Meta: Add minimalistic formatting/style checks

This patch adds a modified copy of the Serenity repo style checking
scripts.

Author: faxe1008

Meta/check-newlines-at-eof.py

@@ -0,0 +1,67 @@
+#!/usr/bin/env python3
+
+import os
+import re
+import subprocess
+import sys
+
+
+RE_RELEVANT_FILE_EXTENSION = re.compile('\\.(cpp|h|gml|html|js|css|sh|py|json|txt)$')
+
+
+def should_check_file(filename):
+    if not RE_RELEVANT_FILE_EXTENSION.search(filename):
+        return False
+    if filename.endswith('.txt'):
+        return 'CMake' in filename
+    return True
+
+
+def find_files_here_or_argv():
+    if len(sys.argv) > 1:
+        raw_list = sys.argv[1:]
+    else:
+        process = subprocess.run(["git", "ls-files"], check=True, capture_output=True)
+        raw_list = process.stdout.decode().strip('\n').split('\n')
+
+    return filter(should_check_file, raw_list)
+
+
+def run():
+    """Check files checked in to git for trailing newlines at end of file."""
+    no_newline_at_eof_errors = []
+    blank_lines_at_eof_errors = []
+
+    did_fail = False
+    for filename in find_files_here_or_argv():
+        with open(filename, "r") as f:
+            f.seek(0, os.SEEK_END)
+
+            f.seek(f.tell() - 1, os.SEEK_SET)
+            if f.read(1) != '\n':
+                did_fail = True
+                no_newline_at_eof_errors.append(filename)
+                continue
+
+            while True:
+                f.seek(f.tell() - 2, os.SEEK_SET)
+                char = f.read(1)
+                if not char.isspace():
+                    break
+                if char == '\n':
+                    did_fail = True
+                    blank_lines_at_eof_errors.append(filename)
+                    break
+
+    if no_newline_at_eof_errors:
+        print("Files with no newline at the end:", " ".join(no_newline_at_eof_errors))
+    if blank_lines_at_eof_errors:
+        print("Files that have blank lines at the end:", " ".join(blank_lines_at_eof_errors))
+
+    if did_fail:
+        sys.exit(1)
+
+
+if __name__ == '__main__':
+    os.chdir(os.path.dirname(__file__) + "/..")
+    run()

Meta/check-style.py

@@ -0,0 +1,86 @@
+#!/usr/bin/env python3
+
+import os
+import re
+import subprocess
+import sys
+
+# Ensure copyright headers match this format and are followed by a blank line:
+# /*
+#  * Copyright (c) YYYY(-YYYY), Whatever
+#  * ... more of these ...
+#  *
+#  * SPDX-License-Identifier: BSD-2-Clause
+#  */
+GOOD_LICENSE_HEADER_PATTERN = re.compile(
+    '^/\\*\n' +
+    '( \\* Copyright \\(c\\) [0-9]{4}(-[0-9]{4})?, .*\n)+' +
+    ' \\*\n' +
+    ' \\* SPDX-License-Identifier: BSD-2-Clause\n' +
+    ' \\*/\n' +
+    '\n')
+LICENSE_HEADER_CHECK_EXCLUDES = {}
+
+# We check that "#pragma once" is present
+PRAGMA_ONCE_STRING = '#pragma once'
+PRAGMA_ONCE_CHECK_EXCLUDES = {}
+
+# We make sure that there's a blank line before and after pragma once
+GOOD_PRAGMA_ONCE_PATTERN = re.compile('(^|\\S\n\n)#pragma once(\n\n\\S.|$)')
+
+
+def should_check_file(filename):
+    if not filename.endswith('.cpp') and not filename.endswith('.h'):
+        return False
+    return True
+
+
+def find_files_here_or_argv():
+    if len(sys.argv) > 1:
+        raw_list = sys.argv[1:]
+    else:
+        process = subprocess.run(["git", "ls-files"], check=True, capture_output=True)
+        raw_list = process.stdout.decode().strip('\n').split('\n')
+
+    return filter(should_check_file, raw_list)
+
+
+def run():
+    errors_license = []
+    errors_pragma_once_bad = []
+    errors_pragma_once_missing = []
+
+    for filename in find_files_here_or_argv():
+        with open(filename, "r") as f:
+            file_content = f.read()
+        if not any(filename.startswith(forbidden_prefix) for forbidden_prefix in LICENSE_HEADER_CHECK_EXCLUDES):
+            if not GOOD_LICENSE_HEADER_PATTERN.search(file_content):
+                errors_license.append(filename)
+        if filename.endswith('.h'):
+            if any(filename.startswith(forbidden_prefix) for forbidden_prefix in PRAGMA_ONCE_CHECK_EXCLUDES):
+                # File was excluded
+                pass
+            elif GOOD_PRAGMA_ONCE_PATTERN.search(file_content):
+                # Excellent, the formatting is correct.
+                pass
+            elif PRAGMA_ONCE_STRING in file_content:
+                # Bad, the '#pragma once' is present but it's formatted wrong.
+                errors_pragma_once_bad.append(filename)
+            else:
+                # Bad, the '#pragma once' is missing completely.
+                errors_pragma_once_missing.append(filename)
+
+    if errors_license:
+        print("Files with bad licenses:", " ".join(errors_license))
+    if errors_pragma_once_missing:
+        print("Files without #pragma once:", " ".join(errors_pragma_once_missing))
+    if errors_pragma_once_bad:
+        print("Files with a bad #pragma once:", " ".join(errors_pragma_once_bad))
+
+    if errors_license or errors_pragma_once_missing or errors_pragma_once_bad:
+        sys.exit(1)
+
+
+if __name__ == '__main__':
+    os.chdir(os.path.dirname(__file__) + "/..")
+    run()

Meta/lint-ci.sh

@@ -0,0 +1,36 @@
+#!/usr/bin/env bash
+
+set -e
+
+script_path=$(cd -P -- "$(dirname -- "$0")" && pwd -P)
+cd "${script_path}/.." || exit 1
+
+RED='\033[0;31m'
+GREEN='\033[0;32m'
+NC='\033[0m' # No Color
+
+FAILURES=0
+
+set +e
+
+for cmd in \
+        Meta/check-newlines-at-eof.py \
+        Meta/check-style.py; do
+    echo "Running ${cmd}"
+    if time "${cmd}" "$@"; then
+        echo -e "[${GREEN}OK${NC}]: ${cmd}"
+    else
+        echo -e "[${RED}FAIL${NC}]: ${cmd}"
+        ((FAILURES+=1))
+    fi
+done
+
+echo "Running Meta/lint-clang-format.sh"
+if time Meta/lint-clang-format.sh --overwrite-inplace "$@" && git diff --exit-code; then
+    echo -e "[${GREEN}OK${NC}]: Meta/lint-clang-format.sh"
+else
+    echo -e "[${RED}FAIL${NC}]: Meta/lint-clang-format.sh"
+    ((FAILURES+=1))
+fi
+
+exit "${FAILURES}"

Meta/lint-clang-format.sh

@@ -0,0 +1,54 @@
+#!/usr/bin/env bash
+
+set -e
+
+script_path=$(cd -P -- "$(dirname -- "$0")" && pwd -P)
+cd "${script_path}/.." || exit 1
+
+if [ "$#" -eq "1" ]; then
+    mapfile -t files < <(
+        git ls-files -- \
+            '*.cpp' \
+            '*.h'
+    )
+else
+    files=()
+    for file in "${@:2}"; do
+        if [[ "${file}" == *".cpp" || "${file}" == *".h" ]]; then
+            files+=("${file}")
+        fi
+    done
+fi
+
+if (( ${#files[@]} )); then
+    CLANG_FORMAT=false
+    if command -v clang-format-14 >/dev/null 2>&1 ; then
+        CLANG_FORMAT=clang-format-14
+    elif command -v clang-format >/dev/null 2>&1 ; then
+        CLANG_FORMAT=clang-format
+        if ! "${CLANG_FORMAT}" --version | awk '{ if (substr($NF, 1, index($NF, ".") - 1) < 14) exit 1; }'; then
+            echo "You are using '$("${CLANG_FORMAT}" --version)', which appears to not be clang-format 14 or later."
+            echo "It is very likely that the resulting changes are not what you wanted."
+        fi
+    else
+        echo "clang-format-14 is not available, but C or C++ files need linting! Either skip this script, or install clang-format-14."
+        echo "(If you install a package 'clang-format', please make sure it's version 14 or later.)"
+        exit 1
+    fi
+
+    if [ "$#" -gt "0" ] && [ "--overwrite-inplace" = "$1" ] ; then
+        true # The only way to run this script.
+    else
+        # Note that this branch also covers --help, -h, -help, -?, etc.
+        echo "USAGE: $0 --overwrite-inplace"
+        echo "The argument is necessary to make you aware that this *will* overwrite your local files."
+        exit 1
+    fi
+
+    echo "Using ${CLANG_FORMAT}"
+
+    "${CLANG_FORMAT}" -style=file -i "${files[@]}"
+    echo "Maybe some files have changed. Sorry, but clang-format doesn't indicate what happened."
+else
+    echo "No .cpp or .h files to check."
+fi

Meta/lint-commit.sh

@@ -0,0 +1,61 @@
+#!/usr/bin/env bash
+
+# the file containing the commit message is passed as the first argument
+commit_file="$1"
+commit_message=$(cat "$commit_file")
+
+error() {
+  echo -e "\033[0;31m$1:\033[0m"
+  echo "$commit_message"
+  exit 1
+}
+
+# fail if the commit message contains windows style line breaks (carriage returns)
+if grep -q -U $'\x0D' "$commit_file"; then
+  error "Commit message contains CRLF line breaks (only unix-style LF linebreaks are allowed)"
+fi
+
+line_number=0
+while read -r line; do
+  # break on git cut line, used by git commit --verbose
+  if [[ "$line" == "# ------------------------ >8 ------------------------" ]]; then
+    break
+  fi
+
+  # ignore comment lines
+  [[ "$line" =~ ^#.* ]] && continue
+  # ignore overlong 'fixup!' commit descriptions
+  [[ "$line" =~ ^fixup!\ .* ]] && continue
+
+  ((line_number += 1))
+  line_length=${#line}
+
+  if [[ $line_number -eq 2 ]] && [[ $line_length -ne 0 ]]; then
+    error "Empty line between commit title and body is missing"
+  fi
+
+  category_pattern='^(Revert "|\S+: )'
+  if [[ $line_number -eq 1 ]] && (echo "$line" | grep -E -v -q "$category_pattern"); then
+    error "Missing category in commit title (if this is a fix up of a previous commit, it should be squashed)"
+  fi
+
+  title_case_pattern="^\S.*?: [A-Z0-9]"
+  if [[ $line_number -eq 1 ]] && (echo "$line" | grep -E -v -q "$title_case_pattern"); then
+    error "First word of commit after the subsystem is not capitalized"
+  fi
+
+  if [[ $line_number -eq 1 ]] && [[ "$line" =~ \.$ ]]; then
+    error "Commit title ends in a period"
+  fi
+
+  url_pattern="([a-z]+:\/\/)?(([a-zA-Z0-9_]|-)+\.)+[a-z]{2,}(:\d+)?([a-zA-Z_0-9@:%\+.~\?&\/=]|-)+"
+  if [[ $line_length -gt 72 ]] && (echo "$line" | grep -E -v -q "$url_pattern"); then
+    error "Commit message lines are too long (maximum allowed is 72 characters)"
+  fi
+
+  if [[ "$line" == "Signed-off-by: "* ]]; then
+    error "Commit body contains a Signed-off-by tag"
+  fi
+
+done <"$commit_file"
+exit 0