ci(check-files): replace bash with faster python

Apply some tricks to speed up this lookup. Use proper regex escaping on
file names to prevent false positives. Also filter out files to prevent
picking up self-references in comments.

Signed-off-by: Randolph Sapp <[email protected]>

Author: StaticRocket

.github/workflows/check-files.yml

@@ -32,23 +32,23 @@ jobs:
           git fetch --no-tags --depth=1 origin master
           git switch master
 
-      - name: Run check-files.sh
+      - name: Run check_files.py
         run: |
           # Disable color output
           export NO_COLOR=true
 
           # Run the test
-          bin/delta.sh -a master -b pr -- ./bin/check-files.sh
+          bin/delta.sh -a master -b pr -- ./bin/check_files.py
 
           # Prepare summary
           WARNING_COUNT=$(wc -l < _new-warn.log)
           if [ "$WARNING_COUNT" -gt "0" ]; then
-            echo "New unreachable files found with check-files.sh:"
+            echo "New unreachable files found with check_files.py:"
             echo '```text'
             cat _new-warn.log
             echo '```'
           else
-            echo "No new unreachable files found with check-files.sh"
+            echo "No new unreachable files found with check_files.py"
           fi >> "$GITHUB_STEP_SUMMARY"
 
           # Prepare the artifacts

bin/check-files.sh

@@ -0,0 +1,82 @@
+#!/usr/bin/env python3
+
+"""Tool to check that all files are being used
+
+SPDX-License-Identifier: MIT
+Copyright (C) 2025 Texas Instruments Incorporated - https://www.ti.com
+"""
+
+import logging
+import re
+from pathlib import Path
+
+logger = logging.getLogger(__name__)
+
+SOURCE_PATH = Path("source/")
+RST_SOURCE = set(SOURCE_PATH.glob("**/*.rst"))
+IGNORED = re.compile(r"([^_].*\.rst)|(version\.txt)")
+
+
+def get_names(base):
+    """Get a set of file names to check for, ignoring anything in that matches the IGNORED regex.
+
+    :param base: Pathlib path to directory to search
+    :return: Set of string path names
+    """
+    files_to_check = set()
+    for file in base.glob("**/*"):
+        if file.is_dir():
+            continue
+
+        name = file.name
+        if IGNORED.match(name):
+            logger.debug("Ignored: %s", name)
+            continue
+
+        files_to_check.add(name)
+    return files_to_check
+
+
+def check_file(string, file):
+    """Check to see if the given string appears in the file.
+
+    :param string: String to look up
+    :param file: Pathlib path to file
+    :return: Boolean based on presence of string
+    """
+    pattern = re.compile(re.escape(string))
+    text = file.read_text(encoding="utf-8")
+    for _ in pattern.finditer(text):
+        return True
+    return False
+
+
+def check_all(string):
+    """Use an scan for any matches in RST_SOURCE files. Do not look for matches in the file itself.
+    That last bit is particularly relevant for RST files that exist to be included in other files.
+
+    :param string: String to look up
+    :return: Boolean based on presence of string in any other files
+    """
+    for file in RST_SOURCE:
+        if file == string:
+            continue
+
+        if check_file(string, file):
+            return True
+    return False
+
+
+def main():
+    """Main CLI entrypoint"""
+    logging.basicConfig(level=logging.INFO)
+
+    files_to_check = get_names(SOURCE_PATH)
+    for filename in files_to_check:
+        if check_all(filename):
+            continue
+        logging.info("File not used: %s", filename)
+
+
+if __name__ == "__main__":
+    main()