ClickHouse/utils/check-style/check-style

#!/usr/bin/env bash

# For code formatting we have clang-format.
#
# But it's not sane to apply clang-format for whole code base,
#  because it sometimes makes worse for properly formatted files.
#
# It's only reasonable to blindly apply clang-format only in cases
#  when the code is likely to be out of style.
#
# For this purpose we have a script that will use very primitive heuristics
#  (simple regexps) to check if the code is likely to have basic style violations.
#  and then to run formatter only for the specified files.

ROOT_PATH=$(git rev-parse --show-toplevel)
EXCLUDE_DIRS='build/|integration/|widechar_width/|glibc-compatibility/|memcpy/|consistent-hashing'

find $ROOT_PATH/{dbms,base} -name '*.h' -or -name '*.cpp' 2>/dev/null |
    grep -vP $EXCLUDE_DIRS |
    xargs grep $@ -P '((class|struct|namespace|enum|if|for|while|else|throw|switch).*|\)(\s*const)?(\s*override)?\s*)\{$|\s$|\t|^ {1,3}[^\* ]\S|\t|^\s*(if|else if|if constexpr|else if constexpr|for|while|catch|switch)\(|\( [^\s\\]|\S \)' |
# a curly brace not in a new line, but not for the case of C++11 init or agg. initialization | trailing whitespace | number of ws not a multiple of 4, but not in the case of comment continuation | a tab character | missing whitespace after for/if/while... before opening brace | whitespaces inside braces
    grep -v -P '(//|:\s+\*|\$\(\()| \)"'
# single-line comment | continuation of a multiline comment | a typical piece of embedded shell code | something like ending of raw string literal

# // namespace comments are unneeded
find $ROOT_PATH/{dbms,base} -name '*.h' -or -name '*.cpp' 2>/dev/null |
    grep -vP $EXCLUDE_DIRS |
    xargs grep $@ -P '}\s*//+\s*namespace\s*'

# Broken symlinks
find -L $ROOT_PATH -type l 2>/dev/null | grep -v contrib && echo "^ Broken symlinks found"

# Double whitespaces
find $ROOT_PATH/{dbms,base} -name '*.h' -or -name '*.cpp' 2>/dev/null | while read i; do $ROOT_PATH/utils/check-style/double-whitespaces.pl < $i || echo -e "^ File $i contains double whitespaces\n"; done