Apply pyupgrade to project

[jdufresne]

Update patterns to modern Python syntax and features. Drops unnecessary
legacy syntax that was required for Python 2 (removed in
9c3db1a).

Details on pyupgrade can be found at:
https://github.com/asottile/pyupgrade

[ikamensh]

This change kinda makes it less readable - have to look into the expression to see if it's a dict or set.

[jdufresne]

I suppose that is a matter of familiarity. Lots of Python projects I've worked on use both set and dict comprehension. Once the syntax is understood and familiar it is readable.

If the project has a policy against this syntax, let me know and I can revert those lines.

[DimitriPapadopoulos]

The list()/set() part is a duplicate of #2085 and the # -*- coding: utf-8 -*- one is a duplicate of #2100.

Perhaps the list()/set() part is the most debatable. Links to relevant DeepSource.io pages:

• 7. Not using literal syntax to initialize empty list/dict/tuple

  It is relatively slower to initialize an empty dictionary by calling dict() than using the empty literal, because the name dict must be looked up in the global scope in case it has been rebound. Same goes for the other two types — list() and tuple().

• Consider using literal syntax to create the data structure — PTC-W0019

  DESCRIPTION

  Using the literal syntax can give minor performance bumps compared to using function calls to create dict, list and tuple.
  [...]
  This is because here, the name dict must be looked up in the global scope in case it has been rebound. Same goes for the other two types list() and tuple().

[DimitriPapadopoulos]

I would revert these two lines in this merge request. Let's discuss them in #2085.
Thinking about it, list()/set() are easy to understand because they are English words, while the []/{} "syntax needs to be understood". While both are acceptable, []/{} reminds me of the arcane syntax of Perl. It reminds me that years ago, I would struggle to understand my own commented Perl scripts a few months after having written them, because of the syntax.

[DimitriPapadopoulos]

Not sure about this one either. What's wrong with the explicit 'r'? It echoes the 'w' below.

[jdufresne]

'r' is the default, so is unnecessary:

https://docs.python.org/3/library/functions.html#open

It is conventional within the larger Python community to omit it. The goal of this PR is to update the code to follow wider community conventions.

[DimitriPapadopoulos]

I would limit this merge request to:

• remove # -*- coding: utf-8 -*-
• *remove (object)
• change ..._regex_def = u"... to ..._regex_def = "...

[jdufresne]

I would limit this merge request to ...

I'm going to leave this as-is unless changes are requested by a maintainer. I think using a tool, such as pyupgrade to get all changes at once simplifies the PR and history. If any of the PRs you listed lands early, I'll rebase and re-run the tool.

[larsoner]

I'm going to leave this as-is unless changes are requested by a maintainer. I think using a tool, such as pyupgrade to get all changes at once simplifies the PR and history. If any of the PRs you listed lands early, I'll rebase and re-run the tool.

I lean toward this as well, and I think we can revert the list/set stuff elsewhere if we want. I don't have a strong preference for either syntax personally, so let's decide in #2085

[larsoner]

Thanks @jdufresne