DB access without object-relation mapping?

[kj]

Python noob here.

I want to write a script that creates and populates a simple Postgres
database.

The word on the street is to use something like SQLAlchemy for
database access in Python, but my experience in the past with
packages that perform automated SQL generation has been awful, so
I always return to lighter-weight solutions that allow me to write
my own SQL. (E.g. when coding in Perl I've used Perl's DBI package
and drivers, rather than the likes of Class::DBI.) So what's the
standard Python way to send SQL directly to a Postgres database
and get back results?

TIA!

kynn

--
NOTE: In my address everything before the first period is backwards;
and the last period, and everything after it, should be discarded.

[Tim Henderson]

On Jul 29, 11:20*am, kj <[e-mail]:

Python noob here.

I want to write a script that creates and populates a simple Postgres
database.

The word on the street is to use something like SQLAlchemy for
database access in Python, but my experience in the past with
packages that perform automated SQL generation has been awful, so
I always return to lighter-weight solutions that allow me to write
my own SQL. *(E.g. when coding in Perl I've used Perl's DBI package
and drivers, rather than the likes of Class::DBI.) *So what's the
standard Python way to send SQL directly to a Postgres database
and get back results?

TIA!

kynn

--
NOTE: In my address everything before the first period is backwards;
and the last period, and everything after it, should be discarded.

Hi,

I believe there are a couple of options but pyscopg, and PyGreSQL seem
to be popular.

Tim

[Paul Boddie]

On 29 Jul, 17:20, kj <[e-mail]:

>
*So what's the
standard Python way to send SQL directly to a Postgres database
and get back results?

Take a look at this page:

http://wiki.python.org/moin/DatabaseInterfaces

I've used psycopg2 and pyPgSQL successfully, although pg_proboscis
looks very interesting, too.

Paul

[kj]

In <cb**********************************@56g2000hsm.g ooglegroups.comTim Henderson <ti******@gmail.comwrites:

>I believe there are a couple of options but pyscopg, and PyGreSQL seem
to be popular.

Great. Thanks!

kynn
--
NOTE: In my address everything before the first period is backwards;
and the last period, and everything after it, should be discarded.

[Ben Finney]

kj <so***@987jk.com.invalidwrites:

The word on the street is to use something like SQLAlchemy for
database access in Python, but my experience in the past with
packages that perform automated SQL generation has been awful, so I
always return to lighter-weight solutions that allow me to write my
own SQL.

Using SQLAlchemy, you *can* write your own SQL (with its 'text' query
type to feed explicit SQL text on the connection); you just don't
*need* to for the majority of simple and not-so-simple cases.

I've been using SQLAlchemy for some time and have found its
combination of "make the simple cases simple" and "allow full control
when needed" to be close to ideal. I can use exactly the same
interface to leverage the SQL generation for mundane tasks and to
craft custom weird SQL that I need.

Combine that with active, level-headed maintenance, and comprehensive
documentation, and I don't even need to consider using the ORM layer
:-)

--
\ “Know what I hate most? Rhetorical questions.� —Henry N. Camp |
`\ |
_o__) |
Ben Finney