remove regex dependency

CHANGES.md

@@ -8,7 +8,7 @@
   cell magics were tokenized, leading to possible indentation errors e.g. with
   `%%writefile`. (#2630)
 - Fix Python 3.10 support on platforms without ProcessPoolExecutor (#2631)
-- Reduce usage of the `regex` dependency (#2644)
+- Remove dependency on `regex` (#2644) (#2663)
 - Fix `match` statements with open sequence subjects, like `match a, b:` or
   `match a, *b:` (#2639) (#2659)
 - Fix `match`/`case` statements that contain `match`/`case` soft keywords multiple

Pipfile

@@ -42,7 +42,6 @@ platformdirs= ">=2"
 click = ">=8.0.0"
 mypy_extensions = ">=0.4.3"
 pathspec = ">=0.8.1"
-regex = ">=2021.4.4"
 tomli = ">=0.2.6, <2.0.0"
 typed-ast = "==1.4.3"
 typing_extensions = {markers = "python_version < '3.10'", version = ">=3.10.0.0"}

docs/integrations/editors.md

@@ -204,30 +204,28 @@ Traceback (most recent call last):
 ImportError: /home/gui/.vim/black/lib/python3.7/site-packages/typed_ast/_ast3.cpython-37m-x86_64-linux-gnu.so: undefined symbool: PyExc_KeyboardInterrupt
 ```
 
-Then you need to install `typed_ast` and `regex` directly from the source code. The
-error happens because `pip` will download [Python wheels](https://pythonwheels.com/) if
-they are available. Python wheels are a new standard of distributing Python packages and
-packages that have Cython and extensions written in C are already compiled, so the
-installation is much more faster. The problem here is that somehow the Python
-environment inside Vim does not match with those already compiled C extensions and these
-kind of errors are the result. Luckily there is an easy fix: installing the packages
-from the source code.
-
-The two packages that cause the problem are:
-
-- [regex](https://pypi.org/project/regex/)
+Then you need to install `typed_ast` directly from the source code. The error happens
+because `pip` will download [Python wheels](https://pythonwheels.com/) if they are
+available. Python wheels are a new standard of distributing Python packages and packages
+that have Cython and extensions written in C are already compiled, so the installation
+is much more faster. The problem here is that somehow the Python environment inside Vim
+does not match with those already compiled C extensions and these kind of errors are the
+result. Luckily there is an easy fix: installing the packages from the source code.
+
+The package that causes problems is:
+
 - [typed-ast](https://pypi.org/project/typed-ast/)
 
 Now remove those two packages:
 
 ```console
-$ pip uninstall regex typed-ast -y
+$ pip uninstall typed-ast -y
 ```
 
 And now you can install them with:
 
 ```console
-$ pip install --no-binary :all: regex typed-ast
+$ pip install --no-binary :all: typed-ast
 ```
 
 The C extensions will be compiled and now Vim's Python environment will match. Note that
@@ -237,7 +235,7 @@ Ubuntu/Debian do `sudo apt-get install build-essential python3-dev`).
 If you later want to update _Black_, you should do it like this:
 
 ```console
-$ pip install -U black --no-binary regex,typed-ast
+$ pip install -U black --no-binary typed-ast
 ```
 
 ### With ALE

setup.py

@@ -101,7 +101,6 @@ def find_python_files(base: Path) -> List[Path]:
         "platformdirs>=2",
         "tomli>=0.2.6,<2.0.0",
         "typed-ast>=1.4.2; python_version < '3.8' and implementation_name == 'cpython'",
-        "regex>=2021.4.4",
         "pathspec>=0.9.0, <1",
         "dataclasses>=0.6; python_version < '3.7'",
         "typing_extensions>=3.10.0.0",

src/black/trans.py

@@ -4,7 +4,7 @@
 from abc import ABC, abstractmethod
 from collections import defaultdict
 from dataclasses import dataclass
-import regex as re  # We need recursive patterns here (?R)
+import re
 from typing import (
     Any,
     Callable,
@@ -453,7 +453,7 @@ def make_naked(string: str, string_prefix: str) -> str:
             # with 'f'...
             if "f" in prefix and "f" not in next_prefix:
                 # Then we must escape any braces contained in this substring.
-                SS = re.subf(r"(\{|\})", "{1}{1}", SS)
+                SS = re.sub(r"(\{|\})", r"\1\1", SS)
 
             NSS = make_naked(SS, next_prefix)