From af9b6253b12c5381378412319cf423ffa6d42530 Mon Sep 17 00:00:00 2001 From: YunaiV Date: Sun, 31 May 2026 17:49:23 +0800 Subject: [PATCH] fix(sql): 修复 convertor 索引解析兼容问题 --- sql/tools/convertor.py | 107 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++----------------- 1 file changed, 90 insertions(+), 17 deletions(-) diff --git a/sql/tools/convertor.py b/sql/tools/convertor.py index 3a8b9f3..3d61e0e 100644 --- a/sql/tools/convertor.py +++ b/sql/tools/convertor.py @@ -77,6 +77,9 @@ def load_and_clean(sql_file: str) -> str: class Convertor(ABC): + # 不同数据库的关键字不完全一致；子类按需声明需要转义的列名。 + reserved_column_names = set() + def __init__(self, src: str, db_type) -> None: self.src = src self.db_type = db_type @@ -179,6 +182,31 @@ class Convertor(ABC): """ return "" + def escape_column_name(self, name: str) -> str: + """转义目标库保留字列名，例如 Oracle / Kingbase 的 level。""" + + column_name = name.lower() + if column_name in self.reserved_column_names: + return f'"{column_name}"' + return column_name + + def escape_insert_columns(self, insert_script: str) -> str: + """INSERT 显式列清单需要和 CREATE / COMMENT 使用同一套列名转义。""" + + match = re.match( + r"(INSERT INTO\s+\S+\s*\()([^)]+)(\)\s+VALUES\s+[\s\S]*)", + insert_script, + flags=re.IGNORECASE, + ) + if not match: + return insert_script + + columns = [ + self.escape_column_name(column.strip()) + for column in match.group(2).split(",") + ] + return f"{match.group(1)}{', '.join(columns)}{match.group(3)}" + @staticmethod def inserts(table_name: str, script_content: str) -> Generator: PREFIX = f"INSERT INTO `{table_name}`" @@ -204,26 +232,65 @@ class Convertor(ABC): Generator[str]: create index 语句 """ - def generate_columns(columns): - keys = [ - f"{col['name'].lower()}{' ' + col['order'].lower() if col['order'] != 'ASC' else ''}" - for col in columns[0] - ] - return ", ".join(keys) - - for no, index in enumerate(ddl["index"], 1): - columns = generate_columns(index["columns"]) + for no, index in enumerate(ddl.get("index", []), 1): + columns = ", ".join(Convertor.index_columns(index.get("columns", []))) + if not columns: + continue table_name = ddl["table_name"].lower() yield f"CREATE INDEX idx_{table_name}_{no:02d} ON {table_name} ({columns})" @staticmethod + def index_columns(columns) -> list: + """兼容 simple-ddl-parser 不同版本的索引列结构。""" + + keys = [] + + def append(name, order="ASC"): + if not name: + return + column_name = str(name).strip("`").lower() + column_order = str(order or "ASC").upper() + if column_order == "DESC": + keys.append(f"{column_name} desc") + else: + keys.append(column_name) + + def visit(value): + # 普通索引常见结构：[[{'name': 'user_id', 'order': 'ASC'}]] + if isinstance(value, (list, tuple)): + for item in value: + visit(item) + return + if isinstance(value, dict): + name = value.get("name") + if isinstance(name, (dict, list, tuple)): + visit(name) + return + append(name, value.get("order", "ASC")) + return + # 唯一索引在部分版本中会被解析成 ['mobile', 'ASC', 'tenant_id', 'ASC']。 + if isinstance(value, str): + token = value.strip("`") + order = token.upper() + if order in ("ASC", "DESC"): + if order == "DESC" and keys and not keys[-1].endswith(" desc"): + keys[-1] = f"{keys[-1]} desc" + return + append(token) + + visit(columns) + return keys + + @staticmethod def unique_index(ddl: Dict) -> Generator: if "constraints" in ddl and "uniques" in ddl["constraints"]: uk_list = ddl["constraints"]["uniques"] for uk in uk_list: table_name = ddl["table_name"] uk_name = uk["constraint_name"] - uk_columns = uk["columns"] + uk_columns = Convertor.index_columns(uk["columns"]) + if not uk_columns: + continue yield table_name, uk_name, uk_columns @staticmethod @@ -381,7 +448,7 @@ class PostgreSQLConvertor(Convertor): ) nullable = "NULL" if col["nullable"] else "NOT NULL" default = f"DEFAULT {col['default']}" if col["default"] is not None else "" - return f"{name} {full_type} {nullable} {default}" + return f"{self.escape_column_name(name)} {full_type} {nullable} {default}" table_name = ddl["table_name"].lower() columns = [f"{_generate_column(col).strip()}" for col in ddl["columns"]] @@ -406,7 +473,7 @@ CREATE TABLE {table_name} ( for column in table_ddl["columns"]: table_comment = column["comment"] script += ( - f"COMMENT ON COLUMN {table_ddl['table_name']}.{column['name']} IS '{table_comment}';" + f"COMMENT ON COLUMN {table_ddl['table_name']}.{self.escape_column_name(column['name'])} IS '{table_comment}';" + "\n" ) @@ -435,6 +502,7 @@ CREATE TABLE {table_name} ( """生成 insert 语句，以及根据最后的 insert id+1 生成 Sequence""" inserts = list(Convertor.inserts(table_name, self.content)) + inserts = [self.escape_insert_columns(s) for s in inserts] # 转换 MySQL 字符串转义为 PostgreSQL 格式：\\ -> \，\' -> '' inserts = [re.sub(r"\\\\|\\'", lambda m: "\\" if m.group() == "\\\\" else "''", s) for s in inserts] ## 生成 insert 脚本 @@ -482,6 +550,8 @@ INSERT INTO dual VALUES (1); class OracleConvertor(Convertor): + reserved_column_names = {"level", "size"} + def __init__(self, src): super().__init__(src, "Oracle") @@ -526,10 +596,8 @@ class OracleConvertor(Convertor): # Oracle的 INSERT '' 不能通过NOT NULL校验，因此对文字类型字段覆写为 NULL nullable = "NULL" if type in ("varchar", "text", "longtext") else nullable default = f"DEFAULT {col['default']}" if col["default"] is not None else "" - # Oracle 中 size 不能作为字段名 - field_name = '"size"' if name == "size" else name # Oracle DEFAULT 定义在 NULLABLE 之前 - return f"{field_name} {full_type} {default} {nullable}" + return f"{self.escape_column_name(name)} {full_type} {default} {nullable}" table_name = ddl["table_name"].lower() columns = [f"{generate_column(col).strip()}" for col in ddl["columns"]] @@ -554,7 +622,7 @@ CREATE TABLE {table_name} ( for column in table_ddl["columns"]: table_comment = column["comment"] script += ( - f"COMMENT ON COLUMN {table_ddl['table_name']}.{column['name']} IS '{table_comment}';" + f"COMMENT ON COLUMN {table_ddl['table_name']}.{self.escape_column_name(column['name'])} IS '{table_comment}';" + "\n" ) @@ -586,6 +654,7 @@ CREATE TABLE {table_name} ( """拷贝 INSERT 语句""" inserts = [] for insert_script in Convertor.inserts(table_name, self.content): + insert_script = self.escape_insert_columns(insert_script) # 对日期数据添加 TO_DATE 转换 insert_script = re.sub( r"('\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}')", @@ -907,6 +976,8 @@ SET IDENTITY_INSERT {table_name.lower()} OFF; class KingbaseConvertor(PostgreSQLConvertor): + reserved_column_names = {"level"} + def __init__(self, src): super().__init__(src) self.db_type = "Kingbase" @@ -925,7 +996,7 @@ class KingbaseConvertor(PostgreSQLConvertor): if full_type == "text": nullable = "NULL" default = f"DEFAULT {col['default']}" if col["default"] is not None else "" - return f"{name} {full_type} {nullable} {default}" + return f"{self.escape_column_name(name)} {full_type} {nullable} {default}" table_name = ddl["table_name"].lower() columns = [f"{_generate_column(col).strip()}" for col in ddl["columns"]] @@ -945,6 +1016,8 @@ CREATE TABLE {table_name} ( class OpengaussConvertor(KingbaseConvertor): + reserved_column_names = set() + def __init__(self, src): super().__init__(src) self.db_type = "OpenGauss" -- libgit2 0.21.4