David Stockwell wrote:
> FWIW this is what I'm going to do for enumerated types in python.
> Its not perfect but it will make it fairly easy to get at column names
> so I can build SQL statements on the fly and still have fairly easy to
> maintain code
>
> #setup stuff
> NAME_COL, ADDRESS_COL, CITY_COL, ZIPCODE_COL, \
> STATE_COL, COUNTRY_COL, SS_COL, CAT_COL, \
> DATE_COL, SALARY_COL = range(10)
>
> mycol = {
> NAME_COL: " NAME ", ADDRESS_COL: " ADDRESS ",
> CITY_COL: " CITY ", ZIPCODE_COL: " ZIPCODE ",
> STATE_COL:" STATE ", COUNTRY_COL: " COUNTRY ",
> SS_COL: " SS ", CAT_COL: " CAT ",
> DATE_COL: " DATE ", SALARY_COL: " SALARY " }
> # Use these for indexing by column name
>
> # demonstration on how to 'get'
>
> print mycol[CITY_COL]
I don't know much about SQL, so this may be a dumb question, but... Is it a
requirement that you have indices for the columns as well as names? IOW,
does it matter that NAME_COL is 0, ADDRESS_COL is 1, etc.? Or is the end
result you're after to have NAME_COL be a way to get to the string " NAME "?
If you don't need the numeric indices, there are a couple of other
approaches you could take, both inspired by Peter Hansen's nifty bit of
code:
First, since you're putting CITY_COL and so forth in the global namespace
anyway, you could simply define them as the strings:
def makeColumnNames( names ):
class container: pass
cols = container()
for c in names.split():
cols.__dict__[c] = ' %s ' % c
return cols
cols = makeColumnNames(
'NAME ADDRESS CITY ZIPCODE STATE COUNTRY SS CAT DATE SALARY' )
print "'%s'" % cols.CITY
Or, you could avoid cluttering up the global namespace with those names,
avoid the need for the _COL suffix, and use the cleaner cols.CITY notation
instead of cols[CITY_COL]:
def makeColumnNames( names ):
class container: pass
cols = container()
for c in names.split():
cols.__dict__[c] = ' %s ' % c
return cols
cols = makeColumnNames(
'NAME ADDRESS CITY ZIPCODE STATE COUNTRY SS CAT DATE SALARY' )
print "'%s'" % cols.CITY
If you do need the numeric indices as well, it would be easy to come up with
a variation of one of the above approaches that gives you both the indices
and the names in a convenient way.
-Mike