mirror of
https://github.com/TriM-Organization/LiteyukiBot-TriM.git
synced 2024-11-11 01:27:29 +08:00
feat: 新的orm框架
This commit is contained in:
parent
45c3bf3a5e
commit
c2e43fff7a
330
liteyuki/utils/data_new.py
Normal file
330
liteyuki/utils/data_new.py
Normal file
@ -0,0 +1,330 @@
|
||||
import copy
|
||||
import json
|
||||
import os
|
||||
import pickle
|
||||
import sqlite3
|
||||
import types
|
||||
from types import NoneType
|
||||
from collections.abc import Iterable
|
||||
from pydantic import BaseModel, Field
|
||||
from typing import Any
|
||||
|
||||
LOG_OUT = True
|
||||
|
||||
|
||||
def log(*args, **kwargs):
|
||||
if LOG_OUT:
|
||||
print(*args, **kwargs)
|
||||
|
||||
|
||||
class LiteModel(BaseModel):
|
||||
TABLE_NAME: str = None
|
||||
id: int = None
|
||||
|
||||
|
||||
class Database:
|
||||
def __init__(self, db_name: str):
|
||||
|
||||
if os.path.dirname(db_name) != "" and not os.path.exists(os.path.dirname(db_name)):
|
||||
os.makedirs(os.path.dirname(db_name))
|
||||
|
||||
self.db_name = db_name
|
||||
self.conn = sqlite3.connect(db_name)
|
||||
self.cursor = self.conn.cursor()
|
||||
|
||||
def first(self, model: LiteModel, condition: str, *args: Any, default: Any = None) -> LiteModel | Any | None:
|
||||
"""查询第一个
|
||||
Args:
|
||||
model: 数据模型实例
|
||||
condition: 查询条件,不给定则查询所有
|
||||
*args: 参数化查询参数
|
||||
default: 默认值
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
all_results = self.all(model, condition, *args, default=default)
|
||||
return all_results[0] if all_results else default
|
||||
|
||||
def all(self, model: LiteModel, condition: str = "", *args: Any, default: Any = None) -> list[LiteModel] | list[Any] | None:
|
||||
"""查询所有
|
||||
Args:
|
||||
model: 数据模型实例
|
||||
condition: 查询条件,不给定则查询所有
|
||||
*args: 参数化查询参数
|
||||
default: 默认值
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
table_name = model.TABLE_NAME
|
||||
model_type = type(model)
|
||||
if not table_name:
|
||||
raise ValueError(f"数据模型{model_type.__name__}未提供表名")
|
||||
|
||||
condition = f"WHERE {condition}"
|
||||
|
||||
results = self.cursor.execute(f"SELECT * FROM {table_name} {condition}", args).fetchall()
|
||||
fields = [description[0] for description in self.cursor.description]
|
||||
if not results:
|
||||
return default
|
||||
else:
|
||||
return [model_type(**self._load(dict(zip(fields, result)))) for result in results]
|
||||
|
||||
def upsert(self, *args: LiteModel):
|
||||
"""增/改操作
|
||||
Args:
|
||||
*args:
|
||||
|
||||
Returns:
|
||||
"""
|
||||
table_list = [item[0] for item in self.cursor.execute("SELECT name FROM sqlite_master WHERE type='table'").fetchall()]
|
||||
for model in args:
|
||||
if not model.TABLE_NAME:
|
||||
raise ValueError(f"数据模型 {model.__class__.__name__} 未提供表名")
|
||||
elif model.TABLE_NAME not in table_list:
|
||||
raise ValueError(f"数据模型 {model.__class__.__name__} 的表 {model.TABLE_NAME} 不存在,请先迁移")
|
||||
else:
|
||||
self._save(model.model_dump(by_alias=True))
|
||||
|
||||
def _save(self, obj: Any) -> Any:
|
||||
# obj = copy.deepcopy(obj)
|
||||
if isinstance(obj, dict):
|
||||
table_name = obj.get("TABLE_NAME")
|
||||
row_id = obj.get("id")
|
||||
new_obj = {}
|
||||
for field, value in obj.items():
|
||||
if isinstance(value, self.ITERABLE_TYPE):
|
||||
new_obj[self._get_stored_field_prefix(value) + field] = self._save(value) # self._save(value) # -> bytes
|
||||
elif isinstance(value, self.BASIC_TYPE):
|
||||
new_obj[field] = value
|
||||
else:
|
||||
raise ValueError(f"数据模型{table_name}包含不支持的数据类型,字段:{field} 值:{value} 值类型:{type(value)}")
|
||||
if table_name:
|
||||
fields, values = [], []
|
||||
for n_field, n_value in new_obj.items():
|
||||
if n_field not in ["TABLE_NAME", "id"]:
|
||||
fields.append(n_field)
|
||||
values.append(n_value)
|
||||
# 移除TABLE_NAME和id
|
||||
fields = list(fields)
|
||||
values = list(values)
|
||||
if row_id is not None:
|
||||
# 如果 _id 不为空,将 'id' 插入到字段列表的开始
|
||||
fields.insert(0, 'id')
|
||||
# 将 _id 插入到值列表的开始
|
||||
values.insert(0, row_id)
|
||||
fields = ', '.join([f'"{field}"' for field in fields])
|
||||
placeholders = ', '.join('?' for _ in values)
|
||||
self.cursor.execute(f"INSERT OR REPLACE INTO {table_name}({fields}) VALUES ({placeholders})", tuple(values))
|
||||
self.conn.commit()
|
||||
foreign_id = self.cursor.execute("SELECT last_insert_rowid()").fetchone()[0]
|
||||
return f"{self.FOREIGN_KEY_PREFIX}{foreign_id}@{table_name}" # -> FOREIGN_KEY_123456@{table_name} id@{table_name}
|
||||
else:
|
||||
return pickle.dumps(new_obj) # -> bytes
|
||||
elif isinstance(obj, (list, set, tuple)):
|
||||
obj_type = type(obj) # 到时候转回去
|
||||
new_obj = []
|
||||
for item in obj:
|
||||
if isinstance(item, self.ITERABLE_TYPE):
|
||||
new_obj.append(self._save(item))
|
||||
elif isinstance(item, self.BASIC_TYPE):
|
||||
new_obj.append(item)
|
||||
else:
|
||||
raise ValueError(f"数据模型包含不支持的数据类型,值:{item} 值类型:{type(item)}")
|
||||
return pickle.dumps(obj_type(new_obj)) # -> bytes
|
||||
else:
|
||||
raise ValueError(f"数据模型包含不支持的数据类型,值:{obj} 值类型:{type(obj)}")
|
||||
|
||||
def _load(self, obj: Any) -> Any:
|
||||
|
||||
if isinstance(obj, dict):
|
||||
|
||||
new_obj = {}
|
||||
|
||||
for field, value in obj.items():
|
||||
|
||||
field: str
|
||||
|
||||
if field.startswith(self.BYTES_PREFIX):
|
||||
|
||||
new_obj[field.replace(self.BYTES_PREFIX, "")] = self._load(pickle.loads(value))
|
||||
|
||||
elif field.startswith(self.FOREIGN_KEY_PREFIX):
|
||||
|
||||
new_obj[field.replace(self.FOREIGN_KEY_PREFIX, "")] = self._load(self._get_foreign_data(value))
|
||||
|
||||
else:
|
||||
new_obj[field] = value
|
||||
return new_obj
|
||||
elif isinstance(obj, (list, set, tuple)):
|
||||
|
||||
print(" - Load as List")
|
||||
|
||||
new_obj = []
|
||||
for item in obj:
|
||||
|
||||
print(" - Loading Item", item)
|
||||
|
||||
if isinstance(item, bytes):
|
||||
|
||||
# 对bytes进行尝试解析,解析失败则返回原始bytes
|
||||
try:
|
||||
new_obj.append(self._load(pickle.loads(item)))
|
||||
except Exception as e:
|
||||
new_obj.append(self._load(item))
|
||||
|
||||
print(" - Load as Bytes | Result:", new_obj[-1])
|
||||
|
||||
elif isinstance(item, str) and item.startswith(self.FOREIGN_KEY_PREFIX):
|
||||
new_obj.append(self._load(self._get_foreign_data(item)))
|
||||
else:
|
||||
new_obj.append(self._load(item))
|
||||
return new_obj
|
||||
else:
|
||||
return obj
|
||||
|
||||
def delete(self, model: LiteModel, condition: str, *args: Any):
|
||||
pass
|
||||
|
||||
def auto_migrate(self, *args: LiteModel):
|
||||
|
||||
"""
|
||||
自动迁移模型
|
||||
Args:
|
||||
*args: 模型类实例化对象,支持空默认值,不支持嵌套迁移
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
for model in args:
|
||||
if not model.TABLE_NAME:
|
||||
raise ValueError(f"数据模型{type(model).__name__}未提供表名")
|
||||
|
||||
# 若无则创建表
|
||||
self.cursor.execute(
|
||||
f'CREATE TABLE IF NOT EXISTS "{model.TABLE_NAME}" (id INTEGER PRIMARY KEY AUTOINCREMENT)'
|
||||
)
|
||||
|
||||
# 获取表结构,field -> SqliteType
|
||||
new_structure = {}
|
||||
for n_field, n_value in model.model_dump(by_alias=True).items():
|
||||
if n_field not in ["TABLE_NAME", "id"]:
|
||||
new_structure[self._get_stored_field_prefix(n_value) + n_field] = self._get_stored_type(n_value)
|
||||
|
||||
# 原有的字段列表
|
||||
existing_structure = dict([(column[1], column[2]) for column in self.cursor.execute(f'PRAGMA table_info({model.TABLE_NAME})').fetchall()])
|
||||
# 检测缺失字段,由于SQLite是动态类型,所以不需要检测类型
|
||||
for n_field, n_type in new_structure.items():
|
||||
if n_field not in existing_structure.keys() and n_field.lower() not in ["id", "table_name"]:
|
||||
self.cursor.execute(
|
||||
f'ALTER TABLE "{model.TABLE_NAME}" ADD COLUMN "{n_field}" {n_type}'
|
||||
)
|
||||
|
||||
# 检测多余字段进行删除
|
||||
for e_field in existing_structure.keys():
|
||||
if e_field not in new_structure.keys() and e_field.lower() not in ['id']:
|
||||
self.cursor.execute(
|
||||
f'ALTER TABLE "{model.TABLE_NAME}" DROP COLUMN "{e_field}"'
|
||||
)
|
||||
|
||||
self.conn.commit()
|
||||
# 已完成
|
||||
|
||||
def _get_stored_field_prefix(self, value) -> str:
|
||||
"""根据类型获取存储字段前缀,一定在后加上字段名
|
||||
* -> ""
|
||||
Args:
|
||||
value: 储存的值
|
||||
|
||||
Returns:
|
||||
Sqlite3存储字段
|
||||
"""
|
||||
|
||||
if isinstance(value, LiteModel) or isinstance(value, dict) and "TABLE_NAME" in value:
|
||||
return self.FOREIGN_KEY_PREFIX
|
||||
elif type(value) in self.ITERABLE_TYPE:
|
||||
return self.BYTES_PREFIX
|
||||
return ""
|
||||
|
||||
def _get_stored_type(self, value) -> str:
|
||||
"""获取存储类型
|
||||
|
||||
Args:
|
||||
value: 储存的值
|
||||
|
||||
Returns:
|
||||
Sqlite3存储类型
|
||||
"""
|
||||
if isinstance(value, dict) and "TABLE_NAME" in value:
|
||||
# 是一个模型字典,储存外键
|
||||
return "INTEGER"
|
||||
return self.TYPE_MAPPING.get(type(value), "TEXT")
|
||||
|
||||
def _get_foreign_data(self, foreign_value: str) -> dict:
|
||||
"""
|
||||
获取外键数据
|
||||
Args:
|
||||
foreign_value:
|
||||
|
||||
Returns:
|
||||
|
||||
"""
|
||||
foreign_value = foreign_value.replace(self.FOREIGN_KEY_PREFIX, "")
|
||||
table_name = foreign_value.split("@")[-1]
|
||||
foreign_id = foreign_value.split("@")[0]
|
||||
fields = [description[1] for description in self.cursor.execute(f"PRAGMA table_info({table_name})").fetchall()]
|
||||
result = self.cursor.execute(f"SELECT * FROM {table_name} WHERE id = ?", (foreign_id,)).fetchone()
|
||||
return dict(zip(fields, result))
|
||||
|
||||
TYPE_MAPPING = {
|
||||
int : "INTEGER",
|
||||
float : "REAL",
|
||||
str : "TEXT",
|
||||
bool : "INTEGER",
|
||||
bytes : "BLOB",
|
||||
NoneType : "NULL",
|
||||
# dict : "TEXT",
|
||||
# list : "TEXT",
|
||||
# tuple : "TEXT",
|
||||
# set : "TEXT",
|
||||
|
||||
dict : "BLOB", # LITEYUKIDICT{key_name}
|
||||
list : "BLOB", # LITEYUKILIST{key_name}
|
||||
tuple : "BLOB", # LITEYUKITUPLE{key_name}
|
||||
set : "BLOB", # LITEYUKISET{key_name}
|
||||
LiteModel: "INTEGER" # FOREIGN_KEY_{table_name}
|
||||
}
|
||||
|
||||
# 基础类型
|
||||
BASIC_TYPE = (int, float, str, bool, bytes, NoneType)
|
||||
# 可序列化类型
|
||||
ITERABLE_TYPE = (dict, list, tuple, set, LiteModel)
|
||||
|
||||
# 外键前缀
|
||||
FOREIGN_KEY_PREFIX = "FOREIGN_KEY_"
|
||||
# 转换为的字节前缀
|
||||
BYTES_PREFIX = "PICKLE_BYTES_"
|
||||
|
||||
|
||||
def check_sqlite_keyword(name):
|
||||
sqlite_keywords = [
|
||||
"ABORT", "ACTION", "ADD", "AFTER", "ALL", "ALTER", "ANALYZE", "AND", "AS", "ASC",
|
||||
"ATTACH", "AUTOINCREMENT", "BEFORE", "BEGIN", "BETWEEN", "BY", "CASCADE", "CASE",
|
||||
"CAST", "CHECK", "COLLATE", "COLUMN", "COMMIT", "CONFLICT", "CONSTRAINT", "CREATE",
|
||||
"CROSS", "CURRENT_DATE", "CURRENT_TIME", "CURRENT_TIMESTAMP", "DATABASE", "DEFAULT",
|
||||
"DEFERRABLE", "DEFERRED", "DELETE", "DESC", "DETACH", "DISTINCT", "DROP", "EACH",
|
||||
"ELSE", "END", "ESCAPE", "EXCEPT", "EXCLUSIVE", "EXISTS", "EXPLAIN", "FAIL", "FOR",
|
||||
"FOREIGN", "FROM", "FULL", "GLOB", "GROUP", "HAVING", "IF", "IGNORE", "IMMEDIATE",
|
||||
"IN", "INDEX", "INDEXED", "INITIALLY", "INNER", "INSERT", "INSTEAD", "INTERSECT",
|
||||
"INTO", "IS", "ISNULL", "JOIN", "KEY", "LEFT", "LIKE", "LIMIT", "MATCH", "NATURAL",
|
||||
"NO", "NOT", "NOTNULL", "NULL", "OF", "OFFSET", "ON", "OR", "ORDER", "OUTER", "PLAN",
|
||||
"PRAGMA", "PRIMARY", "QUERY", "RAISE", "RECURSIVE", "REFERENCES", "REGEXP", "REINDEX",
|
||||
"RELEASE", "RENAME", "REPLACE", "RESTRICT", "RIGHT", "ROLLBACK", "ROW", "SAVEPOINT",
|
||||
"SELECT", "SET", "TABLE", "TEMP", "TEMPORARY", "THEN", "TO", "TRANSACTION", "TRIGGER",
|
||||
"UNION", "UNIQUE", "UPDATE", "USING", "VACUUM", "VALUES", "VIEW", "VIRTUAL", "WHEN",
|
||||
"WHERE", "WITH", "WITHOUT"
|
||||
]
|
||||
return True
|
||||
# if name.upper() in sqlite_keywords:
|
||||
# raise ValueError(f"'{name}' 是SQLite保留字,不建议使用,请更换名称")
|
Loading…
Reference in New Issue
Block a user