tldr/scripts/check-pr.sh

166 lines
5.2 KiB
Bash
Raw Normal View History

#!/usr/bin/env bash
# SPDX-License-Identifier: MIT
2020-06-09 15:03:35 +01:00
# This script is executed by GitHub Actions for every pull request opened.
# It currently accomplishes the following objectives:
#
# 1. Detect pages that were just copied (i.e. cp pages/{common,linux}/7z.md).
# 2. Detect pages that were added in a platform specific directory although
# they already exist under 'common'.
# 3. Detect pages that were added in the 'common' platform although they
# already exist under a platform specific directory.
# 4. Detect pages that do not exist as English pages yet.
2023-11-14 15:36:02 +00:00
# 5. Detect outdated pages. A page is marked as outdated when the number of
# commands differ from the number of commands in the English page or the
# contents of the commands differ from the English page.
# 6. Detect other miscellaneous anomalies in the pages folder.
#
# Results are printed to stdout, logs and errors to stderr.
#
# NOTE: must be run from the repository root directory to correctly work!
# NOTE: no `set -e`, failure of this script should not invalidate the build.
# Check for duplicated pages.
function check_duplicates {
local page=$1 # page path in the format 'platform/pagename.md'
local parts
local other
readarray -td'/' parts < <(echo -n "$page")
local platform=${parts[0]}
local file=${parts[1]}
case "$platform" in
common) # check if page already exists in other platforms
2019-11-18 22:11:58 +00:00
for other in ${PLATFORMS/common/}; do
if [[ -f "pages/$other/$file" ]]; then
2019-11-18 22:11:58 +00:00
printf "\x2d $MSG_EXISTS" "$page" "$other"
fi
done
;;
*) # check if page already exists under common
if [[ -f "pages/common/$file" ]]; then
2019-11-18 22:11:58 +00:00
printf "\x2d $MSG_EXISTS" "$page" 'common'
fi
;;
esac
}
function check_missing_english_page() {
2023-11-14 15:36:02 +00:00
local page=$1
local english_page="pages/${page#pages*\/}"
2023-11-14 15:36:02 +00:00
if [[ "$page" = "$english_page" ]]; then
return 1
fi
if [[ ! -f "$english_page" ]]; then
printf "\x2d $MSG_NOT_EXISTS" "$page" "$english_page"
fi
}
2023-11-14 15:36:02 +00:00
function check_outdated_page() {
local page=$1
local english_page="pages/${page#pages*\/}"
local command_regex='^`[^`]\+`$'
2023-11-14 15:36:02 +00:00
if [[ "$page" = "$english_page" ]] || [[ ! -f "$english_page" ]]; then
return 1
fi
local english_commands=$(grep -c $command_regex "$english_page")
mapfile -t stripped_english_commands < <(grep $command_regex "$english_page" | sed 's/{{[^}]*}}/{{}}/g' | sed 's/"[^"]*"/""/g' | sed "s/'[^']*'//g" | sed 's/`//g')
local commands=$(grep -c $command_regex $page)
mapfile -t stripped_commands < <(grep $command_regex "$page" | sed 's/{{[^}]*}}/{{}}/g' | sed 's/"[^"]*"/""/g' | sed "s/'[^']*'//g" | sed 's/`//g')
local english_commands_as_string=$(printf "%s\n" "${stripped_english_commands[*]}")
local commands_as_string=$(printf "%s\n" "${stripped_commands[*]}")
if [[ $english_commands != $commands ]]; then
printf "\x2d $MSG_OUTDATED" "$page" "based on number of commands"
elif [[ "$english_commands_as_string" != "$commands_as_string" ]]; then
printf "\x2d $MSG_OUTDATED" "$page" "based on the command contents itself"
fi
}
# Look at git diff and check for copied/duplicated pages.
function check_diff {
local git_diff
local line
local entry
2023-11-14 15:36:02 +00:00
git_diff=$(git diff --name-status --find-copies-harder --diff-filter=ACM origin/main -- pages*/)
if [[ -n $git_diff ]]; then
echo -e "Check PR: git diff:\n$git_diff" >&2
else
2019-11-18 22:11:58 +00:00
echo 'Check PR: git diff looks fine, no interesting changes detected.' >&2
return 0
fi
while read line; do
readarray -td$'\t' entry < <(echo -n "$line")
local change="${entry[0]}"
local file1="${entry[1]}"
local file2="${entry[2]}"
case "$change" in
C*) # file2 is a copy of file1
local percentage=${change#C}
percentage=${percentage#0}
percentage=${percentage#0}
2019-11-18 22:11:58 +00:00
printf "\x2d $MSG_IS_COPY" "$file2" "$file1" "$percentage"
;;
2023-11-14 15:36:02 +00:00
A|M) # file1 was newly added or modified
check_duplicates "$file1"
check_missing_english_page "$file1"
2023-11-14 15:36:02 +00:00
check_outdated_page "$file1"
;;
esac
done <<< "$git_diff"
}
# Recursively check the pages/ folder for anomalies.
function check_structure {
2019-11-18 22:11:58 +00:00
for platform in $PLATFORMS; do
if [[ ! -d "pages/$platform" ]]; then
2019-11-18 22:11:58 +00:00
printf "\x2d $MSG_NOT_DIR" "pages/$platform"
else
2019-11-18 22:11:58 +00:00
for page in "pages/$platform"/*; do
if [[ ! -f $page ]]; then
2019-11-18 22:11:58 +00:00
printf "\x2d $MSG_NOT_FILE" "$page"
elif [[ ${page:(-3)} != ".md" ]]; then
2019-11-18 22:11:58 +00:00
printf "\x2d $MSG_NOT_MD" "$page"
fi
done
fi
done
}
###################################
# MAIN
###################################
2019-11-18 22:11:58 +00:00
MSG_EXISTS='The page `%s` already exists under the `%s` platform.\n'
MSG_NOT_EXISTS='The page `%s` does not exists as English page `%s` yet.\n'
2023-11-14 15:36:02 +00:00
MSG_OUTDATED='The page `%s` is outdated, %s.\n'
2019-11-18 22:11:58 +00:00
MSG_IS_COPY='The page `%s` seems to be a copy of `%s` (%d%% matching).\n'
MSG_NOT_DIR='The file `%s` does not look like a directory.\n'
MSG_NOT_FILE='The file `%s` does not look like a regular file.\n'
MSG_NOT_MD='The file `%s` does not have a `.md` extension.\n'
PLATFORMS=$(ls pages/)
2019-11-18 22:11:58 +00:00
if [[ $CI == true && $GITHUB_REPOSITORY == "tldr-pages/tldr" && $PULL_REQUEST_ID != "" ]]; then
check_diff
check_structure
else
echo 'Not a pull request, refusing to run.' >&2
exit 0
fi