On Friday, September 9, 2016 at 5:58:16 PM UTC+5:30, Chris Angelico wrote:
> On Fri, Sep 9, 2016 at 10:21 PM, Frank Millman wrote:
> > I am building a series of JOINs for a SQL statement. Some of them can be
> > nested - table A is joined from table B is joined from table C. In other
> > parts of the same SQL statement, table A could be joined from table D which
> > is joined from table E.
> >
> > For each potential candidate, I need to check if I have already built the
> > JOIN. If so, just use it. If not, build it, and update a dictionary to
> > indicate that it has been built.
> >
> > The key to the dictionary is the path to the joined table, which could be
> > several levels deep. I build up a tuple to represent the path. Each table
> > needs two pieces of information to identify it, so I use a tuple for each
> > identifier.
> >
> > The value portion of the dictionary is simply the alias generated for the
> > joined table. I use the alias in the body of the SQL statement.
>
> Sounds fair enough. A tuple is a pretty reasonable way to do this.
>
> The other approach would be to craft the aliases directly, as strings.
> For instance, your first example could be JOIN_A_B_C, and your second
> JOIN_A_D_E. (Sort the table names alphabetically before "_".join()ing
> them if you want order to be insignificant.)
If the requirement is to have ('A', 'B') == ('B', 'A')
the correct data structure to use is set(['A','B'])
In this case frozenset for immutability or allowed-for-dict-key requirements
[Disclaimer: Ive never used it like this so dont know rough edges; just
talking ‘theoretically’!!]
--
https://mail.python.org/mailman/listinfo/python-list
