pandas.read_sql_table 生成表未找到错误

Question

我正在尝试通过 python 访问我的 Oracle 11g(r2) Express Edition 数据库。我特别想从其中一个表 HISTORY_FULLNESS 创建一个 pandas DataFrame。但是，当我尝试使用 pandas.read_sql_table() 函数时，它给了我一个 Table Not Found 错误。在下面的代码中，我复制了引擎正在执行的查询，结果中显然有 HISTORY_FULLNESS。任何线索我做错了什么？

我有以下代码（***添加到混淆 ip 和密码）：

### Imports ###
import pandas as pd
import numpy as np
import cx_Oracle
import pandas.io.sql as psql
import sqlalchemy

# Open connection to Oracle Database
OpsDBengine = sqlalchemy.create_engine('oracle://OPSDBADMIN:***@***.***.***.***:1521/', echo=True)

connection = OpsDBengine.connect()
result = connection.execute("SELECT table_name FROM all_tables WHERE nvl(tablespace_name, 'no tablespace') NOT IN ('SYSTEM', 'SYSAUX') AND OWNER = 'OPSDBADMIN' AND IOT_NAME IS NULL")
for row in result:
    print "tablename: ", row['table_name']


testDF = pd.read_sql_table(table_name = 'HISTORY_FULLNESS', con=OpsDBengine, schema='OPSDBADMIN', index_col = 'FULLNESSRECORDID')

输出如下：

2015-02-19 15:23:58,714 INFO sqlalchemy.engine.base.Engine SELECT USER FROM DUAL
INFO:sqlalchemy.engine.base.Engine:SELECT USER FROM DUAL
2015-02-19 15:23:58,716 INFO sqlalchemy.engine.base.Engine {}
INFO:sqlalchemy.engine.base.Engine:{}
2015-02-19 15:23:58,740 INFO sqlalchemy.engine.base.Engine SELECT CAST('test plain returns' AS VARCHAR(60 CHAR)) AS anon_1 FROM DUAL
INFO:sqlalchemy.engine.base.Engine:SELECT CAST('test plain returns' AS VARCHAR(60 CHAR)) AS anon_1 FROM DUAL
2015-02-19 15:23:58,740 INFO sqlalchemy.engine.base.Engine {}
INFO:sqlalchemy.engine.base.Engine:{}
2015-02-19 15:23:58,746 INFO sqlalchemy.engine.base.Engine SELECT CAST('test unicode returns' AS NVARCHAR2(60)) AS anon_1 FROM DUAL
INFO:sqlalchemy.engine.base.Engine:SELECT CAST('test unicode returns' AS NVARCHAR2(60)) AS anon_1 FROM DUAL
2015-02-19 15:23:58,749 INFO sqlalchemy.engine.base.Engine {}
INFO:sqlalchemy.engine.base.Engine:{}
2015-02-19 15:23:58,766 INFO sqlalchemy.engine.base.Engine SELECT table_name FROM all_tables WHERE nvl(tablespace_name, 'no tablespace') NOT IN ('SYSTEM', 'SYSAUX') AND OWNER = 'OPSDBADMIN' AND IOT_NAME IS NULL
INFO:sqlalchemy.engine.base.Engine:SELECT table_name FROM all_tables WHERE nvl(tablespace_name, 'no tablespace') NOT IN ('SYSTEM', 'SYSAUX') AND OWNER = 'OPSDBADMIN' AND IOT_NAME IS NULL
2015-02-19 15:23:58,766 INFO sqlalchemy.engine.base.Engine {}
INFO:sqlalchemy.engine.base.Engine:{}
tablename:  SAFETY_CASEFILE
tablename:  HTMLDB_PLAN_TABLE
tablename:  HISTORY_FULLNESS
tablename:  HR_EMPLOYEE_PHOTOS
tablename:  HR_MANAGER_DEPARTMENT_GROUPS
tablename:  DB_DEPARTMENT_GROUPS
tablename:  HR_MANAGER_DEPT_ASSIGNMENTS
tablename:  HR_MANAGER_SHIFT_ASSIGNMENTS
tablename:  HR_EMPLOYEE_ROLE
2015-02-19 15:23:58,825 INFO sqlalchemy.engine.base.Engine SELECT table_name FROM all_tables WHERE nvl(tablespace_name, 'no tablespace') NOT IN ('SYSTEM', 'SYSAUX') AND OWNER = :owner AND IOT_NAME IS NULL
INFO:sqlalchemy.engine.base.Engine:SELECT table_name FROM all_tables WHERE nvl(tablespace_name, 'no tablespace') NOT IN ('SYSTEM', 'SYSAUX') AND OWNER = :owner AND IOT_NAME IS NULL
2015-02-19 15:23:58,825 INFO sqlalchemy.engine.base.Engine {'owner': u'OPSDBADMIN'}
INFO:sqlalchemy.engine.base.Engine:{'owner': u'OPSDBADMIN'}
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-9-8f6c6d72ed22> in <module>()
     15 
     16 
---> 17 testDF = pd.read_sql_table(table_name = 'HISTORY_FULLNESS', con=OpsDBengine, schema='OPSDBADMIN', index_col = 'FULLNESSRECORDID')

C:\Anaconda\lib\site-packages\pandas\io\sql.pyc in read_sql_table(table_name, con, schema, index_col, coerce_float, parse_dates, columns, chunksize)
    334         meta.reflect(only=[table_name])
    335     except sqlalchemy.exc.InvalidRequestError:
--> 336         raise ValueError("Table %s not found" % table_name)
    337 
    338     pandas_sql = SQLDatabase(con, meta=meta)

ValueError: Table HISTORY_FULLNESS not found

Answer 1

有点老问题，但可以帮助其他人：

您可以使用两种方法来做到这一点：

1. testDF = pd.read_sql_query("SELECT * FROM HISTORY_FULLNESS", con=OpsDBengine)
2. output = con.execute("SELECT * FROM HISTORY_FULLNESS"); testDF = pd.DataFrame(outpt.fetchall());

Answer 2

我意识到我正在发布一个非常古老的问题，但我在 oracle 上遇到了同样的问题。您是否尝试过将表名设为小写？这对我有用。

https://github.com/pandas-dev/pandas/issues/11258

### Imports ###
import pandas as pd
import numpy as np
import cx_Oracle
import pandas.io.sql as psql
import sqlalchemy

# Open connection to Oracle Database
OpsDBengine = 
sqlalchemy.create_engine('oracle://OPSDBADMIN:***@***.***.***.***:1521/', echo=True)

connection = OpsDBengine.connect()
result = connection.execute("SELECT table_name FROM all_tables WHERE nvl(tablespace_name, 'no tablespace') NOT IN ('SYSTEM', 'SYSAUX') AND OWNER = 'OPSDBADMIN' AND IOT_NAME IS NULL")
for row in result:
print "tablename: ", row['table_name']


testDF = pd.read_sql_table(table_name = 'HISTORY_FULLNESS'.lower(), con=OpsDBengine, schema='OPSDBADMIN', index_col = 'FULLNESSRECORDID')