.js` → `insert-*.jsonl` の順で適用する scenario を CI に乗せる
+- source DB に代表的な index を作り、`dump_schema` が `createIndex` 行を実際に吐く状態を作る
+- 生成された createIndex 行が mongosh で実際に通ること、target 側で index が round-trip することを検証
+- 既存の snapshot test (`spec/insert_output_snapshot_spec.rb`) でも createIndex 行を固定化
+
+## 変更内容
+
+### scenario 層
+| パス | 変更 |
+|---|---|
+| `scenario/setup_with_mongodb.rb` | seed 流し込みの後に 3 種類の代表的 index を作る (unique `shops.name` / plain `users.email` / 複合 `orders.shop_id+user_id`) |
+| `scenario/import_with_mongodb.rb` | `--no-drop` と `--input-dir DIR` フラグを追加。from-clean は drop すると schema.js が作った index ごと消えてしまうため |
+| `scenario/verify_with_mongodb.rb` | `--with-indexes` で target collection の index を assert (default scenario では import 時に drop されるのでスキップ) |
+| `scenario/test_with_mongodb_from_clean.sh` (新規) | `mongosh dropDatabase` → exwiw 実行 → `mongosh insert-000-schema.js` → `import --no-drop --input-dir tmp/mongodb-clean` → `verify --with-indexes` |
+| `.github/workflows/scenario.yml` | with_mongodb job に `mongodb-mongosh` install ステップと `test_with_mongodb_from_clean.sh` 実行ステップを追加。apt repo の codename は `jammy` 固定 (ubuntu-latest が noble に上がる前提) |
+
+### snapshot test 層
+| パス | 変更 |
+|---|---|
+| `spec/support/bootstrap_databases.rb` | scenario と同じ 3 index を bootstrap で作る |
+| `spec/insert_output_snapshots/mongodb/insert-000-schema.js` | 3 つの `db.getCollection(...).createIndex(...)` 行が追加される形で再生成 |
+
+## 設計上の判断
+
+- **unique index は `users.email` ではなく `shops.name` に貼る**: seed の `users.email`
+  は `user1@example.com` が 5 shop に重複するので unique にできない。`shops.name`
+  ("Shop 1".."Shop 5") は seed 上一意なので unique 可。
+- **既存 scenario への副作用を最小化**: `import_with_mongodb.rb` のデフォルト挙動は変えず
+  `--no-drop` フラグで opt-in。既存 `test_with_mongodb.sh` は無修正で動く。
+- **verify を 2 用途で兼用**: `--with-indexes` 切り替えで from-clean のみ index を見る。
+  既存 scenario は drop→insert で index が無くなるため index 検証はスキップ。
+- **CI への mongosh install**: `mongo:7` service container には mongosh があるが、
+  ubuntu-latest 上の `mongosh` コマンドは別。MongoDB の apt repo (`mongodb-mongosh`
+  パッケージ) を入れる。codename は `jammy` 固定 (MongoDB 7.0 repo が noble を
+  carry していない時期があるため)。
+- **snapshot fixture を indexes 入りに更新**: bootstrap_databases.rb と
+  setup_with_mongodb.rb で同じ index を作るので、snapshot test と scenario test の
+  期待値が分岐しない。
+
+## Verification
+
+- `bash scenario/test_with_mongodb.sh` 既存 scenario 維持を確認 ✓
+- `bash scenario/test_with_mongodb_from_clean.sh` 新規 scenario 通過を確認
+  (indexes round-trip OK) ✓
+- `bundle exec rspec` 全 153 examples / 0 failures ✓
+- `tmp/mongodb-clean/insert-000-schema.js` を目視で確認:
+  ```js
+  db.getCollection("shops").createIndex({"name":1}, {"unique":true,"name":"idx_shops_name"});
+  db.getCollection("users").createIndex({"email":1}, {"name":"idx_users_email"});
+  db.getCollection("orders").createIndex({"shop_id":1,"user_id":1}, {"name":"idx_orders_shop_user"});
+  ```
+
+## 留意点
+
+- `import_with_mongodb.rb` のフラグ解析は手書きの ARGV パース。引数が増えるなら
+  OptionParser 化を検討する余地あり (現状は 2 フラグなので過剰)。
+- ubuntu-latest が将来 codename を変えても apt repo の `jammy` 指定は壊れない想定だが、
+  MongoDB 8.x へ移行する際は repo URL の `7.0` も更新が必要。
+- 既存 issue #16 のスコープは MongoDB のみ。SQL 系の from_clean は別 PR で導入済み。

data/lib/exwiw/adapter/mongodb_adapter.rb

@@ -3,10 +3,9 @@
 require 'json'
 
 # NOTE: This adapter consumes MongodbCollectionConfig (`fields` instead of
-# the SQL adapters' `columns`). It assumes a "flat" document schema where
-# references between collections are expressed as scalar foreign keys
-# (e.g. `shop_id` on `users`); the forward fan-out strategy here cannot
-# follow references that live inside embedded structures.
+# `columns`, plus `embedded_in`). Top-level collections are dumped as one
+# jsonl per collection; configs marked `embedded_in` are not dumped on their
+# own — their masking rules apply to subdocuments inside the parent.
 module Exwiw
   module Adapter
     class MongodbAdapter < Base
@@ -19,8 +18,32 @@ module Exwiw
         @state = {}
       end
 
-      def build_query(config, dump_target, _config_by_name)
+      def dumpable?(config)
+        !config.embedded?
+      end
+
+      def validate_as_dump_target!(config)
+        return unless config.embedded?
+
+        raise NotImplementedError,
+              "dump_target '#{config.name}' is an embedded MongodbCollectionConfig; " \
+              "specify a top-level collection instead."
+      end
+
+      def build_query(config, dump_target, config_by_name)
+        if config.embedded?
+          raise NotImplementedError,
+                "MongodbAdapter#build_query was called with embedded config '#{config.name}'. " \
+                "Embedded configs are masked through the parent collection."
+        end
+
         reject_filter!(config)
+        # Stash the embedded-children index for the matching to_bulk_insert call
+        # below. The Adapter contract does not pass config_by_name to
+        # to_bulk_insert (SQL adapters don't need it), so we rely on the Runner
+        # invariant that build_query is always called before to_bulk_insert for
+        # the same config.
+        @embedded_children_by_parent = index_embedded_children(config_by_name)
 
         filter =
           if config.name == dump_target.table_name
@@ -57,9 +80,15 @@ module Exwiw
         docs
       end
 
+      # NOTE: relies on @embedded_children_by_parent set by a prior build_query
+      # call for the same config. This implicit ordering exists because the
+      # Adapter contract intentionally does not thread config_by_name through
+      # to_bulk_insert (SQL adapters don't need it). Safe in Runner, fragile in
+      # tests — call build_query first.
       def to_bulk_insert(rows, config)
         rows.map do |doc|
           apply_replace_with!(doc, config)
+          apply_embedded_masking!(doc, config)
           JSON.generate(extended_json(doc))
         end.join("\n")
       end
@@ -72,6 +101,47 @@ module Exwiw
         'jsonl'
       end
 
+      def schema_output_extension
+        'js'
+      end
+
+      # Index options copied through to the emitted createIndex call. Anything
+      # else (`v`, `ns`, server-internal fields) is dropped — they would either
+      # be rejected by createIndex or are not portable across mongod versions.
+      INDEX_OPTION_ALLOWLIST = %w[
+        unique sparse hidden expireAfterSeconds collation
+        partialFilterExpression wildcardProjection
+      ].freeze
+
+      def dump_schema(ordered_tables, output_path)
+        require 'json'
+
+        collections = ordered_tables.reject(&:embedded?)
+
+        File.open(output_path, 'w') do |file|
+          file.puts("// Auto-generated by exwiw. Apply with: mongosh \"$MONGODB_URI\" #{File.basename(output_path)}")
+          file.puts
+
+          collections.each do |config|
+            name = config.name
+            file.puts(%(try { db.createCollection(#{JSON.generate(name)}); } catch (e) { if (e.code !== 48) throw e; }))
+          end
+          file.puts
+
+          collections.each do |config|
+            name = config.name
+            indexes = db[name].indexes.to_a.reject { |idx| idx['name'] == '_id_' }
+            indexes.each do |idx|
+              key = idx['key']
+              opts = idx.slice(*INDEX_OPTION_ALLOWLIST)
+              opts['name'] = idx['name'] if idx['name']
+              file.puts(%(db.getCollection(#{JSON.generate(name)}).createIndex(#{JSON.generate(key)}, #{JSON.generate(opts)});))
+            end
+          end
+        end
+        @logger.info("  Wrote schema for #{collections.size} collection(s) to #{output_path}.")
+      end
+
       def supports_bulk_delete?
         false
       end
@@ -96,6 +166,14 @@ module Exwiw
               "collection-level `filter` is not supported by MongodbAdapter (collection: #{config.name})"
       end
 
+      private def index_embedded_children(config_by_name)
+        config_by_name.each_value.with_object({}) do |child, acc|
+          next unless child.embedded?
+
+          (acc[child.embedded_in.collection_name] ||= []) << child
+        end
+      end
+
       private def build_projection(config)
         projection = {}
         # Always include primary key so masking templates referencing it work,
@@ -104,6 +182,11 @@ module Exwiw
         config.fields.each do |field|
           projection[field.name] = 1
         end
+        # Pull in paths owned by configs that mark themselves embedded in this
+        # collection, so the masker sees the subdocuments.
+        embedded_children_of(config).each do |child|
+          projection[child.embedded_in.path] = 1
+        end
         projection
       end
 
@@ -118,6 +201,32 @@ module Exwiw
         end
       end
 
+      private def apply_embedded_masking!(doc, parent_config)
+        embedded_children_of(parent_config).each do |child|
+          walk(doc, child.embedded_in.path) do |subdoc|
+            apply_replace_with!(subdoc, child)
+            apply_embedded_masking!(subdoc, child)
+          end
+        end
+      end
+
+      private def embedded_children_of(parent_config)
+        @embedded_children_by_parent.fetch(parent_config.name, [])
+      end
+
+      private def walk(doc, dotted_path)
+        segments = dotted_path.split(".")
+        *prefix, last = segments
+        container = prefix.reduce(doc) { |acc, seg| acc.is_a?(Hash) ? acc[seg] : nil }
+        return unless container.is_a?(Hash)
+
+        value = container[last]
+        case value
+        when Array then value.each { |sub| yield sub if sub.is_a?(Hash) }
+        when Hash  then yield value
+        end
+      end
+
       private def extended_json(doc)
         if doc.respond_to?(:as_extended_json)
           doc.as_extended_json(mode: :relaxed)

data/lib/exwiw/adapter/mysql2_adapter.rb

@@ -14,6 +14,57 @@ module Exwiw
         connection.query(sql, cast: false, as: :array).to_a
       end
 
+      def dump_schema(ordered_tables, output_path)
+        require 'open3'
+
+        table_names = ordered_tables.map(&:name)
+        if table_names.empty?
+          File.write(output_path, "-- Auto-generated by exwiw. No tables in scope.\n")
+          return
+        end
+
+        cmd = [
+          'mysqldump',
+          "--host=#{@connection_config.host}",
+          "--port=#{@connection_config.port}",
+          "--user=#{@connection_config.user}",
+          '--no-data',
+          '--skip-add-drop-table',
+          # `--skip-comments` only suppresses the dump's header lines
+          # (e.g. `-- MySQL dump ...`, server version banner). Column and
+          # table `COMMENT '...'` clauses are emitted inline inside
+          # CREATE TABLE statements and are NOT affected, so this flag is
+          # purely about reducing noise in the generated file.
+          '--skip-comments',
+          '--skip-set-charset',
+          # Suppress `SET @@GLOBAL.GTID_PURGED=...` from the dump. It is intended
+          # for replication setup and breaks when the target already has GTIDs
+          # (ERROR 3546: added gtid set must not overlap with @@GLOBAL.GTID_EXECUTED).
+          '--set-gtid-purged=OFF',
+          '--compact',
+          @connection_config.database_name,
+          *table_names,
+        ]
+        env = { 'MYSQL_PWD' => @connection_config.password.to_s }
+
+        @logger.debug("  Running mysqldump for #{table_names.size} table(s)...")
+        stdout, stderr, status = Open3.capture3(env, *cmd)
+        unless status.success?
+          if stderr.include?('command not found') || stderr.empty?
+            raise "Failed to run `mysqldump`. Ensure the mysql client is installed and on PATH. stderr: #{stderr}"
+          end
+          raise "mysqldump failed (exit #{status.exitstatus}): #{stderr}"
+        end
+
+        idempotent = DdlPostprocessor.add_if_not_exists_to_create_table(stdout)
+
+        File.open(output_path, 'w') do |file|
+          file.puts("-- Auto-generated by exwiw via mysqldump. Idempotent CREATE TABLE statements for mysql.")
+          file.write(idempotent)
+        end
+        @logger.info("  Wrote schema for #{table_names.size} table(s) to #{output_path}.")
+      end
+
       def to_bulk_insert(results, table)
         table_name = table.name
 

data/lib/exwiw/adapter/postgresql_adapter.rb

@@ -14,6 +14,51 @@ module Exwiw
         connection.exec(sql).values
       end
 
+      def dump_schema(ordered_tables, output_path)
+        require 'open3'
+
+        table_names = ordered_tables.map(&:name)
+        if table_names.empty?
+          File.write(output_path, "-- Auto-generated by exwiw. No tables in scope.\n")
+          return
+        end
+
+        cmd = [
+          'pg_dump',
+          "--host=#{@connection_config.host}",
+          "--port=#{@connection_config.port}",
+          "--username=#{@connection_config.user}",
+          '--schema-only',
+          '--no-owner',
+          '--no-acl',
+          *table_names.flat_map { |t| ['--table', t] },
+          @connection_config.database_name,
+        ]
+        env = { 'PGPASSWORD' => @connection_config.password.to_s }
+
+        @logger.debug("  Running pg_dump for #{table_names.size} table(s)...")
+        stdout, stderr, status = Open3.capture3(env, *cmd)
+        unless status.success?
+          if stderr.include?('command not found') || stderr.empty?
+            raise "Failed to run `pg_dump`. Ensure the postgresql client is installed and on PATH. stderr: #{stderr}"
+          end
+          raise "pg_dump failed (exit #{status.exitstatus}): #{stderr}"
+        end
+
+        idempotent = stdout
+        idempotent = DdlPostprocessor.add_if_not_exists_to_create_schema(idempotent)
+        idempotent = DdlPostprocessor.add_if_not_exists_to_create_sequence(idempotent)
+        idempotent = DdlPostprocessor.add_if_not_exists_to_create_table(idempotent)
+        idempotent = DdlPostprocessor.add_if_not_exists_to_create_index(idempotent)
+        idempotent = DdlPostprocessor.wrap_add_constraint_in_do_block(idempotent)
+
+        File.open(output_path, 'w') do |file|
+          file.puts("-- Auto-generated by exwiw via pg_dump. Idempotent DDL for postgresql.")
+          file.write(idempotent)
+        end
+        @logger.info("  Wrote schema for #{table_names.size} table(s) to #{output_path}.")
+      end
+
       def to_bulk_insert(results, table)
         table_name = table.name
 
@@ -29,6 +74,36 @@ module Exwiw
         "INSERT INTO #{table_name} (#{column_names}) VALUES\n#{values};"
       end
 
+      # Transcribe the FROM-side sequence cursor backing `table.primary_key`
+      # onto the import target. Without this, importing into a clean DB leaves
+      # the sequence at 1 while the inserted rows occupy higher IDs, so the
+      # next default-PK INSERT collides. We query FROM's `last_value` /
+      # `is_called` directly (matching what pg_dump emits) rather than using
+      # MAX(pk), so a subsetted dump still preserves the source's "next id".
+      # Returns nil for non-auto-increment PKs (pg_get_serial_sequence -> NULL).
+      #
+      # Scope: ONLY the sequence attached to the primary key is synced. If a
+      # table has additional auto-increment columns (e.g. a non-PK SERIAL),
+      # those sequences are NOT transcribed and a subsequent default-value
+      # INSERT on them can collide. Rails-managed schemas don't hit this
+      # because only `id` is auto-increment, but bare PostgreSQL schemas may.
+      def post_insert_sql(table)
+        pk = table.primary_key
+        return nil if pk.nil? || pk.empty?
+
+        seq_name = connection
+          .exec_params("SELECT pg_get_serial_sequence($1, $2)", [table.name, pk])
+          .values.dig(0, 0)
+        return nil if seq_name.nil?
+
+        last_value, is_called = connection
+          .exec("SELECT last_value, is_called FROM #{seq_name}")
+          .values.first
+        is_called_sql = (is_called == 't' || is_called == true) ? 'true' : 'false'
+
+        "SELECT pg_catalog.setval('#{escape_single_quote(seq_name)}', #{last_value}, #{is_called_sql});"
+      end
+
       def to_bulk_delete(select_query_ast, table)
         raise NotImplementedError unless select_query_ast.is_a?(Exwiw::QueryAst::Select)
 

data/lib/exwiw/adapter/sqlite3_adapter.rb

@@ -14,6 +14,47 @@ module Exwiw
         connection.execute(sql)
       end
 
+      def dump_schema(ordered_tables, output_path)
+        @logger.debug("  Reading schema from sqlite_master...")
+        target_names = ordered_tables.map(&:name)
+        # `sqlite_master` row order preserves table creation order, which is also
+        # the dependency order produced by ActiveRecord-style migrations. To respect
+        # the caller-provided order, we partition tables / their owned indexes by
+        # ordered_tables.
+        all = connection.execute(<<~SQL)
+          SELECT type, name, tbl_name, sql FROM sqlite_master
+          WHERE sql IS NOT NULL AND name NOT LIKE 'sqlite_%'
+        SQL
+
+        tables_by_name   = all.select { |type, _, _, _| type == 'table'   }.to_h { |_, name, _, sql| [name, sql] }
+        indexes_by_owner = all.select { |type, _, _, _| type == 'index'   }.group_by { |_, _, tbl, _| tbl }
+        triggers_by_owner = all.select { |type, _, _, _| type == 'trigger' }.group_by { |_, _, tbl, _| tbl }
+
+        statements = []
+        target_names.each do |name|
+          table_sql = tables_by_name[name]
+          next unless table_sql
+
+          statements << finalize_stmt(DdlPostprocessor.add_if_not_exists_to_create_table(table_sql.strip))
+          (indexes_by_owner[name] || []).each do |_, _, _, idx_sql|
+            statements << finalize_stmt(DdlPostprocessor.add_if_not_exists_to_create_index(idx_sql.strip))
+          end
+          (triggers_by_owner[name] || []).each do |_, _, _, trg_sql|
+            statements << finalize_stmt(trg_sql.strip)
+          end
+        end
+
+        File.open(output_path, 'w') do |file|
+          file.puts("-- Auto-generated by exwiw. Idempotent CREATE statements for sqlite3.")
+          file.puts(statements.join("\n"))
+        end
+        @logger.info("  Wrote #{statements.size} schema statement(s) to #{output_path}.")
+      end
+
+      private def finalize_stmt(stmt)
+        stmt.end_with?(';') ? stmt : "#{stmt};"
+      end
+
       def to_bulk_insert(results, table)
         table_name = table.name
 

data/lib/exwiw/adapter.rb

@@ -30,10 +30,46 @@ module Exwiw
         'sql'
       end
 
+      # File extension used for the leading `insert-000-schema.*` file.
+      # SQL adapters emit `.sql` (CREATE TABLE IF NOT EXISTS ...);
+      # MongodbAdapter overrides to `.js` (mongosh-runnable createCollection / createIndex).
+      def schema_output_extension
+        'sql'
+      end
+
+      # Write the leading schema-creation file for this adapter to `output_path`.
+      # Default is a no-op; subclasses override to emit idempotent DDL so the
+      # generated dump can be applied to an empty database.
+      #
+      # @param ordered_tables [Array] table configs in dependency order
+      # @param output_path [String] absolute path to write to
+      def dump_schema(ordered_tables, output_path)
+      end
+
       # Whether this adapter emits delete-NNN-*.sql files.
       def supports_bulk_delete?
         true
       end
+
+      # Whether the given config produces its own dump output and needs an
+      # independent processing pass. SQL adapters always do; non-SQL adapters
+      # may exclude e.g. embedded subdocument configs.
+      def dumpable?(_config)
+        true
+      end
+
+      # Hook for adapter-specific validation when this config is used as the
+      # dump_target. Default: nothing to validate.
+      def validate_as_dump_target!(_config)
+      end
+
+      # Optional SQL appended to the per-table insert-NNN-<table>.* file after
+      # the bulk INSERT statements. Use to bring side-state in sync with the
+      # explicit IDs that were just inserted (e.g. PostgreSQL sequences).
+      # Default: nil (nothing appended).
+      def post_insert_sql(_table)
+        nil
+      end
     end
 
     # @params [Exwiw::QueryAst] query_ast

data/lib/exwiw/ddl_postprocessor.rb

@@ -0,0 +1,61 @@
+# frozen_string_literal: true
+
+module Exwiw
+  # Rewrites raw CREATE statements emitted by mysqldump / pg_dump /
+  # sqlite_master.sql into idempotent forms so the generated
+  # `insert-000-schema.sql` file can be re-applied without error.
+  module DdlPostprocessor
+    module_function
+
+    # `CREATE TABLE [name]` → `CREATE TABLE IF NOT EXISTS [name]`.
+    # `TEMP` / `TEMPORARY` variants and already-IF-NOT-EXISTS lines are skipped.
+    def add_if_not_exists_to_create_table(sql)
+      sql.gsub(/\bCREATE\s+TABLE\b(?!\s+IF\s+NOT\s+EXISTS)/i) do |m|
+        "#{m} IF NOT EXISTS"
+      end
+    end
+
+    # `CREATE [UNIQUE] INDEX [name]` → `CREATE [UNIQUE] INDEX IF NOT EXISTS [name]`.
+    # Use only for databases that support it (PostgreSQL, SQLite). MySQL does NOT
+    # support `CREATE INDEX IF NOT EXISTS` — do not call from the MySQL adapter.
+    def add_if_not_exists_to_create_index(sql)
+      sql.gsub(/\bCREATE(\s+UNIQUE)?\s+INDEX\b(?!\s+IF\s+NOT\s+EXISTS)/i) do
+        unique = Regexp.last_match(1) || ""
+        "CREATE#{unique} INDEX IF NOT EXISTS"
+      end
+    end
+
+    # `CREATE SCHEMA [name]` → `CREATE SCHEMA IF NOT EXISTS [name]`.
+    def add_if_not_exists_to_create_schema(sql)
+      sql.gsub(/\bCREATE\s+SCHEMA\b(?!\s+IF\s+NOT\s+EXISTS)/i) do |m|
+        "#{m} IF NOT EXISTS"
+      end
+    end
+
+    # `CREATE SEQUENCE [name]` → `CREATE SEQUENCE IF NOT EXISTS [name]`.
+    def add_if_not_exists_to_create_sequence(sql)
+      sql.gsub(/\bCREATE\s+SEQUENCE\b(?!\s+IF\s+NOT\s+EXISTS)/i) do |m|
+        "#{m} IF NOT EXISTS"
+      end
+    end
+
+    # `ALTER TABLE ... ADD CONSTRAINT ...;` is not idempotent on its own.
+    # PostgreSQL's PL/pgSQL has no IF-NOT-EXISTS clause for ADD CONSTRAINT, so wrap
+    # each statement in a DO block that swallows `duplicate_object`.
+    # Matches only statements whose ALTER TABLE clause leads directly into ADD CONSTRAINT
+    # (no intervening ALTER COLUMN / DROP / etc) so that unrelated ALTER TABLE statements
+    # in the same dump are not absorbed.
+    ADD_CONSTRAINT_RE = /^[ \t]*ALTER\s+TABLE\s+(?:ONLY\s+)?[^\s;,]+\s+(?:\n[ \t]*)?ADD\s+CONSTRAINT\b[^;]*;/m.freeze
+
+    def wrap_add_constraint_in_do_block(sql)
+      sql.gsub(ADD_CONSTRAINT_RE) do |stmt|
+        <<~SQL.chomp
+          DO $exwiw$ BEGIN
+            #{stmt.strip}
+          EXCEPTION WHEN duplicate_object THEN NULL;
+          END $exwiw$;
+        SQL
+      end
+    end
+  end
+end

data/lib/exwiw/embedded_in.rb

@@ -0,0 +1,14 @@
+# frozen_string_literal: true
+
+module Exwiw
+  class EmbeddedIn
+    include Serdes
+
+    attribute :collection_name, String
+    attribute :path, String
+
+    def self.from_symbol_keys(hash)
+      from(hash.transform_keys(&:to_s))
+    end
+  end
+end

data/lib/exwiw/mongodb_collection_config.rb

@@ -14,8 +14,34 @@ module Exwiw
     attribute :fields, array(MongodbField)
     attribute :bulk_insert_chunk_size, optional(Integer), skip_serializing_if_nil: true
 
+    # Marks this config as physically embedded inside another collection's
+    # documents. When set, this config is not processed as a standalone dump
+    # unit; its masking rules are applied to the parent's subdocuments at
+    # `path`.
+    attribute :embedded_in, optional(EmbeddedIn), skip_serializing_if_nil: true
+
+    def self.from(obj)
+      instance = super
+      instance.__send__(:validate_embedded!)
+      instance
+    end
+
     def self.from_symbol_keys(hash)
       from(JSON.parse(hash.to_json))
     end
+
+    def embedded?
+      !embedded_in.nil?
+    end
+
+    private def validate_embedded!
+      return unless embedded?
+      return if belongs_tos.empty?
+
+      raise ArgumentError,
+            "MongodbCollectionConfig '#{name}' is embedded_in '#{embedded_in.collection_name}'; " \
+            "belongs_tos must be empty (cross-collection refs from inside embedded arrays " \
+            "are not supported)."
+    end
   end
 end

data/lib/exwiw/runner.rb

@@ -20,16 +20,24 @@ module Exwiw
 
     def run
       adapter = Adapter.build(@connection_config, @logger)
-      tables = load_table_config(adapter.class.table_config_class)
+      configs = load_table_config(adapter.class.table_config_class)
+
+      table_by_name = configs.each_with_object({}) { |config, hash| hash[config.name] = config }
+
+      target = table_by_name[@dump_target.table_name]
+      adapter.validate_as_dump_target!(target) if target
 
       @logger.info("Determining table processing order...")
-      ordered_table_names = DetermineTableProcessingOrder.run(tables)
+      ordered_table_names = DetermineTableProcessingOrder.run(configs.select { |c| adapter.dumpable?(c) })
 
       if !Dir.exist?(@output_dir)
         FileUtils.mkdir_p(@output_dir)
       end
 
-      table_by_name = tables.each_with_object({}) { |table, hash| hash[table.name] = table }
+      ordered_tables = ordered_table_names.map { |n| table_by_name.fetch(n) }
+      schema_path = File.join(@output_dir, "insert-000-schema.#{adapter.schema_output_extension}")
+      @logger.info("Writing schema to #{schema_path}...")
+      adapter.dump_schema(ordered_tables, schema_path)
 
       total_size = ordered_table_names.size
       ordered_table_names.each_with_index do |table_name, idx|
@@ -54,6 +62,8 @@ module Exwiw
         insert_idx = (idx + 1).to_s.rjust(3, '0')
         File.open(File.join(@output_dir, "insert-#{insert_idx}-#{table_name}.#{adapter.output_extension}"), 'w') do |file|
           file.puts(insert_sql)
+          post = adapter.post_insert_sql(table)
+          file.puts(post) if post
         end
 
         if adapter.supports_bulk_delete?
@@ -78,14 +88,5 @@ module Exwiw
         klass.from(json)
       end
     end
-
-    private def build_adapter
-      case @connection_config["adapter"]
-      when "sqlite3"
-        Sqlite3Adapter.new(@connection_config)
-      else
-        raise "Unsupported adapter"
-      end
-    end
   end
 end

data/lib/exwiw/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module Exwiw
-  VERSION = "0.1.6"
+  VERSION = "0.1.8"
 end

data/lib/exwiw.rb

@@ -8,8 +8,10 @@ require "serdes"
 require_relative "exwiw/belongs_to"
 require_relative "exwiw/table_column"
 require_relative "exwiw/table_config"
+require_relative "exwiw/embedded_in"
 require_relative "exwiw/mongodb_field"
 require_relative "exwiw/mongodb_collection_config"
+require_relative "exwiw/ddl_postprocessor"
 require_relative "exwiw/adapter"
 require_relative "exwiw/adapter/sqlite3_adapter"
 require_relative "exwiw/adapter/mysql2_adapter"

data/mise.toml

@@ -0,0 +1,6 @@
+[env]
+# Prepend scenario/bin so `pg_dump` resolves to the wrapper that delegates to
+# the postgres container (compose.yml). exwiw's PostgreSQL adapter shells out
+# to pg_dump, which requires a server/client major-version match — the dev DB
+# is postgres:17 while host clients are often older (e.g. Homebrew pg14).
+_.path = ["./scenario/bin"]

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: exwiw
 version: !ruby/object:Gem::Version
-  version: 0.1.6
+  version: 0.1.8
 platform: ruby
 authors:
 - Shia
@@ -32,10 +32,11 @@ executables:
 extensions: []
 extra_rdoc_files: []
 files:
-- ".rubocop.yml"
 - CHANGELOG.md
 - LICENSE.txt
 - README.md
+- docs/plans/2026-05-15-insert-000-schema-file.md
+- docs/plans/2026-05-16-mongodb-from-clean-scenario.md
 - exe/exwiw
 - lib/exwiw.rb
 - lib/exwiw/adapter.rb
@@ -45,7 +46,9 @@ files:
 - lib/exwiw/adapter/sqlite3_adapter.rb
 - lib/exwiw/belongs_to.rb
 - lib/exwiw/cli.rb
+- lib/exwiw/ddl_postprocessor.rb
 - lib/exwiw/determine_table_processing_order.rb
+- lib/exwiw/embedded_in.rb
 - lib/exwiw/mongo_query.rb
 - lib/exwiw/mongodb_collection_config.rb
 - lib/exwiw/mongodb_field.rb
@@ -58,6 +61,7 @@ files:
 - lib/exwiw/table_config.rb
 - lib/exwiw/version.rb
 - lib/tasks/exwiw.rake
+- mise.toml
 homepage: https://github.com/riseshia/exwiw
 licenses:
 - MIT
@@ -79,7 +83,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.6.9
+rubygems_version: 4.0.10
 specification_version: 4
 summary: Ruby gem that allows you to export records from a database to a dump file.
 test_files: []

data/.rubocop.yml

@@ -1,10 +0,0 @@
-plugins:
-  - rubocop-greppable_rails
-
-AllCops:
-  TargetRubyVersion: 3.3
-  DisabledByDefault: true
-  SuggestExtensions: false
-
-GreppableRails/UseInlineAccessModifier:
-  Enabled: true