Fix UTF-8 validation in static checks

Use isutf8 instead of recode to detect invalid UTF-8 sequences.

Also add the necessary dependencies to run the static checks locally
using act (https://github.com/nektos/act) with the Medium size image.

(cherry picked from commit 13d9a8c88f)
This commit is contained in:
Gergely Kis 2022-09-07 23:12:00 +02:00 committed by Rémi Verschelde
parent 2bb8183011
commit 47bc0f649d
2 changed files with 23 additions and 13 deletions

View File

@ -24,8 +24,8 @@ jobs:
- name: Install dependencies - name: Install dependencies
run: | run: |
sudo apt-get install -qq dos2unix recode clang-format-13 libxml2-utils sudo apt-get install -qq dos2unix clang-format-13 libxml2-utils python3-pip moreutils
sudo update-alternatives --remove-all clang-format sudo update-alternatives --remove-all clang-format || true
sudo update-alternatives --install /usr/bin/clang-format clang-format /usr/bin/clang-format-13 100 sudo update-alternatives --install /usr/bin/clang-format clang-format /usr/bin/clang-format-13 100
sudo pip3 install black==22.3.0 pygments sudo pip3 install black==22.3.0 pygments

View File

@ -5,8 +5,8 @@
# run before them. # run before them.
# We need dos2unix and recode. # We need dos2unix and recode.
if [ ! -x "$(command -v dos2unix)" -o ! -x "$(command -v recode)" ]; then if [ ! -x "$(command -v dos2unix)" -o ! -x "$(command -v isutf8)" ]; then
printf "Install 'dos2unix' and 'recode' to use this script.\n" printf "Install 'dos2unix' and 'isutf8' (from the moreutils package) to use this script.\n"
fi fi
set -uo pipefail set -uo pipefail
@ -36,7 +36,7 @@ while IFS= read -rd '' f; do
continue continue
fi fi
# Ensure that files are UTF-8 formatted. # Ensure that files are UTF-8 formatted.
recode UTF-8 "$f" 2> /dev/null isutf8 "$f" >> utf8-validation.txt 2>&1
# Ensure that files have LF line endings and do not contain a BOM. # Ensure that files have LF line endings and do not contain a BOM.
dos2unix "$f" 2> /dev/null dos2unix "$f" 2> /dev/null
# Remove trailing space characters and ensures that files end # Remove trailing space characters and ensures that files end
@ -48,17 +48,27 @@ done
git diff --color > patch.patch git diff --color > patch.patch
# If no patch has been generated all is OK, clean up, and exit. # If no UTF-8 violations were collected and no patch has been
if [ ! -s patch.patch ] ; then # generated all is OK, clean up, and exit.
if [ ! -s utf8-validation.txt ] && [ ! -s patch.patch ] ; then
printf "Files in this commit comply with the formatting rules.\n" printf "Files in this commit comply with the formatting rules.\n"
rm -f patch.patch rm -f patch.patch utf8-validation.txt
exit 0 exit 0
fi fi
# A patch has been created, notify the user, clean up, and exit. # Violations detected, notify the user, clean up, and exit.
if [ -s utf8-validation.txt ]
then
printf "\n*** The following files contain invalid UTF-8 character sequences:\n\n"
cat utf8-validation.txt
fi
if [ -s patch.patch ]
then
printf "\n*** The following differences were found between the code " printf "\n*** The following differences were found between the code "
printf "and the formatting rules:\n\n" printf "and the formatting rules:\n\n"
cat patch.patch cat patch.patch
fi
rm -f utf8-validation.txt patch.patch
printf "\n*** Aborting, please fix your commit(s) with 'git commit --amend' or 'git rebase -i <hash>'\n" printf "\n*** Aborting, please fix your commit(s) with 'git commit --amend' or 'git rebase -i <hash>'\n"
rm -f patch.patch
exit 1 exit 1