Question on Expression object

[c-jamie]

Hi Toby,

Thanks for your work on this. It’s an awesome achievement for the analytics community!

I had a question on the expression system logic.

What is the logic when deciding to make a child expression associated with the this key in args or the expression / expressions key?

As a concrete example.

In [3]: sel = parse_one("select a,b,c from d where p='q'")
In [4]: whe= sel.args.get('where')

I would now expect the EQ expression child to be in whe.expression, but it’s not, it’s accessible via whe.this

In [5]: whe.expression 
Out[5]:
In [6]: whe.this
Out[6]:
 (EQ this: 
  (COLUMN this: 
    (IDENTIFIER this: p, quoted: False)), expression: 
  (LITERAL this: q, is_string: True))

However for the children of the EQ expression, the left child is available via this and the right child is available via .expression

In [14]: whe.this.this
Out[14]: 
(COLUMN this: 
  (IDENTIFIER this: p, quoted: False))

In [15]: whe.this.expression
Out[15]: (LITERAL this: q, is_string: True)

Relatedly:

1. What’s the meaning of the this keyword for an Expression object?
2. What is the logic which governs when the expression \ expressions keyword is populated for an Expression object?

Thanks!

[tobymao]

this is the default argument of expression, expression is usually used when there's another argument

so where.this is the inner. for eq, this is left and expression is right. you can also call left and right on any binary operation.

checkout out sqlglot/expressions.py for more details