diff --git a/.flake8 b/.flake8
index 7bc346a09c1..85f51cf9f05 100644
--- a/.flake8
+++ b/.flake8
@@ -1,6 +1,6 @@
 [flake8]
 # B905 should be enabled when we drop support for 3.9
-ignore = E203, E266, E501, W503, B905, B907
+ignore = E203, E266, E501, E704, W503, B905, B907
 # line length is intentionally set to 80 here because black uses Bugbear
 # See https://black.readthedocs.io/en/stable/the_black_code_style/current_style.html#line-length for more details
 max-line-length = 80
diff --git a/.github/workflows/changelog.yml b/.github/workflows/changelog.yml
index b3e1f0b9024..a1804597d7d 100644
--- a/.github/workflows/changelog.yml
+++ b/.github/workflows/changelog.yml
@@ -14,7 +14,7 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Grep CHANGES.md for PR number
         if: contains(github.event.pull_request.labels.*.name, 'skip news') != true
diff --git a/.github/workflows/diff_shades.yml b/.github/workflows/diff_shades.yml
index d685ef9456d..97db907abc8 100644
--- a/.github/workflows/diff_shades.yml
+++ b/.github/workflows/diff_shades.yml
@@ -19,7 +19,7 @@ jobs:
       matrix: ${{ steps.set-config.outputs.matrix }}
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: actions/setup-python@v4
         with:
           python-version: "*"
@@ -44,7 +44,7 @@ jobs:
       HATCH_BUILD_HOOKS_ENABLE: "1"
       # Clang is less picky with the C code it's given than gcc (and may
       # generate faster binaries too).
-      CC: clang-12
+      CC: clang-14
     strategy:
       fail-fast: false
       matrix:
@@ -52,7 +52,7 @@ jobs:
 
     steps:
       - name: Checkout this repository (full clone)
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           # The baseline revision could be rather old so a full clone is ideal.
           fetch-depth: 0
diff --git a/.github/workflows/diff_shades_comment.yml b/.github/workflows/diff_shades_comment.yml
index 22c293f91d2..b86bd93410e 100644
--- a/.github/workflows/diff_shades_comment.yml
+++ b/.github/workflows/diff_shades_comment.yml
@@ -12,7 +12,7 @@ jobs:
   comment:
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - uses: actions/setup-python@v4
         with:
           python-version: "*"
diff --git a/.github/workflows/doc.yml b/.github/workflows/doc.yml
index fc94dea62d9..fa3d87c70f5 100644
--- a/.github/workflows/doc.yml
+++ b/.github/workflows/doc.yml
@@ -21,7 +21,7 @@ jobs:
 
     runs-on: ${{ matrix.os }}
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up latest Python
         uses: actions/setup-python@v4
diff --git a/.github/workflows/docker.yml b/.github/workflows/docker.yml
index 8baace940ba..ee858236fcf 100644
--- a/.github/workflows/docker.yml
+++ b/.github/workflows/docker.yml
@@ -16,16 +16,16 @@ jobs:
     runs-on: ubuntu-latest
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Set up QEMU
-        uses: docker/setup-qemu-action@v2
+        uses: docker/setup-qemu-action@v3
 
       - name: Set up Docker Buildx
-        uses: docker/setup-buildx-action@v2
+        uses: docker/setup-buildx-action@v3
 
       - name: Login to DockerHub
-        uses: docker/login-action@v2
+        uses: docker/login-action@v3
         with:
           username: ${{ secrets.DOCKERHUB_USERNAME }}
           password: ${{ secrets.DOCKERHUB_TOKEN }}
@@ -36,7 +36,7 @@ jobs:
           latest_non_release)" >> $GITHUB_ENV
 
       - name: Build and push
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5
         with:
           context: .
           platforms: linux/amd64,linux/arm64
@@ -47,7 +47,7 @@ jobs:
         if:
           ${{ github.event_name == 'release' && github.event.action == 'published' &&
           !github.event.release.prerelease }}
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5
         with:
           context: .
           platforms: linux/amd64,linux/arm64
@@ -58,7 +58,7 @@ jobs:
         if:
           ${{ github.event_name == 'release' && github.event.action == 'published' &&
           github.event.release.prerelease }}
-        uses: docker/build-push-action@v4
+        uses: docker/build-push-action@v5
         with:
           context: .
           platforms: linux/amd64,linux/arm64
diff --git a/.github/workflows/fuzz.yml b/.github/workflows/fuzz.yml
index 4439148a1c7..1b5a50c0e0b 100644
--- a/.github/workflows/fuzz.yml
+++ b/.github/workflows/fuzz.yml
@@ -25,7 +25,7 @@ jobs:
         python-version: ["3.8", "3.9", "3.10", "3.11"]
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up Python ${{ matrix.python-version }}
         uses: actions/setup-python@v4
diff --git a/.github/workflows/lint.yml b/.github/workflows/lint.yml
index 064d4745a53..3eaf5785f5a 100644
--- a/.github/workflows/lint.yml
+++ b/.github/workflows/lint.yml
@@ -14,7 +14,7 @@ jobs:
     runs-on: ubuntu-latest
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Assert PR target is main
         if: github.event_name == 'pull_request' && github.repository == 'psf/black'
diff --git a/.github/workflows/pypi_upload.yml b/.github/workflows/pypi_upload.yml
index ab2c6402c23..a57013d67c1 100644
--- a/.github/workflows/pypi_upload.yml
+++ b/.github/workflows/pypi_upload.yml
@@ -1,9 +1,12 @@
-name: Build wheels and publish to PyPI
+name: Build and publish
 
 on:
   release:
     types: [published]
   pull_request:
+  push:
+    branches:
+      - main
 
 permissions:
   contents: read
@@ -12,9 +15,10 @@ jobs:
   main:
     name: sdist + pure wheel
     runs-on: ubuntu-latest
+    if: github.event_name == 'release'
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up latest Python
         uses: actions/setup-python@v4
@@ -35,34 +39,58 @@ jobs:
           TWINE_PASSWORD: ${{ secrets.PYPI_TOKEN }}
         run: twine upload --verbose -u '__token__' dist/*
 
+  generate_wheels_matrix:
+    name: generate wheels matrix
+    runs-on: ubuntu-latest
+    outputs:
+      include: ${{ steps.set-matrix.outputs.include }}
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install cibuildwheel and pypyp
+        run: |
+          pipx install cibuildwheel==2.15.0
+          pipx install pypyp==1
+      - name: generate matrix
+        if: github.event_name != 'pull_request'
+        run: |
+          {
+            cibuildwheel --print-build-identifiers --platform linux \
+            | pyp 'json.dumps({"only": x, "os": "ubuntu-latest"})' \
+            && cibuildwheel --print-build-identifiers --platform macos \
+            | pyp 'json.dumps({"only": x, "os": "macos-latest"})' \
+            && cibuildwheel --print-build-identifiers --platform windows \
+            | pyp 'json.dumps({"only": x, "os": "windows-latest"})'
+          } | pyp 'json.dumps(list(map(json.loads, lines)))' > /tmp/matrix
+        env:
+          CIBW_ARCHS_LINUX: x86_64
+          CIBW_ARCHS_MACOS: x86_64 arm64
+          CIBW_ARCHS_WINDOWS: AMD64
+      - name: generate matrix (PR)
+        if: github.event_name == 'pull_request'
+        run: |
+          cibuildwheel --print-build-identifiers --platform linux \
+          | pyp 'json.dumps({"only": x, "os": "ubuntu-latest"})' \
+          | pyp 'json.dumps(list(map(json.loads, lines)))' > /tmp/matrix
+        env:
+          CIBW_BUILD: "cp38-* cp311-*"
+          CIBW_ARCHS_LINUX: x86_64
+      - id: set-matrix
+        run: echo "include=$(cat /tmp/matrix)" | tee -a $GITHUB_OUTPUT
+
   mypyc:
-    name: mypyc wheels (${{ matrix.name }})
+    name: mypyc wheels ${{ matrix.only }}
+    needs: generate_wheels_matrix
     runs-on: ${{ matrix.os }}
     strategy:
       fail-fast: false
       matrix:
-        include:
-          - os: ubuntu-latest
-            name: linux-x86_64
-          - os: windows-2019
-            name: windows-amd64
-          - os: macos-11
-            name: macos-x86_64
-            macos_arch: "x86_64"
-          - os: macos-11
-            name: macos-arm64
-            macos_arch: "arm64"
-          - os: macos-11
-            name: macos-universal2
-            macos_arch: "universal2"
+        include: ${{ fromJson(needs.generate_wheels_matrix.outputs.include) }}
 
     steps:
-      - uses: actions/checkout@v3
-
-      - name: Build wheels via cibuildwheel
-        uses: pypa/cibuildwheel@v2.15.0
-        env:
-          CIBW_ARCHS_MACOS: "${{ matrix.macos_arch }}"
+      - uses: actions/checkout@v4
+      - uses: pypa/cibuildwheel@v2.16.2
+        with:
+          only: ${{ matrix.only }}
 
       - name: Upload wheels as workflow artifacts
         uses: actions/upload-artifact@v3
@@ -80,12 +108,13 @@ jobs:
     name: Update stable branch
     needs: [main, mypyc]
     runs-on: ubuntu-latest
+    if: github.event_name == 'release'
     permissions:
       contents: write
 
     steps:
       - name: Checkout stable branch
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
         with:
           ref: stable
           fetch-depth: 0
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
index 7daa31ee903..1f33f2b814f 100644
--- a/.github/workflows/test.yml
+++ b/.github/workflows/test.yml
@@ -35,7 +35,7 @@ jobs:
         os: [ubuntu-latest, macOS-latest, windows-latest]
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up Python ${{ matrix.python-version }}
         uses: actions/setup-python@v4
@@ -75,7 +75,7 @@ jobs:
 
     runs-on: ubuntu-latest
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
       - name: Send finished signal to Coveralls
         uses: AndreMiras/coveralls-python-action@8799c9f4443ac4201d2e2f2c725d577174683b99
         with:
@@ -93,7 +93,7 @@ jobs:
         os: [ubuntu-latest, macOS-latest]
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up latest Python
         uses: actions/setup-python@v4
diff --git a/.github/workflows/upload_binary.yml b/.github/workflows/upload_binary.yml
index 22535a64c67..bb19d48158c 100644
--- a/.github/workflows/upload_binary.yml
+++ b/.github/workflows/upload_binary.yml
@@ -29,7 +29,7 @@ jobs:
             executable_mime: "application/x-mach-binary"
 
     steps:
-      - uses: actions/checkout@v3
+      - uses: actions/checkout@v4
 
       - name: Set up latest Python
         uses: actions/setup-python@v4
diff --git a/.gitignore b/.gitignore
index 249499b135e..4a4f1b738ad 100644
--- a/.gitignore
+++ b/.gitignore
@@ -4,6 +4,7 @@
 _build
 .DS_Store
 .vscode
+.python-version
 docs/_static/pypi.svg
 .tox
 __pycache__
diff --git a/.pre-commit-config.yaml b/.pre-commit-config.yaml
index 6301526a445..623e661ac07 100644
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@@ -39,10 +39,11 @@ repos:
         exclude: ^src/blib2to3/
 
   - repo: https://github.com/pre-commit/mirrors-mypy
-    rev: v1.5.0
+    rev: v1.5.1
     hooks:
       - id: mypy
         exclude: ^docs/conf.py
+        args: ["--config-file", "pyproject.toml"]
         additional_dependencies:
           - types-PyYAML
           - tomli >= 0.2.6, < 2.0.0
@@ -51,9 +52,13 @@ repos:
           - platformdirs >= 2.1.0
           - pytest
           - hypothesis
+          - aiohttp >= 3.7.4
+          - types-commonmark
+          - urllib3
+          - hypothesmith
 
   - repo: https://github.com/pre-commit/mirrors-prettier
-    rev: v3.0.1
+    rev: v3.0.3
     hooks:
       - id: prettier
         exclude: \.github/workflows/diff_shades\.yml
diff --git a/CHANGES.md b/CHANGES.md
index a68106ad23f..2a50e456550 100644
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -47,6 +47,43 @@
 <!-- Major changes to documentation and policies. Small docs changes
      don't need a changelog entry. -->
 
+## 23.10.0
+
+### Stable style
+
+- Fix comments getting removed from inside parenthesized strings (#3909)
+
+### Preview style
+
+- Fix long lines with power operators getting split before the line length (#3942)
+- Long type hints are now wrapped in parentheses and properly indented when split across
+  multiple lines (#3899)
+- Magic trailing commas are now respected in return types. (#3916)
+- Require one empty line after module-level docstrings. (#3932)
+- Treat raw triple-quoted strings as docstrings (#3947)
+
+### Configuration
+
+- Fix cache versioning logic when `BLACK_CACHE_DIR` is set (#3937)
+
+### Parser
+
+- Fix bug where attributes named `type` were not acccepted inside `match` statements
+  (#3950)
+- Add support for PEP 695 type aliases containing lambdas and other unusual expressions
+  (#3949)
+
+### Output
+
+- Black no longer attempts to provide special errors for attempting to format Python 2
+  code (#3933)
+- Black will more consistently print stacktraces on internal errors in verbose mode
+  (#3938)
+
+### Integrations
+
+- The action output displayed in the job summary is now wrapped in Markdown (#3914)
+
 ## 23.9.1
 
 Due to various issues, the previous release (23.9.0) did not include compiled mypyc
diff --git a/Dockerfile b/Dockerfile
index 4e8f12f9798..a9e0ea5081e 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -1,4 +1,4 @@
-FROM python:3-slim AS builder
+FROM python:3.11-slim AS builder
 
 RUN mkdir /src
 COPY . /src/
@@ -10,7 +10,7 @@ RUN . /opt/venv/bin/activate && pip install --no-cache-dir --upgrade pip setupto
     && cd /src \
     && pip install --no-cache-dir .[colorama,d]
 
-FROM python:3-slim
+FROM python:3.11-slim
 
 # copy only Python packages to limit the image size
 COPY --from=builder /opt/venv /opt/venv
diff --git a/README.md b/README.md
index b257c333f0d..cad8184f7bc 100644
--- a/README.md
+++ b/README.md
@@ -38,7 +38,7 @@ Try it out now using the [Black Playground](https://black.vercel.app). Watch the
 
 ### Installation
 
-_Black_ can be installed by running `pip install black`. It requires Python 3.7+ to run.
+_Black_ can be installed by running `pip install black`. It requires Python 3.8+ to run.
 If you want to format Jupyter Notebooks, install with `pip install "black[jupyter]"`.
 
 If you can't wait for the latest _hotness_ and want to install from GitHub, use:
diff --git a/action.yml b/action.yml
index 282fca43dea..8b698ae3c80 100644
--- a/action.yml
+++ b/action.yml
@@ -36,10 +36,24 @@ runs:
     - name: black
       run: |
         if [ "$RUNNER_OS" == "Windows" ]; then
-          python $GITHUB_ACTION_PATH/action/main.py | tee -a $GITHUB_STEP_SUMMARY
+          runner="python"
         else
-          python3 $GITHUB_ACTION_PATH/action/main.py | tee -a $GITHUB_STEP_SUMMARY
+          runner="python3"
         fi
+
+        out=$(${runner} $GITHUB_ACTION_PATH/action/main.py)
+        exit_code=$?
+
+        # Display the raw output in the step
+        echo "${out}"
+
+        # Display the Markdown output in the job summary
+        echo "\`\`\`python" >> $GITHUB_STEP_SUMMARY
+        echo "${out}" >> $GITHUB_STEP_SUMMARY
+        echo "\`\`\`" >> $GITHUB_STEP_SUMMARY
+
+        # Exit with the exit-code returned by Black
+        exit ${exit_code}
       env:
         # TODO: Remove once https://github.com/actions/runner/issues/665 is fixed.
         INPUT_OPTIONS: ${{ inputs.options }}
diff --git a/autoload/black.vim b/autoload/black.vim
index 4eb9b25db28..051fea05c3b 100644
--- a/autoload/black.vim
+++ b/autoload/black.vim
@@ -75,8 +75,8 @@ def _initialize_black_env(upgrade=False):
       return True
 
   pyver = sys.version_info[:3]
-  if pyver < (3, 7):
-    print("Sorry, Black requires Python 3.7+ to run.")
+  if pyver < (3, 8):
+    print("Sorry, Black requires Python 3.8+ to run.")
     return False
 
   from pathlib import Path
diff --git a/docs/conf.py b/docs/conf.py
index f7cf1b42842..6b645435325 100644
--- a/docs/conf.py
+++ b/docs/conf.py
@@ -210,6 +210,13 @@ def make_pypi_svg(version: str) -> None:
 
 autodoc_member_order = "bysource"
 
+#  -- sphinx-copybutton configuration ----------------------------------------
+copybutton_prompt_text = (
+    r">>> |\.\.\. |> |\$ |\# | In \[\d*\]: | {2,5}\.\.\.: | {5,8}: "
+)
+copybutton_prompt_is_regexp = True
+copybutton_remove_prompts = True
+
 # -- Options for intersphinx extension ---------------------------------------
 
 # Example configuration for intersphinx: refer to the Python standard library.
diff --git a/docs/contributing/the_basics.md b/docs/contributing/the_basics.md
index 40d233257e3..bc1680eecfd 100644
--- a/docs/contributing/the_basics.md
+++ b/docs/contributing/the_basics.md
@@ -37,6 +37,63 @@ the root of the black repo:
 (.venv)$ tox -e run_self
 ```
 
+### Development
+
+Further examples of invoking the tests
+
+```console
+# Run all of the above mentioned, in parallel
+(.venv)$ tox --parallel=auto
+
+# Run tests on a specific python version
+(.venv)$ tox -e py39
+
+# pass arguments to pytest
+(.venv)$ tox -e py -- --no-cov
+
+# print full tree diff, see documentation below
+(.venv)$ tox -e py -- --print-full-tree
+
+# disable diff printing, see documentation below
+(.venv)$ tox -e py -- --print-tree-diff=False
+```
+
+### Testing
+
+All aspects of the _Black_ style should be tested. Normally, tests should be created as
+files in the `tests/data/cases` directory. These files consist of up to three parts:
+
+- A line that starts with `# flags: ` followed by a set of command-line options. For
+  example, if the line is `# flags: --preview --skip-magic-trailing-comma`, the test
+  case will be run with preview mode on and the magic trailing comma off. The options
+  accepted are mostly a subset of those of _Black_ itself, except for the
+  `--minimum-version=` flag, which should be used when testing a grammar feature that
+  works only in newer versions of Python. This flag ensures that we don't try to
+  validate the AST on older versions and tests that we autodetect the Python version
+  correctly when the feature is used. For the exact flags accepted, see the function
+  `get_flags_parser` in `tests/util.py`. If this line is omitted, the default options
+  are used.
+- A block of Python code used as input for the formatter.
+- The line `# output`, followed by the output of _Black_ when run on the previous block.
+  If this is omitted, the test asserts that _Black_ will leave the input code unchanged.
+
+_Black_ has two pytest command-line options affecting test files in `tests/data/` that
+are split into an input part, and an output part, separated by a line with`# output`.
+These can be passed to `pytest` through `tox`, or directly into pytest if not using
+`tox`.
+
+#### `--print-full-tree`
+
+Upon a failing test, print the full concrete syntax tree (CST) as it is after processing
+the input ("actual"), and the tree that's yielded after parsing the output ("expected").
+Note that a test can fail with different output with the same CST. This used to be the
+default, but now defaults to `False`.
+
+#### `--print-tree-diff`
+
+Upon a failing test, print the diff of the trees as described above. This is the
+default. To turn it off pass `--print-tree-diff=False`.
+
 ### News / Changelog Requirement
 
 `Black` has CI that will check for an entry corresponding to your PR in `CHANGES.md`. If
diff --git a/docs/faq.md b/docs/faq.md
index 8941ca3fe4d..c62e1b504b5 100644
--- a/docs/faq.md
+++ b/docs/faq.md
@@ -86,7 +86,7 @@ disabled-by-default counterpart W504. E203 should be disabled while changes are
 
 ## Which Python versions does Black support?
 
-Currently the runtime requires Python 3.7-3.11. Formatting is supported for files
+Currently the runtime requires Python 3.8-3.11. Formatting is supported for files
 containing syntax from Python 3.3 to 3.11. We promise to support at least all Python
 versions that have not reached their end of life. This is the case for both running
 _Black_ and formatting code.
@@ -95,7 +95,7 @@ Support for formatting Python 2 code was removed in version 22.0. While we've ma
 plans to stop supporting older Python 3 minor versions immediately, their support might
 also be removed some time in the future without a deprecation period.
 
-Runtime support for 3.6 was removed in version 22.10.0.
+Runtime support for 3.7 was removed in version 23.7.0.
 
 ## Why does my linter or typechecker complain after I format my code?
 
@@ -107,7 +107,7 @@ codebase with _Black_.
 
 ## Can I run Black with PyPy?
 
-Yes, there is support for PyPy 3.7 and higher.
+Yes, there is support for PyPy 3.8 and higher.
 
 ## Why does Black not detect syntax errors in my code?
 
diff --git a/docs/getting_started.md b/docs/getting_started.md
index 33fb2f978bb..15b7646a509 100644
--- a/docs/getting_started.md
+++ b/docs/getting_started.md
@@ -16,7 +16,7 @@ Also, you can try out _Black_ online for minimal fuss on the
 
 ## Installation
 
-_Black_ can be installed by running `pip install black`. It requires Python 3.7+ to run.
+_Black_ can be installed by running `pip install black`. It requires Python 3.8+ to run.
 If you want to format Jupyter Notebooks, install with `pip install "black[jupyter]"`.
 
 If you can't wait for the latest _hotness_ and want to install from GitHub, use:
diff --git a/docs/guides/using_black_with_other_tools.md b/docs/guides/using_black_with_other_tools.md
index 6c6fbb88174..22c641a7420 100644
--- a/docs/guides/using_black_with_other_tools.md
+++ b/docs/guides/using_black_with_other_tools.md
@@ -173,7 +173,7 @@ limit of `88`, _Black_'s default. This explains `max-line-length = 88`.
 ```ini
 [flake8]
 max-line-length = 88
-extend-ignore = E203
+extend-ignore = E203, E704
 ```
 
 </details>
diff --git a/docs/integrations/editors.md b/docs/integrations/editors.md
index cebe2b0721e..7d056160fcb 100644
--- a/docs/integrations/editors.md
+++ b/docs/integrations/editors.md
@@ -236,7 +236,7 @@ Configuration:
 
 #### Installation
 
-This plugin **requires Vim 7.0+ built with Python 3.7+ support**. It needs Python 3.7 to
+This plugin **requires Vim 7.0+ built with Python 3.8+ support**. It needs Python 3.8 to
 be able to run _Black_ inside the Vim process which is much faster than calling an
 external command.
 
@@ -288,8 +288,8 @@ $ git checkout origin/stable -b stable
 ##### Arch Linux
 
 On Arch Linux, the plugin is shipped with the
-[`python-black`](https://archlinux.org/packages/community/any/python-black/) package, so
-you can start using it in Vim after install with no additional setup.
+[`python-black`](https://archlinux.org/packages/extra/any/python-black/) package, so you
+can start using it in Vim after install with no additional setup.
 
 ##### Vim 8 Native Plugin Management
 
@@ -391,7 +391,7 @@ close and reopen your File, _Black_ will be done with its job.
 
 - Use the
   [Python extension](https://marketplace.visualstudio.com/items?itemName=ms-python.python)
-  ([instructions](https://code.visualstudio.com/docs/python/editing#_formatting)).
+  ([instructions](https://code.visualstudio.com/docs/python/formatting)).
 
 - Alternatively the pre-release
   [Black Formatter](https://marketplace.visualstudio.com/items?itemName=ms-python.black-formatter)
@@ -399,9 +399,10 @@ close and reopen your File, _Black_ will be done with its job.
   server for Black. Formatting is much more responsive using this extension, **but the
   minimum supported version of Black is 22.3.0**.
 
-## SublimeText 3
+## SublimeText
 
-Use [sublack plugin](https://github.com/jgirardet/sublack).
+For SublimeText 3, use [sublack plugin](https://github.com/jgirardet/sublack). For
+higher versions, it is recommended to use [LSP](#python-lsp-server) as documented below.
 
 ## Python LSP Server
 
diff --git a/docs/integrations/source_version_control.md b/docs/integrations/source_version_control.md
index 2afcc02f3cd..16354f849ba 100644
--- a/docs/integrations/source_version_control.md
+++ b/docs/integrations/source_version_control.md
@@ -8,7 +8,7 @@ Use [pre-commit](https://pre-commit.com/). Once you
 repos:
   # Using this mirror lets us use mypyc-compiled black, which is about 2x faster
   - repo: https://github.com/psf/black-pre-commit-mirror
-    rev: 23.9.1
+    rev: 23.10.0
     hooks:
       - id: black
         # It is recommended to specify the latest version of Python
@@ -35,7 +35,7 @@ include Jupyter Notebooks. To use this hook, simply replace the hook's `id: blac
 repos:
   # Using this mirror lets us use mypyc-compiled black, which is about 2x faster
   - repo: https://github.com/psf/black-pre-commit-mirror
-    rev: 23.9.1
+    rev: 23.10.0
     hooks:
       - id: black-jupyter
         # It is recommended to specify the latest version of Python
diff --git a/docs/requirements.txt b/docs/requirements.txt
index b8bab4393ff..b5b9e22fc84 100644
--- a/docs/requirements.txt
+++ b/docs/requirements.txt
@@ -1,9 +1,9 @@
 # Used by ReadTheDocs; pinned requirements for stability.
 
 myst-parser==2.0.0
-Sphinx==7.2.3
+Sphinx==7.2.6
 # Older versions break Sphinx even though they're declared to be supported.
-docutils==0.19
+docutils==0.20.1
 sphinxcontrib-programoutput==0.17
 sphinx_copybutton==0.5.2
 furo==2023.9.10
diff --git a/docs/the_black_code_style/current_style.md b/docs/the_black_code_style/current_style.md
index e1a8078bf2c..ff757a8276b 100644
--- a/docs/the_black_code_style/current_style.md
+++ b/docs/the_black_code_style/current_style.md
@@ -173,7 +173,7 @@ If you use Flake8, you have a few options:
    max-line-length = 80
    ...
    select = C,E,F,W,B,B950
-   extend-ignore = E203, E501
+   extend-ignore = E203, E501, E704
    ```
 
    The rationale for E950 is explained in
@@ -184,7 +184,7 @@ If you use Flake8, you have a few options:
    ```ini
    [flake8]
    max-line-length = 88
-   extend-ignore = E203
+   extend-ignore = E203, E704
    ```
 
 An explanation of why E203 is disabled can be found in the [Slices section](#slices) of
diff --git a/docs/usage_and_configuration/the_basics.md b/docs/usage_and_configuration/the_basics.md
index 57fafd87654..5b132a95eae 100644
--- a/docs/usage_and_configuration/the_basics.md
+++ b/docs/usage_and_configuration/the_basics.md
@@ -52,18 +52,19 @@ See also [the style documentation](labels/line-length).
 
 #### `-t`, `--target-version`
 
-Python versions that should be supported by Black's output. You should include all
-versions that your code supports. If you support Python 3.7 through 3.10, you should
-write:
+Python versions that should be supported by Black's output. You can run `black --help`
+and look for the `--target-version` option to see the full list of supported versions.
+You should include all versions that your code supports. If you support Python 3.8
+through 3.11, you should write:
 
 ```console
-$ black -t py37 -t py38 -t py39 -t py310
+$ black -t py38 -t py39 -t py310 -t py311
 ```
 
 In a [configuration file](#configuration-via-a-file), you can write:
 
 ```toml
-target-version = ["py37", "py38", "py39", "py310"]
+target-version = ["py38", "py39", "py310", "py311"]
 ```
 
 _Black_ uses this option to decide what grammar to use to parse your code. In addition,
@@ -193,8 +194,8 @@ configuration file for consistent results across environments.
 
 ```console
 $ black --version
-black, 23.9.1 (compiled: yes)
-$ black --required-version 23.9.1 -c "format = 'this'"
+black, 23.10.0 (compiled: yes)
+$ black --required-version 23.10.0 -c "format = 'this'"
 format = "this"
 $ black --required-version 31.5b2 -c "still = 'beta?!'"
 Oh no! 💥 💔 💥 The required version does not match the running version!
@@ -285,7 +286,7 @@ You can check the version of _Black_ you have installed using the `--version` fl
 
 ```console
 $ black --version
-black, 23.9.1
+black, 23.10.0
 ```
 
 #### `--config`
diff --git a/mypy.ini b/mypy.ini
deleted file mode 100644
index 95ec22d65be..00000000000
--- a/mypy.ini
+++ /dev/null
@@ -1,41 +0,0 @@
-[mypy]
-# Specify the target platform details in config, so your developers are
-# free to run mypy on Windows, Linux, or macOS and get consistent
-# results.
-python_version=3.8
-
-mypy_path=src
-
-show_column_numbers=True
-show_error_codes=True
-
-# be strict
-strict=True
-
-# except for...
-no_implicit_reexport = False
-
-# Unreachable blocks have been an issue when compiling mypyc, let's try
-# to avoid 'em in the first place.
-warn_unreachable=True
-
-[mypy-blib2to3.driver.*]
-ignore_missing_imports = True
-
-[mypy-IPython.*]
-ignore_missing_imports = True
-
-[mypy-colorama.*]
-ignore_missing_imports = True
-
-[mypy-pathspec.*]
-ignore_missing_imports = True
-
-[mypy-tokenize_rt.*]
-ignore_missing_imports = True
-
-[mypy-uvloop.*]
-ignore_missing_imports = True
-
-[mypy-_black_version.*]
-ignore_missing_imports = True
diff --git a/pyproject.toml b/pyproject.toml
index 159907ac8ec..8c55076e4c9 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -137,6 +137,7 @@ exclude = [
   # Compiled modules can't be run directly and that's a problem here:
   "/src/black/__main__.py",
 ]
+mypy-args = ["--ignore-missing-imports"]
 options = { debug_level = "0" }
 
 [tool.cibuildwheel]
@@ -145,8 +146,8 @@ build-verbosity = 1
 # - Python: CPython 3.8+ only
 # - Architecture (64-bit only): amd64 / x86_64, universal2, and arm64
 # - OS: Linux (no musl), Windows, and macOS
-build = "cp3*-*"
-skip = ["*-manylinux_i686", "*-musllinux_*", "*-win32", "pp-*", "cp312-*"]
+build = "cp3*"
+skip = ["*-manylinux_i686", "*-musllinux_*", "*-win32", "pp*", "cp312-*"]
 # This is the bare minimum needed to run the test suite. Pulling in the full
 # test_requirements.txt would download a bunch of other packages not necessary
 # here and would slow down the testing step a fair bit.
@@ -223,3 +224,26 @@ omit = [
 ]
 [tool.coverage.run]
 relative_files = true
+
+[tool.mypy]
+# Specify the target platform details in config, so your developers are
+# free to run mypy on Windows, Linux, or macOS and get consistent
+# results.
+python_version = "3.8"
+mypy_path = "src"
+strict = true
+# Unreachable blocks have been an issue when compiling mypyc, let's try to avoid 'em in the first place.
+warn_unreachable = true
+implicit_reexport = true
+show_error_codes = true
+show_column_numbers = true
+
+[[tool.mypy.overrides]]
+module = ["pathspec.*", "IPython.*", "colorama.*", "tokenize_rt.*", "uvloop.*", "_black_version.*"]
+ignore_missing_imports = true
+
+# CI only checks src/, but in case users are running LSP or similar we explicitly ignore
+# errors in test data files.
+[[tool.mypy.overrides]]
+module = ["tests.data.*"]
+ignore_errors = true
diff --git a/scripts/check_pre_commit_rev_in_example.py b/scripts/check_pre_commit_rev_in_example.py
index 9560b3b8401..107c6444dca 100644
--- a/scripts/check_pre_commit_rev_in_example.py
+++ b/scripts/check_pre_commit_rev_in_example.py
@@ -14,7 +14,7 @@
 
 import commonmark
 import yaml
-from bs4 import BeautifulSoup
+from bs4 import BeautifulSoup  # type: ignore[import]
 
 
 def main(changes: str, source_version_control: str) -> None:
diff --git a/scripts/check_version_in_basics_example.py b/scripts/check_version_in_basics_example.py
index 7f559b3aee1..0f42bafe334 100644
--- a/scripts/check_version_in_basics_example.py
+++ b/scripts/check_version_in_basics_example.py
@@ -8,7 +8,7 @@
 import sys
 
 import commonmark
-from bs4 import BeautifulSoup
+from bs4 import BeautifulSoup  # type: ignore[import]
 
 
 def main(changes: str, the_basics: str) -> None:
diff --git a/scripts/diff_shades_gha_helper.py b/scripts/diff_shades_gha_helper.py
index 7a58fbe9b28..895516deb51 100644
--- a/scripts/diff_shades_gha_helper.py
+++ b/scripts/diff_shades_gha_helper.py
@@ -119,7 +119,7 @@ def main() -> None:
 @main.command("config", help="Acquire run configuration and metadata.")
 @click.argument("event", type=click.Choice(["push", "pull_request"]))
 def config(event: Literal["push", "pull_request"]) -> None:
-    import diff_shades
+    import diff_shades  # type: ignore[import]
 
     if event == "push":
         jobs = [{"mode": "preview-changes", "force-flag": "--force-preview-style"}]
diff --git a/scripts/fuzz.py b/scripts/fuzz.py
index 25362c927d4..929d3eac4f5 100644
--- a/scripts/fuzz.py
+++ b/scripts/fuzz.py
@@ -21,7 +21,7 @@
     max_examples=1000,  # roughly 1k tests/minute, or half that under coverage
     derandomize=True,  # deterministic mode to avoid CI flakiness
     deadline=None,  # ignore Hypothesis' health checks; we already know that
-    suppress_health_check=HealthCheck.all(),  # this is slow and filter-heavy.
+    suppress_health_check=list(HealthCheck),  # this is slow and filter-heavy.
 )
 @given(
     # Note that while Hypothesmith might generate code unlike that written by
@@ -80,7 +80,7 @@ def test_idempotent_any_syntatically_valid_python(
     try:
         import sys
 
-        import atheris
+        import atheris  # type: ignore[import]
     except ImportError:
         pass
     else:
diff --git a/scripts/make_width_table.py b/scripts/make_width_table.py
index 30fd32c34b0..3c7cae60f7f 100644
--- a/scripts/make_width_table.py
+++ b/scripts/make_width_table.py
@@ -15,11 +15,12 @@
     pip install -U wcwidth
 
 """
+
 import sys
 from os.path import basename, dirname, join
 from typing import Iterable, Tuple
 
-import wcwidth
+import wcwidth  # type: ignore[import]
 
 
 def make_width_table() -> Iterable[Tuple[int, int, int]]:
diff --git a/src/black/cache.py b/src/black/cache.py
index 77f66cc34a9..6baa096baca 100644
--- a/src/black/cache.py
+++ b/src/black/cache.py
@@ -1,4 +1,5 @@
 """Caching of formatted files with feature-based invalidation."""
+
 import hashlib
 import os
 import pickle
@@ -36,8 +37,9 @@ def get_cache_dir() -> Path:
     repeated calls.
     """
     # NOTE: Function mostly exists as a clean way to test getting the cache directory.
-    default_cache_dir = user_cache_dir("black", version=__version__)
+    default_cache_dir = user_cache_dir("black")
     cache_dir = Path(os.environ.get("BLACK_CACHE_DIR", default_cache_dir))
+    cache_dir = cache_dir / __version__
     return cache_dir
 
 
diff --git a/src/black/concurrency.py b/src/black/concurrency.py
index ce016578399..55c96b66c86 100644
--- a/src/black/concurrency.py
+++ b/src/black/concurrency.py
@@ -9,6 +9,7 @@
 import os
 import signal
 import sys
+import traceback
 from concurrent.futures import Executor, ProcessPoolExecutor, ThreadPoolExecutor
 from multiprocessing import Manager
 from pathlib import Path
@@ -170,8 +171,10 @@ async def schedule_formatting(
             src = tasks.pop(task)
             if task.cancelled():
                 cancelled.append(task)
-            elif task.exception():
-                report.failed(src, str(task.exception()))
+            elif exc := task.exception():
+                if report.verbose:
+                    traceback.print_exception(type(exc), exc, exc.__traceback__)
+                report.failed(src, str(exc))
             else:
                 changed = Changed.YES if task.result() else Changed.NO
                 # If the file was written back or was successfully checked as
diff --git a/src/black/debug.py b/src/black/debug.py
index 150b44842dd..cebc48765ba 100644
--- a/src/black/debug.py
+++ b/src/black/debug.py
@@ -1,5 +1,5 @@
-from dataclasses import dataclass
-from typing import Iterator, TypeVar, Union
+from dataclasses import dataclass, field
+from typing import Any, Iterator, List, TypeVar, Union
 
 from black.nodes import Visitor
 from black.output import out
@@ -14,26 +14,33 @@
 @dataclass
 class DebugVisitor(Visitor[T]):
     tree_depth: int = 0
+    list_output: List[str] = field(default_factory=list)
+    print_output: bool = True
+
+    def out(self, message: str, *args: Any, **kwargs: Any) -> None:
+        self.list_output.append(message)
+        if self.print_output:
+            out(message, *args, **kwargs)
 
     def visit_default(self, node: LN) -> Iterator[T]:
         indent = " " * (2 * self.tree_depth)
         if isinstance(node, Node):
             _type = type_repr(node.type)
-            out(f"{indent}{_type}", fg="yellow")
+            self.out(f"{indent}{_type}", fg="yellow")
             self.tree_depth += 1
             for child in node.children:
                 yield from self.visit(child)
 
             self.tree_depth -= 1
-            out(f"{indent}/{_type}", fg="yellow", bold=False)
+            self.out(f"{indent}/{_type}", fg="yellow", bold=False)
         else:
             _type = token.tok_name.get(node.type, str(node.type))
-            out(f"{indent}{_type}", fg="blue", nl=False)
+            self.out(f"{indent}{_type}", fg="blue", nl=False)
             if node.prefix:
                 # We don't have to handle prefixes for `Node` objects since
                 # that delegates to the first child anyway.
-                out(f" {node.prefix!r}", fg="green", bold=False, nl=False)
-            out(f" {node.value!r}", fg="blue", bold=False)
+                self.out(f" {node.prefix!r}", fg="green", bold=False, nl=False)
+            self.out(f" {node.value!r}", fg="blue", bold=False)
 
     @classmethod
     def show(cls, code: Union[str, Leaf, Node]) -> None:
diff --git a/src/black/linegen.py b/src/black/linegen.py
index 507e860190f..d12ca39d037 100644
--- a/src/black/linegen.py
+++ b/src/black/linegen.py
@@ -1,6 +1,7 @@
 """
 Generating lines of code.
 """
+
 import sys
 from dataclasses import replace
 from enum import Enum, auto
@@ -397,6 +398,24 @@ def visit_factor(self, node: Node) -> Iterator[Line]:
             node.insert_child(index, Node(syms.atom, [lpar, operand, rpar]))
         yield from self.visit_default(node)
 
+    def visit_tname(self, node: Node) -> Iterator[Line]:
+        """
+        Add potential parentheses around types in function parameter lists to be made
+        into real parentheses in case the type hint is too long to fit on a line
+        Examples:
+        def foo(a: int, b: float = 7): ...
+
+        ->
+
+        def foo(a: (int), b: (float) = 7): ...
+        """
+        if Preview.parenthesize_long_type_hints in self.mode:
+            assert len(node.children) == 3
+            if maybe_make_parens_invisible_in_atom(node.children[2], parent=node):
+                wrap_in_parentheses(node, node.children[2], visible=False)
+
+        yield from self.visit_default(node)
+
     def visit_STRING(self, leaf: Leaf) -> Iterator[Line]:
         if Preview.hex_codes_in_unicode_sequences in self.mode:
             normalize_unicode_escape_sequences(leaf)
@@ -498,7 +517,14 @@ def __post_init__(self) -> None:
         self.visit_except_clause = partial(v, keywords={"except"}, parens={"except"})
         self.visit_with_stmt = partial(v, keywords={"with"}, parens={"with"})
         self.visit_classdef = partial(v, keywords={"class"}, parens=Ø)
-        self.visit_expr_stmt = partial(v, keywords=Ø, parens=ASSIGNMENTS)
+
+        # When this is moved out of preview, add ":" directly to ASSIGNMENTS in nodes.py
+        if Preview.parenthesize_long_type_hints in self.mode:
+            assignments = ASSIGNMENTS | {":"}
+        else:
+            assignments = ASSIGNMENTS
+        self.visit_expr_stmt = partial(v, keywords=Ø, parens=assignments)
+
         self.visit_return_stmt = partial(v, keywords={"return"}, parens={"return"})
         self.visit_import_from = partial(v, keywords=Ø, parens={"import"})
         self.visit_del_stmt = partial(v, keywords=Ø, parens={"del"})
@@ -510,6 +536,17 @@ def __post_init__(self) -> None:
         self.visit_case_block = self.visit_match_case
 
 
+def _hugging_power_ops_line_to_string(
+    line: Line,
+    features: Collection[Feature],
+    mode: Mode,
+) -> Optional[str]:
+    try:
+        return line_to_string(next(hug_power_op(line, features, mode)))
+    except CannotTransform:
+        return None
+
+
 def transform_line(
     line: Line, mode: Mode, features: Collection[Feature] = ()
 ) -> Iterator[Line]:
@@ -525,6 +562,14 @@ def transform_line(
 
     line_str = line_to_string(line)
 
+    # We need the line string when power operators are hugging to determine if we should
+    # split the line. Default to line_str, if no power operator are present on the line.
+    line_str_hugging_power_ops = (
+        (_hugging_power_ops_line_to_string(line, features, mode) or line_str)
+        if Preview.fix_power_op_line_length in mode
+        else line_str
+    )
+
     ll = mode.line_length
     sn = mode.string_normalization
     string_merge = StringMerger(ll, sn)
@@ -538,7 +583,7 @@ def transform_line(
         and not line.should_split_rhs
         and not line.magic_trailing_comma
         and (
-            is_line_short_enough(line, mode=mode, line_str=line_str)
+            is_line_short_enough(line, mode=mode, line_str=line_str_hugging_power_ops)
             or line.contains_unsplittable_type_ignore()
         )
         and not (line.inside_brackets and line.contains_standalone_comments())
@@ -548,7 +593,7 @@ def transform_line(
             transformers = [string_merge, string_paren_strip]
         else:
             transformers = []
-    elif line.is_def:
+    elif line.is_def and not should_split_funcdef_with_rhs(line, mode):
         transformers = [left_hand_split]
     else:
 
@@ -627,6 +672,40 @@ def _rhs(
         yield line
 
 
+def should_split_funcdef_with_rhs(line: Line, mode: Mode) -> bool:
+    """If a funcdef has a magic trailing comma in the return type, then we should first
+    split the line with rhs to respect the comma.
+    """
+    if Preview.respect_magic_trailing_comma_in_return_type not in mode:
+        return False
+
+    return_type_leaves: List[Leaf] = []
+    in_return_type = False
+
+    for leaf in line.leaves:
+        if leaf.type == token.COLON:
+            in_return_type = False
+        if in_return_type:
+            return_type_leaves.append(leaf)
+        if leaf.type == token.RARROW:
+            in_return_type = True
+
+    # using `bracket_split_build_line` will mess with whitespace, so we duplicate a
+    # couple lines from it.
+    result = Line(mode=line.mode, depth=line.depth)
+    leaves_to_track = get_leaves_inside_matching_brackets(return_type_leaves)
+    for leaf in return_type_leaves:
+        result.append(
+            leaf,
+            preformatted=True,
+            track_bracket=id(leaf) in leaves_to_track,
+        )
+
+    # we could also return true if the line is too long, and the return type is longer
+    # than the param list. Or if `should_split_rhs` returns True.
+    return result.magic_trailing_comma is not None
+
+
 class _BracketSplitComponent(Enum):
     head = auto()
     body = auto()
@@ -1368,7 +1447,7 @@ def maybe_make_parens_invisible_in_atom(
     Returns whether the node should itself be wrapped in invisible parentheses.
     """
     if (
-        node.type != syms.atom
+        node.type not in (syms.atom, syms.expr)
         or is_empty_tuple(node)
         or is_one_tuple(node)
         or (is_yield(node) and parent.type != syms.expr_stmt)
@@ -1392,6 +1471,7 @@ def maybe_make_parens_invisible_in_atom(
             syms.except_clause,
             syms.funcdef,
             syms.with_stmt,
+            syms.tname,
             # these ones aren't useful to end users, but they do please fuzzers
             syms.for_stmt,
             syms.del_stmt,
diff --git a/src/black/lines.py b/src/black/lines.py
index 71b657a0654..48fde888208 100644
--- a/src/black/lines.py
+++ b/src/black/lines.py
@@ -193,11 +193,16 @@ def is_class_paren_empty(self) -> bool:
     @property
     def is_triple_quoted_string(self) -> bool:
         """Is the line a triple quoted string?"""
-        return (
-            bool(self)
-            and self.leaves[0].type == token.STRING
-            and self.leaves[0].value.startswith(('"""', "'''"))
-        )
+        if not self or self.leaves[0].type != token.STRING:
+            return False
+        value = self.leaves[0].value
+        if value.startswith(('"""', "'''")):
+            return True
+        if Preview.accept_raw_docstrings in self.mode and value.startswith(
+            ("r'''", 'r"""', "R'''", 'R"""')
+        ):
+            return True
+        return False
 
     @property
     def opens_block(self) -> bool:
@@ -550,6 +555,15 @@ def maybe_empty_lines(self, current_line: Line) -> LinesBlock:
             if self.previous_line is None
             else before - previous_after
         )
+        if (
+            Preview.module_docstring_newlines in current_line.mode
+            and self.previous_block
+            and self.previous_block.previous_block is None
+            and len(self.previous_block.original_line.leaves) == 1
+            and self.previous_block.original_line.is_triple_quoted_string
+        ):
+            before = 1
+
         block = LinesBlock(
             mode=self.mode,
             previous_block=self.previous_block,
diff --git a/src/black/mode.py b/src/black/mode.py
index 8a855ac495a..309f22dae94 100644
--- a/src/black/mode.py
+++ b/src/black/mode.py
@@ -180,11 +180,16 @@ class Preview(Enum):
     # for https://github.com/psf/black/issues/3117 to be fixed.
     string_processing = auto()
     parenthesize_conditional_expressions = auto()
+    parenthesize_long_type_hints = auto()
+    respect_magic_trailing_comma_in_return_type = auto()
     skip_magic_trailing_comma_in_subscript = auto()
     wrap_long_dict_values_in_parens = auto()
     wrap_multiple_context_managers_in_parens = auto()
     dummy_implementations = auto()
     walrus_subscript = auto()
+    module_docstring_newlines = auto()
+    accept_raw_docstrings = auto()
+    fix_power_op_line_length = auto()
 
 
 class Deprecated(UserWarning):
diff --git a/src/black/numerics.py b/src/black/numerics.py
index 879e5b2cf36..67ac8595fcc 100644
--- a/src/black/numerics.py
+++ b/src/black/numerics.py
@@ -1,6 +1,7 @@
 """
 Formatting numeric literals.
 """
+
 from blib2to3.pytree import Leaf
 
 
diff --git a/src/black/parsing.py b/src/black/parsing.py
index e98e019cac6..ea282d1805c 100644
--- a/src/black/parsing.py
+++ b/src/black/parsing.py
@@ -1,9 +1,10 @@
 """
 Parse Python code and perform AST validation.
 """
+
 import ast
 import sys
-from typing import Final, Iterable, Iterator, List, Set, Tuple
+from typing import Iterable, Iterator, List, Set, Tuple
 
 from black.mode import VERSION_TO_FEATURES, Feature, TargetVersion, supports_feature
 from black.nodes import syms
@@ -14,8 +15,6 @@
 from blib2to3.pgen2.tokenize import TokenError
 from blib2to3.pytree import Leaf, Node
 
-PY2_HINT: Final = "Python 2 support was removed in version 22.0."
-
 
 class InvalidInput(ValueError):
     """Raised when input source code fails all parse attempts."""
@@ -26,9 +25,9 @@ def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
         # No target_version specified, so try all grammars.
         return [
             # Python 3.7-3.9
-            pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords,
+            pygram.python_grammar_async_keywords,
             # Python 3.0-3.6
-            pygram.python_grammar_no_print_statement_no_exec_statement,
+            pygram.python_grammar,
             # Python 3.10+
             pygram.python_grammar_soft_keywords,
         ]
@@ -39,12 +38,10 @@ def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
         target_versions, Feature.ASYNC_IDENTIFIERS
     ) and not supports_feature(target_versions, Feature.PATTERN_MATCHING):
         # Python 3.7-3.9
-        grammars.append(
-            pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords
-        )
+        grammars.append(pygram.python_grammar_async_keywords)
     if not supports_feature(target_versions, Feature.ASYNC_KEYWORDS):
         # Python 3.0-3.6
-        grammars.append(pygram.python_grammar_no_print_statement_no_exec_statement)
+        grammars.append(pygram.python_grammar)
     if any(Feature.PATTERN_MATCHING in VERSION_TO_FEATURES[v] for v in target_versions):
         # Python 3.10+
         grammars.append(pygram.python_grammar_soft_keywords)
@@ -89,14 +86,6 @@ def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -
         # Choose the latest version when raising the actual parsing error.
         assert len(errors) >= 1
         exc = errors[max(errors)]
-
-        if matches_grammar(src_txt, pygram.python_grammar) or matches_grammar(
-            src_txt, pygram.python_grammar_no_print_statement
-        ):
-            original_msg = exc.args[0]
-            msg = f"{original_msg}\n{PY2_HINT}"
-            raise InvalidInput(msg) from None
-
         raise exc from None
 
     if isinstance(result, Leaf):
diff --git a/src/black/report.py b/src/black/report.py
index a507671e4c0..89899f2f389 100644
--- a/src/black/report.py
+++ b/src/black/report.py
@@ -1,6 +1,7 @@
 """
 Summarize Black runs to users.
 """
+
 from dataclasses import dataclass
 from enum import Enum
 from pathlib import Path
diff --git a/src/black/rusty.py b/src/black/rusty.py
index 84a80b5a2c2..ebd4c052d1f 100644
--- a/src/black/rusty.py
+++ b/src/black/rusty.py
@@ -2,6 +2,7 @@
 
 See https://doc.rust-lang.org/book/ch09-00-error-handling.html.
 """
+
 from typing import Generic, TypeVar, Union
 
 T = TypeVar("T")
diff --git a/src/black/trans.py b/src/black/trans.py
index daed26427d7..a2bff7f227a 100644
--- a/src/black/trans.py
+++ b/src/black/trans.py
@@ -1,6 +1,7 @@
 """
 String transformers that can split and merge strings.
 """
+
 import re
 from abc import ABC, abstractmethod
 from collections import defaultdict
@@ -942,6 +943,9 @@ def _transform_to_new_line(
                 LL[lpar_or_rpar_idx].remove()  # Remove lpar.
                 replace_child(LL[idx], string_leaf)
                 new_line.append(string_leaf)
+                # replace comments
+                old_comments = new_line.comments.pop(id(LL[idx]), [])
+                new_line.comments.setdefault(id(string_leaf), []).extend(old_comments)
             else:
                 LL[lpar_or_rpar_idx].remove()  # This is a rpar.
 
diff --git a/src/blackd/__init__.py b/src/blackd/__init__.py
index 6b0f3d33295..972f24181cb 100644
--- a/src/blackd/__init__.py
+++ b/src/blackd/__init__.py
@@ -74,7 +74,9 @@ def main(bind_host: str, bind_port: int) -> None:
     app = make_app()
     ver = black.__version__
     black.out(f"blackd version {ver} listening on {bind_host} port {bind_port}")
-    web.run_app(app, host=bind_host, port=bind_port, handle_signals=True, print=None)
+    # TODO: aiohttp had an incorrect annotation for `print` argument,
+    #  It'll be fixed once aiohttp releases that code
+    web.run_app(app, host=bind_host, port=bind_port, handle_signals=True, print=None)  # type: ignore[arg-type]
 
 
 def make_app() -> web.Application:
diff --git a/src/blib2to3/Grammar.txt b/src/blib2to3/Grammar.txt
index e48e66363fb..5db78723cec 100644
--- a/src/blib2to3/Grammar.txt
+++ b/src/blib2to3/Grammar.txt
@@ -80,8 +80,8 @@ vfplist: vfpdef (',' vfpdef)* [',']
 
 stmt: simple_stmt | compound_stmt
 simple_stmt: small_stmt (';' small_stmt)* [';'] NEWLINE
-small_stmt: (type_stmt | expr_stmt | print_stmt  | del_stmt | pass_stmt | flow_stmt |
-             import_stmt | global_stmt | exec_stmt | assert_stmt)
+small_stmt: (type_stmt | expr_stmt | del_stmt | pass_stmt | flow_stmt |
+             import_stmt | global_stmt | assert_stmt)
 expr_stmt: testlist_star_expr (annassign | augassign (yield_expr|testlist) |
                      ('=' (yield_expr|testlist_star_expr))*)
 annassign: ':' test ['=' (yield_expr|testlist_star_expr)]
@@ -89,8 +89,6 @@ testlist_star_expr: (test|star_expr) (',' (test|star_expr))* [',']
 augassign: ('+=' | '-=' | '*=' | '@=' | '/=' | '%=' | '&=' | '|=' | '^=' |
             '<<=' | '>>=' | '**=' | '//=')
 # For normal and annotated assignments, additional restrictions enforced by the interpreter
-print_stmt: 'print' ( [ test (',' test)* [','] ] |
-                      '>>' test [ (',' test)+ [','] ] )
 del_stmt: 'del' exprlist
 pass_stmt: 'pass'
 flow_stmt: break_stmt | continue_stmt | return_stmt | raise_stmt | yield_stmt
@@ -109,9 +107,8 @@ import_as_names: import_as_name (',' import_as_name)* [',']
 dotted_as_names: dotted_as_name (',' dotted_as_name)*
 dotted_name: NAME ('.' NAME)*
 global_stmt: ('global' | 'nonlocal') NAME (',' NAME)*
-exec_stmt: 'exec' expr ['in' test [',' test]]
 assert_stmt: 'assert' test [',' test]
-type_stmt: "type" NAME [typeparams] '=' expr
+type_stmt: "type" NAME [typeparams] '=' test
 
 compound_stmt: if_stmt | while_stmt | for_stmt | try_stmt | with_stmt | funcdef | classdef | decorated | async_stmt | match_stmt
 async_stmt: ASYNC (funcdef | with_stmt | for_stmt)
diff --git a/src/blib2to3/pgen2/parse.py b/src/blib2to3/pgen2/parse.py
index 299cc24a15f..ad51a3dad08 100644
--- a/src/blib2to3/pgen2/parse.py
+++ b/src/blib2to3/pgen2/parse.py
@@ -211,6 +211,7 @@ def __init__(self, grammar: Grammar, convert: Optional[Convert] = None) -> None:
         # See note in docstring above. TL;DR this is ignored.
         self.convert = convert or lam_sub
         self.is_backtracking = False
+        self.last_token: Optional[int] = None
 
     def setup(self, proxy: "TokenProxy", start: Optional[int] = None) -> None:
         """Prepare for parsing.
@@ -236,6 +237,7 @@ def setup(self, proxy: "TokenProxy", start: Optional[int] = None) -> None:
         self.rootnode: Optional[NL] = None
         self.used_names: Set[str] = set()
         self.proxy = proxy
+        self.last_token = None
 
     def addtoken(self, type: int, value: str, context: Context) -> bool:
         """Add a token; return True iff this is the end of the program."""
@@ -317,6 +319,7 @@ def _addtoken(self, ilabel: int, type: int, value: str, context: Context) -> boo
                         dfa, state, node = self.stack[-1]
                         states, first = dfa
                     # Done with this token
+                    self.last_token = type
                     return False
 
             else:
@@ -343,9 +346,23 @@ def classify(self, type: int, value: str, context: Context) -> List[int]:
                 return [self.grammar.keywords[value]]
             elif value in self.grammar.soft_keywords:
                 assert type in self.grammar.tokens
+                # Current soft keywords (match, case, type) can only appear at the
+                # beginning of a statement. So as a shortcut, don't try to treat them
+                # like keywords in any other context.
+                # ('_' is also a soft keyword in the real grammar, but for our grammar
+                # it's just an expression, so we don't need to treat it specially.)
+                if self.last_token not in (
+                    None,
+                    token.INDENT,
+                    token.DEDENT,
+                    token.NEWLINE,
+                    token.SEMI,
+                    token.COLON,
+                ):
+                    return [self.grammar.tokens[type]]
                 return [
-                    self.grammar.soft_keywords[value],
                     self.grammar.tokens[type],
+                    self.grammar.soft_keywords[value],
                 ]
 
         ilabel = self.grammar.tokens.get(type)
diff --git a/src/blib2to3/pygram.py b/src/blib2to3/pygram.py
index c30c630e816..2b43b4c112b 100644
--- a/src/blib2to3/pygram.py
+++ b/src/blib2to3/pygram.py
@@ -63,7 +63,6 @@ class _python_symbols(Symbols):
     encoding_decl: int
     eval_input: int
     except_clause: int
-    exec_stmt: int
     expr: int
     expr_stmt: int
     exprlist: int
@@ -97,7 +96,6 @@ class _python_symbols(Symbols):
     pattern: int
     patterns: int
     power: int
-    print_stmt: int
     raise_stmt: int
     return_stmt: int
     shift_expr: int
@@ -153,22 +151,16 @@ class _pattern_symbols(Symbols):
 
 
 python_grammar: Grammar
-python_grammar_no_print_statement: Grammar
-python_grammar_no_print_statement_no_exec_statement: Grammar
-python_grammar_no_print_statement_no_exec_statement_async_keywords: Grammar
-python_grammar_no_exec_statement: Grammar
-pattern_grammar: Grammar
+python_grammar_async_keywords: Grammar
 python_grammar_soft_keywords: Grammar
-
+pattern_grammar: Grammar
 python_symbols: _python_symbols
 pattern_symbols: _pattern_symbols
 
 
 def initialize(cache_dir: Union[str, "os.PathLike[str]", None] = None) -> None:
     global python_grammar
-    global python_grammar_no_print_statement
-    global python_grammar_no_print_statement_no_exec_statement
-    global python_grammar_no_print_statement_no_exec_statement_async_keywords
+    global python_grammar_async_keywords
     global python_grammar_soft_keywords
     global python_symbols
     global pattern_grammar
@@ -180,38 +172,25 @@ def initialize(cache_dir: Union[str, "os.PathLike[str]", None] = None) -> None:
         os.path.dirname(__file__), "PatternGrammar.txt"
     )
 
-    # Python 2
     python_grammar = driver.load_packaged_grammar("blib2to3", _GRAMMAR_FILE, cache_dir)
-    python_grammar.version = (2, 0)
+    assert "print" not in python_grammar.keywords
+    assert "exec" not in python_grammar.keywords
 
     soft_keywords = python_grammar.soft_keywords.copy()
     python_grammar.soft_keywords.clear()
 
     python_symbols = _python_symbols(python_grammar)
 
-    # Python 2 + from __future__ import print_function
-    python_grammar_no_print_statement = python_grammar.copy()
-    del python_grammar_no_print_statement.keywords["print"]
-
     # Python 3.0-3.6
-    python_grammar_no_print_statement_no_exec_statement = python_grammar.copy()
-    del python_grammar_no_print_statement_no_exec_statement.keywords["print"]
-    del python_grammar_no_print_statement_no_exec_statement.keywords["exec"]
-    python_grammar_no_print_statement_no_exec_statement.version = (3, 0)
+    python_grammar.version = (3, 0)
 
     # Python 3.7+
-    python_grammar_no_print_statement_no_exec_statement_async_keywords = (
-        python_grammar_no_print_statement_no_exec_statement.copy()
-    )
-    python_grammar_no_print_statement_no_exec_statement_async_keywords.async_keywords = (
-        True
-    )
-    python_grammar_no_print_statement_no_exec_statement_async_keywords.version = (3, 7)
+    python_grammar_async_keywords = python_grammar.copy()
+    python_grammar_async_keywords.async_keywords = True
+    python_grammar_async_keywords.version = (3, 7)
 
     # Python 3.10+
-    python_grammar_soft_keywords = (
-        python_grammar_no_print_statement_no_exec_statement_async_keywords.copy()
-    )
+    python_grammar_soft_keywords = python_grammar_async_keywords.copy()
     python_grammar_soft_keywords.soft_keywords = soft_keywords
     python_grammar_soft_keywords.version = (3, 10)
 
diff --git a/tests/conftest.py b/tests/conftest.py
index 67517268d1b..1a0dd747d8e 100644
--- a/tests/conftest.py
+++ b/tests/conftest.py
@@ -1 +1,28 @@
+import pytest
+
 pytest_plugins = ["tests.optional"]
+
+PRINT_FULL_TREE: bool = False
+PRINT_TREE_DIFF: bool = True
+
+
+def pytest_addoption(parser: pytest.Parser) -> None:
+    parser.addoption(
+        "--print-full-tree",
+        action="store_true",
+        default=False,
+        help="print full syntax trees on failed tests",
+    )
+    parser.addoption(
+        "--print-tree-diff",
+        action="store_true",
+        default=True,
+        help="print diff of syntax trees on failed tests",
+    )
+
+
+def pytest_configure(config: pytest.Config) -> None:
+    global PRINT_FULL_TREE
+    global PRINT_TREE_DIFF
+    PRINT_FULL_TREE = config.getoption("--print-full-tree")
+    PRINT_TREE_DIFF = config.getoption("--print-tree-diff")
diff --git a/tests/data/simple_cases/attribute_access_on_number_literals.py b/tests/data/cases/attribute_access_on_number_literals.py
similarity index 100%
rename from tests/data/simple_cases/attribute_access_on_number_literals.py
rename to tests/data/cases/attribute_access_on_number_literals.py
diff --git a/tests/data/simple_cases/beginning_backslash.py b/tests/data/cases/beginning_backslash.py
similarity index 100%
rename from tests/data/simple_cases/beginning_backslash.py
rename to tests/data/cases/beginning_backslash.py
diff --git a/tests/data/simple_cases/bracketmatch.py b/tests/data/cases/bracketmatch.py
similarity index 100%
rename from tests/data/simple_cases/bracketmatch.py
rename to tests/data/cases/bracketmatch.py
diff --git a/tests/data/simple_cases/class_blank_parentheses.py b/tests/data/cases/class_blank_parentheses.py
similarity index 100%
rename from tests/data/simple_cases/class_blank_parentheses.py
rename to tests/data/cases/class_blank_parentheses.py
diff --git a/tests/data/simple_cases/class_methods_new_line.py b/tests/data/cases/class_methods_new_line.py
similarity index 100%
rename from tests/data/simple_cases/class_methods_new_line.py
rename to tests/data/cases/class_methods_new_line.py
diff --git a/tests/data/simple_cases/collections.py b/tests/data/cases/collections.py
similarity index 100%
rename from tests/data/simple_cases/collections.py
rename to tests/data/cases/collections.py
diff --git a/tests/data/simple_cases/comment_after_escaped_newline.py b/tests/data/cases/comment_after_escaped_newline.py
similarity index 100%
rename from tests/data/simple_cases/comment_after_escaped_newline.py
rename to tests/data/cases/comment_after_escaped_newline.py
diff --git a/tests/data/simple_cases/comments.py b/tests/data/cases/comments.py
similarity index 100%
rename from tests/data/simple_cases/comments.py
rename to tests/data/cases/comments.py
diff --git a/tests/data/simple_cases/comments2.py b/tests/data/cases/comments2.py
similarity index 100%
rename from tests/data/simple_cases/comments2.py
rename to tests/data/cases/comments2.py
diff --git a/tests/data/simple_cases/comments3.py b/tests/data/cases/comments3.py
similarity index 100%
rename from tests/data/simple_cases/comments3.py
rename to tests/data/cases/comments3.py
diff --git a/tests/data/simple_cases/comments4.py b/tests/data/cases/comments4.py
similarity index 100%
rename from tests/data/simple_cases/comments4.py
rename to tests/data/cases/comments4.py
diff --git a/tests/data/simple_cases/comments5.py b/tests/data/cases/comments5.py
similarity index 100%
rename from tests/data/simple_cases/comments5.py
rename to tests/data/cases/comments5.py
diff --git a/tests/data/simple_cases/comments6.py b/tests/data/cases/comments6.py
similarity index 100%
rename from tests/data/simple_cases/comments6.py
rename to tests/data/cases/comments6.py
diff --git a/tests/data/simple_cases/comments8.py b/tests/data/cases/comments8.py
similarity index 100%
rename from tests/data/simple_cases/comments8.py
rename to tests/data/cases/comments8.py
diff --git a/tests/data/simple_cases/comments9.py b/tests/data/cases/comments9.py
similarity index 100%
rename from tests/data/simple_cases/comments9.py
rename to tests/data/cases/comments9.py
diff --git a/tests/data/simple_cases/comments_non_breaking_space.py b/tests/data/cases/comments_non_breaking_space.py
similarity index 100%
rename from tests/data/simple_cases/comments_non_breaking_space.py
rename to tests/data/cases/comments_non_breaking_space.py
diff --git a/tests/data/simple_cases/composition.py b/tests/data/cases/composition.py
similarity index 100%
rename from tests/data/simple_cases/composition.py
rename to tests/data/cases/composition.py
diff --git a/tests/data/simple_cases/composition_no_trailing_comma.py b/tests/data/cases/composition_no_trailing_comma.py
similarity index 100%
rename from tests/data/simple_cases/composition_no_trailing_comma.py
rename to tests/data/cases/composition_no_trailing_comma.py
diff --git a/tests/data/conditional_expression.py b/tests/data/cases/conditional_expression.py
similarity index 99%
rename from tests/data/conditional_expression.py
rename to tests/data/cases/conditional_expression.py
index 620a12dc986..c30cd76c791 100644
--- a/tests/data/conditional_expression.py
+++ b/tests/data/cases/conditional_expression.py
@@ -1,3 +1,4 @@
+# flags: --preview
 long_kwargs_single_line = my_function(
     foo="test, this is a sample value",
     bar=some_long_value_name_foo_bar_baz if some_boolean_variable else some_fallback_value_foo_bar_baz,
diff --git a/tests/data/simple_cases/docstring.py b/tests/data/cases/docstring.py
similarity index 100%
rename from tests/data/simple_cases/docstring.py
rename to tests/data/cases/docstring.py
diff --git a/tests/data/simple_cases/docstring_no_extra_empty_line_before_eof.py b/tests/data/cases/docstring_no_extra_empty_line_before_eof.py
similarity index 100%
rename from tests/data/simple_cases/docstring_no_extra_empty_line_before_eof.py
rename to tests/data/cases/docstring_no_extra_empty_line_before_eof.py
diff --git a/tests/data/miscellaneous/docstring_no_string_normalization.py b/tests/data/cases/docstring_no_string_normalization.py
similarity index 98%
rename from tests/data/miscellaneous/docstring_no_string_normalization.py
rename to tests/data/cases/docstring_no_string_normalization.py
index a90b578f09a..4ec6b8a0153 100644
--- a/tests/data/miscellaneous/docstring_no_string_normalization.py
+++ b/tests/data/cases/docstring_no_string_normalization.py
@@ -1,3 +1,4 @@
+# flags: --skip-string-normalization
 class ALonelyClass:
     '''
     A multiline class docstring.
diff --git a/tests/data/simple_cases/docstring_preview.py b/tests/data/cases/docstring_preview.py
similarity index 100%
rename from tests/data/simple_cases/docstring_preview.py
rename to tests/data/cases/docstring_preview.py
diff --git a/tests/data/miscellaneous/docstring_preview_no_string_normalization.py b/tests/data/cases/docstring_preview_no_string_normalization.py
similarity index 88%
rename from tests/data/miscellaneous/docstring_preview_no_string_normalization.py
rename to tests/data/cases/docstring_preview_no_string_normalization.py
index 338cc01d33e..712c7364f51 100644
--- a/tests/data/miscellaneous/docstring_preview_no_string_normalization.py
+++ b/tests/data/cases/docstring_preview_no_string_normalization.py
@@ -1,3 +1,4 @@
+# flags: --preview --skip-string-normalization
 def do_not_touch_this_prefix():
     R"""There was a bug where docstring prefixes would be normalized even with -S."""
 
diff --git a/tests/data/simple_cases/empty_lines.py b/tests/data/cases/empty_lines.py
similarity index 100%
rename from tests/data/simple_cases/empty_lines.py
rename to tests/data/cases/empty_lines.py
diff --git a/tests/data/simple_cases/expression.diff b/tests/data/cases/expression.diff
similarity index 100%
rename from tests/data/simple_cases/expression.diff
rename to tests/data/cases/expression.diff
diff --git a/tests/data/simple_cases/expression.py b/tests/data/cases/expression.py
similarity index 100%
rename from tests/data/simple_cases/expression.py
rename to tests/data/cases/expression.py
diff --git a/tests/data/simple_cases/fmtonoff.py b/tests/data/cases/fmtonoff.py
similarity index 100%
rename from tests/data/simple_cases/fmtonoff.py
rename to tests/data/cases/fmtonoff.py
diff --git a/tests/data/simple_cases/fmtonoff2.py b/tests/data/cases/fmtonoff2.py
similarity index 100%
rename from tests/data/simple_cases/fmtonoff2.py
rename to tests/data/cases/fmtonoff2.py
diff --git a/tests/data/simple_cases/fmtonoff3.py b/tests/data/cases/fmtonoff3.py
similarity index 100%
rename from tests/data/simple_cases/fmtonoff3.py
rename to tests/data/cases/fmtonoff3.py
diff --git a/tests/data/simple_cases/fmtonoff4.py b/tests/data/cases/fmtonoff4.py
similarity index 100%
rename from tests/data/simple_cases/fmtonoff4.py
rename to tests/data/cases/fmtonoff4.py
diff --git a/tests/data/simple_cases/fmtonoff5.py b/tests/data/cases/fmtonoff5.py
similarity index 100%
rename from tests/data/simple_cases/fmtonoff5.py
rename to tests/data/cases/fmtonoff5.py
diff --git a/tests/data/simple_cases/fmtpass_imports.py b/tests/data/cases/fmtpass_imports.py
similarity index 100%
rename from tests/data/simple_cases/fmtpass_imports.py
rename to tests/data/cases/fmtpass_imports.py
diff --git a/tests/data/simple_cases/fmtskip.py b/tests/data/cases/fmtskip.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip.py
rename to tests/data/cases/fmtskip.py
diff --git a/tests/data/simple_cases/fmtskip2.py b/tests/data/cases/fmtskip2.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip2.py
rename to tests/data/cases/fmtskip2.py
diff --git a/tests/data/simple_cases/fmtskip3.py b/tests/data/cases/fmtskip3.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip3.py
rename to tests/data/cases/fmtskip3.py
diff --git a/tests/data/simple_cases/fmtskip4.py b/tests/data/cases/fmtskip4.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip4.py
rename to tests/data/cases/fmtskip4.py
diff --git a/tests/data/simple_cases/fmtskip5.py b/tests/data/cases/fmtskip5.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip5.py
rename to tests/data/cases/fmtskip5.py
diff --git a/tests/data/simple_cases/fmtskip6.py b/tests/data/cases/fmtskip6.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip6.py
rename to tests/data/cases/fmtskip6.py
diff --git a/tests/data/simple_cases/fmtskip7.py b/tests/data/cases/fmtskip7.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip7.py
rename to tests/data/cases/fmtskip7.py
diff --git a/tests/data/simple_cases/fmtskip8.py b/tests/data/cases/fmtskip8.py
similarity index 100%
rename from tests/data/simple_cases/fmtskip8.py
rename to tests/data/cases/fmtskip8.py
diff --git a/tests/data/simple_cases/fstring.py b/tests/data/cases/fstring.py
similarity index 100%
rename from tests/data/simple_cases/fstring.py
rename to tests/data/cases/fstring.py
diff --git a/tests/data/cases/funcdef_return_type_trailing_comma.py b/tests/data/cases/funcdef_return_type_trailing_comma.py
new file mode 100644
index 00000000000..9b9b9c673de
--- /dev/null
+++ b/tests/data/cases/funcdef_return_type_trailing_comma.py
@@ -0,0 +1,301 @@
+# flags: --preview --minimum-version=3.10
+# normal, short, function definition
+def foo(a, b) -> tuple[int, float]: ...
+
+
+# normal, short, function definition w/o return type
+def foo(a, b): ...
+
+
+# no splitting
+def foo(a: A, b: B) -> list[p, q]:
+    pass
+
+
+# magic trailing comma in param list
+def foo(a, b,): ...
+
+
+# magic trailing comma in nested params in param list
+def foo(a, b: tuple[int, float,]): ...
+
+
+# magic trailing comma in return type, no params
+def a() -> tuple[
+    a,
+    b,
+]: ...
+
+
+# magic trailing comma in return type, params
+def foo(a: A, b: B) -> list[
+    p,
+    q,
+]:
+    pass
+
+
+# magic trailing comma in param list and in return type
+def foo(
+    a: a,
+    b: b,
+) -> list[
+    a,
+    a,
+]:
+    pass
+
+
+# long function definition, param list is longer
+def aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(
+    bbbbbbbbbbbbbbbbbb,
+) -> cccccccccccccccccccccccccccccc: ...
+
+
+# long function definition, return type is longer
+# this should maybe split on rhs?
+def aaaaaaaaaaaaaaaaa(bbbbbbbbbbbbbbbbbb) -> list[
+    Ccccccccccccccccccccccccccccccccccccccccccccccccccc, Dddddd
+]: ...
+
+
+# long return type, no param list
+def foo() -> list[
+    Loooooooooooooooooooooooooooooooooooong,
+    Loooooooooooooooooooong,
+    Looooooooooooong,
+]: ...
+
+
+# long function name, no param list, no return value
+def thiiiiiiiiiiiiiiiiiis_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiis_veeeeeeeeeeeeeeeeeeeeeeery_looooooong():
+    pass
+
+
+# long function name, no param list
+def thiiiiiiiiiiiiiiiiiis_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiis_veeeeeeeeeeeeeeeeeeeeeeery_looooooong() -> (
+    list[int, float]
+): ...
+
+
+# long function name, no return value
+def thiiiiiiiiiiiiiiiiiis_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiis_veeeeeeeeeeeeeeeeeeeeeeery_looooooong(
+    a, b
+): ...
+
+
+# unskippable type hint (??)
+def foo(a) -> list[aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa]:  # type: ignore
+    pass
+
+
+def foo(a) -> list[
+    aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+]:  # abpedeifnore
+    pass
+
+def foo(a, b: list[Bad],): ... # type: ignore
+
+# don't lose any comments (no magic)
+def foo( # 1
+    a, # 2
+    b) -> list[ # 3
+               a, # 4
+               b]: # 5
+        ... # 6
+
+
+# don't lose any comments (param list magic)
+def foo( # 1
+    a, # 2
+    b,) -> list[ # 3
+               a, # 4
+               b]: # 5
+        ... # 6
+
+
+# don't lose any comments (return type magic)
+def foo( # 1
+    a, # 2
+    b) -> list[ # 3
+               a, # 4
+               b,]: # 5
+        ... # 6
+
+
+# don't lose any comments (both magic)
+def foo( # 1
+    a, # 2
+    b,) -> list[ # 3
+               a, # 4
+               b,]: # 5
+        ... # 6
+
+# real life example
+def SimplePyFn(
+    context: hl.GeneratorContext,
+    buffer_input: Buffer[UInt8, 2],
+    func_input: Buffer[Int32, 2],
+    float_arg: Scalar[Float32],
+    offset: int = 0,
+) -> tuple[
+    Buffer[UInt8, 2],
+    Buffer[UInt8, 2],
+]: ...
+# output
+# normal, short, function definition
+def foo(a, b) -> tuple[int, float]: ...
+
+
+# normal, short, function definition w/o return type
+def foo(a, b): ...
+
+
+# no splitting
+def foo(a: A, b: B) -> list[p, q]:
+    pass
+
+
+# magic trailing comma in param list
+def foo(
+    a,
+    b,
+): ...
+
+
+# magic trailing comma in nested params in param list
+def foo(
+    a,
+    b: tuple[
+        int,
+        float,
+    ],
+): ...
+
+
+# magic trailing comma in return type, no params
+def a() -> tuple[
+    a,
+    b,
+]: ...
+
+
+# magic trailing comma in return type, params
+def foo(a: A, b: B) -> list[
+    p,
+    q,
+]:
+    pass
+
+
+# magic trailing comma in param list and in return type
+def foo(
+    a: a,
+    b: b,
+) -> list[
+    a,
+    a,
+]:
+    pass
+
+
+# long function definition, param list is longer
+def aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa(
+    bbbbbbbbbbbbbbbbbb,
+) -> cccccccccccccccccccccccccccccc: ...
+
+
+# long function definition, return type is longer
+# this should maybe split on rhs?
+def aaaaaaaaaaaaaaaaa(
+    bbbbbbbbbbbbbbbbbb,
+) -> list[Ccccccccccccccccccccccccccccccccccccccccccccccccccc, Dddddd]: ...
+
+
+# long return type, no param list
+def foo() -> list[
+    Loooooooooooooooooooooooooooooooooooong,
+    Loooooooooooooooooooong,
+    Looooooooooooong,
+]: ...
+
+
+# long function name, no param list, no return value
+def thiiiiiiiiiiiiiiiiiis_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiis_veeeeeeeeeeeeeeeeeeeeeeery_looooooong():
+    pass
+
+
+# long function name, no param list
+def thiiiiiiiiiiiiiiiiiis_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiis_veeeeeeeeeeeeeeeeeeeeeeery_looooooong() -> (
+    list[int, float]
+): ...
+
+
+# long function name, no return value
+def thiiiiiiiiiiiiiiiiiis_iiiiiiiiiiiiiiiiiiiiiiiiiiiiiis_veeeeeeeeeeeeeeeeeeeeeeery_looooooong(
+    a, b
+): ...
+
+
+# unskippable type hint (??)
+def foo(a) -> list[aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa]:  # type: ignore
+    pass
+
+
+def foo(
+    a,
+) -> list[
+    aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa, aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+]:  # abpedeifnore
+    pass
+
+
+def foo(
+    a,
+    b: list[Bad],
+): ...  # type: ignore
+
+
+# don't lose any comments (no magic)
+def foo(a, b) -> list[a, b]:  # 1  # 2  # 3  # 4  # 5
+    ...  # 6
+
+
+# don't lose any comments (param list magic)
+def foo(  # 1
+    a,  # 2
+    b,
+) -> list[a, b]:  # 3  # 4  # 5
+    ...  # 6
+
+
+# don't lose any comments (return type magic)
+def foo(a, b) -> list[  # 1  # 2  # 3
+    a,  # 4
+    b,
+]:  # 5
+    ...  # 6
+
+
+# don't lose any comments (both magic)
+def foo(  # 1
+    a,  # 2
+    b,
+) -> list[  # 3
+    a,  # 4
+    b,
+]:  # 5
+    ...  # 6
+
+
+# real life example
+def SimplePyFn(
+    context: hl.GeneratorContext,
+    buffer_input: Buffer[UInt8, 2],
+    func_input: Buffer[Int32, 2],
+    float_arg: Scalar[Float32],
+    offset: int = 0,
+) -> tuple[
+    Buffer[UInt8, 2],
+    Buffer[UInt8, 2],
+]: ...
diff --git a/tests/data/simple_cases/function.py b/tests/data/cases/function.py
similarity index 100%
rename from tests/data/simple_cases/function.py
rename to tests/data/cases/function.py
diff --git a/tests/data/simple_cases/function2.py b/tests/data/cases/function2.py
similarity index 100%
rename from tests/data/simple_cases/function2.py
rename to tests/data/cases/function2.py
diff --git a/tests/data/simple_cases/function_trailing_comma.py b/tests/data/cases/function_trailing_comma.py
similarity index 100%
rename from tests/data/simple_cases/function_trailing_comma.py
rename to tests/data/cases/function_trailing_comma.py
diff --git a/tests/data/simple_cases/ignore_pyi.py b/tests/data/cases/ignore_pyi.py
similarity index 97%
rename from tests/data/simple_cases/ignore_pyi.py
rename to tests/data/cases/ignore_pyi.py
index 3ef61079bfe..4fae7530eb9 100644
--- a/tests/data/simple_cases/ignore_pyi.py
+++ b/tests/data/cases/ignore_pyi.py
@@ -1,3 +1,4 @@
+# flags: --pyi
 def f():  # type: ignore
     ...
 
diff --git a/tests/data/simple_cases/import_spacing.py b/tests/data/cases/import_spacing.py
similarity index 100%
rename from tests/data/simple_cases/import_spacing.py
rename to tests/data/cases/import_spacing.py
diff --git a/tests/data/miscellaneous/linelength6.py b/tests/data/cases/linelength6.py
similarity index 80%
rename from tests/data/miscellaneous/linelength6.py
rename to tests/data/cases/linelength6.py
index 4fb342726f5..158038bf960 100644
--- a/tests/data/miscellaneous/linelength6.py
+++ b/tests/data/cases/linelength6.py
@@ -1,3 +1,4 @@
+# flags: --line-length=6
 # Regression test for #3427, which reproes only with line length <= 6
 def f():
     """
diff --git a/tests/data/miscellaneous/long_strings_flag_disabled.py b/tests/data/cases/long_strings_flag_disabled.py
similarity index 100%
rename from tests/data/miscellaneous/long_strings_flag_disabled.py
rename to tests/data/cases/long_strings_flag_disabled.py
diff --git a/tests/data/cases/module_docstring_1.py b/tests/data/cases/module_docstring_1.py
new file mode 100644
index 00000000000..d5897b4db60
--- /dev/null
+++ b/tests/data/cases/module_docstring_1.py
@@ -0,0 +1,26 @@
+# flags: --preview
+"""Single line module-level docstring should be followed by single newline."""
+
+
+
+
+a = 1
+
+
+"""I'm just a string so should be followed by 2 newlines."""
+
+
+
+
+b = 2
+
+# output
+"""Single line module-level docstring should be followed by single newline."""
+
+a = 1
+
+
+"""I'm just a string so should be followed by 2 newlines."""
+
+
+b = 2
diff --git a/tests/data/cases/module_docstring_2.py b/tests/data/cases/module_docstring_2.py
new file mode 100644
index 00000000000..e1f81b4d76b
--- /dev/null
+++ b/tests/data/cases/module_docstring_2.py
@@ -0,0 +1,68 @@
+# flags: --preview
+"""I am a very helpful module docstring.
+
+Lorem ipsum dolor sit amet, consectetur adipiscing elit,
+sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
+Ut enim ad minim veniam,
+quis nostrud exercitation ullamco laboris
+nisi ut aliquip ex ea commodo consequat.
+Duis aute irure dolor in reprehenderit in voluptate
+velit esse cillum dolore eu fugiat nulla pariatur.
+Excepteur sint occaecat cupidatat non proident,
+sunt in culpa qui officia deserunt mollit anim id est laborum.
+"""
+
+
+
+
+a = 1
+
+
+"""Look at me I'm a docstring...
+
+............................................................
+............................................................
+............................................................
+............................................................
+............................................................
+............................................................
+............................................................
+........................................................NOT!
+"""
+
+
+
+
+b = 2
+
+# output
+"""I am a very helpful module docstring.
+
+Lorem ipsum dolor sit amet, consectetur adipiscing elit,
+sed do eiusmod tempor incididunt ut labore et dolore magna aliqua.
+Ut enim ad minim veniam,
+quis nostrud exercitation ullamco laboris
+nisi ut aliquip ex ea commodo consequat.
+Duis aute irure dolor in reprehenderit in voluptate
+velit esse cillum dolore eu fugiat nulla pariatur.
+Excepteur sint occaecat cupidatat non proident,
+sunt in culpa qui officia deserunt mollit anim id est laborum.
+"""
+
+a = 1
+
+
+"""Look at me I'm a docstring...
+
+............................................................
+............................................................
+............................................................
+............................................................
+............................................................
+............................................................
+............................................................
+........................................................NOT!
+"""
+
+
+b = 2
diff --git a/tests/data/cases/module_docstring_3.py b/tests/data/cases/module_docstring_3.py
new file mode 100644
index 00000000000..0631e136a3d
--- /dev/null
+++ b/tests/data/cases/module_docstring_3.py
@@ -0,0 +1,8 @@
+# flags: --preview
+"""Single line module-level docstring should be followed by single newline."""
+a = 1
+
+# output
+"""Single line module-level docstring should be followed by single newline."""
+
+a = 1
diff --git a/tests/data/cases/module_docstring_4.py b/tests/data/cases/module_docstring_4.py
new file mode 100644
index 00000000000..515174dcc04
--- /dev/null
+++ b/tests/data/cases/module_docstring_4.py
@@ -0,0 +1,9 @@
+# flags: --preview
+"""Single line module-level docstring should be followed by single newline."""
+
+a = 1
+
+# output
+"""Single line module-level docstring should be followed by single newline."""
+
+a = 1
diff --git a/tests/data/simple_cases/multiline_consecutive_open_parentheses_ignore.py b/tests/data/cases/multiline_consecutive_open_parentheses_ignore.py
similarity index 100%
rename from tests/data/simple_cases/multiline_consecutive_open_parentheses_ignore.py
rename to tests/data/cases/multiline_consecutive_open_parentheses_ignore.py
diff --git a/tests/data/miscellaneous/nested_stub.pyi b/tests/data/cases/nested_stub.py
similarity index 97%
rename from tests/data/miscellaneous/nested_stub.pyi
rename to tests/data/cases/nested_stub.py
index 15e69d854db..b81549ec115 100644
--- a/tests/data/miscellaneous/nested_stub.pyi
+++ b/tests/data/cases/nested_stub.py
@@ -1,3 +1,4 @@
+# flags: --pyi --preview
 import sys
 
 class Outer:
diff --git a/tests/data/py_36/numeric_literals.py b/tests/data/cases/numeric_literals.py
similarity index 91%
rename from tests/data/py_36/numeric_literals.py
rename to tests/data/cases/numeric_literals.py
index 254da68d330..99669328744 100644
--- a/tests/data/py_36/numeric_literals.py
+++ b/tests/data/cases/numeric_literals.py
@@ -1,5 +1,3 @@
-#!/usr/bin/env python3.6
-
 x = 123456789
 x = 123456
 x = .1
@@ -21,9 +19,6 @@
 
 # output
 
-
-#!/usr/bin/env python3.6
-
 x = 123456789
 x = 123456
 x = 0.1
diff --git a/tests/data/py_36/numeric_literals_skip_underscores.py b/tests/data/cases/numeric_literals_skip_underscores.py
similarity index 80%
rename from tests/data/py_36/numeric_literals_skip_underscores.py
rename to tests/data/cases/numeric_literals_skip_underscores.py
index e345bb90276..6d60bdbb34d 100644
--- a/tests/data/py_36/numeric_literals_skip_underscores.py
+++ b/tests/data/cases/numeric_literals_skip_underscores.py
@@ -1,5 +1,3 @@
-#!/usr/bin/env python3.6
-
 x = 123456789
 x = 1_2_3_4_5_6_7
 x = 1E+1
@@ -11,8 +9,6 @@
 
 # output
 
-#!/usr/bin/env python3.6
-
 x = 123456789
 x = 1_2_3_4_5_6_7
 x = 1e1
diff --git a/tests/data/simple_cases/one_element_subscript.py b/tests/data/cases/one_element_subscript.py
similarity index 100%
rename from tests/data/simple_cases/one_element_subscript.py
rename to tests/data/cases/one_element_subscript.py
diff --git a/tests/data/py_310/parenthesized_context_managers.py b/tests/data/cases/parenthesized_context_managers.py
similarity index 95%
rename from tests/data/py_310/parenthesized_context_managers.py
rename to tests/data/cases/parenthesized_context_managers.py
index 1ef09a1bd34..16645a18baa 100644
--- a/tests/data/py_310/parenthesized_context_managers.py
+++ b/tests/data/cases/parenthesized_context_managers.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 with (CtxManager() as example):
     ...
 
diff --git a/tests/data/py_310/pattern_matching_complex.py b/tests/data/cases/pattern_matching_complex.py
similarity index 96%
rename from tests/data/py_310/pattern_matching_complex.py
rename to tests/data/cases/pattern_matching_complex.py
index 97ee194fd39..10b4d26e289 100644
--- a/tests/data/py_310/pattern_matching_complex.py
+++ b/tests/data/cases/pattern_matching_complex.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 # Cases sampled from Lib/test/test_patma.py
 
 # case black_test_patma_098
@@ -142,3 +143,7 @@
         y = 1
     case []:
         y = 2
+# issue 3790
+match (X.type, Y):
+    case _:
+        pass
diff --git a/tests/data/py_310/pattern_matching_extras.py b/tests/data/cases/pattern_matching_extras.py
similarity index 98%
rename from tests/data/py_310/pattern_matching_extras.py
rename to tests/data/cases/pattern_matching_extras.py
index 0242d264e5b..1e1481d7bbe 100644
--- a/tests/data/py_310/pattern_matching_extras.py
+++ b/tests/data/cases/pattern_matching_extras.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 import match
 
 match something:
diff --git a/tests/data/py_310/pattern_matching_generic.py b/tests/data/cases/pattern_matching_generic.py
similarity index 98%
rename from tests/data/py_310/pattern_matching_generic.py
rename to tests/data/cases/pattern_matching_generic.py
index 00a0e4a677d..4b4d45f0bff 100644
--- a/tests/data/py_310/pattern_matching_generic.py
+++ b/tests/data/cases/pattern_matching_generic.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 re.match()
 match = a
 with match() as match:
diff --git a/tests/data/py_310/pattern_matching_simple.py b/tests/data/cases/pattern_matching_simple.py
similarity index 98%
rename from tests/data/py_310/pattern_matching_simple.py
rename to tests/data/cases/pattern_matching_simple.py
index 5ed62415a4b..6fa2000f0de 100644
--- a/tests/data/py_310/pattern_matching_simple.py
+++ b/tests/data/cases/pattern_matching_simple.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 # Cases sampled from PEP 636 examples
 
 match command.split():
diff --git a/tests/data/py_310/pattern_matching_style.py b/tests/data/cases/pattern_matching_style.py
similarity index 97%
rename from tests/data/py_310/pattern_matching_style.py
rename to tests/data/cases/pattern_matching_style.py
index 8e18ce2ada6..2ee6ea2b6e9 100644
--- a/tests/data/py_310/pattern_matching_style.py
+++ b/tests/data/cases/pattern_matching_style.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 match something:
     case b(): print(1+1)
     case c(
diff --git a/tests/data/cases/pep604_union_types_line_breaks.py b/tests/data/cases/pep604_union_types_line_breaks.py
new file mode 100644
index 00000000000..fee2b840494
--- /dev/null
+++ b/tests/data/cases/pep604_union_types_line_breaks.py
@@ -0,0 +1,188 @@
+# flags: --preview --minimum-version=3.10
+# This has always worked
+z= Loooooooooooooooooooooooong | Loooooooooooooooooooooooong | Loooooooooooooooooooooooong | Loooooooooooooooooooooooong
+
+# "AnnAssign"s now also work
+z: Loooooooooooooooooooooooong | Loooooooooooooooooooooooong | Loooooooooooooooooooooooong | Loooooooooooooooooooooooong
+z: (Short
+    | Short2
+    | Short3
+    | Short4)
+z: (int)
+z: ((int))
+
+
+z: Loooooooooooooooooooooooong | Loooooooooooooooooooooooong | Loooooooooooooooooooooooong | Loooooooooooooooooooooooong = 7
+z: (Short
+    | Short2
+    | Short3
+    | Short4) = 8
+z: (int) = 2.3
+z: ((int)) = foo()
+
+# In case I go for not enforcing parantheses, this might get improved at the same time
+x = (
+    z
+    == 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999,
+    y
+    == 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999,
+)
+
+x = (
+    z == (9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999),
+    y == (9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999),
+)
+
+# handle formatting of "tname"s in parameter list
+
+# remove unnecessary paren
+def foo(i: (int)) -> None: ...
+
+
+# this is a syntax error in the type annotation according to mypy, but it's not invalid *python* code, so make sure we don't mess with it and make it so.
+def foo(i: (int,)) -> None: ...
+
+def foo(
+    i: int,
+    x: Loooooooooooooooooooooooong
+    | Looooooooooooooooong
+    | Looooooooooooooooooooong
+    | Looooooong,
+    *,
+    s: str,
+) -> None:
+    pass
+
+
+@app.get("/path/")
+async def foo(
+    q: str
+    | None = Query(None, title="Some long title", description="Some long description")
+):
+    pass
+
+
+def f(
+    max_jobs: int
+    | None = Option(
+        None, help="Maximum number of jobs to launch. And some additional text."
+        ),
+    another_option: bool = False
+    ):
+    ...
+
+
+# output
+# This has always worked
+z = (
+    Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+)
+
+# "AnnAssign"s now also work
+z: (
+    Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+)
+z: Short | Short2 | Short3 | Short4
+z: int
+z: int
+
+
+z: (
+    Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+    | Loooooooooooooooooooooooong
+) = 7
+z: Short | Short2 | Short3 | Short4 = 8
+z: int = 2.3
+z: int = foo()
+
+# In case I go for not enforcing parantheses, this might get improved at the same time
+x = (
+    z
+    == 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999
+    | 9999999999999999999999999999999999999999,
+    y
+    == 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999
+    + 9999999999999999999999999999999999999999,
+)
+
+x = (
+    z
+    == (
+        9999999999999999999999999999999999999999
+        | 9999999999999999999999999999999999999999
+        | 9999999999999999999999999999999999999999
+        | 9999999999999999999999999999999999999999
+    ),
+    y
+    == (
+        9999999999999999999999999999999999999999
+        + 9999999999999999999999999999999999999999
+        + 9999999999999999999999999999999999999999
+        + 9999999999999999999999999999999999999999
+    ),
+)
+
+# handle formatting of "tname"s in parameter list
+
+
+# remove unnecessary paren
+def foo(i: int) -> None: ...
+
+
+# this is a syntax error in the type annotation according to mypy, but it's not invalid *python* code, so make sure we don't mess with it and make it so.
+def foo(i: (int,)) -> None: ...
+
+
+def foo(
+    i: int,
+    x: (
+        Loooooooooooooooooooooooong
+        | Looooooooooooooooong
+        | Looooooooooooooooooooong
+        | Looooooong
+    ),
+    *,
+    s: str,
+) -> None:
+    pass
+
+
+@app.get("/path/")
+async def foo(
+    q: str | None = Query(
+        None, title="Some long title", description="Some long description"
+    )
+):
+    pass
+
+
+def f(
+    max_jobs: int | None = Option(
+        None, help="Maximum number of jobs to launch. And some additional text."
+    ),
+    another_option: bool = False,
+): ...
diff --git a/tests/data/py_38/pep_570.py b/tests/data/cases/pep_570.py
similarity index 95%
rename from tests/data/py_38/pep_570.py
rename to tests/data/cases/pep_570.py
index ca8f7ab1d95..2641c2b970e 100644
--- a/tests/data/py_38/pep_570.py
+++ b/tests/data/cases/pep_570.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.8
 def positional_only_arg(a, /):
     pass
 
diff --git a/tests/data/py_38/pep_572.py b/tests/data/cases/pep_572.py
similarity index 96%
rename from tests/data/py_38/pep_572.py
rename to tests/data/cases/pep_572.py
index d41805f1cb1..742b6d5b7e4 100644
--- a/tests/data/py_38/pep_572.py
+++ b/tests/data/cases/pep_572.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.8
 (a := 1)
 (a := a)
 if (match := pattern.search(data)) is None:
diff --git a/tests/data/fast/pep_572_do_not_remove_parens.py b/tests/data/cases/pep_572_do_not_remove_parens.py
similarity index 96%
rename from tests/data/fast/pep_572_do_not_remove_parens.py
rename to tests/data/cases/pep_572_do_not_remove_parens.py
index 05619ddcc2b..08dba3ffdf9 100644
--- a/tests/data/fast/pep_572_do_not_remove_parens.py
+++ b/tests/data/cases/pep_572_do_not_remove_parens.py
@@ -1,3 +1,4 @@
+# flags: --fast
 # Most of the following examples are really dumb, some of them aren't even accepted by Python,
 # we're fixing them only so fuzzers (which follow the grammar which actually allows these
 # examples matter of fact!) don't yell at us :p
diff --git a/tests/data/py_310/pep_572_py310.py b/tests/data/cases/pep_572_py310.py
similarity index 93%
rename from tests/data/py_310/pep_572_py310.py
rename to tests/data/cases/pep_572_py310.py
index cb82b2d23f8..9f999deeb89 100644
--- a/tests/data/py_310/pep_572_py310.py
+++ b/tests/data/cases/pep_572_py310.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 # Unparenthesized walruses are now allowed in indices since Python 3.10.
 x[a:=0]
 x[a:=0, b:=1]
diff --git a/tests/data/py_39/pep_572_py39.py b/tests/data/cases/pep_572_py39.py
similarity index 89%
rename from tests/data/py_39/pep_572_py39.py
rename to tests/data/cases/pep_572_py39.py
index b8b081b8c45..d1614624d99 100644
--- a/tests/data/py_39/pep_572_py39.py
+++ b/tests/data/cases/pep_572_py39.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.9
 # Unparenthesized walruses are now allowed in set literals & set comprehensions
 # since Python 3.9
 {x := 1, 2, 3}
diff --git a/tests/data/py_38/pep_572_remove_parens.py b/tests/data/cases/pep_572_remove_parens.py
similarity index 98%
rename from tests/data/py_38/pep_572_remove_parens.py
rename to tests/data/cases/pep_572_remove_parens.py
index b952b2940c5..24f1ac29168 100644
--- a/tests/data/py_38/pep_572_remove_parens.py
+++ b/tests/data/cases/pep_572_remove_parens.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.8
 if (foo := 0):
     pass
 
diff --git a/tests/data/simple_cases/pep_604.py b/tests/data/cases/pep_604.py
similarity index 100%
rename from tests/data/simple_cases/pep_604.py
rename to tests/data/cases/pep_604.py
diff --git a/tests/data/py_311/pep_646.py b/tests/data/cases/pep_646.py
similarity index 98%
rename from tests/data/py_311/pep_646.py
rename to tests/data/cases/pep_646.py
index e843ecf39d8..92b568a379c 100644
--- a/tests/data/py_311/pep_646.py
+++ b/tests/data/cases/pep_646.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.11
 A[*b]
 A[*b] = 1
 A
diff --git a/tests/data/py_311/pep_654.py b/tests/data/cases/pep_654.py
similarity index 96%
rename from tests/data/py_311/pep_654.py
rename to tests/data/cases/pep_654.py
index 387c0816f4b..12e49180e41 100644
--- a/tests/data/py_311/pep_654.py
+++ b/tests/data/cases/pep_654.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.11
 try:
     raise OSError("blah")
 except* ExceptionGroup as e:
diff --git a/tests/data/py_311/pep_654_style.py b/tests/data/cases/pep_654_style.py
similarity index 98%
rename from tests/data/py_311/pep_654_style.py
rename to tests/data/cases/pep_654_style.py
index 9fc7c0c84db..0d34650e098 100644
--- a/tests/data/py_311/pep_654_style.py
+++ b/tests/data/cases/pep_654_style.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.11
 try:
     raise OSError("blah")
 except               * ExceptionGroup as e:
diff --git a/tests/data/miscellaneous/power_op_newline.py b/tests/data/cases/power_op_newline.py
similarity index 73%
rename from tests/data/miscellaneous/power_op_newline.py
rename to tests/data/cases/power_op_newline.py
index 85d434d63f6..d9b31403c9d 100644
--- a/tests/data/miscellaneous/power_op_newline.py
+++ b/tests/data/cases/power_op_newline.py
@@ -1,3 +1,4 @@
+# flags: --line-length=0
 importA;()<<0**0#
 
 # output
diff --git a/tests/data/simple_cases/power_op_spacing.py b/tests/data/cases/power_op_spacing.py
similarity index 95%
rename from tests/data/simple_cases/power_op_spacing.py
rename to tests/data/cases/power_op_spacing.py
index c95fa788fc3..b3ef0aae084 100644
--- a/tests/data/simple_cases/power_op_spacing.py
+++ b/tests/data/cases/power_op_spacing.py
@@ -29,6 +29,13 @@ def function_dont_replace_spaces():
 p = {(k, k**2): v**2 for k, v in pairs}
 q = [10**i for i in range(6)]
 r = x**y
+s = 1 ** 1
+t = (
+    1
+    ** 1
+    **1
+    ** 1
+)
 
 a = 5.0**~4.0
 b = 5.0 ** f()
@@ -47,6 +54,13 @@ def function_dont_replace_spaces():
 o = settings(max_examples=10**6.0)
 p = {(k, k**2): v**2.0 for k, v in pairs}
 q = [10.5**i for i in range(6)]
+s = 1.0 ** 1.0
+t = (
+    1.0
+    ** 1.0
+    **1.0
+    ** 1.0
+)
 
 
 # WE SHOULD DEFINITELY NOT EAT THESE COMMENTS (https://github.com/psf/black/issues/2873)
@@ -97,6 +111,8 @@ def function_dont_replace_spaces():
 p = {(k, k**2): v**2 for k, v in pairs}
 q = [10**i for i in range(6)]
 r = x**y
+s = 1**1
+t = 1**1**1**1
 
 a = 5.0**~4.0
 b = 5.0 ** f()
@@ -115,6 +131,8 @@ def function_dont_replace_spaces():
 o = settings(max_examples=10**6.0)
 p = {(k, k**2): v**2.0 for k, v in pairs}
 q = [10.5**i for i in range(6)]
+s = 1.0**1.0
+t = 1.0**1.0**1.0**1.0
 
 
 # WE SHOULD DEFINITELY NOT EAT THESE COMMENTS (https://github.com/psf/black/issues/2873)
diff --git a/tests/data/simple_cases/prefer_rhs_split_reformatted.py b/tests/data/cases/prefer_rhs_split_reformatted.py
similarity index 100%
rename from tests/data/simple_cases/prefer_rhs_split_reformatted.py
rename to tests/data/cases/prefer_rhs_split_reformatted.py
diff --git a/tests/data/preview/async_stmts.py b/tests/data/cases/preview_async_stmts.py
similarity index 93%
rename from tests/data/preview/async_stmts.py
rename to tests/data/cases/preview_async_stmts.py
index fe9594b2164..0a7671be5a6 100644
--- a/tests/data/preview/async_stmts.py
+++ b/tests/data/cases/preview_async_stmts.py
@@ -1,3 +1,4 @@
+# flags: --preview
 async def func() -> (int):
     return 0
 
diff --git a/tests/data/preview/cantfit.py b/tests/data/cases/preview_cantfit.py
similarity index 99%
rename from tests/data/preview/cantfit.py
rename to tests/data/cases/preview_cantfit.py
index 0849374f776..d5da6654f0c 100644
--- a/tests/data/preview/cantfit.py
+++ b/tests/data/cases/preview_cantfit.py
@@ -1,3 +1,4 @@
+# flags: --preview
 # long variable name
 this_is_a_ridiculously_long_name_and_nobody_in_their_right_mind_would_use_one_like_it = 0
 this_is_a_ridiculously_long_name_and_nobody_in_their_right_mind_would_use_one_like_it = 1  # with a comment
diff --git a/tests/data/preview/comments7.py b/tests/data/cases/preview_comments7.py
similarity index 92%
rename from tests/data/preview/comments7.py
rename to tests/data/cases/preview_comments7.py
index 8b1224017e5..006d4f7266f 100644
--- a/tests/data/preview/comments7.py
+++ b/tests/data/cases/preview_comments7.py
@@ -1,3 +1,4 @@
+# flags: --preview
 from .config import (
     Any,
     Bool,
@@ -131,6 +132,18 @@ def test_fails_invalid_post_data(
 
 square = Square(4) # type: Optional[Square]
 
+# Regression test for https://github.com/psf/black/issues/3756.
+[
+    (
+        "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"  # aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+    ),
+]
+[
+    (  # aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+        "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"  # aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+    ),
+]
+
 # output
 
 from .config import (
@@ -282,3 +295,15 @@ def test_fails_invalid_post_data(
 
 
 square = Square(4)  # type: Optional[Square]
+
+# Regression test for https://github.com/psf/black/issues/3756.
+[
+    (  # aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+        "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
+    ),
+]
+[
+    (  # aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+        "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"  # aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa
+    ),
+]
diff --git a/tests/data/preview_context_managers/targeting_py38.py b/tests/data/cases/preview_context_managers_38.py
similarity index 96%
rename from tests/data/preview_context_managers/targeting_py38.py
rename to tests/data/cases/preview_context_managers_38.py
index f125cdffb8a..719d94fdcc5 100644
--- a/tests/data/preview_context_managers/targeting_py38.py
+++ b/tests/data/cases/preview_context_managers_38.py
@@ -1,3 +1,4 @@
+# flags: --preview --minimum-version=3.8
 with \
      make_context_manager1() as cm1, \
      make_context_manager2() as cm2, \
diff --git a/tests/data/preview_context_managers/targeting_py39.py b/tests/data/cases/preview_context_managers_39.py
similarity index 98%
rename from tests/data/preview_context_managers/targeting_py39.py
rename to tests/data/cases/preview_context_managers_39.py
index c9fcf9c8ba2..589e00ad187 100644
--- a/tests/data/preview_context_managers/targeting_py39.py
+++ b/tests/data/cases/preview_context_managers_39.py
@@ -1,3 +1,4 @@
+# flags: --preview --minimum-version=3.9
 with \
      make_context_manager1() as cm1, \
      make_context_manager2() as cm2, \
diff --git a/tests/data/preview_context_managers/auto_detect/features_3_10.py b/tests/data/cases/preview_context_managers_autodetect_310.py
similarity index 93%
rename from tests/data/preview_context_managers/auto_detect/features_3_10.py
rename to tests/data/cases/preview_context_managers_autodetect_310.py
index 1458df1cb41..a9e31076f03 100644
--- a/tests/data/preview_context_managers/auto_detect/features_3_10.py
+++ b/tests/data/cases/preview_context_managers_autodetect_310.py
@@ -1,3 +1,4 @@
+# flags: --preview --minimum-version=3.10
 # This file uses pattern matching introduced in Python 3.10.
 
 
diff --git a/tests/data/preview_context_managers/auto_detect/features_3_11.py b/tests/data/cases/preview_context_managers_autodetect_311.py
similarity index 92%
rename from tests/data/preview_context_managers/auto_detect/features_3_11.py
rename to tests/data/cases/preview_context_managers_autodetect_311.py
index f83c5330ab3..af1e83fe74c 100644
--- a/tests/data/preview_context_managers/auto_detect/features_3_11.py
+++ b/tests/data/cases/preview_context_managers_autodetect_311.py
@@ -1,3 +1,4 @@
+# flags: --preview --minimum-version=3.11
 # This file uses except* clause in Python 3.11.
 
 
diff --git a/tests/data/preview_context_managers/auto_detect/features_3_8.py b/tests/data/cases/preview_context_managers_autodetect_38.py
similarity index 98%
rename from tests/data/preview_context_managers/auto_detect/features_3_8.py
rename to tests/data/cases/preview_context_managers_autodetect_38.py
index 79e438b995e..25217a40604 100644
--- a/tests/data/preview_context_managers/auto_detect/features_3_8.py
+++ b/tests/data/cases/preview_context_managers_autodetect_38.py
@@ -1,3 +1,4 @@
+# flags: --preview
 # This file doesn't use any Python 3.9+ only grammars.
 
 
diff --git a/tests/data/preview_context_managers/auto_detect/features_3_9.py b/tests/data/cases/preview_context_managers_autodetect_39.py
similarity index 93%
rename from tests/data/preview_context_managers/auto_detect/features_3_9.py
rename to tests/data/cases/preview_context_managers_autodetect_39.py
index 0d28f993108..3f72e48db9d 100644
--- a/tests/data/preview_context_managers/auto_detect/features_3_9.py
+++ b/tests/data/cases/preview_context_managers_autodetect_39.py
@@ -1,3 +1,4 @@
+# flags: --preview --minimum-version=3.9
 # This file uses parenthesized context managers introduced in Python 3.9.
 
 
diff --git a/tests/data/preview/dummy_implementations.py b/tests/data/cases/preview_dummy_implementations.py
similarity index 98%
rename from tests/data/preview/dummy_implementations.py
rename to tests/data/cases/preview_dummy_implementations.py
index e07c25ed129..98b69bf87b2 100644
--- a/tests/data/preview/dummy_implementations.py
+++ b/tests/data/cases/preview_dummy_implementations.py
@@ -1,3 +1,4 @@
+# flags: --preview
 from typing import NoReturn, Protocol, Union, overload
 
 
diff --git a/tests/data/preview/format_unicode_escape_seq.py b/tests/data/cases/preview_format_unicode_escape_seq.py
similarity index 96%
rename from tests/data/preview/format_unicode_escape_seq.py
rename to tests/data/cases/preview_format_unicode_escape_seq.py
index 3440696c303..65c3d8d166e 100644
--- a/tests/data/preview/format_unicode_escape_seq.py
+++ b/tests/data/cases/preview_format_unicode_escape_seq.py
@@ -1,3 +1,4 @@
+# flags: --preview
 x = "\x1F"
 x = "\\x1B"
 x = "\\\x1B"
diff --git a/tests/data/preview/long_dict_values.py b/tests/data/cases/preview_long_dict_values.py
similarity index 99%
rename from tests/data/preview/long_dict_values.py
rename to tests/data/cases/preview_long_dict_values.py
index 4c515180028..fbbacd13d1d 100644
--- a/tests/data/preview/long_dict_values.py
+++ b/tests/data/cases/preview_long_dict_values.py
@@ -1,3 +1,4 @@
+# flags: --preview
 my_dict = {
     "something_something":
         r"Lorem ipsum dolor sit amet, an sed convenire eloquentiam \t"
diff --git a/tests/data/preview/long_strings.py b/tests/data/cases/preview_long_strings.py
similarity index 99%
rename from tests/data/preview/long_strings.py
rename to tests/data/cases/preview_long_strings.py
index 059148729d5..5519f098774 100644
--- a/tests/data/preview/long_strings.py
+++ b/tests/data/cases/preview_long_strings.py
@@ -1,3 +1,4 @@
+# flags: --preview
 x = "This is a really long string that can't possibly be expected to fit all together on one line. In fact it may even take up three or more lines... like four or five... but probably just three."
 
 x += "This is a really long string that can't possibly be expected to fit all together on one line. In fact it may even take up three or more lines... like four or five... but probably just three."
diff --git a/tests/data/preview/long_strings__east_asian_width.py b/tests/data/cases/preview_long_strings__east_asian_width.py
similarity index 96%
rename from tests/data/preview/long_strings__east_asian_width.py
rename to tests/data/cases/preview_long_strings__east_asian_width.py
index fb66a78ed8b..d190f422a60 100644
--- a/tests/data/preview/long_strings__east_asian_width.py
+++ b/tests/data/cases/preview_long_strings__east_asian_width.py
@@ -1,3 +1,4 @@
+# flags: --preview
 # The following strings do not have not-so-many chars, but are long enough
 # when these are rendered in a monospace font (if the renderer respects
 # Unicode East Asian Width properties).
diff --git a/tests/data/preview/long_strings__edge_case.py b/tests/data/cases/preview_long_strings__edge_case.py
similarity index 99%
rename from tests/data/preview/long_strings__edge_case.py
rename to tests/data/cases/preview_long_strings__edge_case.py
index 2bc0b6ed328..a8e8971968c 100644
--- a/tests/data/preview/long_strings__edge_case.py
+++ b/tests/data/cases/preview_long_strings__edge_case.py
@@ -1,3 +1,4 @@
+# flags: --preview
 some_variable = "This string is long but not so long that it needs to be split just yet"
 some_variable = 'This string is long but not so long that it needs to be split just yet'
 some_variable = "This string is long, just long enough that it needs to be split, u get?"
diff --git a/tests/data/preview/long_strings__regression.py b/tests/data/cases/preview_long_strings__regression.py
similarity index 99%
rename from tests/data/preview/long_strings__regression.py
rename to tests/data/cases/preview_long_strings__regression.py
index 5f0646e6029..40d5e745cc8 100644
--- a/tests/data/preview/long_strings__regression.py
+++ b/tests/data/cases/preview_long_strings__regression.py
@@ -1,3 +1,4 @@
+# flags: --preview
 class A:
     def foo():
         result = type(message)("")
diff --git a/tests/data/preview/long_strings__type_annotations.py b/tests/data/cases/preview_long_strings__type_annotations.py
similarity index 95%
rename from tests/data/preview/long_strings__type_annotations.py
rename to tests/data/cases/preview_long_strings__type_annotations.py
index 41d7ee2b67b..8beb877bdd1 100644
--- a/tests/data/preview/long_strings__type_annotations.py
+++ b/tests/data/cases/preview_long_strings__type_annotations.py
@@ -1,3 +1,4 @@
+# flags: --preview
 def func(
     arg1,
     arg2,
@@ -54,6 +55,6 @@ def func(
 
 
 def func(
-    argument: ("int |" "str"),
+    argument: "int |" "str",
 ) -> Set["int |" " str"]:
     pass
diff --git a/tests/data/preview/multiline_strings.py b/tests/data/cases/preview_multiline_strings.py
similarity index 99%
rename from tests/data/preview/multiline_strings.py
rename to tests/data/cases/preview_multiline_strings.py
index bb517d128e2..dec4ef2e548 100644
--- a/tests/data/preview/multiline_strings.py
+++ b/tests/data/cases/preview_multiline_strings.py
@@ -1,3 +1,4 @@
+# flags: --preview
 """cow
 say""",
 call(3, "dogsay", textwrap.dedent("""dove
diff --git a/tests/data/preview/no_blank_line_before_docstring.py b/tests/data/cases/preview_no_blank_line_before_docstring.py
similarity index 97%
rename from tests/data/preview/no_blank_line_before_docstring.py
rename to tests/data/cases/preview_no_blank_line_before_docstring.py
index a37362de100..303035a7efb 100644
--- a/tests/data/preview/no_blank_line_before_docstring.py
+++ b/tests/data/cases/preview_no_blank_line_before_docstring.py
@@ -1,3 +1,4 @@
+# flags: --preview
 def line_before_docstring():
 
     """Please move me up"""
diff --git a/tests/data/preview/pep_572.py b/tests/data/cases/preview_pep_572.py
similarity index 75%
rename from tests/data/preview/pep_572.py
rename to tests/data/cases/preview_pep_572.py
index a50e130ad9c..8e801ff6cdc 100644
--- a/tests/data/preview/pep_572.py
+++ b/tests/data/cases/preview_pep_572.py
@@ -1,3 +1,4 @@
+# flags: --preview
 x[(a:=0):]
 x[:(a:=0)]
 
diff --git a/tests/data/preview/percent_precedence.py b/tests/data/cases/preview_percent_precedence.py
similarity index 96%
rename from tests/data/preview/percent_precedence.py
rename to tests/data/cases/preview_percent_precedence.py
index b895443fb46..aeaf450ff5e 100644
--- a/tests/data/preview/percent_precedence.py
+++ b/tests/data/cases/preview_percent_precedence.py
@@ -1,3 +1,4 @@
+# flags: --preview
 ("" % a) ** 2
 ("" % a)[0]
 ("" % a)()
diff --git a/tests/data/cases/preview_power_op_spacing.py b/tests/data/cases/preview_power_op_spacing.py
new file mode 100644
index 00000000000..650c6fecb20
--- /dev/null
+++ b/tests/data/cases/preview_power_op_spacing.py
@@ -0,0 +1,97 @@
+# flags: --preview
+a = 1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1
+b = 1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1
+c = 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1 ** 1
+d = 1**1 ** 1**1 ** 1**1 ** 1**1 ** 1**1**1 ** 1 ** 1**1 ** 1**1**1**1**1 ** 1 ** 1**1**1 **1**1** 1 ** 1 ** 1
+e = 𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟
+f = 𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟
+
+a = 1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0
+b = 1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0
+c = 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0 ** 1.0
+d = 1.0**1.0 ** 1.0**1.0 ** 1.0**1.0 ** 1.0**1.0 ** 1.0**1.0**1.0 ** 1.0 ** 1.0**1.0 ** 1.0**1.0**1.0
+
+# output
+a = 1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1
+b = (
+    1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+    ** 1
+)
+c = 1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1
+d = 1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1**1
+e = 𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟**𨉟
+f = (
+    𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+    ** 𨉟
+)
+
+a = 1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0
+b = (
+    1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+    ** 1.0
+)
+c = 1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0
+d = 1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0**1.0
diff --git a/tests/data/preview/prefer_rhs_split.py b/tests/data/cases/preview_prefer_rhs_split.py
similarity index 99%
rename from tests/data/preview/prefer_rhs_split.py
rename to tests/data/cases/preview_prefer_rhs_split.py
index a809eacc773..c732c33b53a 100644
--- a/tests/data/preview/prefer_rhs_split.py
+++ b/tests/data/cases/preview_prefer_rhs_split.py
@@ -1,3 +1,4 @@
+# flags: --preview
 first_item, second_item = (
     some_looooooooong_module.some_looooooooooooooong_function_name(
         first_argument, second_argument, third_argument
diff --git a/tests/data/cases/preview_return_annotation_brackets_string.py b/tests/data/cases/preview_return_annotation_brackets_string.py
new file mode 100644
index 00000000000..fea0ea6839a
--- /dev/null
+++ b/tests/data/cases/preview_return_annotation_brackets_string.py
@@ -0,0 +1,24 @@
+# flags: --preview
+# Long string example
+def frobnicate() -> "ThisIsTrulyUnreasonablyExtremelyLongClassName | list[ThisIsTrulyUnreasonablyExtremelyLongClassName]":
+    pass
+
+# splitting the string breaks if there's any parameters
+def frobnicate(a) -> "ThisIsTrulyUnreasonablyExtremelyLongClassName | list[ThisIsTrulyUnreasonablyExtremelyLongClassName]":
+    pass
+
+# output
+
+# Long string example
+def frobnicate() -> (
+    "ThisIsTrulyUnreasonablyExtremelyLongClassName |"
+    " list[ThisIsTrulyUnreasonablyExtremelyLongClassName]"
+):
+    pass
+
+
+# splitting the string breaks if there's any parameters
+def frobnicate(
+    a,
+) -> "ThisIsTrulyUnreasonablyExtremelyLongClassName | list[ThisIsTrulyUnreasonablyExtremelyLongClassName]":
+    pass
diff --git a/tests/data/preview/trailing_comma.py b/tests/data/cases/preview_trailing_comma.py
similarity index 97%
rename from tests/data/preview/trailing_comma.py
rename to tests/data/cases/preview_trailing_comma.py
index 5b09c664606..bba7e7ad16d 100644
--- a/tests/data/preview/trailing_comma.py
+++ b/tests/data/cases/preview_trailing_comma.py
@@ -1,3 +1,4 @@
+# flags: --preview
 e = {
     "a": fun(msg, "ts"),
     "longggggggggggggggid": ...,
diff --git a/tests/data/preview_py_310/pep_572.py b/tests/data/cases/py310_pep572.py
similarity index 77%
rename from tests/data/preview_py_310/pep_572.py
rename to tests/data/cases/py310_pep572.py
index 78d4e9e4506..172be3898d6 100644
--- a/tests/data/preview_py_310/pep_572.py
+++ b/tests/data/cases/py310_pep572.py
@@ -1,3 +1,4 @@
+# flags: --preview --minimum-version=3.10
 x[a:=0]
 x[a := 0]
 x[a := 0, b := 1]
diff --git a/tests/data/py_37/python37.py b/tests/data/cases/python37.py
similarity index 95%
rename from tests/data/py_37/python37.py
rename to tests/data/cases/python37.py
index dab8b404a73..3f61106c45d 100644
--- a/tests/data/py_37/python37.py
+++ b/tests/data/cases/python37.py
@@ -1,4 +1,4 @@
-#!/usr/bin/env python3.7
+# flags: --minimum-version=3.7
 
 
 def f():
@@ -33,9 +33,6 @@ def make_arange(n):
 # output
 
 
-#!/usr/bin/env python3.7
-
-
 def f():
     return (i * 2 async for i in arange(42))
 
diff --git a/tests/data/py_38/python38.py b/tests/data/cases/python38.py
similarity index 93%
rename from tests/data/py_38/python38.py
rename to tests/data/cases/python38.py
index 63b0588bc27..919ea6aeed4 100644
--- a/tests/data/py_38/python38.py
+++ b/tests/data/cases/python38.py
@@ -1,4 +1,4 @@
-#!/usr/bin/env python3.8
+# flags: --minimum-version=3.8
 
 
 def starred_return():
@@ -22,9 +22,6 @@ def t():
 # output
 
 
-#!/usr/bin/env python3.8
-
-
 def starred_return():
     my_list = ["value2", "value3"]
     return "value1", *my_list
diff --git a/tests/data/py_39/python39.py b/tests/data/cases/python39.py
similarity index 92%
rename from tests/data/py_39/python39.py
rename to tests/data/cases/python39.py
index ae67c2257eb..1b9536c1529 100644
--- a/tests/data/py_39/python39.py
+++ b/tests/data/cases/python39.py
@@ -1,4 +1,4 @@
-#!/usr/bin/env python3.9
+# flags: --minimum-version=3.9
 
 @relaxed_decorator[0]
 def f():
@@ -14,10 +14,6 @@ def f():
 
 # output
 
-
-#!/usr/bin/env python3.9
-
-
 @relaxed_decorator[0]
 def f():
     ...
diff --git a/tests/data/simple_cases/remove_await_parens.py b/tests/data/cases/remove_await_parens.py
similarity index 100%
rename from tests/data/simple_cases/remove_await_parens.py
rename to tests/data/cases/remove_await_parens.py
diff --git a/tests/data/simple_cases/remove_except_parens.py b/tests/data/cases/remove_except_parens.py
similarity index 100%
rename from tests/data/simple_cases/remove_except_parens.py
rename to tests/data/cases/remove_except_parens.py
diff --git a/tests/data/simple_cases/remove_for_brackets.py b/tests/data/cases/remove_for_brackets.py
similarity index 100%
rename from tests/data/simple_cases/remove_for_brackets.py
rename to tests/data/cases/remove_for_brackets.py
diff --git a/tests/data/simple_cases/remove_newline_after_code_block_open.py b/tests/data/cases/remove_newline_after_code_block_open.py
similarity index 100%
rename from tests/data/simple_cases/remove_newline_after_code_block_open.py
rename to tests/data/cases/remove_newline_after_code_block_open.py
diff --git a/tests/data/py_310/remove_newline_after_match.py b/tests/data/cases/remove_newline_after_match.py
similarity index 94%
rename from tests/data/py_310/remove_newline_after_match.py
rename to tests/data/cases/remove_newline_after_match.py
index f7bcfbf27a2..fe6592b664d 100644
--- a/tests/data/py_310/remove_newline_after_match.py
+++ b/tests/data/cases/remove_newline_after_match.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 def http_status(status):
 
     match status:
diff --git a/tests/data/simple_cases/remove_parens.py b/tests/data/cases/remove_parens.py
similarity index 100%
rename from tests/data/simple_cases/remove_parens.py
rename to tests/data/cases/remove_parens.py
diff --git a/tests/data/py_39/remove_with_brackets.py b/tests/data/cases/remove_with_brackets.py
similarity index 98%
rename from tests/data/py_39/remove_with_brackets.py
rename to tests/data/cases/remove_with_brackets.py
index ea58ab93a16..3ee64902a30 100644
--- a/tests/data/py_39/remove_with_brackets.py
+++ b/tests/data/cases/remove_with_brackets.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.9
 with (open("bla.txt")):
     pass
 
diff --git a/tests/data/simple_cases/return_annotation_brackets.py b/tests/data/cases/return_annotation_brackets.py
similarity index 92%
rename from tests/data/simple_cases/return_annotation_brackets.py
rename to tests/data/cases/return_annotation_brackets.py
index 265c30220d8..8509ecdb92c 100644
--- a/tests/data/simple_cases/return_annotation_brackets.py
+++ b/tests/data/cases/return_annotation_brackets.py
@@ -87,6 +87,11 @@ def foo() -> tuple[loooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
 def foo() -> tuple[int, int, int,]:
     return 2
 
+# Magic trailing comma example, with params
+# this is broken - the trailing comma is transferred to the param list. Fixed in preview
+def foo(a,b) -> tuple[int, int, int,]:
+    return 2
+
 # output
 # Control
 def double(a: int) -> int:
@@ -208,3 +213,11 @@ def foo() -> (
     ]
 ):
     return 2
+
+
+# Magic trailing comma example, with params
+# this is broken - the trailing comma is transferred to the param list. Fixed in preview
+def foo(
+    a, b
+) -> tuple[int, int, int,]:
+    return 2
diff --git a/tests/data/simple_cases/skip_magic_trailing_comma.py b/tests/data/cases/skip_magic_trailing_comma.py
similarity index 97%
rename from tests/data/simple_cases/skip_magic_trailing_comma.py
rename to tests/data/cases/skip_magic_trailing_comma.py
index c020db79864..4dda5df40f0 100644
--- a/tests/data/simple_cases/skip_magic_trailing_comma.py
+++ b/tests/data/cases/skip_magic_trailing_comma.py
@@ -1,3 +1,4 @@
+# flags: --skip-magic-trailing-comma
 # We should not remove the trailing comma in a single-element subscript.
 a: tuple[int,]
 b = tuple[int,]
diff --git a/tests/data/simple_cases/slices.py b/tests/data/cases/slices.py
similarity index 100%
rename from tests/data/simple_cases/slices.py
rename to tests/data/cases/slices.py
diff --git a/tests/data/py_310/starred_for_target.py b/tests/data/cases/starred_for_target.py
similarity index 92%
rename from tests/data/py_310/starred_for_target.py
rename to tests/data/cases/starred_for_target.py
index 8fc8e059ed3..13e517816d6 100644
--- a/tests/data/py_310/starred_for_target.py
+++ b/tests/data/cases/starred_for_target.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.10
 for x in *a, *b:
     print(x)
 
diff --git a/tests/data/simple_cases/string_prefixes.py b/tests/data/cases/string_prefixes.py
similarity index 100%
rename from tests/data/simple_cases/string_prefixes.py
rename to tests/data/cases/string_prefixes.py
diff --git a/tests/data/miscellaneous/stub.pyi b/tests/data/cases/stub.py
similarity index 99%
rename from tests/data/miscellaneous/stub.pyi
rename to tests/data/cases/stub.py
index af2cd2c2c02..f3828d55ba2 100644
--- a/tests/data/miscellaneous/stub.pyi
+++ b/tests/data/cases/stub.py
@@ -1,3 +1,4 @@
+# flags: --pyi
 X: int
 
 def f(): ...
diff --git a/tests/data/simple_cases/torture.py b/tests/data/cases/torture.py
similarity index 100%
rename from tests/data/simple_cases/torture.py
rename to tests/data/cases/torture.py
diff --git a/tests/data/simple_cases/trailing_comma_optional_parens1.py b/tests/data/cases/trailing_comma_optional_parens1.py
similarity index 100%
rename from tests/data/simple_cases/trailing_comma_optional_parens1.py
rename to tests/data/cases/trailing_comma_optional_parens1.py
diff --git a/tests/data/simple_cases/trailing_comma_optional_parens2.py b/tests/data/cases/trailing_comma_optional_parens2.py
similarity index 100%
rename from tests/data/simple_cases/trailing_comma_optional_parens2.py
rename to tests/data/cases/trailing_comma_optional_parens2.py
diff --git a/tests/data/simple_cases/trailing_comma_optional_parens3.py b/tests/data/cases/trailing_comma_optional_parens3.py
similarity index 100%
rename from tests/data/simple_cases/trailing_comma_optional_parens3.py
rename to tests/data/cases/trailing_comma_optional_parens3.py
diff --git a/tests/data/simple_cases/trailing_commas_in_leading_parts.py b/tests/data/cases/trailing_commas_in_leading_parts.py
similarity index 100%
rename from tests/data/simple_cases/trailing_commas_in_leading_parts.py
rename to tests/data/cases/trailing_commas_in_leading_parts.py
diff --git a/tests/data/simple_cases/tricky_unicode_symbols.py b/tests/data/cases/tricky_unicode_symbols.py
similarity index 100%
rename from tests/data/simple_cases/tricky_unicode_symbols.py
rename to tests/data/cases/tricky_unicode_symbols.py
diff --git a/tests/data/simple_cases/tupleassign.py b/tests/data/cases/tupleassign.py
similarity index 100%
rename from tests/data/simple_cases/tupleassign.py
rename to tests/data/cases/tupleassign.py
diff --git a/tests/data/cases/type_aliases.py b/tests/data/cases/type_aliases.py
new file mode 100644
index 00000000000..7c2009e8202
--- /dev/null
+++ b/tests/data/cases/type_aliases.py
@@ -0,0 +1,30 @@
+# flags: --minimum-version=3.12
+
+type A=int
+type Gen[T]=list[T]
+type Alias[T]=lambda: T
+type And[T]=T and T
+type IfElse[T]=T if T else T
+type One = int; type Another = str
+class X: type InClass = int
+
+type = aliased
+print(type(42))
+
+# output
+
+type A = int
+type Gen[T] = list[T]
+type Alias[T] = lambda: T
+type And[T] = T and T
+type IfElse[T] = T if T else T
+type One = int
+type Another = str
+
+
+class X:
+    type InClass = int
+
+
+type = aliased
+print(type(42))
diff --git a/tests/data/type_comments/type_comment_syntax_error.py b/tests/data/cases/type_comment_syntax_error.py
similarity index 100%
rename from tests/data/type_comments/type_comment_syntax_error.py
rename to tests/data/cases/type_comment_syntax_error.py
diff --git a/tests/data/py_312/type_params.py b/tests/data/cases/type_params.py
similarity index 97%
rename from tests/data/py_312/type_params.py
rename to tests/data/cases/type_params.py
index 5f8ec43267c..720a775ef31 100644
--- a/tests/data/py_312/type_params.py
+++ b/tests/data/cases/type_params.py
@@ -1,3 +1,4 @@
+# flags: --minimum-version=3.12
 def func  [T ](): pass
 async def func [ T ] (): pass
 class C[ T ] : pass
diff --git a/tests/data/simple_cases/whitespace.py b/tests/data/cases/whitespace.py
similarity index 100%
rename from tests/data/simple_cases/whitespace.py
rename to tests/data/cases/whitespace.py
diff --git a/tests/data/miscellaneous/force_pyi.py b/tests/data/miscellaneous/force_pyi.py
index 07ed93c6879..40caf30a983 100644
--- a/tests/data/miscellaneous/force_pyi.py
+++ b/tests/data/miscellaneous/force_pyi.py
@@ -1,3 +1,4 @@
+# flags: --pyi
 from typing import Union
 
 @bird
diff --git a/tests/data/miscellaneous/string_quotes.py b/tests/data/miscellaneous/string_quotes.py
index 3384241f4ad..6ec088ac79b 100644
--- a/tests/data/miscellaneous/string_quotes.py
+++ b/tests/data/miscellaneous/string_quotes.py
@@ -1,4 +1,5 @@
 ''''''
+
 '\''
 '"'
 "'"
@@ -59,6 +60,7 @@
 # output
 
 """"""
+
 "'"
 '"'
 "'"
diff --git a/tests/data/preview/return_annotation_brackets_string.py b/tests/data/preview/return_annotation_brackets_string.py
deleted file mode 100644
index 6978829fd5c..00000000000
--- a/tests/data/preview/return_annotation_brackets_string.py
+++ /dev/null
@@ -1,12 +0,0 @@
-# Long string example
-def frobnicate() -> "ThisIsTrulyUnreasonablyExtremelyLongClassName | list[ThisIsTrulyUnreasonablyExtremelyLongClassName]":
-    pass
-
-# output
-
-# Long string example
-def frobnicate() -> (
-    "ThisIsTrulyUnreasonablyExtremelyLongClassName |"
-    " list[ThisIsTrulyUnreasonablyExtremelyLongClassName]"
-):
-    pass
diff --git a/tests/data/py_312/type_aliases.py b/tests/data/py_312/type_aliases.py
deleted file mode 100644
index 84e07e50fe2..00000000000
--- a/tests/data/py_312/type_aliases.py
+++ /dev/null
@@ -1,13 +0,0 @@
-type A=int
-type Gen[T]=list[T]
-
-type = aliased
-print(type(42))
-
-# output
-
-type A = int
-type Gen[T] = list[T]
-
-type = aliased
-print(type(42))
diff --git a/tests/data/raw_docstring.py b/tests/data/raw_docstring.py
new file mode 100644
index 00000000000..751fd3201df
--- /dev/null
+++ b/tests/data/raw_docstring.py
@@ -0,0 +1,32 @@
+# flags: --preview --skip-string-normalization
+class C:
+
+    r"""Raw"""
+
+def f():
+
+    r"""Raw"""
+
+class SingleQuotes:
+
+
+    r'''Raw'''
+
+class UpperCaseR:
+    R"""Raw"""
+
+# output
+class C:
+    r"""Raw"""
+
+
+def f():
+    r"""Raw"""
+
+
+class SingleQuotes:
+    r'''Raw'''
+
+
+class UpperCaseR:
+    R"""Raw"""
diff --git a/tests/optional.py b/tests/optional.py
index 8a39cc440a6..3f5277b6b03 100644
--- a/tests/optional.py
+++ b/tests/optional.py
@@ -26,7 +26,7 @@
     from pytest import StashKey
 except ImportError:
     # pytest < 7
-    from _pytest.store import StoreKey as StashKey  # type: ignore[no-redef]
+    from _pytest.store import StoreKey as StashKey  # type: ignore[import, no-redef]
 
 log = logging.getLogger(__name__)
 
diff --git a/tests/test_black.py b/tests/test_black.py
index d22b6859607..537ca80d432 100644
--- a/tests/test_black.py
+++ b/tests/test_black.py
@@ -9,7 +9,6 @@
 import re
 import sys
 import types
-import unittest
 from concurrent.futures import ThreadPoolExecutor
 from contextlib import contextmanager, redirect_stderr
 from dataclasses import replace
@@ -188,7 +187,9 @@ def test_experimental_string_processing_warns(self) -> None:
         )
 
     def test_piping(self) -> None:
-        source, expected = read_data_from_file(PROJECT_ROOT / "src/black/__init__.py")
+        _, source, expected = read_data_from_file(
+            PROJECT_ROOT / "src/black/__init__.py"
+        )
         result = BlackRunner().invoke(
             black.main,
             [
@@ -210,8 +211,8 @@ def test_piping_diff(self) -> None:
             r"(STDIN|STDOUT)\t\d\d\d\d-\d\d-\d\d \d\d:\d\d:\d\d\.\d\d\d\d\d\d"
             r"\+\d\d:\d\d"
         )
-        source, _ = read_data("simple_cases", "expression.py")
-        expected, _ = read_data("simple_cases", "expression.diff")
+        source, _ = read_data("cases", "expression.py")
+        expected, _ = read_data("cases", "expression.diff")
         args = [
             "-",
             "--fast",
@@ -228,7 +229,7 @@ def test_piping_diff(self) -> None:
         self.assertEqual(expected, actual)
 
     def test_piping_diff_with_color(self) -> None:
-        source, _ = read_data("simple_cases", "expression.py")
+        source, _ = read_data("cases", "expression.py")
         args = [
             "-",
             "--fast",
@@ -264,7 +265,7 @@ def _test_wip(self) -> None:
         black.assert_stable(source, actual, black.FileMode())
 
     def test_pep_572_version_detection(self) -> None:
-        source, _ = read_data("py_38", "pep_572")
+        source, _ = read_data("cases", "pep_572")
         root = black.lib2to3_parse(source)
         features = black.get_features_used(root)
         self.assertIn(black.Feature.ASSIGNMENT_EXPRESSIONS, features)
@@ -273,7 +274,7 @@ def test_pep_572_version_detection(self) -> None:
 
     def test_pep_695_version_detection(self) -> None:
         for file in ("type_aliases", "type_params"):
-            source, _ = read_data("py_312", file)
+            source, _ = read_data("cases", file)
             root = black.lib2to3_parse(source)
             features = black.get_features_used(root)
             self.assertIn(black.Feature.TYPE_PARAMS, features)
@@ -281,7 +282,7 @@ def test_pep_695_version_detection(self) -> None:
             self.assertIn(black.TargetVersion.PY312, versions)
 
     def test_expression_ff(self) -> None:
-        source, expected = read_data("simple_cases", "expression.py")
+        source, expected = read_data("cases", "expression.py")
         tmp_file = Path(black.dump_to_file(source))
         try:
             self.assertTrue(ff(tmp_file, write_back=black.WriteBack.YES))
@@ -294,8 +295,8 @@ def test_expression_ff(self) -> None:
             black.assert_stable(source, actual, DEFAULT_MODE)
 
     def test_expression_diff(self) -> None:
-        source, _ = read_data("simple_cases", "expression.py")
-        expected, _ = read_data("simple_cases", "expression.diff")
+        source, _ = read_data("cases", "expression.py")
+        expected, _ = read_data("cases", "expression.diff")
         tmp_file = Path(black.dump_to_file(source))
         diff_header = re.compile(
             rf"{re.escape(str(tmp_file))}\t\d\d\d\d-\d\d-\d\d "
@@ -320,8 +321,8 @@ def test_expression_diff(self) -> None:
             self.assertEqual(expected, actual, msg)
 
     def test_expression_diff_with_color(self) -> None:
-        source, _ = read_data("simple_cases", "expression.py")
-        expected, _ = read_data("simple_cases", "expression.diff")
+        source, _ = read_data("cases", "expression.py")
+        expected, _ = read_data("cases", "expression.diff")
         tmp_file = Path(black.dump_to_file(source))
         try:
             result = BlackRunner().invoke(
@@ -340,7 +341,7 @@ def test_expression_diff_with_color(self) -> None:
         self.assertIn("\033[0m", actual)
 
     def test_detect_pos_only_arguments(self) -> None:
-        source, _ = read_data("py_38", "pep_570")
+        source, _ = read_data("cases", "pep_570")
         root = black.lib2to3_parse(source)
         features = black.get_features_used(root)
         self.assertIn(black.Feature.POS_ONLY_ARGUMENTS, features)
@@ -402,7 +403,7 @@ def test_skip_source_first_line_when_mixing_newlines(self) -> None:
             self.assertEqual(test_file.read_bytes(), expected)
 
     def test_skip_magic_trailing_comma(self) -> None:
-        source, _ = read_data("simple_cases", "expression")
+        source, _ = read_data("cases", "expression")
         expected, _ = read_data(
             "miscellaneous", "expression_skip_magic_trailing_comma.diff"
         )
@@ -434,7 +435,7 @@ def test_skip_magic_trailing_comma(self) -> None:
     @patch("black.dump_to_file", dump_to_stderr)
     def test_async_as_identifier(self) -> None:
         source_path = get_case_path("miscellaneous", "async_as_identifier")
-        source, expected = read_data_from_file(source_path)
+        _, source, expected = read_data_from_file(source_path)
         actual = fs(source)
         self.assertFormatEqual(expected, actual)
         major, minor = sys.version_info[:2]
@@ -448,8 +449,8 @@ def test_async_as_identifier(self) -> None:
 
     @patch("black.dump_to_file", dump_to_stderr)
     def test_python37(self) -> None:
-        source_path = get_case_path("py_37", "python37")
-        source, expected = read_data_from_file(source_path)
+        source_path = get_case_path("cases", "python37")
+        _, source, expected = read_data_from_file(source_path)
         actual = fs(source)
         self.assertFormatEqual(expected, actual)
         major, minor = sys.version_info[:2]
@@ -885,7 +886,7 @@ def test_get_features_used(self) -> None:
         self.assertEqual(black.get_features_used(node), {Feature.NUMERIC_UNDERSCORES})
         node = black.lib2to3_parse("123456\n")
         self.assertEqual(black.get_features_used(node), set())
-        source, expected = read_data("simple_cases", "function")
+        source, expected = read_data("cases", "function")
         node = black.lib2to3_parse(source)
         expected_features = {
             Feature.TRAILING_COMMA_IN_CALL,
@@ -895,7 +896,7 @@ def test_get_features_used(self) -> None:
         self.assertEqual(black.get_features_used(node), expected_features)
         node = black.lib2to3_parse(expected)
         self.assertEqual(black.get_features_used(node), expected_features)
-        source, expected = read_data("simple_cases", "expression")
+        source, expected = read_data("cases", "expression")
         node = black.lib2to3_parse(source)
         self.assertEqual(black.get_features_used(node), set())
         node = black.lib2to3_parse(expected)
@@ -1047,9 +1048,10 @@ def test_endmarker(self) -> None:
         self.assertEqual(len(n.children), 1)
         self.assertEqual(n.children[0].type, black.token.ENDMARKER)
 
+    @patch("tests.conftest.PRINT_FULL_TREE", True)
+    @patch("tests.conftest.PRINT_TREE_DIFF", False)
     @pytest.mark.incompatible_with_mypyc
-    @unittest.skipIf(os.environ.get("SKIP_AST_PRINT"), "user set SKIP_AST_PRINT")
-    def test_assertFormatEqual(self) -> None:
+    def test_assertFormatEqual_print_full_tree(self) -> None:
         out_lines = []
         err_lines = []
 
@@ -1068,6 +1070,29 @@ def err(msg: str, **kwargs: Any) -> None:
         self.assertIn("Actual tree:", out_str)
         self.assertEqual("".join(err_lines), "")
 
+    @patch("tests.conftest.PRINT_FULL_TREE", False)
+    @patch("tests.conftest.PRINT_TREE_DIFF", True)
+    @pytest.mark.incompatible_with_mypyc
+    def test_assertFormatEqual_print_tree_diff(self) -> None:
+        out_lines = []
+        err_lines = []
+
+        def out(msg: str, **kwargs: Any) -> None:
+            out_lines.append(msg)
+
+        def err(msg: str, **kwargs: Any) -> None:
+            err_lines.append(msg)
+
+        with patch("black.output._out", out), patch("black.output._err", err):
+            with self.assertRaises(AssertionError):
+                self.assertFormatEqual("j = [1, 2, 3]\n", "j = [1, 2, 3,]\n")
+
+        out_str = "".join(out_lines)
+        self.assertIn("Tree Diff:", out_str)
+        self.assertIn("+          COMMA", out_str)
+        self.assertIn("+ ','", out_str)
+        self.assertEqual("".join(err_lines), "")
+
     @event_loop()
     @patch("concurrent.futures.ProcessPoolExecutor", MagicMock(side_effect=OSError))
     def test_works_in_mono_process_only_environment(self) -> None:
@@ -1086,7 +1111,7 @@ def test_check_diff_use_together(self) -> None:
             src1 = get_case_path("miscellaneous", "string_quotes")
             self.invokeBlack([str(src1), "--diff", "--check"], exit_code=1)
             # Files which will not be reformatted.
-            src2 = get_case_path("simple_cases", "composition")
+            src2 = get_case_path("cases", "composition")
             self.invokeBlack([str(src2), "--diff", "--check"])
             # Multi file command.
             self.invokeBlack([str(src1), str(src2), "--diff", "--check"], exit_code=1)
@@ -1307,7 +1332,7 @@ def test_reformat_one_with_stdin_and_existing_path(self) -> None:
             report = MagicMock()
             # Even with an existing file, since we are forcing stdin, black
             # should output to stdout and not modify the file inplace
-            p = THIS_DIR / "data" / "simple_cases" / "collections.py"
+            p = THIS_DIR / "data" / "cases" / "collections.py"
             # Make sure is_file actually returns True
             self.assertTrue(p.is_file())
             path = Path(f"__BLACK_STDIN_FILENAME__{p}")
@@ -1938,11 +1963,11 @@ def test_get_cache_dir(
         # If BLACK_CACHE_DIR is not set, use user_cache_dir
         monkeypatch.delenv("BLACK_CACHE_DIR", raising=False)
         with patch_user_cache_dir:
-            assert get_cache_dir() == workspace1
+            assert get_cache_dir().parent == workspace1
 
         # If it is set, use the path provided in the env var.
         monkeypatch.setenv("BLACK_CACHE_DIR", str(workspace2))
-        assert get_cache_dir() == workspace2
+        assert get_cache_dir().parent == workspace2
 
     def test_cache_broken_file(self) -> None:
         mode = DEFAULT_MODE
diff --git a/tests/test_blackd.py b/tests/test_blackd.py
index dd2126e6bc2..59703036dc0 100644
--- a/tests/test_blackd.py
+++ b/tests/test_blackd.py
@@ -31,7 +31,7 @@ def unittest_run_loop(func, *args, **kwargs):
 
 
 @pytest.mark.blackd
-class BlackDTestCase(AioHTTPTestCase):  # type: ignore[misc]
+class BlackDTestCase(AioHTTPTestCase):
     def test_blackd_main(self) -> None:
         with patch("blackd.web.run_app"):
             result = CliRunner().invoke(blackd.main, [])
@@ -104,7 +104,7 @@ async def check(header_value: str, expected_status: int = 400) -> None:
 
     @unittest_run_loop
     async def test_blackd_pyi(self) -> None:
-        source, expected = read_data("miscellaneous", "stub.pyi")
+        source, expected = read_data("cases", "stub.py")
         response = await self.client.post(
             "/", data=source, headers={blackd.PYTHON_VARIANT_HEADER: "pyi"}
         )
diff --git a/tests/test_format.py b/tests/test_format.py
index f3db423b637..4e863c6c54b 100644
--- a/tests/test_format.py
+++ b/tests/test_format.py
@@ -1,4 +1,3 @@
-import re
 from dataclasses import replace
 from typing import Any, Iterator
 from unittest.mock import patch
@@ -6,13 +5,13 @@
 import pytest
 
 import black
+from black.mode import TargetVersion
 from tests.util import (
-    DEFAULT_MODE,
-    PY36_VERSIONS,
     all_data_cases,
     assert_format,
     dump_to_stderr,
     read_data,
+    read_data_with_mode,
 )
 
 
@@ -22,61 +21,33 @@ def patch_dump_to_file(request: Any) -> Iterator[None]:
         yield
 
 
-def check_file(
-    subdir: str, filename: str, mode: black.Mode, *, data: bool = True
-) -> None:
-    source, expected = read_data(subdir, filename, data=data)
-    assert_format(source, expected, mode, fast=False)
+def check_file(subdir: str, filename: str, *, data: bool = True) -> None:
+    args, source, expected = read_data_with_mode(subdir, filename, data=data)
+    assert_format(
+        source,
+        expected,
+        args.mode,
+        fast=args.fast,
+        minimum_version=args.minimum_version,
+    )
+    if args.minimum_version is not None:
+        major, minor = args.minimum_version
+        target_version = TargetVersion[f"PY{major}{minor}"]
+        mode = replace(args.mode, target_versions={target_version})
+        assert_format(
+            source, expected, mode, fast=args.fast, minimum_version=args.minimum_version
+        )
 
 
 @pytest.mark.filterwarnings("ignore:invalid escape sequence.*:DeprecationWarning")
-@pytest.mark.parametrize("filename", all_data_cases("simple_cases"))
+@pytest.mark.parametrize("filename", all_data_cases("cases"))
 def test_simple_format(filename: str) -> None:
-    magic_trailing_comma = filename != "skip_magic_trailing_comma"
-    mode = black.Mode(
-        magic_trailing_comma=magic_trailing_comma, is_pyi=filename.endswith("_pyi")
-    )
-    check_file("simple_cases", filename, mode)
-
-
-@pytest.mark.parametrize("filename", all_data_cases("preview"))
-def test_preview_format(filename: str) -> None:
-    check_file("preview", filename, black.Mode(preview=True))
-
-
-def test_preview_context_managers_targeting_py38() -> None:
-    source, expected = read_data("preview_context_managers", "targeting_py38.py")
-    mode = black.Mode(preview=True, target_versions={black.TargetVersion.PY38})
-    assert_format(source, expected, mode, minimum_version=(3, 8))
-
-
-def test_preview_context_managers_targeting_py39() -> None:
-    source, expected = read_data("preview_context_managers", "targeting_py39.py")
-    mode = black.Mode(preview=True, target_versions={black.TargetVersion.PY39})
-    assert_format(source, expected, mode, minimum_version=(3, 9))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("preview_py_310"))
-def test_preview_python_310(filename: str) -> None:
-    source, expected = read_data("preview_py_310", filename)
-    mode = black.Mode(target_versions={black.TargetVersion.PY310}, preview=True)
-    assert_format(source, expected, mode, minimum_version=(3, 10))
-
-
-@pytest.mark.parametrize(
-    "filename", all_data_cases("preview_context_managers/auto_detect")
-)
-def test_preview_context_managers_auto_detect(filename: str) -> None:
-    match = re.match(r"features_3_(\d+)", filename)
-    assert match is not None, "Unexpected filename format: %s" % filename
-    source, expected = read_data("preview_context_managers/auto_detect", filename)
-    mode = black.Mode(preview=True)
-    assert_format(source, expected, mode, minimum_version=(3, int(match.group(1))))
+    check_file("cases", filename)
 
 
 # =============== #
-# Complex cases
-# ============= #
+# Unusual cases
+# =============== #
 
 
 def test_empty() -> None:
@@ -84,48 +55,6 @@ def test_empty() -> None:
     assert_format(source, expected)
 
 
-@pytest.mark.parametrize("filename", all_data_cases("py_36"))
-def test_python_36(filename: str) -> None:
-    source, expected = read_data("py_36", filename)
-    mode = black.Mode(target_versions=PY36_VERSIONS)
-    assert_format(source, expected, mode, minimum_version=(3, 6))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("py_37"))
-def test_python_37(filename: str) -> None:
-    source, expected = read_data("py_37", filename)
-    mode = black.Mode(target_versions={black.TargetVersion.PY37})
-    assert_format(source, expected, mode, minimum_version=(3, 7))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("py_38"))
-def test_python_38(filename: str) -> None:
-    source, expected = read_data("py_38", filename)
-    mode = black.Mode(target_versions={black.TargetVersion.PY38})
-    assert_format(source, expected, mode, minimum_version=(3, 8))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("py_39"))
-def test_python_39(filename: str) -> None:
-    source, expected = read_data("py_39", filename)
-    mode = black.Mode(target_versions={black.TargetVersion.PY39})
-    assert_format(source, expected, mode, minimum_version=(3, 9))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("py_310"))
-def test_python_310(filename: str) -> None:
-    source, expected = read_data("py_310", filename)
-    mode = black.Mode(target_versions={black.TargetVersion.PY310})
-    assert_format(source, expected, mode, minimum_version=(3, 10))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("py_310"))
-def test_python_310_without_target_version(filename: str) -> None:
-    source, expected = read_data("py_310", filename)
-    mode = black.Mode()
-    assert_format(source, expected, mode, minimum_version=(3, 10))
-
-
 def test_patma_invalid() -> None:
     source, expected = read_data("miscellaneous", "pattern_matching_invalid")
     mode = black.Mode(target_versions={black.TargetVersion.PY310})
@@ -133,88 +62,3 @@ def test_patma_invalid() -> None:
         assert_format(source, expected, mode, minimum_version=(3, 10))
 
     exc_info.match("Cannot parse: 10:11")
-
-
-@pytest.mark.parametrize("filename", all_data_cases("py_311"))
-def test_python_311(filename: str) -> None:
-    source, expected = read_data("py_311", filename)
-    mode = black.Mode(target_versions={black.TargetVersion.PY311})
-    assert_format(source, expected, mode, minimum_version=(3, 11))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("py_312"))
-def test_python_312(filename: str) -> None:
-    source, expected = read_data("py_312", filename)
-    mode = black.Mode(target_versions={black.TargetVersion.PY312})
-    assert_format(source, expected, mode, minimum_version=(3, 12))
-
-
-@pytest.mark.parametrize("filename", all_data_cases("fast"))
-def test_fast_cases(filename: str) -> None:
-    source, expected = read_data("fast", filename)
-    assert_format(source, expected, fast=True)
-
-
-def test_python_2_hint() -> None:
-    with pytest.raises(black.parsing.InvalidInput) as exc_info:
-        assert_format("print 'daylily'", "print 'daylily'")
-    exc_info.match(black.parsing.PY2_HINT)
-
-
-@pytest.mark.filterwarnings("ignore:invalid escape sequence.*:DeprecationWarning")
-def test_docstring_no_string_normalization() -> None:
-    """Like test_docstring but with string normalization off."""
-    source, expected = read_data("miscellaneous", "docstring_no_string_normalization")
-    mode = replace(DEFAULT_MODE, string_normalization=False)
-    assert_format(source, expected, mode)
-
-
-def test_docstring_line_length_6() -> None:
-    """Like test_docstring but with line length set to 6."""
-    source, expected = read_data("miscellaneous", "linelength6")
-    mode = black.Mode(line_length=6)
-    assert_format(source, expected, mode)
-
-
-def test_preview_docstring_no_string_normalization() -> None:
-    """
-    Like test_docstring but with string normalization off *and* the preview style
-    enabled.
-    """
-    source, expected = read_data(
-        "miscellaneous", "docstring_preview_no_string_normalization"
-    )
-    mode = replace(DEFAULT_MODE, string_normalization=False, preview=True)
-    assert_format(source, expected, mode)
-
-
-def test_long_strings_flag_disabled() -> None:
-    """Tests for turning off the string processing logic."""
-    source, expected = read_data("miscellaneous", "long_strings_flag_disabled")
-    mode = replace(DEFAULT_MODE, experimental_string_processing=False)
-    assert_format(source, expected, mode)
-
-
-def test_stub() -> None:
-    mode = replace(DEFAULT_MODE, is_pyi=True)
-    source, expected = read_data("miscellaneous", "stub.pyi")
-    assert_format(source, expected, mode)
-
-
-def test_nested_stub() -> None:
-    mode = replace(DEFAULT_MODE, is_pyi=True, preview=True)
-    source, expected = read_data("miscellaneous", "nested_stub.pyi")
-    assert_format(source, expected, mode)
-
-
-def test_power_op_newline() -> None:
-    # requires line_length=0
-    source, expected = read_data("miscellaneous", "power_op_newline")
-    assert_format(source, expected, mode=black.Mode(line_length=0))
-
-
-def test_type_comment_syntax_error() -> None:
-    """Test that black is able to format python code with type comment syntax errors."""
-    source, expected = read_data("type_comments", "type_comment_syntax_error")
-    assert_format(source, expected)
-    black.assert_equivalent(source, expected)
diff --git a/tests/util.py b/tests/util.py
index 967d576fafe..a31ae0992c2 100644
--- a/tests/util.py
+++ b/tests/util.py
@@ -1,17 +1,23 @@
+import argparse
+import functools
 import os
+import shlex
 import sys
 import unittest
 from contextlib import contextmanager
-from dataclasses import replace
+from dataclasses import dataclass, field, replace
 from functools import partial
 from pathlib import Path
 from typing import Any, Iterator, List, Optional, Tuple
 
 import black
+from black.const import DEFAULT_LINE_LENGTH
 from black.debug import DebugVisitor
 from black.mode import TargetVersion
 from black.output import diff, err, out
 
+from . import conftest
+
 PYTHON_SUFFIX = ".py"
 ALLOWED_SUFFIXES = (PYTHON_SUFFIX, ".pyi", ".out", ".diff", ".ipynb")
 
@@ -33,23 +39,42 @@
 fs = partial(black.format_str, mode=DEFAULT_MODE)
 
 
+@dataclass
+class TestCaseArgs:
+    mode: black.Mode = field(default_factory=black.Mode)
+    fast: bool = False
+    minimum_version: Optional[Tuple[int, int]] = None
+
+
 def _assert_format_equal(expected: str, actual: str) -> None:
-    if actual != expected and not os.environ.get("SKIP_AST_PRINT"):
+    if actual != expected and (conftest.PRINT_FULL_TREE or conftest.PRINT_TREE_DIFF):
         bdv: DebugVisitor[Any]
-        out("Expected tree:", fg="green")
+        actual_out: str = ""
+        expected_out: str = ""
+        if conftest.PRINT_FULL_TREE:
+            out("Expected tree:", fg="green")
         try:
             exp_node = black.lib2to3_parse(expected)
-            bdv = DebugVisitor()
+            bdv = DebugVisitor(print_output=conftest.PRINT_FULL_TREE)
             list(bdv.visit(exp_node))
+            expected_out = "\n".join(bdv.list_output)
         except Exception as ve:
             err(str(ve))
-        out("Actual tree:", fg="red")
+        if conftest.PRINT_FULL_TREE:
+            out("Actual tree:", fg="red")
         try:
             exp_node = black.lib2to3_parse(actual)
-            bdv = DebugVisitor()
+            bdv = DebugVisitor(print_output=conftest.PRINT_FULL_TREE)
             list(bdv.visit(exp_node))
+            actual_out = "\n".join(bdv.list_output)
         except Exception as ve:
             err(str(ve))
+        if conftest.PRINT_TREE_DIFF:
+            out("Tree Diff:")
+            out(
+                diff(expected_out, actual_out, "expected tree", "actual tree")
+                or "Trees do not differ"
+            )
 
     if actual != expected:
         out(diff(expected, actual, "expected", "actual"))
@@ -164,18 +189,85 @@ def get_case_path(
     return case_path
 
 
+def read_data_with_mode(
+    subdir_name: str, name: str, data: bool = True
+) -> Tuple[TestCaseArgs, str, str]:
+    """read_data_with_mode('test_name') -> Mode(), 'input', 'output'"""
+    return read_data_from_file(get_case_path(subdir_name, name, data))
+
+
 def read_data(subdir_name: str, name: str, data: bool = True) -> Tuple[str, str]:
     """read_data('test_name') -> 'input', 'output'"""
-    return read_data_from_file(get_case_path(subdir_name, name, data))
+    _, input, output = read_data_with_mode(subdir_name, name, data)
+    return input, output
+
+
+def _parse_minimum_version(version: str) -> Tuple[int, int]:
+    major, minor = version.split(".")
+    return int(major), int(minor)
 
 
-def read_data_from_file(file_name: Path) -> Tuple[str, str]:
+@functools.lru_cache()
+def get_flags_parser() -> argparse.ArgumentParser:
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--target-version",
+        action="append",
+        type=lambda val: TargetVersion[val.upper()],
+        default=(),
+    )
+    parser.add_argument("--line-length", default=DEFAULT_LINE_LENGTH, type=int)
+    parser.add_argument(
+        "--skip-string-normalization", default=False, action="store_true"
+    )
+    parser.add_argument("--pyi", default=False, action="store_true")
+    parser.add_argument("--ipynb", default=False, action="store_true")
+    parser.add_argument(
+        "--skip-magic-trailing-comma", default=False, action="store_true"
+    )
+    parser.add_argument("--preview", default=False, action="store_true")
+    parser.add_argument("--fast", default=False, action="store_true")
+    parser.add_argument(
+        "--minimum-version",
+        type=_parse_minimum_version,
+        default=None,
+        help=(
+            "Minimum version of Python where this test case is parseable. If this is"
+            " set, the test case will be run twice: once with the specified"
+            " --target-version, and once with --target-version set to exactly the"
+            " specified version. This ensures that Black's autodetection of the target"
+            " version works correctly."
+        ),
+    )
+    return parser
+
+
+def parse_mode(flags_line: str) -> TestCaseArgs:
+    parser = get_flags_parser()
+    args = parser.parse_args(shlex.split(flags_line))
+    mode = black.Mode(
+        target_versions=set(args.target_version),
+        line_length=args.line_length,
+        string_normalization=not args.skip_string_normalization,
+        is_pyi=args.pyi,
+        is_ipynb=args.ipynb,
+        magic_trailing_comma=not args.skip_magic_trailing_comma,
+        preview=args.preview,
+    )
+    return TestCaseArgs(mode=mode, fast=args.fast, minimum_version=args.minimum_version)
+
+
+def read_data_from_file(file_name: Path) -> Tuple[TestCaseArgs, str, str]:
     with open(file_name, "r", encoding="utf8") as test:
         lines = test.readlines()
     _input: List[str] = []
     _output: List[str] = []
     result = _input
+    mode = TestCaseArgs()
     for line in lines:
+        if not _input and line.startswith("# flags: "):
+            mode = parse_mode(line[len("# flags: ") :])
+            continue
         line = line.replace(EMPTY_LINE, "")
         if line.rstrip() == "# output":
             result = _output
@@ -185,7 +277,7 @@ def read_data_from_file(file_name: Path) -> Tuple[str, str]:
     if _input and not _output:
         # If there's no output marker, treat the entire file as already pre-formatted.
         _output = _input[:]
-    return "".join(_input).strip() + "\n", "".join(_output).strip() + "\n"
+    return mode, "".join(_input).strip() + "\n", "".join(_output).strip() + "\n"
 
 
 def read_jupyter_notebook(subdir_name: str, name: str, data: bool = True) -> str:
diff --git a/tox.ini b/tox.ini
index d34dbbc71db..018cef993c0 100644
--- a/tox.ini
+++ b/tox.ini
@@ -1,6 +1,6 @@
 [tox]
 isolated_build = true
-envlist = {,ci-}py{37,38,39,310,311,py3},fuzz,run_self
+envlist = {,ci-}py{38,39,310,311,py3},fuzz,run_self
 
 [testenv]
 setenv =