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/{src,base,programs,utils} -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*)\{$|\s$|^ {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 | 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

# Tabs
find $ROOT_PATH/{src,base,programs,utils} -name '*.h' -or -name '*.cpp' 2>/dev/null |
    grep -vP $EXCLUDE_DIRS |
    xargs grep $@ -F $'\t'

# // namespace comments are unneeded
find $ROOT_PATH/{src,base,programs,utils} -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/{src,base,programs,utils} -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

# Unused ErrorCodes
# NOTE: to fix automatically, replace echo with:
# sed -i "/extern const int $code/d" $file
find $ROOT_PATH/{src,base,programs,utils} -name '*.h' -or -name '*.cpp' | xargs grep -l -P 'extern const int [_A-Z]+' | while read file; do grep -P 'extern const int [_A-Z]+;' $file | sed -r -e 's/^.*?extern const int ([_A-Z]+);.*?$/\1/' | while read code; do grep -q "ErrorCodes::$code" $file || echo "ErrorCode $code is defined but not used in file $file"; done; done

# Undefined ErrorCodes
# NOTE: to fix automatically, replace echo with:
# ( grep -q -F 'namespace ErrorCodes' $file && sed -i -r "0,/(\s*)extern const int [_A-Z]+/s//\1extern const int $code;\n&/" $file || awk '{ print; if (ns == 1) { ns = 2 }; if (ns == 2) { ns = 0; print "namespace ErrorCodes\n{\n    extern const int '$code';\n}" } }; /namespace DB/ { ns = 1; };' < $file > ${file}.tmp && mv ${file}.tmp $file )
find $ROOT_PATH/{src,base,programs,utils} -name '*.h' -or -name '*.cpp' | xargs grep -l -P 'ErrorCodes::[_A-Z]+' | while read file; do grep -P 'ErrorCodes::[_A-Z]+' $file | sed -r -e 's/^.*?ErrorCodes::([_A-Z]+).*?$/\1/' | while read code; do grep -q "extern const int $code" $file || echo "ErrorCode $code is used in file $file but not defined"; done; done

# Duplicate ErrorCodes
find $ROOT_PATH/{src,base,programs,utils} -name '*.h' -or -name '*.cpp' | xargs grep -l -P 'ErrorCodes::[_A-Z]+' | while read file; do grep -P 'extern const int [_A-Z]+;' $file | sort | uniq -c | grep -v -P ' +1 ' && echo "Duplicate ErrorCode in file $file"; done

# Three or more consecutive empty lines
find $ROOT_PATH/{src,base,programs,utils} -name '*.h' -or -name '*.cpp' | while read file; do awk '/^$/ { ++i; if (i > 2) { print "More than two consecutive empty lines in file '$file'" } } /./ { i = 0 }' $file; done

# Broken XML files (requires libxml2-utils)
find $ROOT_PATH/{src,base,programs,utils} -name '*.xml' | xargs xmllint --noout --nonet

# Machine translation to Russian is strictly prohibited
find $ROOT_PATH/docs/ru -name '*.md' | xargs grep -l -F 'machine_translated: true'

# Tests should not be named with "fail" in their names. It makes looking at the results less convenient.
find $ROOT_PATH/tests/queries -iname '*fail*' | grep . && echo 'Tests should not be named with "fail" in their names. It makes looking at the results less convenient when you search for "fail" substring in browser.'

# All the submodules should be from https://github.com/
find $ROOT_PATH -name '.gitmodules' | while read i; do grep -F 'url = ' $i | grep -v -F 'https://github.com/' && echo 'All the submodules should be from https://github.com/'; done

# There shouldn't be any code snippets under GPL or LGPL
find $ROOT_PATH/{src,base,programs} -name '*.h' -or -name '*.cpp' 2>/dev/null | xargs grep -i -F 'General Public License' && echo "There shouldn't be any code snippets under GPL or LGPL"