PRESUBMIT.py

# Copyright 2022 The Dawn & Tint Authors
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# 1. Redistributions of source code must retain the above copyright notice, this
#    list of conditions and the following disclaimer.
#
# 2. Redistributions in binary form must reproduce the above copyright notice,
#    this list of conditions and the following disclaimer in the documentation
#    and/or other materials provided with the distribution.
#
# 3. Neither the name of the copyright holder nor the names of its
#    contributors may be used to endorse or promote products derived from
#    this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE
# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

import hashlib
import dataclasses
import pathlib
import re
import sys
from typing import Optional
from typing import Sequence
from typing import Tuple

PRESUBMIT_VERSION = '2.0.0'

NONINCLUSIVE_LANGUAGE_REGEXES = [
    re.compile(reg) for reg in [
        r"(?i)black[-_]?list",
        r"(?i)white[-_]?list",
        r"(?i)gr[ea]y[-_]?list",
        r"(?i)(first class citizen)",
        r"(?i)black[-_]?hat",
        r"(?i)white[-_]?hat",
        r"(?i)gr[ea]y[-_]?hat",
        r"(?i)master",
        r"(?i)slave",
        r"(?i)\bhim\b",
        r"(?i)\bhis\b",
        r"(?i)\bshe\b",
        r"(?i)\bher\b",
        r"(?i)\bguys\b",
        r"(?i)\bhers\b",
        r"(?i)\bman\b",
        r"(?i)\bwoman\b",
        r"(?i)\she\s",
        r"(?i)\she$",
        r"(?i)^he\s",
        r"(?i)^he$",
        r"(?i)\she['|\u2019]d\s",
        r"(?i)\she['|\u2019]d$",
        r"(?i)^he['|\u2019]d\s",
        r"(?i)^he['|\u2019]d$",
        r"(?i)\she['|\u2019]s\s",
        r"(?i)\she['|\u2019]s$",
        r"(?i)^he['|\u2019]s\s",
        r"(?i)^he['|\u2019]s$",
        r"(?i)\she['|\u2019]ll\s",
        r"(?i)\she['|\u2019]ll$",
        r"(?i)^he['|\u2019]ll\s",
        r"(?i)^he['|\u2019]ll$",
        r"(?i)grandfather",
        r"(?i)\bmitm\b",
        r"(?i)\bcrazy\b",
        r"(?i)\binsane\b",
        r"(?i)\bblind\sto\b",
        r"(?i)\bflying\sblind\b",
        r"(?i)\bblind\seye\b",
        r"(?i)\bcripple\b",
        r"(?i)\bcrippled\b",
        r"(?i)\bdumb\b",
        r"(?i)\bdummy\b",
        r"(?i)\bparanoid\b",
        r"(?i)\bsane\b",
        r"(?i)\bsanity\b",
        r"(?i)red[-_]?line",
    ]
]

LINT_FILTERS = []


# Copied from Chrome's BanRule.
@dataclasses.dataclass
class BanRule:
    # String pattern. If the pattern begins with a slash, the pattern will be
    # treated as a regular expression instead.
    pattern: str

    # Explanation as a sequence of strings. Each string in the sequence will be
    # printed on its own line.
    explanation: Tuple[str, ...]

    # Whether or not to treat this ban as a fatal error.
    treat_as_error: bool = False

    # Paths that should be excluded from the ban check. Each string is a regular
    # expression that will be matched against the path of the file being checked
    # relative to the root of the source tree.
    excluded_paths: Optional[Sequence[str]] = None

    # If True, surfaces any violation as a Gerrit comment on the CL after
    # running the CQ.
    surface_as_gerrit_lint: Optional[bool] = None


# Configuration for banned patterns checks.
_BANNED_CPP_PATTERNS: Sequence[BanRule] = (
    BanRule(
        pattern=
        r'/\bTINT_BEGIN_DISABLE_WARNING\s*\(\s*(UNSAFE_BUFFER_USAGE|UNSAFE_BUFFER_USAGE_IN_CONTAINER)\s*\)',
        excluded_paths=(
            r'^src/tint/lang/spirv/reader/parser/parser\.cc$',
            r'^src/tint/lang/glsl/validate/validate\.cc$',
        ),
        explanation=(
            'Do not introduce new instances of TINT_BEGIN_DISABLE_WARNING(UNSAFE_BUFFER_USAGE) ',
            'or TINT_BEGIN_DISABLE_WARNING(UNSAFE_BUFFER_USAGE_IN_CONTAINER). ',
            'Use DAWN_UNSAFE_BUFFERS with a // SAFETY: comment instead, ',
            'or rewrite to be safe.',
        ),
        treat_as_error=False,
        surface_as_gerrit_lint=True,
    ),
    BanRule(
        pattern=r'/\bDAWN_UNSAFE_TODO\b',
        explanation=(
            'Do not introduce new instances of DAWN_UNSAFE_TODO. ',
            'Use DAWN_UNSAFE_BUFFERS with a // SAFETY: comment instead, ',
            'or rewrite to be safe.',
        ),
        treat_as_error=False,
        surface_as_gerrit_lint=True,
    ),
    BanRule(
        pattern=r'/#pragma\s+allow_unsafe_buffers\b',
        explanation=(
            '#pragma allow_unsafe_buffers is discouraged. Prefer using ',
            'DAWN_UNSAFE_BUFFERS with a // SAFETY: comment for ',
            'specific blocks, or rewrite to be safe.',
        ),
        treat_as_error=False,
        surface_as_gerrit_lint=True,
    ),
    BanRule(
        pattern=r'__EMSCRIPTEN__\b',
        excluded_paths=(r'^src/utils/platform\.h$', ),
        explanation='Use DAWN_PLATFORM_IS(EMSCRIPTEN) instead where possible.',
        treat_as_error=False,
        surface_as_gerrit_lint=True,
    ),
    BanRule(
        pattern=r'/\b(EXPECT_DEATH|EXPECT_DEBUG_DEATH)\b',
        excluded_paths=(r'^src/utils/gtest\.h$', ),
        explanation=(
            'Use EXPECT_DEATH_IF_SUPPORTED or ',
            'DAWN_EXPECT_DEBUG_DEATH_IF_SUPPORTED instead.',
        ),
        treat_as_error=True,
        surface_as_gerrit_lint=True,
    ),
    BanRule(
        pattern=r'/\b(ASSERT_DEATH|ASSERT_DEBUG_DEATH)\b',
        excluded_paths=(r'^src/utils/gtest\.h$', ),
        explanation=(
            'Use ASSERT_DEATH_IF_SUPPORTED or ',
            'DAWN_ASSERT_DEBUG_DEATH_IF_SUPPORTED instead.',
        ),
        treat_as_error=True,
        surface_as_gerrit_lint=True,
    ),
    BanRule(
        pattern=r'runtime/explicit',
        explanation='Use explicit(false) instead of NOLINT(runtime/explicit).',
        treat_as_error=True,
        surface_as_gerrit_lint=True,
    ),
    BanRule(
        pattern=
        r'/misc-explicit-constructor|cppcoreguidelines-explicit-constructor',
        # We would prefer to use explicit(false) here too, but it doesn't work
        # to suppress the clang-tidy warning on operators.
        # There's no ban rule on google-explicit-constructor because it's hard
        # to implement while allowing it on operators.
        explanation=(
            'Use explicit(false) for constructors, and ',
            'NOLINTNEXTLINE(google-explicit-constructor) for operators.',
        ),
        treat_as_error=True,
        surface_as_gerrit_lint=True,
    ),
)

EXPECTED_LICENSE_TEXT = {
    "//": [
        "//",
        "// Redistribution and use in source and binary forms, with or without",
        "// modification, are permitted provided that the following conditions are met:",
        "//",
        "// 1. Redistributions of source code must retain the above copyright notice, this",
        "//    list of conditions and the following disclaimer.",
        "//",
        "// 2. Redistributions in binary form must reproduce the above copyright notice,",
        "//    this list of conditions and the following disclaimer in the documentation",
        "//    and/or other materials provided with the distribution.",
        "//",
        "// 3. Neither the name of the copyright holder nor the names of its",
        "//    contributors may be used to endorse or promote products derived from",
        "//    this software without specific prior written permission.",
        "//",
        "// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS \"AS IS\"",
        "// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE",
        "// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE",
        "// DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE",
        "// FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL",
        "// DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR",
        "// SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER",
        "// CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,",
        "// OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE",
        "// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.",
    ],
    "//*": [
        "//*",
        "//* Redistribution and use in source and binary forms, with or without",
        "//* modification, are permitted provided that the following conditions are met:",
        "//*",
        "//* 1. Redistributions of source code must retain the above copyright notice, this",
        "//*    list of conditions and the following disclaimer.",
        "//*",
        "//* 2. Redistributions in binary form must reproduce the above copyright notice,",
        "//*    this list of conditions and the following disclaimer in the documentation",
        "//*    and/or other materials provided with the distribution.",
        "//*",
        "//* 3. Neither the name of the copyright holder nor the names of its",
        "//*    contributors may be used to endorse or promote products derived from",
        "//*    this software without specific prior written permission.",
        "//*",
        "//* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS \"AS IS\"",
        "//* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE",
        "//* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE",
        "//* DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE",
        "//* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL",
        "//* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR",
        "//* SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER",
        "//* CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,",
        "//* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE",
        "//* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.",
    ],
    "#": [
        "#",
        "# Redistribution and use in source and binary forms, with or without",
        "# modification, are permitted provided that the following conditions are met:",
        "#",
        "# 1. Redistributions of source code must retain the above copyright notice, this",
        "#    list of conditions and the following disclaimer.",
        "#",
        "# 2. Redistributions in binary form must reproduce the above copyright notice,",
        "#    this list of conditions and the following disclaimer in the documentation",
        "#    and/or other materials provided with the distribution.",
        "#",
        "# 3. Neither the name of the copyright holder nor the names of its",
        "#    contributors may be used to endorse or promote products derived from",
        "#    this software without specific prior written permission.",
        "#",
        "# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS \"AS IS\"",
        "# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE",
        "# IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE",
        "# DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE",
        "# FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL",
        "# DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR",
        "# SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER",
        "# CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,",
        "# OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE",
        "# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.",
    ]
}

# 'go fmt' has slightly different spacing rules relative to our C++
# style, so .go files need their own variant
EXPECTED_LICENSE_TEXT_GO = [
    "//",
    "// Redistribution and use in source and binary forms, with or without",
    "// modification, are permitted provided that the following conditions are met:",
    "//",
    "//  1. Redistributions of source code must retain the above copyright notice, this",
    "//     list of conditions and the following disclaimer.",
    "//",
    "//  2. Redistributions in binary form must reproduce the above copyright notice,",
    "//     this list of conditions and the following disclaimer in the documentation",
    "//     and/or other materials provided with the distribution.",
    "//",
    "//  3. Neither the name of the copyright holder nor the names of its",
    "//     contributors may be used to endorse or promote products derived from",
    "//     this software without specific prior written permission.",
    "//",
    "// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS \"AS IS\"",
    "// AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE",
    "// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE",
    "// DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE",
    "// FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL",
    "// DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR",
    "// SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER",
    "// CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,",
    "// OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE",
    "// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.",
]


def _NonInclusiveFileFilter(file):
    """Filters files that are exempt from the non-inclusive language check."""
    filter_list = [
        "Doxyfile",  # References to main pages
        "PRESUBMIT.py",  # Non-inclusive language check data
        "PRESUBMIT.py.tint",  # Non-inclusive language check data
        "docs/dawn/debug_markers.md",  # External URL
        "docs/dawn/infra.md",  # Infra settings
        "docs/tint/spirv-input-output-variables.md",  # External URL
        "infra/config/global/generated/cr-buildbucket.cfg",  # Infra settings
        "infra/config/global/main.star",  # Infra settings
        "infra/kokoro/windows/build.bat",  # External URL
        "src/dawn/common/GPUInfo.cpp",  # External URL
        "src/dawn/common/ThreadLocal.cpp",  # External URL
        "src/dawn/native/CommandEncoder.cpp",  # External URL
        "src/dawn/native/metal/BackendMTL.mm",  # OSX Constant
        "src/dawn/native/metal/PhysicalDeviceMTL.mm",  # OSX deprecated API
        "src/dawn/native/vulkan/SamplerVk.cpp",  # External URL
        "src/dawn/native/vulkan/TextureVk.cpp",  # External URL
        "src/tools/src/cmd/run-cts/main.go",  # Terminal type name
        "src/dawn/samples/ComputeBoids.cpp",  # External URL
        "src/dawn/tests/end2end/DepthBiasTests.cpp",  # External URL
        "src/tint/transform/canonicalize_entry_point_io.cc",  # External URL
        "test/tint/samples/compute_boids.wgsl",  # External URL
        "third_party/directx-shader-compiler/BUILD.gn",  # Third party file
        "third_party/directx-shader-compiler/LICENSE",  # Third party file
        "third_party/directx-shader-compiler/build/message_compiler.py",  # Third party file
        "third_party/EGL-Registry/src/api/KHR/khrplatform.h",  # Third party file
        "tools/roll-all",  # Branch name
        "tools/src/container/key.go",  # External URL
        "go.sum",  # External URL
    ]
    return file.LocalPath().replace('\\', '/') not in filter_list


def CheckNonInclusiveLanguage(input_api, output_api):
    """Checks the files for non-inclusive language."""
    matches = []
    for f in input_api.AffectedFiles(include_deletes=False,
                                     file_filter=_NonInclusiveFileFilter):
        for line_num, line in enumerate(f.NewContents(), start=1):
            for reg in NONINCLUSIVE_LANGUAGE_REGEXES:
                if match := reg.search(line):
                    matches.append(
                        f"{f.LocalPath()} ({line_num}): found non-inclusive language: {match.group(0)}"
                    )

    if matches:
        return [
            output_api.PresubmitPromptWarning("Non-inclusive language found:",
                                              items=matches)
        ]

    return []


def _CalculateEnumeratedEntriesAndTypes(lines):
    """Returns a dictionary of enumerated entries, and a list of all the 'types' encountered.

    The implemented parsing is unsophisticated, and assumes a readable/well-formed .proto file.
    Things like unmatched '{}'s will cause a crash. Missing ';'s or writing something like `} message Foo {` will also
    cause misbehaviour.
    Constructs like this are normally bad style, so if really needed, adding support for them is left as an exercise for
    the reader.
    """
    push_re = re.compile(r'(\w+) {(.*)')
    value_re = re.compile(r'(\w+) = (\d+);(.*)')
    reserved_re = re.compile(r'^\s*reserved\s+(.*);(.*)')
    number_re = re.compile(r'\d+')
    pop_re = re.compile(r'}(.*)')

    prefix_stack = []
    prefix_str = ""
    enumerated_entries = {}
    oneof_scopes = set()
    types = []
    for l in lines:
        l = l.strip().rstrip()
        l = l.split("//", 1)[0]
        while l:
            if match := re.search(push_re, l):
                if l[:match.start()].strip() == "oneof":
                    oneof_scopes.add('.'.join(prefix_stack + [match.group(1)]))
                prefix_stack.append(match.group(1))
                prefix_str = '.'.join(prefix_stack)
                types.append(prefix_str)
                l = match.group(2)
                continue
            if match := re.search(reserved_re, l):
                new_numbers = number_re.findall(match.group(0))
                reserved_numbers = enumerated_entries.get(
                    f"{prefix_str}.reserved", [])
                reserved_numbers.extend(new_numbers)
                enumerated_entries[f"{prefix_str}.reserved"] = reserved_numbers
                l = match.group(1)
                continue
            if match := re.search(value_re, l):
                enumerated_entries[
                    f"{prefix_str}.{match.group(1)}"] = match.group(2)
                l = match.group(2)
                continue
            if match := re.search(pop_re, l):
                prefix_stack.pop()
                prefix_str = '.'.join(prefix_stack)
                l = match.group(1)
                continue
            l = ""

    return enumerated_entries, types, oneof_scopes


def CheckIRBinaryCompatibility(input_api, output_api):
    """Checks for changes to ir.proto that may cause compatibility issues"""
    proto_file = None
    old_entries, old_types = {}, []
    new_entries, new_types = {}, []
    for file in input_api.AffectedFiles(
            include_deletes=False,
            file_filter=lambda f: f.LocalPath().replace(
                '\\', '/') == "src/tint/utils/protos/ir/ir.proto"):
        if proto_file:
            return [
                output_api.PresubmitError(
                    f"Unexpectedly found more than one ir.proto in change, [{file.AbsoluteLocalPath()}, {proto_file}]"
                )
            ]
        proto_file = file.AbsoluteLocalPath()
        old_entries, old_types, old_oneofs = _CalculateEnumeratedEntriesAndTypes(
            file.OldContents())
        new_entries, new_types, new_oneofs = _CalculateEnumeratedEntriesAndTypes(
            file.NewContents())

    changes = []

    for k in old_entries:
        if k not in new_entries:
            entry_prefix = k.rsplit('.', 1)[0]
            reserved = new_entries.get(f"{entry_prefix}.reserved", [])
            if old_entries[k] in reserved:
                continue

            if entry_prefix in old_oneofs:
                parent_prefix = entry_prefix.rsplit('.', 1)[0]
                reserved = new_entries.get(f"{parent_prefix}.reserved", [])
                if old_entries[k] in reserved:
                    continue

            changes.append(
                f"entry '{k}' has been removed without reserving, old={old_entries[k]}"
            )
            continue

        if old_entries[k] != new_entries[k]:
            changes.append(
                f"entry '{k}' has changed, old={old_entries[k]}, new={new_entries[k]}"
            )
            continue

    for s in old_types:
        if s not in new_types:
            changes.append(f"type '{s}' has been removed")

    if changes:
        return [
            output_api.PresubmitError(
                "Incompatible changes detected in ir.proto", items=changes)
        ]
    return []


def CheckNoStaleGen(input_api, output_api):
    """Checks that Tint generated files are not stale."""
    sys.path += [input_api.change.RepositoryRoot()]

    import go_presubmit_support

    results = []
    try:
        go = go_presubmit_support.go_path(input_api)
        input_api.subprocess.check_call_out(
            [go, "run", "tools/src/cmd/gen/main.go", "--check-stale"],
            stdout=input_api.subprocess.PIPE,
            stderr=input_api.subprocess.PIPE,
            cwd=input_api.change.RepositoryRoot())
    except input_api.subprocess.CalledProcessError as e:
        if input_api.is_committing:
            results.append(output_api.PresubmitError('%s' % (e, )))
        else:
            results.append(output_api.PresubmitPromptWarning('%s' % (e, )))
    return results


def CheckWebgpuHeaderDiff(input_api, output_api):
    """Checks that generated WebGPU C Headers are not stale."""
    results = []
    try:
        input_api.subprocess.check_call_out(
            [sys.executable, "third_party/webgpu-headers/cli", "check"],
            stdout=input_api.subprocess.PIPE,
            stderr=input_api.subprocess.PIPE,
            cwd=input_api.change.RepositoryRoot())
    except input_api.subprocess.CalledProcessError as e:
        if input_api.is_committing:
            results.append(output_api.PresubmitError('%s' % (e, )))
        else:
            results.append(output_api.PresubmitPromptWarning('%s' % (e, )))
    return results


def _HasNoStrayWhitespaceFilter(file):
    """Filters files that are exempt from the canned no stray whitespace check."""
    filter_list = [
        "third_party/webgpu-headers/webgpu.h.diff",  # Generated diff file
    ]
    return file.LocalPath().replace('\\', '/') not in filter_list


def _CheckCopyrightYear(input_api, output_api, f, new_contents_lines,
                        current_year):
    copyright_regex = re.compile(r'Copyright (\d{4})')
    for line in new_contents_lines:
        if match := copyright_regex.search(line):
            year = int(match.group(1))
            if year != current_year:
                return [
                    output_api.PresubmitPromptWarning(
                        f'{f.LocalPath()}: Copyright year is {year}, should be {current_year} as this is a new file.'
                    )
                ]
            return []
    return [
        output_api.PresubmitPromptWarning(
            f'{f.LocalPath()}: No copyright header found.')
    ]


def _CheckCopyrightText(input_api, output_api, f, new_contents_lines):
    # Find the copyright line and check the lines after it.
    copyright_regex = re.compile(r'Copyright (\d{4})')
    for i, line in enumerate(new_contents_lines):
        if match := copyright_regex.search(line):
            # Determine the comment prefix.
            prefix = "//"
            if line.strip().startswith("//*"):
                prefix = "//*"
            if line.strip().startswith("#"):
                prefix = "#"

            expected = EXPECTED_LICENSE_TEXT.get(prefix)
            if f.LocalPath().endswith('.go') and prefix == "//":
                expected = EXPECTED_LICENSE_TEXT_GO

            if not expected:
                return []

            # Check the next len(expected) lines.
            actual_license_text = new_contents_lines[i + 1:i + 1 +
                                                     len(expected)]
            if actual_license_text != expected:
                return [
                    output_api.PresubmitPromptWarning(
                        f'{f.LocalPath()}: Copyright license text is incorrect.'
                    )
                ]
            return []
    return []


def _CheckCopyright(input_api, output_api):
    """Checks that newly added files have a correct copyright year and license text."""
    current_year = int(input_api.time.strftime('%Y'))

    errors = []

    # Use a list for deleted contents to handle multiple files with the same
    # content being renamed.
    deleted_files_hashes = []
    for f in input_api.AffectedFiles(include_deletes=True):
        if f.Action() == 'D':
            deleted_files_hashes.append(
                hashlib.sha256(''.join(
                    f.OldContents()).encode('utf-8')).hexdigest())

    for f in input_api.AffectedFiles(include_deletes=False):
        path = f.LocalPath().replace('\\', '/')
        if not (path.endswith(
            ('.h', '.cc', '.cpp', '.mm', '.gn', '.gni', '.cmake',
             'CMakeLists.txt', 'BUILD.bazel', '.py', '.go', '.star', '.js',
             '.mjs', '.sh', '.def', '.idl'))):
            continue

        new_contents_lines = list(f.NewContents())

        # Always check for the copyright license text body if the file is modified.
        errors.extend(
            _CheckCopyrightText(input_api, output_api, f, new_contents_lines))

        # Only check for the copyright year if the file is newly added.
        if f.Action() == 'A':
            new_content_hash = hashlib.sha256(
                ''.join(new_contents_lines).encode('utf-8')).hexdigest()

            # If the file is a rename, we don't check for the copyright year.
            # A rename is detected if a file with the same content is also
            # deleted in the same changelist.
            is_rename = False
            if new_content_hash in deleted_files_hashes:
                deleted_files_hashes.remove(new_content_hash)
                is_rename = True

            if not is_rename:
                errors.extend(
                    _CheckCopyrightYear(input_api, output_api, f,
                                        new_contents_lines, current_year))

    return errors


def CheckPresubmitTests(input_api, output_api):
    """Test PRESUBMIT.py during presubmit."""
    # TODO(crbug.com/517626950): consider running the tests only when
    # PRESUBMIT.py has been modified. For now, this is very cheap, so it
    # doesn't matter.
    return input_api.RunTests(
        input_api.canned_checks.GetUnitTestsInDirectory(
            input_api,
            output_api,
            input_api.PresubmitLocalPath(),
            files_to_check=[r'^PRESUBMIT_test\.py$']))


def CheckUnsafeBuffersSafetyComments(input_api, output_api):
    """Checks that DAWN_UNSAFE_BUFFERS is accompanied by a
    // SAFETY: comment.
    """
    # We only check C++ source files.
    exts = ('.h', '.cc', '.cpp', '.mm')
    file_filter = lambda f: f.LocalPath().endswith(exts)

    unsafe_buffers_regex = re.compile(r'\bDAWN_UNSAFE_BUFFERS\b')
    safety_comment_regex = re.compile(r'//.*\bSAFETY\b')

    problems = []
    locations = []

    for f in input_api.AffectedFiles(include_deletes=False,
                                     file_filter=file_filter):
        lines = f.NewContents()
        for line_num, line in enumerate(lines, start=1):
            if line.strip().startswith('//'):
                continue
            if unsafe_buffers_regex.search(line):
                # Check if safety comment is on the same line.
                if safety_comment_regex.search(line):
                    continue

                # Check preceding lines for a SAFETY comment.
                has_safety = False
                for check_line_num in range(line_num - 1, 0, -1):
                    check_line = lines[check_line_num - 1].strip()
                    if not check_line.startswith('//'):
                        # Not a comment line, stop searching.
                        break
                    if safety_comment_regex.search(check_line):
                        has_safety = True
                        break

                if not has_safety:
                    problems.append(
                        f"{f.LocalPath()}:{line_num}: "
                        "DAWN_UNSAFE_BUFFERS usage must be accompanied by a "
                        "// SAFETY: comment.")
                    locations.append(
                        output_api.PresubmitResultLocation(
                            file_path=f.LocalPath(),
                            start_line=line_num,
                            end_line=line_num,
                        ))

    if problems:
        return [
            output_api.PresubmitPromptWarning(
                "DAWN_UNSAFE_BUFFERS usages must be accompanied by a "
                "// SAFETY: comment explaining why they are safe.",
                items=problems,
                locations=locations)
        ]
    return []


def CheckChangeTodoHasOwner(input_api, output_api):
    """
    Checks that TODO comments have the issue number.

    This is similar to the canned check in
    depot_tools/presubmit_canned_checks.py but with Dawn-specific exclusions:
    - Excludes DAWN_UNSAFE_TODO macros.
    - Excludes PRESUBMIT.py and PRESUBMIT_test.py to avoid false positives.
    """
    legacyTODO = '\\s*\\(.+\\)\\s*:'  # TODO(owner/bug):
    modernTODO = ':\\s*[^\\s]+\\s*\\-'  # TODO: owner - description
    unowned_todo = input_api.re.compile('\\bTODO(?!(%s|%s))' %
                                        (legacyTODO, modernTODO))

    # Skip PRESUBMIT files to avoid false positives from regexes and tests.
    def FileFilter(affected_file):
        filename = affected_file.LocalPath().replace('\\', '/')
        if filename in ('PRESUBMIT.py', 'PRESUBMIT_test.py'):
            return False
        return input_api.FilterSourceFile(affected_file)

    errors = []
    for f in input_api.AffectedFiles(include_deletes=False,
                                     file_filter=FileFilter):
        for line_num, line in f.ChangedContents():
            if unowned_todo.search(line):
                errors.append(
                    f'Found TODO with no issue number in {f.LocalPath()}:{line_num}'
                )

    if errors:
        return [output_api.PresubmitPromptWarning('\n'.join(errors))]
    return []


# Copied from Chrome's _GetMessageForMatchingType.
def _GetMessageForMatchingType(input_api, affected_file, line_number, line,
                               ban_rule):
    """
    Helper method for checking for banned constructs.

    Returns an string composed of the name of the file, the line number
    where the match has been found and the additional text passed as
    |message| in case the target type name matches the text inside the
    line passed as parameter.
    """
    result = []

    # Ignore comments containing banned strings (but not NOLINT comments).
    if input_api.re.search(r'^ *//(?! *NOLINT)', line):
        return result
    # A // nocheck comment will bypass this error.
    if line.endswith(' nocheck'):
        return result

    matched = False
    if ban_rule.pattern[0:1] == '/':
        regex = ban_rule.pattern[1:]
        if input_api.re.search(regex, line):
            matched = True
    elif ban_rule.pattern in line:
        matched = True

    if matched:
        result.append('    %s:%d:' % (affected_file.LocalPath(), line_number))
        for line in ban_rule.explanation:
            result.append('      %s' % line)

    return result


# Copied from Chrome's CheckNoBannedPatterns with modifications.
def CheckNoBannedPatterns(input_api, output_api):
    """Make sure that banned patterns are not used."""
    results = []

    def IsExcludedFile(affected_file, excluded_paths):
        if not excluded_paths:
            return False

        local_path = pathlib.Path(affected_file.LocalPath()).as_posix()
        for item in excluded_paths:
            if input_api.re.match(item, local_path):
                return True
        return False

    def CheckForMatch(affected_file, line_num: int, line: str,
                      ban_rule: BanRule):
        if IsExcludedFile(affected_file, ban_rule.excluded_paths):
            return

        message = _GetMessageForMatchingType(input_api, affected_file,
                                             line_num, line, ban_rule)
        if message:
            result_loc = []
            if ban_rule.surface_as_gerrit_lint:
                result_loc.append(
                    output_api.PresubmitResultLocation(
                        file_path=affected_file.LocalPath(),
                        start_line=line_num,
                        end_line=line_num,
                    ))
            if ban_rule.treat_as_error:
                results.append(
                    output_api.PresubmitError('A banned pattern was used.\n' +
                                              '\n'.join(message),
                                              locations=result_loc))
            else:
                results.append(
                    output_api.PresubmitPromptWarning(
                        'A banned pattern was used.\n' + '\n'.join(message),
                        locations=result_loc))

    file_filter = lambda f: f.LocalPath().endswith(
        ('.cc', '.mm', '.cpp', '.h'))
    for f in input_api.AffectedFiles(file_filter=file_filter):
        for line_num, line in f.ChangedContents():
            for ban_rule in _BANNED_CPP_PATTERNS:
                CheckForMatch(f, line_num, line, ban_rule)

    return results


def CheckChange(input_api, output_api):
    results = []
    results.extend(
        input_api.canned_checks.CheckForCommitObjects(input_api, output_api))

    result_factory = output_api.PresubmitPromptWarning
    if input_api.is_committing:
        result_factory = output_api.PresubmitError

    # Check for formatting.
    results.extend(
        input_api.canned_checks.CheckPatchFormatted(
            input_api, output_api, result_factory=result_factory))
    results.extend(
        input_api.canned_checks.CheckGNFormatted(input_api, output_api))
    results.extend(
        input_api.canned_checks.CheckChangeHasNoCrAndHasOnlyOneEol(
            input_api, output_api))
    results.extend(
        input_api.canned_checks.CheckChangeHasNoTabs(input_api, output_api))
    results.extend(
        input_api.canned_checks.CheckChangeHasNoStrayWhitespace(
            input_api,
            output_api,
            source_file_filter=_HasNoStrayWhitespaceFilter))
    results.extend(
        input_api.canned_checks.CheckChangeHasDescription(
            input_api, output_api))
    results.extend(
        input_api.canned_checks.CheckDoNotSubmit(input_api, output_api))
    results.extend(_CheckCopyright(input_api, output_api))
    # Note, the verbose_level here should match what is set in tools/lint so
    # the same set of lint errors are reported on the CQ and Kokoro bots.
    results.extend(
        input_api.canned_checks.CheckChangeLintsClean(
            input_api, output_api, lint_filters=LINT_FILTERS, verbose_level=1))

    return results