py/objcomplex: Allow strings of the form -A+Bj. #7623
Conversation
This works by splitting the string at the non-leading sign.
The following strings will work.
print(complex("1+2j"))
print(complex("+1+2j"))
print(complex("-1+2j"))
Cases without a non-leading sign will be treated as they already were.
Checks are added to force A to be real and B to be imaginary. This
raises exceptions for the same cases as CPython, like Aj+B, even though
MicroPython would otherwise return the correct result.
Signed-off-by: Laurens Valk <laurens@pybricks.com>
64ab364 to
84bb633
Compare
Codecov Report
@@ Coverage Diff @@
## master #7623 +/- ##
==========================================
- Coverage 98.27% 98.27% -0.01%
==========================================
Files 154 154
Lines 19994 20001 +7
==========================================
+ Hits 19650 19656 +6
- Misses 344 345 +1
Continue to review full report at Codecov.
|
|
Thanks! This is +96 bytes on PYBV11. I notice that objcomplex.c is the only place that sets the Here's an attempt at doing this: ae8a38e Comes in at +64 bytes. This also deals with an extra case where there's leading spaces (i.e. the +/- for the real component may not be the first character, e.g. |
|
@jimmo 's version does look slightly more efficient, and it only creates one complex object whereas the version here creates 2 and then adds them (creating a third on the heap). What do you think @laurensvalk ? |
|
Thanks for your quick responses. I'm all for @jimmo's more efficient implementation, so feel free to close this one. For context, I came across this while documenting the MicroPython modules in more detail (still a work in progress) for use with Pybricks. Our plan is to provide typing and autocomplete in our online editor.
|
Correctly raise OS error in socketpool_socket_recv_into()
This reverts commit 7e6e824. Fixes micropython#7770 The change in micropython#7623 needs to be revered; the raise-site added in micropython#7632 is the correct one and the one in socketpool needs to be reverted. This is not affecting 8.0.x because micropython#7623 was not back-ported to there before we realized it was not a full fix. Both micropython#7770 and micropython#7606 should be re-tested. I didn't test.
Allows parsing of a complex string with a similar result as CPython.
Until now, only strings like
3jwere supported, with strings like1+2jgiving a syntax error.The following strings will now also work.
Cases with only a sign at [0] or no sign at all are treated as they already were.
Checks are added to force A to be real and B to be imaginary. This raises exceptions for the same cases as CPython, like Aj+B, even though MicroPython would otherwise return the correct result.
Fixes this MicroPython forum post.
Edit: Fix commit message < 75 characters per line.
Signed-off-by: Laurens Valk laurens@pybricks.com