copy.replace loses slot-backed CSV dialect options #160106

Bug report

copy.replace() loses slot-backed attributes of csv.Dialect subclasses. This can silently change how CSV data is parsed, even when the delimiter was not replaced.

Reproduced with CPython 3.16.0a1. The same implementation is present on main at 31a7d99.

import copy
import csv

class SemicolonDialect(csv.excel):
    __slots__ = ('delimiter',)

    def __init__(self):
        self.delimiter = ';'
        super().__init__()

dialect = SemicolonDialect()
replacement = copy.replace(dialect, quoting=csv.QUOTE_ALL)

print(list(csv.reader(['alpha;beta'], dialect)))
print(list(csv.reader(['alpha;beta'], replacement)))

Actual output:

[['alpha', 'beta']]
[['alpha;beta']]

Expected: both readers produce [['alpha', 'beta']], since only the quoting setting was replaced.

Dialect.__replace__() copies only __dict__, losing attributes stored in slots. The replacement therefore lacks its delimiter, and csv.reader falls back to the default comma delimiter.

I have prepared a local fix and regression tests.

CPython versions tested on:

3.16

Operating systems tested on:

No response

Linked PRs

• gh-160106: Fix csv.Dialect.__replace__() losing slot-backed attributes #160641