Work-arround for automatically pass a port number of 10000

[alex-ber]

First, thank you very much for you effort. I've found it here https://github.com/dropbox/PyHive/issues/32

I'm proposing changing documentation for caveat, beside editing hive.py the user can write the following:

from sqlalchemy import create_engine
import puretransport

from pyhive.sqlalchemy_hive import HiveDialect as _HiveDialect
class _HiveFixedDialect(object):

    def create_connect_args(self, url):
        kwargs = {
            'host': url.host,
            'port': url.port,
            #'port': url.port or 10000,
            'username': url.username,
            'password': url.password,
            'database': url.database or 'default',
        }
        kwargs.update(url.query)
        return [], kwargs

_HiveDialect.create_connect_args = _HiveFixedDialect.create_connect_args


transport = puretransport.transport_factory(host='host',
                                            port=10000,
                                            username='username',
                                            password='secret')

# connection string pattern hive://username@/database_name
engine = create_engine('hive://username@/default',
                       connect_args={'thrift_transport': transport})
result = engine.execute("select * from table limit 1")

Note the line _HiveDialect.create_connect_args = _HiveFixedDialect.create_connect_args. This line effectively replaces create_connect_args() function that inserts default 10000 port to our version that didn't.

Note: you can use also 'hive://' as connection string, if you want to connect to default database.