From 95195b557466708c06f9b631c3909a557dcd5b57 Mon Sep 17 00:00:00 2001
From: Pieter De Gendt <pieter.degendt@basalte.be>
Date: Tue, 15 Oct 2024 11:12:19 +0200
Subject: [PATCH] scripts: ci: check_compliance: Add text encoding check

Add a check for text files to make sure these are encoded in ascii
or utf-8.

Signed-off-by: Pieter De Gendt <pieter.degendt@basalte.be>
---
 .gitignore                     |  1 +
 scripts/ci/check_compliance.py | 27 +++++++++++++++++++++++++++
 2 files changed, 28 insertions(+)

diff --git a/.gitignore b/.gitignore
index f9fa32b7a6b..c33e2d93747 100644
--- a/.gitignore
+++ b/.gitignore
@@ -103,4 +103,5 @@ ModulesMaintainers.txt
 Nits.txt
 Pylint.txt
 SphinxLint.txt
+TextEncoding.txt
 YAMLLint.txt
diff --git a/scripts/ci/check_compliance.py b/scripts/ci/check_compliance.py
index 038370ae3c4..f586b53d147 100755
--- a/scripts/ci/check_compliance.py
+++ b/scripts/ci/check_compliance.py
@@ -1606,6 +1606,33 @@ class KeepSorted(ComplianceTest):
             with open(file, "r") as fp:
                 self.check_file(file, fp)
 
+
+class TextEncoding(ComplianceTest):
+    """
+    Check that any text file is encoded in ascii or utf-8.
+    """
+    name = "TextEncoding"
+    doc = "Check the encoding of text files."
+    path_hint = "<git-top>"
+
+    ALLOWED_CHARSETS = ["us-ascii", "utf-8"]
+
+    def run(self):
+        m = magic.Magic(mime=True, mime_encoding=True)
+
+        for file in get_files(filter="d"):
+            full_path = os.path.join(GIT_TOP, file)
+            mime_type = m.from_file(full_path)
+
+            if not mime_type.startswith("text/"):
+                continue
+
+            # format is "text/<type>; charset=<charset>"
+            if mime_type.rsplit('=')[-1] not in self.ALLOWED_CHARSETS:
+                desc = f"Text file with unsupported encoding: {file} has mime type {mime_type}"
+                self.fmtd_failure("error", "TextEncoding", file, desc=desc)
+
+
 def init_logs(cli_arg):
     # Initializes logging