pyodbc connect string

[Larry Martell]

On Tue, Apr 29, 2014 at 7:14 PM, Ben Finney <ben at benfinney.id.au> wrote:
> Larry Martell <larry.martell at gmail.com> writes:
>
>> I am having a problem building a connect string for pyodbc. It works
>> when everything is hard coded, but if I build the connect string it
>> fails.
>>
>> This works:
>>
>> pyodbc.connect('DRIVER=FreeTDS;' 'SERVER=xx.xx.xx.xx;' 'PORT=1433;'
>> 'DATABASE=blah;' 'UID=foo;' 'PWD=bar;')
>
> This calls the function with a single string,
> "DRIVER=FreeTDS;SERVER=xx.xx.xx.xx;PORT=1433;DATABASE=blah;UID=foo;PWD=bar;".
>
> Remember that consecutive, whitespace-separated, quote-delimited
> fragments specify the construction of a single string literal::
>
>     >>> 'foo'
>     'foo'
>     >>> 'foo' 'bar'
>     'foobar'
>     >>> 'foo' 'bar' 'baz'
>     'foobarbaz'
>
> See the reference for how this concatenation occurs
> <URL:https://docs.python.org/3/reference/lexical_analysis.html#string-literal-concatenation>.
>
>> But this does not:
>>
>> pyodbc.connect(conn_str)
>>
>> Where conn_str is:
>>
>> 'DRIVER=FreeTDS;' 'SERVER=xx.xx.xx.xx;' 'PORT=1433;' 'DATABASE=blah;'
>> 'UID=foo;' 'PWD=bar;'
>
> This string is different, because it contains a whole lot of quotation
> marks and whitespace not in the string you show in the first example.
>
>> conn_str is constructed with:
>>
>>  conn_str = "'DRIVER=%s;' 'SERVER=%s;' 'PORT=%s;' 'DATABASE=%s;'
>> 'UID=%s;' 'PWD=%s;'" \
>>                 % (RECIPE_DB['DRIVER'], RECIPE_DB['SERVER'],
>> RECIPE_DB['PORT'], RECIPE_DB['DATABASE'],
>>                    RECIPE_DB['USER'], RECIPE_DB['PASSWORD'])
>
> Remove the extraneous quotes and whitespace, which were not in the
> original string you showed above.
>
> On a separate point: Since you have string keys for the mapping, you can
> make the interpolation more readable::
>
>     conn_str = (
>             "DRIVER=%(DRIVER)s;SERVER=%(SERVER)s;PORT=%(PORT)s;"
>             "DATABASE=%(DATABASE)s;UID=%(USER)s;PWD=%(PASSWORD)s;"
>             ) % RECIPE_DB
>
> since the named placeholders will be looked up by key in the ‘RECIPE_DB’
> mapping.
>
> There are other improvements to suggest, but I don't want to distract
> from the main point of the post.

Thanks for the explanation.