On Tue, Jan 19, 2016 at 1:39 PM, Jim Nasby <jim.na...@bluetreble.com> wrote:
> Is this odd quoting behavior expected? (Note difference between "'a':1"
> and "'b':'a'")
>
> ~@decina.local/53896# \set df
>> pd.DataFrame.from_dict([{'a':1,'b':'a'},{'a':2,'b':'b'}])
>> ~@decina.local/53896# \echo :df
>> pd.DataFrame.from_dict([{a:1,b:'a'},{a:2,b:'b'}])
>> ~@decina.local/53896#
>
>
Yeah, something odds is going one here, testing on 9.3.10 Ubuntu 14.04
\set df ' " d ' " "
select :'df' => " d " "
Removing the last double-quote results in an "unterminated quoted string"
error
Apparently the quoting rules for \set need to first be discovered and then
added to the documentation.
\set df alpha'beta'delta => alphabetadelta
\set df "alpha'beta'delta" => "alpha'beta'delta"
So adding double-quotes causes the rest of the content to be considered
literally but unfortunately the double-quotes themselves become part of the
value. Single-quotes by themselves get paired up and removed.
[testing some more]
*Got It!*
Its the colon that is allowing the single-quotes around the "value" to
remain. PostgreSQL read " :'a' " and sees a variable that it might need to
resolve - and when it cannot it simply leaves the variable name reference
in place.
If not for the colon all of the single-quotes would have been removed as
delimiters.
I still think some improvement and or outright fixes could be made here but
am going to leave it here for now.
David J.
