From 20bf72751c865e2597ca89e6acb5a8066d2e81f4 Mon Sep 17 00:00:00 2001
From: Gergely Kis <gergely.kis@migeran.com>
Date: Mon, 5 Sep 2022 11:55:51 +0200
Subject: [PATCH] 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.
---
 .github/workflows/static_checks.yml |  4 ++--
 misc/scripts/file_format.sh         | 27 ++++++++++++++++++++-------
 2 files changed, 22 insertions(+), 9 deletions(-)

diff --git a/.github/workflows/static_checks.yml b/.github/workflows/static_checks.yml
index 3e84702d7f7..5b4de06e9e6 100644
--- a/.github/workflows/static_checks.yml
+++ b/.github/workflows/static_checks.yml
@@ -24,8 +24,8 @@ jobs:
 
       - name: Install dependencies
         run: |
-          sudo apt-get install -qq dos2unix recode clang-format-13 libxml2-utils
-          sudo update-alternatives --remove-all clang-format
+          sudo apt-get install -qq dos2unix recode clang-format-13 libxml2-utils python3-pip moreutils
+          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 pip3 install black==22.3.0 pygments pytest
 
diff --git a/misc/scripts/file_format.sh b/misc/scripts/file_format.sh
index 731b3ee0051..1200b96ea0f 100755
--- a/misc/scripts/file_format.sh
+++ b/misc/scripts/file_format.sh
@@ -41,7 +41,7 @@ while IFS= read -rd '' f; do
         continue
     fi
     # 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.
     dos2unix "$f" 2> /dev/null
     # Remove trailing space characters and ensures that files end
@@ -51,15 +51,28 @@ done
 
 diff=$(git diff --color)
 
-# If no diff has been generated all is OK, clean up, and exit.
-if [ -z "$diff" ] ; then
+# If no UTF-8 violations were collected and no diff has been
+# generated all is OK, clean up, and exit.
+if [ ! -s utf8-validation.txt ] && [ -z "$diff" ] ; then
     printf "Files in this commit comply with the formatting rules.\n"
+    rm -f utf8-violations.txt
     exit 0
 fi
 
-# A diff has been created, notify the user, clean up, and exit.
-printf "\n*** The following differences were found between the code "
-printf "and the formatting rules:\n\n"
-echo "$diff"
+# Violations detected, notify the user, clean up, and exit.
+if [ -s utf8-validation.txt ]
+then
+    printf "\n*** The following files contain invalid UTF-8 character sequences:\n\n"
+    cat utf8-validation.txt
+    rm -f utf8-validation.txt
+fi
+
+if [ ! -z "$diff" ]
+then
+    printf "\n*** The following differences were found between the code "
+    printf "and the formatting rules:\n\n"
+    echo "$diff"
+fi
+
 printf "\n*** Aborting, please fix your commit(s) with 'git commit --amend' or 'git rebase -i <hash>'\n"
 exit 1