Block macro outputting a single-item body

[Technologicat]

It would be convenient if a block macro could return, when appropriate for some particular use case, just a single AST node as the transformed body. Currently the return type from a block macro must be list.

A real use case for this is the autocurry macro in unpythonic. (Usage, implementation.)

Minimal (non-)working example:

main.py:

from mwe import macros, mwe

with mwe:
    print("hi")
    print("ho")

mwe.py:

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

from macropy.core.macros import Macros

macros = Macros()

@macros.block
def mwe(tree, **kw):
    return tree[0]  # return just one AST node --> crash

[edit] update link.
[edit] update doc link.

[catb0t]

but then any caller has to have an if statement to ask whether the returned thing was iterable or one node. why not keep consistent return types and always return a List<node>?

[Technologicat]

Hmm, a very pythonic viewpoint.

The reason I opened this ticket is that this also came up over email, and @azazel75 said the behavior I described (a block macro requires returning a list also for a single item) sounded like a bug.

I'd be happy if this was just mentioned in the documentation (somewhere here?), or if the assert failure message explictly said that a list was expected. Currently, it says what it got, but not what it expected. The information is there, but you have to look at the stack trace.

[azazel75]

After reviewing and testing the code, I can say that the block macros type expects the macro to return a sequence of trees, but from my test it doesn't "crash", the return value hits an assertion, as of now..

  File "run.py", line 2, in <module>
    import main
  File "<frozen importlib._bootstrap>", line 971, in _find_and_load
  File "<frozen importlib._bootstrap>", line 951, in _find_and_load_unlocked
  File "<frozen importlib._bootstrap>", line 894, in _find_spec
  File "/home/azazel/wip/macropy/macropy/macropy/core/import_hooks.py", line 147, in find_spec
    code, tree = self.expand_macros(source, origin, spec)
  File "/home/azazel/wip/macropy/macropy/macropy/core/import_hooks.py", line 111, in expand_macros
    tree, source_code, modules).expand_macros()
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 592, in expand_macros
    tree = super().expand_macros(tree)
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 399, in expand_macros
    return self.walk_tree(tree)
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 548, in walk_tree
    self.walk_children(tree)
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 513, in walk_children
    new_value = self.walk_tree(old_value)
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 548, in walk_tree
    self.walk_children(tree)
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 518, in walk_children
    new_t = self.walk_tree(t)
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 539, in walk_tree
    new_tree = self.walk_tree(expand_it.send(new_tree))
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 421, in macro_expand
    mdata = type_it.send(new_tree)
  File "/home/azazel/wip/macropy/macropy/macropy/core/macros.py", line 191, in detect_macro
    assert isinstance(new_tree, list), type(new_tree)
AssertionError: <class '_ast.Expr'>

I may not be the best option to deal with this issue, but it's somewhat clear why the assert raises an error. I will think how to make it more clear

[Technologicat]

Yes, the assert isinstance(new_tree, list) is the very last item in the stack trace, but Python has spoiled me to expect the error message to already contain all necessary information. :)

Maybe something like this?

 assert isinstance(new_tree, list), "expected list, got {}".format(type(new_tree))

[Technologicat]

Update: during my investigation of #21, I noticed that in macropy.core.macros.Block.detect_macro, there is already a special case to automatically wrap a single expression returned by a block macro into an ast.Expr (as required for an expression in a statement position) and then pack that into a list.

However, if the block macro returns a single statement, there's no special case to catch that, and it just errors out.

IMHO, this asymmetry feels unintentional, and likely an oversight from the original authors.

I think this would be more logical if there either were no special cases (Zen of Python 8, not special enough to break the rules), or if both a single expression and a single statement were special-cased (Zen of Python 9, practicality beats purity).

[azazel75]

For me yours is a special case, it takes as list of statements as input it's natural to expect that it emit a list of statement. The ast.Expr trick is there for backward compatibility but no more exceptions will be added, by me.

[Technologicat]

I agree it's fine to expect a list of statements. Knowing that, there's no problem.

I didn't know the ast.Expr trick is for backward compatibility only, as the code itself didn't suggest it is deprecated. Granted, it's not mentioned in the docs.

[Technologicat]

Ping? Should this be mentioned in the docs? I could write a PR with a suggestion?

If you think there's no need to document this, in that case I think it's fine to close this issue.