Use generator to unpack list comprehension #444
Conversation
| if sys.version_info < (3, 8): # pragma: no cover (py38+) | ||
| start = i - 1 | ||
| while not (tokens[start].name == 'OP' and tokens[start].src == '['): | ||
| start -= 1 | ||
| else: # pragma: no cover (<py38) | ||
| start = i |
There was a problem hiding this comment.
I think there might be a helper for this somewhere in pyupgrade already? -- maybe try the set / dict comp rewrites?
There was a problem hiding this comment.
I've had a look at _arg_token_index and victims from pyupgrade._token_helpers, but it doesn't seem so easy to adapt them here.
They cover the case when the node yielded contains a function (such as set) which contains a comprehension. Here, on the other hand, the yielded node is the comprehension itself, and in <py38 I need to go backwards to find the starting bracket (rather than forward to find the first element).
I'll take a closer look / have a harder think anyway
There was a problem hiding this comment.
Having looked at this further, I can't figure out how to use victims here...any hints? Else I'll get back to it later in the week, bit stuck on this
In all the other examples of using victims I could find in pyupgrade, when it's called, you have that e.g. tokens[i] is set or dict, so then tokens[i+1] is ( and you can call victims(tokens, i+1, ...) and find the opening and closing brackets.
Here, however, I have that in <py38, tokens[i] is the element immediately after the opening bracket, while in py38+ tokens[i] is the opening bracket itself.
MarcoGorelli
force-pushed
the
unpack-list-comp
branch
from
612e0f2 to
7025f1d
Compare
Addresses part of #257