BUG: handle no rows returns from a sql query

[jreback]

related to #3523

[jreback]

@jorisvandenbossche @hayd new sql list?

[jorisvandenbossche]

This seems to be solved in master in the meantime (works for both sqlalchemy and sqlite3 fallback):

In [1]: import pandas as pd

In [2]: import sqlite3

In [3]: import sqlalchemy

In [4]: pd.__version__ 
Out[4]: '0.14.1-474-g744daa7'

In [5]: con = sqlite3.connect(':memory:')

In [6]: engine = sqlalchemy.create_engine('sqlite:///:memory:')

In [7]: df = pd.DataFrame({'a':[1,2,3], 'b':[2,3,4]})

In [8]: df.to_sql('test', engine, index=False, if_exists='replace')

In [9]: df.to_sql('test', con, index=False, if_exists='replace')

In [10]: pd.read_sql_query("SELECT * FROM test WHERE a < 0", engine)
Out[10]: 
Empty DataFrame
Columns: [a, b]
Index: []

In [11]: pd.read_sql_query("SELECT * FROM test WHERE a < 0", con)
Out[11]: 
Empty DataFrame
Columns: [a, b]
Index: []