在 sqlalchemy 中定义 ORM 关系

Defining ORM relationships in sqlalchemy

我正一头扎进 sqlalchemy ORM 并试图理解对象之间的关系。

我创建了 3 个 classes 分别代表日志文件 (Log)、从日志文件解析的记录 (LogRecord) 和每个日志记录所属的用户 (User)。

我是这样定义 classes 的:

from sqlalchemy import Table, Column, Integer, String, ForeignKey, create_engine
from sqlalchemy.orm import relationship, backref, sessionmaker
from sqlalchemy.ext.declarative import declarative_base
import random

engine = create_engine('sqlite://')
# Create the session class
Session = sessionmaker(bind=engine)
# Create the session
session = Session()

Base = declarative_base()

class Log(Base):

    __tablename__ = 'logs'

    id = Column(Integer, primary_key=True, unique=True)
    log_name = Column(String, unique=True)
    records = relationship("LogRecord", backref="logs")

    def __repr__(self):
        return '<Log(log_name={:s}, id={:d}>'.format(
            self.log_name, self.id)

class LogRecord(Base):

    __tablename__ = 'log_records'

    id = Column(Integer, primary_key=True, unique=True)
    record_name = Column(String)
    log_id = Column(Integer, ForeignKey('logs.id'))
    user_id = Column(Integer, ForeignKey('user.id'))

    user = relationship("User", backref="log_records")

    def __repr__(self):
        return '<LogRecord(record_name={:s}, id={:d}>'.format(
            self.record_name,
            self.id)

class User(Base):

    __tablename__ = 'user'

    id = Column(Integer, primary_key=True, unique=True)
    name = Column(String, unique=True)

    def __repr__(self):
        return '<User(name={:s}, id={:d}>'.format(self.name, self.id)

然后用一些虚假数据填充它们:

Base.metadata.create_all(engine)

users = ['John', 'Alex', 'Nikki']

# Add 10 log files to Log
for x in range(0,10):
    log = Log(log_name='log{:d}.txt'.format(x))
    session.add(log)

# Add users
for user in users:
    session.add(User(name=user))

# Select all users
users = session.query(User).all()

# Add log records
for log in session.query(Log):
    # Assign a random user to each log record
    user = users[int(random.random() * len(users))]
    for x in range(0,10):
        # Assign a random user to each log record
        user = users[int(random.random() * len(users))]
        r = LogRecord(record_name='{:s}-{:d}'.format(log.log_name, int(random.random()*100)),log_id=log.id,user=user)

        session.add(r)

每个日志文件可以包含一个或多个用户的记录。我想获得一个不同的日志文件列表,其中至少包含来自指定用户的一条记录。我试过了:

In [8]: user = session.query(User).filter(User.name == 'John').one()

In [9]: user
Out[9]: <User(name=John, id=1>

然后尝试使用以下方法查找日志:

In [10]: session.query(Log).filter(Log.records.user == user).all()

但得到了以下信息:

session.query(Log).filter(Log.records.user == user).distinct().all()
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-13-d6d323e178f8> in <module>()
----> 1 session.query(Log).filter(Log.records.user == user).distinct().all()

/Users/kerfoot/code/venvs/sqlalchemy/lib/python2.7/sitepackages/sqlalchemy/orm/attributes.pyc in __getattr__(self, key)
192                     type(self.comparator).__name__,
193                     self,
--> 194                     key)
195             )
196 

AttributeError: Neither 'InstrumentedAttribute' object nor 'Comparator' object associated with Log.records has an attribute 'user'

不确定我的 class defs 或查询本身是否有问题。

非常感谢任何帮助!

您不能对属性属性进行过滤,或者简单地说:表达式中不能有多个点:Log.records.user不工作。

Option-1: 你可以像在 SQL:

中那样做
session.query(Log).join(LogRecord, Log.records).filter(LogRecord.user == user).all()

sqlalchemy 足够聪明 return 你只有 Log 个实例的列表而没有任何重复(即使 SQL 语句可能会导致重复)。

Option-2: 另一种方法是使用 any(...),这在逻辑上 更干净:

session.query(Log).filter(Log.records.any(LogRecord.user == user)).all()

Transcript:
-> get all Log instances,
-> which contains at least one LogRecord,
-> that belongs to given user.