Enhance bytearray_repr with bytes_repr's logic

[zhangyangyu]

BPO	27492
Nosy	@serhiy-storchaka, @zhangyangyu
Files	bytearray_repr.patch _Py_bytes_repr.patch bytearray_repr_v2.patch

^{Note: these values reflect the state of the issue at the time it was migrated and might not reflect the current state.}

Show more details

GitHub fields:

assignee = None
closed_at = None
created_at = <Date 2016-07-12.07:09:41.672>
labels = ['interpreter-core', 'type-feature']
title = "Enhance bytearray_repr with bytes_repr's logic"
updated_at = <Date 2016-08-15.16:28:03.851>
user = 'https://proxy-ga.blitzz.co/proxy/123456/github.com/zhangyangyu'

bugs.python.org fields:

activity = <Date 2016-08-15.16:28:03.851>
actor = 'xiang.zhang'
assignee = 'none'
closed = False
closed_date = None
closer = None
components = ['Interpreter Core']
creation = <Date 2016-07-12.07:09:41.672>
creator = 'xiang.zhang'
dependencies = []
files = ['43697', '43702', '44119']
hgrepos = []
issue_num = 27492
keywords = ['patch']
message_count = 6.0
messages = ['270231', '270241', '270249', '270298', '272747', '272778']
nosy_count = 2.0
nosy_names = ['serhiy.storchaka', 'xiang.zhang']
pr_nums = []
priority = 'normal'
resolution = None
stage = None
status = 'open'
superseder = None
type = 'enhancement'
url = 'https://proxy-ga.blitzz.co/proxy/123456/bugs.python.org/issue27492'
versions = ['Python 3.6']

Linked PRs

• gh-71679: Enhance bytearray_repr with bytes_repr's logic #104582

[zhangyangyu]

Right now bytearray_repr's implementation mimics old string_repr, but it has a drawback: It arbitrarily checks overflow using four times the bytearray object's length, but the representation length of the value can range from 1 to 4. I think we can use bytes_repr's logic which calculates the actual output length.

[serhiy-storchaka]

Bytes and bytearray share a much of code. I think it is possible to share the implementation of reprs. Just add two functions in bytes_methods.c: one counts the size of the repr and determines the quote, and other writes the content of the repr in preallocated buffer.

[zhangyangyu]

I considered that too. But I was not sure what code could go into bytes_methods.c then, Python level methods, all common parts or only tp->methods? I'll try to factor the common part out later.

[zhangyangyu]

Hi, Serhiy. I've tried one version factoring the common parts out. I make them one function but it seems not very elegant. I also think about passing the object and use it to get the type and then determine how to get the string, what the pre/postfix are, and the exception message, but that seems not elegant either. And I don't figure out what the two function's advantage. Let me know your considerations.

BTW, I am still not sure repr could be put into bytes_methods.c. The functions in it are all tp->methods.

[serhiy-storchaka]

The two function's advantage is in the ability to reuse this code for other purposes. For example in _codecs.escape_encode(). But since this is the only place where the same algorithm is used and this functions is not documented and I presume it is not much used, this advantage is pretty small.

The simplest implementation of bytearray.__repr__ is

    def __repr__(self):
        return 'bytearray(%r)' % bytes(self)

It is less efficient than the current implementation or proposed patch, but is much simpler. This approach is used in reprs of set, frozenset, deque, array, BaseException, itemgetter, attrgetter, etc.

It can be more efficient is make bytes.__repr__ accepting not only bytes, but objects supporting the buffer protocol.

[zhangyangyu]

It can be more efficient is make bytes.__repr__ accepting not only bytes, but objects supporting the buffer protocol.

I like this idea. v2 implements this.