我正在通过sqlalchemy处理postgresql数据库。我需要使用postgresql＆＃34; date＆＃34;来处理BC日期。列类型（它支持BC日期http://www.postgresql.org/docs/9.2/static/datatype-datetime.html）。但是虽然我可以以字符串的形式插入BC日期（例如＆＃39; 2000-01-01 BC＆＃39;），但我无法检索它们。试图从数据库中获取BC日期引发值错误：年份超出范围。起初我认为原因是它的类型Date中的sqlalchemy使用python datetime.date模块，它不支持BC日期，但我不完全确定。我试图找到一种方法来映射＆＃34; date＆＃34;在数据库中键入自定义python类（绕过datetime.date模块的使用）但我还没有成功（我找到了在sqlalchemy中为各种处理步骤添加逻辑的方法，比如result_processor或者带有column_expression的bind_expression，但我没有＆ ＃39;设法让它发挥作用。）
以下是重现问题的示例。
from sqlalchemy import create_engine, Column, Integer, String, Date
from sqlalchemy.orm import sessionmaker
from sqlalchemy.ext.declarative import declarative_base
engine = create_engine('postgresql://database:database@localhost:5432/sqlalchemy_test', echo=True)
Base = declarative_base()
class User(Base):
__tablename__ = 'users'
id = Column(Integer, primary_key=True)
name = Column(String)
date = Column(Date)
def __repr__(self):
return "<User(name='%s', date='%s')>" % (self.name, self.date)
Base.metadata.create_all(engine)
Session = sessionmaker(bind=engine)
session = Session()
# remove objects from previous launches
session.query(User).delete()
import datetime
a_date = datetime.date.today()
# Insert valid date in datetime.date format, it works.
ed_user = User(name='ed', date=a_date)
# Insert BC date in string format, it works.
al_user = User(name='al', date='1950-12-16 BC')
session.add_all([ed_user, al_user])
session.commit()
# Insert data via raw sql, it works
conn = engine.connect()
conn.execute('INSERT INTO users (name, date) VALUES (%s, %s)', ("Lu", "0090-04-25 BC"))
# Check that all 3 objects are present
user_names = session.query(User.name).all()
print user_names
# Check entry with AD date, it works.
user = session.query(User).filter_by(name='ed').first()
print '-- user vith valid date: ', user, user.date
# But when trying to fetch date fields in any way, it raises Value error
# Trying to fetch model, it raises Value error
users = session.query(User).all()
print users
# Trying to fetch only field, it raises Value error
user_dates = session.query(User.date).all()
print user_dates
# Even trying to fetch dates in raw sql raises Value error
a = conn.execute('SELECT * FROM users')
print [c for c in a]
输出跟踪：
2014-06-19 16:06:20,466 INFO sqlalchemy.engine.base.Engine select version()
2014-06-19 16:06:20,466 INFO sqlalchemy.engine.base.Engine {}
2014-06-19 16:06:20,468 INFO sqlalchemy.engine.base.Engine select current_schema()
2014-06-19 16:06:20,468 INFO sqlalchemy.engine.base.Engine {}
2014-06-19 16:06:20,470 INFO sqlalchemy.engine.base.Engine SELECT CAST('test plain returns' AS VARCHAR(60)) AS anon_1
2014-06-19 16:06:20,470 INFO sqlalchemy.engine.base.Engine {}
2014-06-19 16:06:20,471 INFO sqlalchemy.engine.base.Engine SELECT CAST('test unicode returns' AS VARCHAR(60)) AS anon_1
2014-06-19 16:06:20,471 INFO sqlalchemy.engine.base.Engine {}
2014-06-19 16:06:20,473 INFO sqlalchemy.engine.base.Engine show standard_conforming_strings
2014-06-19 16:06:20,473 INFO sqlalchemy.engine.base.Engine {}
2014-06-19 16:06:20,475 INFO sqlalchemy.engine.base.Engine select relname from pg_class c join pg_namespace n on n.oid=c.relnamespace where n.nspname=current_schema() and relname=%(name)s
2014-06-19 16:06:20,475 INFO sqlalchemy.engine.base.Engine {'name': u'users'}
2014-06-19 16:06:20,477 INFO sqlalchemy.engine.base.Engine BEGIN (implicit)
2014-06-19 16:06:20,478 INFO sqlalchemy.engine.base.Engine DELETE FROM users
2014-06-19 16:06:20,478 INFO sqlalchemy.engine.base.Engine {}
2014-06-19 16:06:20,480 INFO sqlalchemy.engine.base.Engine INSERT INTO users (name, date) VALUES (%(name)s, %(date)s) RETURNING users.id
2014-06-19 16:06:20,481 INFO sqlalchemy.engine.base.Engine {'date': datetime.date(2014, 6, 19), 'name': 'ed'}
2014-06-19 16:06:20,482 INFO sqlalchemy.engine.base.Engine INSERT INTO users (name, date) VALUES (%(name)s, %(date)s) RETURNING users.id
2014-06-19 16:06:20,482 INFO sqlalchemy.engine.base.Engine {'date': '1950-12-16 BC', 'name': 'al'}
2014-06-19 16:06:20,483 INFO sqlalchemy.engine.base.Engine COMMIT
2014-06-19 16:06:20,494 INFO sqlalchemy.engine.base.Engine INSERT INTO users (name, date) VALUES (%s, %s)
2014-06-19 16:06:20,494 INFO sqlalchemy.engine.base.Engine ('Lu', '0090-04-25 BC')
2014-06-19 16:06:20,495 INFO sqlalchemy.engine.base.Engine COMMIT
2014-06-19 16:06:20,517 INFO sqlalchemy.engine.base.Engine BEGIN (implicit)
2014-06-19 16:06:20,517 INFO sqlalchemy.engine.base.Engine SELECT users.name AS users_name
FROM users
2014-06-19 16:06:20,517 INFO sqlalchemy.engine.base.Engine {}
-- user names: [(u'ed',), (u'al',), (u'Lu',)]
2014-06-19 16:06:20,520 INFO sqlalchemy.engine.base.Engine SELECT users.id AS users_id, users.name AS users_name, users.date AS users_date
FROM users
WHERE users.name = %(name_1)s
LIMIT %(param_1)s
2014-06-19 16:06:20,520 INFO sqlalchemy.engine.base.Engine {'name_1': 'ed', 'param_1': 1}
-- user with valid date: <User(name='ed', date='2014-06-19')> 2014-06-19
2014-06-19 16:06:20,523 INFO sqlalchemy.engine.base.Engine SELECT users.id AS users_id, users.name AS users_name, users.date AS users_date
FROM users
2014-06-19 16:06:20,523 INFO sqlalchemy.engine.base.Engine {}
Traceback (most recent call last):
File "test_script.py", line 49, in <module>
users = session.query(User).all()
File "/home/pavel/.virtualenvs/common/local/lib/python2.7/site-packages/sqlalchemy/orm/query.py", line 2292, in all
return list(self)
File "/home/pavel/.virtualenvs/common/local/lib/python2.7/site-packages/sqlalchemy/orm/loading.py", line 65, in instances
fetch = cursor.fetchall()
File "/home/pavel/.virtualenvs/common/local/lib/python2.7/site-packages/sqlalchemy/engine/result.py", line 788, in fetchall
self.cursor, self.context)
File "/home/pavel/.virtualenvs/common/local/lib/python2.7/site-packages/sqlalchemy/engine/base.py", line 1111, in _handle_dbapi_exception
util.reraise(*exc_info)
File "/home/pavel/.virtualenvs/common/local/lib/python2.7/site-packages/sqlalchemy/engine/result.py", line 782, in fetchall
l = self.process_rows(self._fetchall_impl())
File "/home/pavel/.virtualenvs/common/local/lib/python2.7/site-packages/sqlalchemy/engine/result.py", line 749, in _fetchall_impl
return self.cursor.fetchall()
ValueError: year is out of range
我使用postgresql 9.1.11，sqlalchemy 0.9.4，python 2.7
我的问题是：有没有办法在postgresql中使用BC日期＆＃34; date＆＃34;通过sqlalchemy键入列（至少将其作为字符串检索，但透视将其映射到可以使用BC日期的某个模块，如FlexiDate或astropy.time）？
答案 0 :(得分：3)
我正在挖掘这个问题，我发现这个Value错误异常不是由sqlalchemy引发的，而是由数据库驱动程序（在本例中为psycopg2）引发的。 这里，如果使用从上面创建的数据库执行，此代码将引发相同的值错误异常。
import psycopg2
conn = psycopg2.connect("dbname=sqlalchemy_test user=database password=database host=localhost port=5432")
cursor = conn.cursor()
cursor.execute("SELECT * FROM users")
cursor.fetchall()
输出：
Traceback (most recent call last):
File "test_script.py", line 5, in <module>
values = cursor.fetchall()
ValueError: year is out of range
看起来它试图从数据库中检索的值创建datetime.date对象并失败。
我也试过pg8000方言，它没有引发Value错误，但它吃了BC部分，返回datetime.date对象的年，月和日（我的意思是从数据库行中获取值＆＃39; 1990-公元前11-25＆＃39;它返回datetime.date（1990,11,25），实际上是1990-11-25）。 这里的例子：
import pg8000
conn = pg8000.connect(user='postgres', host='localhost', port=5432,
database='sqlalchemy_test', password='postgres')
cursor = conn.cursor()
cursor.execute('SELECT * FROM users')
values = cursor.fetchall()
print values
print [str(val[2]) for val in values]
输出：
([1, u'ed', datetime.date(2014, 6, 19)], [2, u'al', datetime.date(1950, 12, 16)], [3, u'Lu', datetime.date(90, 4, 25)])
['2014-06-19', '1950-12-16', '0090-04-25']
我也想尝试py-postgresql方言，但它仅适用于python 3+，而不是一个选项（生产使用python 2.7.6进行此项目）。
所以，回答我的问题是：＆＃34;不，在sqlalchemy中是不可能的。＆＃34;，因为数据库文字和python对象之间的类型转换发生在数据库驱动程序中，而不是在sqlalchemy部分。
更新：
虽然在sqlalchemy中确实无法做到这一点，但可以在psycopg2中定义自己的类型转换（并覆盖默认行为），并从数据库返回watever中创建日期类型。这是一个例子：
# Cast PostgreSQL date into string
# http://initd.org/psycopg/docs/advanced.html#type-casting-from-sql-to-python
import psycopg2
BcDate = None
# This function dafines types cast, and as returned database literal is already a string, no
# additional logic required.
def cast_bc_date(value, cursor):
return value
def register_bc_date(connection):
global BcDate
if not BcDate:
cursor = connection.cursor()
cursor.execute('SELECT NULL::date')
psql_date_oid = cursor.description[0][1]
BcDate = psycopg2.extensions.new_type((psql_date_oid,), 'DATE', cast_bc_date)
psycopg2.extensions.register_type(BcDate)
conn = psycopg2.connect(database='sqlalchemy_test', user='database', password='database', host='localhost', port=5432)
register_bc_date(conn)
这个问题的例子就像一个魅力。并在sqlalchemy中开辟了额外数据处理的方法。