feat(d2): database/dialect.py - 多方言 SQL-Java 类型映射

新增 DialectAdapter 抽象基类 + MySQLDialect + PostgreSQLDialect:
- MySQL: 从 template_context.py 抽出来的现有映射 (无行为变化)
- PostgreSQL 12+: 新增,重点覆盖差异类型
  * INT2/INT4/INT8 + SERIAL/BIGSERIAL 系列
  * 原生 BOOLEAN (不走 MySQL 的 TINYINT(1) hack)
  * BYTEA / JSON / JSONB / UUID / INET
  * TIMESTAMPTZ -> OffsetDateTime (跨时区场景)
  * CHARACTER VARYING (VARCHAR 官方全名)
- get_dialect(name) registry,未识别退回 MySQL (向后兼容)

36 个 unit test:
- 11 MySQL 映射 (含 TINYINT(1) 布尔)
- 13 PostgreSQL 映射 (含 SERIAL / TIMESTAMPTZ / JSONB)
- 3 跨方言隔离测试 (确保 PG 没 TINYINT / MySQL 没 BYTEA)
- registry / strip_paren / abstract 校验

dialect.py 100% line coverage。

Why: PostgreSQL 整数族 (INT2/INT4/INT8) + 自增 (SERIAL) + 原生类型 (UUID/
JSONB/TIMESTAMPTZ) 跟 MySQL 差异大,硬塞进 template_context 的映射会乱。
独立成模块后,Oracle/SQLServer 加进来也只是新增一个 adapter。

下一步 (D3) 用 Testcontainers 起 postgres:16,跑出来的真实类型走这套映射。

Author: ZhaoXingPeng

src/dbjavagenix/database/dialect.py

@@ -0,0 +1,322 @@
+"""
+P9.2: Database dialect adapters.
+
+为何独立成模块:
+  - v0.1 ~ v0.2 时期 SQL→Java 类型映射散落在 template_context.py 里,只考虑 MySQL
+  - PostgreSQL 引入后,差异点:
+      * 整数: INT2/INT4/INT8 vs MySQL 的 INT/BIGINT
+      * 序列: SERIAL/BIGSERIAL (PG 自增惯例,无 AUTO_INCREMENT 关键字)
+      * 布尔: PG 原生 BOOLEAN, MySQL 用 TINYINT(1) 模拟
+      * 二进制: BYTEA vs BLOB 系列
+      * JSON: JSON + JSONB (两种,PG 推荐 JSONB)
+      * UUID: PG 原生 UUID, MySQL 没有 → CHAR(36) 存
+      * 时区: TIMESTAMPTZ → OffsetDateTime,普通 TIMESTAMP → LocalDateTime
+  - 把方言策略和模板上下文剥离开,后续要加 Oracle / SQLServer 时加一个 adapter 就行
+
+调用方:
+  - template_context.py (TODO 下个迭代切过去,保持向后兼容)
+  - codegen_tools.py / atomic_codegen_tools.py
+  - 集成测试 (PostgreSQL Testcontainers 跑出来的类型也用同一套映射)
+"""
+
+from __future__ import annotations
+
+import re
+from abc import ABC, abstractmethod
+from typing import Dict
+
+
+_PAREN_PATTERN = re.compile(r"\([^)]*\)")
+
+
+def _strip_paren(db_type: str) -> str:
+    """剥离 `VARCHAR(64)` 这种带长度的括号部分,统一小写无空格。"""
+    return _PAREN_PATTERN.sub("", db_type.upper()).strip()
+
+
+# ============================================================
+# 基类
+# ============================================================
+
+
+class DialectAdapter(ABC):
+    """方言适配器基类。每个 DB 厂家一个子类。"""
+
+    name: str = "abstract"
+
+    @property
+    @abstractmethod
+    def type_to_java(self) -> Dict[str, str]:
+        """SQL base type → Java type. base type 已 strip 括号。"""
+
+    @property
+    @abstractmethod
+    def type_to_jdbc(self) -> Dict[str, str]:
+        """SQL base type → JDBC type 字符串 (用于 MyBatis Plus jdbcType)。"""
+
+    @property
+    def string_types(self) -> set[str]:
+        """该方言下视为 String 的类型集合 (base type)。"""
+        return {k for k, v in self.type_to_java.items() if v == "String"}
+
+    @property
+    def date_types(self) -> set[str]:
+        """日期/时间类型 (用于决定是否 import LocalDate 等)。"""
+        return {
+            k
+            for k, v in self.type_to_java.items()
+            if v in {"LocalDate", "LocalTime", "LocalDateTime", "OffsetDateTime", "Instant"}
+        }
+
+    @property
+    def decimal_types(self) -> set[str]:
+        return {k for k, v in self.type_to_java.items() if v == "BigDecimal"}
+
+    def java_type_for(self, db_type: str) -> str:
+        """主入口: 给一个数据库声明类型,返回 Java 类型 (找不到默认 String)。"""
+        base = _strip_paren(db_type)
+        # TINYINT(1) 是 MySQL 布尔约定 — 单独判
+        if db_type.upper().replace(" ", "") == "TINYINT(1)":
+            return self.type_to_java.get("TINYINT(1)", self.type_to_java.get(base, "String"))
+        return self.type_to_java.get(base, "String")
+
+    def jdbc_type_for(self, db_type: str) -> str:
+        base = _strip_paren(db_type)
+        return self.type_to_jdbc.get(base, "VARCHAR")
+
+    def is_string_type(self, db_type: str) -> bool:
+        return _strip_paren(db_type) in self.string_types
+
+    def is_date_type(self, db_type: str) -> bool:
+        return _strip_paren(db_type) in self.date_types
+
+    def is_decimal_type(self, db_type: str) -> bool:
+        return _strip_paren(db_type) in self.decimal_types
+
+
+# ============================================================
+# MySQL (从 template_context.py 抽出来的现有逻辑)
+# ============================================================
+
+
+class MySQLDialect(DialectAdapter):
+    name = "mysql"
+
+    @property
+    def type_to_java(self) -> Dict[str, str]:
+        return {
+            # 整数
+            "TINYINT": "Byte",
+            "TINYINT(1)": "Boolean",  # 经典 MySQL 布尔约定
+            "SMALLINT": "Short",
+            "MEDIUMINT": "Integer",
+            "INT": "Integer",
+            "INTEGER": "Integer",
+            "BIGINT": "Long",
+            # 浮点
+            "FLOAT": "Float",
+            "DOUBLE": "Double",
+            "DECIMAL": "BigDecimal",
+            "NUMERIC": "BigDecimal",
+            # 字符串
+            "CHAR": "String",
+            "VARCHAR": "String",
+            "TEXT": "String",
+            "LONGTEXT": "String",
+            "MEDIUMTEXT": "String",
+            "TINYTEXT": "String",
+            "NCHAR": "String",
+            "NVARCHAR": "String",
+            "ENUM": "String",
+            "SET": "String",
+            # 日期
+            "DATE": "LocalDate",
+            "TIME": "LocalTime",
+            "DATETIME": "LocalDateTime",
+            "TIMESTAMP": "LocalDateTime",
+            "YEAR": "Integer",
+            # 布尔
+            "BOOLEAN": "Boolean",
+            "BOOL": "Boolean",
+            # 二进制
+            "BINARY": "byte[]",
+            "VARBINARY": "byte[]",
+            "BLOB": "byte[]",
+            "LONGBLOB": "byte[]",
+            "MEDIUMBLOB": "byte[]",
+            "TINYBLOB": "byte[]",
+            # JSON
+            "JSON": "String",
+        }
+
+    @property
+    def type_to_jdbc(self) -> Dict[str, str]:
+        return {
+            "TINYINT": "TINYINT",
+            "SMALLINT": "SMALLINT",
+            "MEDIUMINT": "INTEGER",
+            "INT": "INTEGER",
+            "INTEGER": "INTEGER",
+            "BIGINT": "BIGINT",
+            "FLOAT": "FLOAT",
+            "DOUBLE": "DOUBLE",
+            "DECIMAL": "DECIMAL",
+            "NUMERIC": "NUMERIC",
+            "CHAR": "CHAR",
+            "VARCHAR": "VARCHAR",
+            "TEXT": "LONGVARCHAR",
+            "LONGTEXT": "LONGVARCHAR",
+            "MEDIUMTEXT": "LONGVARCHAR",
+            "TINYTEXT": "VARCHAR",
+            "NCHAR": "NCHAR",
+            "NVARCHAR": "NVARCHAR",
+            "DATE": "DATE",
+            "TIME": "TIME",
+            "DATETIME": "TIMESTAMP",
+            "TIMESTAMP": "TIMESTAMP",
+            "BOOLEAN": "BOOLEAN",
+            "TINYINT(1)": "BOOLEAN",
+            "BLOB": "BLOB",
+            "JSON": "LONGVARCHAR",
+        }
+
+
+# ============================================================
+# PostgreSQL
+# ============================================================
+
+
+class PostgreSQLDialect(DialectAdapter):
+    """PostgreSQL 12+ 方言。
+
+    主要差异 vs MySQL:
+      - 整数族用 INT2/INT4/INT8 (官方别名 SMALLINT/INTEGER/BIGINT 也支持)
+      - SERIAL/BIGSERIAL: 自增列,JDBC 端按 Integer/Long 处理
+      - BOOLEAN 原生支持 (不需要 TINYINT(1) hack)
+      - BYTEA 替代 BLOB
+      - JSONB 推荐 (二进制 JSON,带索引能力);JSON 也保留
+      - UUID 原生支持
+      - TIMESTAMPTZ → OffsetDateTime (跨时区场景必须保留时区)
+      - "CHARACTER VARYING" 是 VARCHAR 的官方全名
+    """
+
+    name = "postgresql"
+
+    @property
+    def type_to_java(self) -> Dict[str, str]:
+        return {
+            # 整数
+            "INT2": "Short",
+            "SMALLINT": "Short",
+            "INT4": "Integer",
+            "INT": "Integer",
+            "INTEGER": "Integer",
+            "INT8": "Long",
+            "BIGINT": "Long",
+            # SERIAL 系列 (PG 自增惯例) - JDBC 上和普通整数一致
+            "SMALLSERIAL": "Short",
+            "SERIAL": "Integer",
+            "BIGSERIAL": "Long",
+            # 浮点
+            "REAL": "Float",
+            "FLOAT4": "Float",
+            "DOUBLE PRECISION": "Double",
+            "FLOAT8": "Double",
+            "NUMERIC": "BigDecimal",
+            "DECIMAL": "BigDecimal",
+            "MONEY": "BigDecimal",
+            # 字符串
+            "CHAR": "String",
+            "CHARACTER": "String",
+            "VARCHAR": "String",
+            "CHARACTER VARYING": "String",
+            "TEXT": "String",
+            "BPCHAR": "String",  # internal name for CHAR
+            # 日期
+            "DATE": "LocalDate",
+            "TIME": "LocalTime",
+            "TIMETZ": "OffsetTime",
+            "TIMESTAMP": "LocalDateTime",
+            "TIMESTAMPTZ": "OffsetDateTime",  # 跨时区必须 OffsetDateTime
+            "TIMESTAMP WITH TIME ZONE": "OffsetDateTime",
+            # 布尔
+            "BOOL": "Boolean",
+            "BOOLEAN": "Boolean",
+            # 二进制
+            "BYTEA": "byte[]",
+            # JSON
+            "JSON": "String",
+            "JSONB": "String",
+            # UUID
+            "UUID": "String",
+            # 网络
+            "INET": "String",
+            "CIDR": "String",
+            # 数组占位 — Java 端建议 String,具体由模板决定
+            "ARRAY": "String",
+        }
+
+    @property
+    def type_to_jdbc(self) -> Dict[str, str]:
+        return {
+            "INT2": "SMALLINT",
+            "SMALLINT": "SMALLINT",
+            "INT4": "INTEGER",
+            "INT": "INTEGER",
+            "INTEGER": "INTEGER",
+            "INT8": "BIGINT",
+            "BIGINT": "BIGINT",
+            "SMALLSERIAL": "SMALLINT",
+            "SERIAL": "INTEGER",
+            "BIGSERIAL": "BIGINT",
+            "REAL": "REAL",
+            "FLOAT4": "REAL",
+            "DOUBLE PRECISION": "DOUBLE",
+            "FLOAT8": "DOUBLE",
+            "NUMERIC": "NUMERIC",
+            "DECIMAL": "DECIMAL",
+            "MONEY": "DECIMAL",
+            "CHAR": "CHAR",
+            "CHARACTER": "CHAR",
+            "BPCHAR": "CHAR",
+            "VARCHAR": "VARCHAR",
+            "CHARACTER VARYING": "VARCHAR",
+            "TEXT": "LONGVARCHAR",
+            "DATE": "DATE",
+            "TIME": "TIME",
+            "TIMETZ": "TIME_WITH_TIMEZONE",
+            "TIMESTAMP": "TIMESTAMP",
+            "TIMESTAMPTZ": "TIMESTAMP_WITH_TIMEZONE",
+            "TIMESTAMP WITH TIME ZONE": "TIMESTAMP_WITH_TIMEZONE",
+            "BOOL": "BOOLEAN",
+            "BOOLEAN": "BOOLEAN",
+            "BYTEA": "BINARY",
+            "JSON": "OTHER",
+            "JSONB": "OTHER",
+            "UUID": "OTHER",
+            "INET": "OTHER",
+            "CIDR": "OTHER",
+        }
+
+
+# ============================================================
+# Registry
+# ============================================================
+
+
+_REGISTRY: Dict[str, DialectAdapter] = {
+    "mysql": MySQLDialect(),
+    "postgresql": PostgreSQLDialect(),
+}
+
+
+def get_dialect(db_type: str) -> DialectAdapter:
+    """根据 DatabaseType 字符串 (mysql / postgresql / ...) 取适配器。
+
+    未识别的 db_type 退回 MySQL — 保持向后兼容,避免老调用方崩。
+    """
+    return _REGISTRY.get(db_type.lower(), _REGISTRY["mysql"])
+
+
+def list_supported_dialects() -> list[str]:
+    return sorted(_REGISTRY.keys())