Set up PyDough AST module with basic abstract classes and simple type verifiers #13
Conversation
| assert pydough_type.json_string == type_string | ||
| assert ( | ||
| repr(pydough_type) == repr_string | ||
| ), "parsed type does not match expected repr() string" |
There was a problem hiding this comment.
For consistency.
tests/test_type_verifiers.py
Outdated
| @pytest.mark.parametrize( | ||
| "verifier, args", | ||
| [pytest.param(AllowAny(), [], id="allow_any-empty_args")], | ||
| ) | ||
| def test_verification(verifier: TypeVerifier, args: List[PyDoughAST]): | ||
| """ | ||
| Checks that verifiers accept certain arguments without raising an exception | ||
| """ | ||
| verifier.accepts(args) |
There was a problem hiding this comment.
A dumb test, but setting up the framework & verifying that the imports don't cause a build error.
| @@ -0,0 +1,38 @@ | |||
| """ | |||
There was a problem hiding this comment.
This file will eventually be filled by a lot of verifiers that borrow a lot of ideas from Calcite, such as:
- Signatures
- Compounds
- Allow a specific type
- Require membership to a family
- Require matches across arguments
- Variadics
- Literal matching (e.g.
DATE('now', 'start of month', '-3 months'))
However, these will usually be loose requirements that represent a minimal amount of universally accepted behavior (sound-but-not-complete), such as:
x + yrequires the arguments to be numericLOWER(...)requires the argument to be a string- Datetime extraction functions require a datetime argument.
- Usually, if the type is unknown, just allow it.
The main thing is making sure that things that should be expressions vs collections are actually expressions vs collections (e.g. no calling COALESCE on tpch.Parts). That kind of check will come into play later.
knassre-bodo
changed the base branch from
main
to
kian/update_metadata_imports
| from abc import ABC, abstractmethod | ||
|
|
||
|
|
||
| class PyDoughAST(ABC): |
There was a problem hiding this comment.
Once I start adding implementations, will create a builder class for ease of testing.
| @@ -0,0 +1,46 @@ | |||
| """ | |||
| TODO: add file-level docstring. | |||
| """ | |||
There was a problem hiding this comment.
Some simple tests for now. Will become far more elaborate once I have node builders.
| def __eq__(self, other): | ||
| return self.equals(other) | ||
|
|
||
| @abstractmethod |
There was a problem hiding this comment.
I assume this is just to make an abstract method since you can't make __eq__ abstract?
There was a problem hiding this comment.
More or less, yes
Defines the PyDough AST module with the following:
PyDoughASTfor all AST nodesPyDoughASTExceptionexception classPyDoughExpressionASTfor all AST expression nodes