Implement class syntax for TypedDict #2808
Conversation
mypy/semanal.py
Outdated
| if any(not isinstance(expr, RefExpr) or | ||
| expr.fullname != 'mypy_extensions.TypedDict' and | ||
| not self.is_typeddict(expr) for expr in defn.base_type_exprs): | ||
| self.fail("All bases of a new TypedDict must be TypedDict's", defn) |
There was a problem hiding this comment.
Grammar nit: Replace TypedDict's with TypedDict types or similar.
mypy/semanal.py
Outdated
| tpdict = None # type: OrderedDict[str, Type] | ||
| for base in typeddict_bases: | ||
| # mypy doesn't yet understand predicates like is_typeddict | ||
| tpdict = base.node.typeddict_type.items # type: ignore |
There was a problem hiding this comment.
I'd recommend using a cast instead of # type: ignore, since it's more specific about where mypy gets confused.
There was a problem hiding this comment.
It required me three casts to convince mypy that everything is OK and it became unreadable, I used asserts instead.
mypy/semanal.py
Outdated
| else: | ||
| name = stmt.lvalues[0].name | ||
| if name in (oldfields or []): | ||
| self.fail('Cannot overwrite TypedDict field {} while extending' |
There was a problem hiding this comment.
Use double quotes around field name (e.g., ... TypedDict field "x" ...).
| # flags: --python-version 3.6 | ||
| from mypy_extensions import TypedDict | ||
|
|
||
| class Point1(TypedDict): |
There was a problem hiding this comment.
Add test case for duplicate fields within a single definition. Example:
class Bad(TypedDict):
x: int
x: int
| fields.append(name) | ||
| types.append(AnyType() if stmt.type is None else self.anal_type(stmt.type)) | ||
| # ...despite possible minor failures that allow further analyzis. | ||
| if name.startswith('_'): |
There was a problem hiding this comment.
Why not allow fields starting with underscores? Existing JSON data may use underscore prefixes, I think.
There was a problem hiding this comment.
I agree this is probably an unnecessary limitation (I copied it from the functional form, and that one probably copied it from namedtuple). Maybe it is worth making a separate PR removing this limitation from both forms?
There was a problem hiding this comment.
The no-underscore-prefix limitation allows extending the syntax of the functional form with new keywords in the future if it becomes useful.
I believe the no-underscore-prefix limitation for namedtuples used a similar rationale.
|
@JukkaL Thank you for review! I implemented changes in a new commit. |
|
Thank you! |
Fixes #2740
This allows the following:
(Side note: I spent an hour figuring out what are anonymous and named TypedDict's, maybe better call them explicit/implicit, or explicit/ihferred, if I understood them correctly?)