Don't touch the style of generated files

Ideally the result of the generator would conform to the code style, but
this would be difficult, especially with respect to the placement of line
breaks in long logical lines. So, to avoid surprises when checking the style
of generated files (which happens in releases and in long-time support
branches), systematically skip generated files.

Signed-off-by: Gilles Peskine <Gilles.Peskine@arm.com>

Author: gilles-peskine-arm

scripts/code_style.py

@@ -22,20 +22,44 @@
 import argparse
 import io
 import os
+import re
 import subprocess
 import sys
-from typing import List
+from typing import FrozenSet, List
 
 UNCRUSTIFY_SUPPORTED_VERSION = "0.75.1"
 CONFIG_FILE = ".uncrustify.cfg"
 UNCRUSTIFY_EXE = "uncrustify"
 UNCRUSTIFY_ARGS = ["-c", CONFIG_FILE]
 STDOUT_UTF8 = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')
 STDERR_UTF8 = io.TextIOWrapper(sys.stderr.buffer, encoding='utf-8')
+CHECK_GENERATED_FILES = "tests/scripts/check-generated-files.sh"
 
 def print_err(*args):
     print("Error: ", *args, file=STDERR_UTF8)
 
+# Match FILENAME(s) in "check SCRIPT (FILENAME...)"
+CHECK_CALL_RE = re.compile(r"\n\s*check\s+[^\s#$&*?;|]+([^\n#$&*?;|]+)",
+                           re.ASCII)
+def list_generated_files() -> FrozenSet[str]:
+    """Return the names of generated files.
+
+    We don't reformat generated files, since the result might be different
+    from the output of the generator. Ideally the result of the generator
+    would conform to the code style, but this would be difficult, especially
+    with respect to the placement of line breaks in long logical lines.
+    """
+    # Parse check-generated-files.sh to get an up-to-date list of
+    # generated files. Read the file rather than calling it so that
+    # this script only depends on Git, Python and uncrustify, and not other
+    # tools such as sh or grep which might not be available on Windows.
+    # This introduces a limitation: check-generated-files.sh must have
+    # the expected format and must list the files explicitly, not through
+    # wildcards or command substitution.
+    content = open(CHECK_GENERATED_FILES, encoding="utf-8").read()
+    checks = re.findall(CHECK_CALL_RE, content)
+    return frozenset(word for s in checks for word in s.split())
+
 def get_src_files() -> List[str]:
     """
     Use git ls-files to get a list of the source files
@@ -52,11 +76,14 @@ def get_src_files() -> List[str]:
         print_err("git ls-files returned: " + str(result.returncode))
         return []
     else:
+        generated_files = list_generated_files()
         src_files = str(result.stdout, "utf-8").split()
-        # Don't correct style for files in 3rdparty/
-        src_files = list(filter( \
-                lambda filename: not filename.startswith("3rdparty/"), \
-                src_files))
+        # Don't correct style for third-party files (and, for simplicity,
+        # companion files in the same subtree), or for automatically
+        # generated files (we're correcting the templates instead).
+        src_files = [filename for filename in src_files
+                     if not (filename.startswith("3rdparty/") or
+                             filename in generated_files)]
         return src_files
 
 def get_uncrustify_version() -> str:

tests/scripts/check-generated-files.sh

@@ -116,6 +116,11 @@ check()
     fi
 }
 
+# Note: if the format of calls to the "check" function changes, update
+# scripts/code_style.py accordingly. For generated C source files (*.h or *.c),
+# the format must be "check SCRIPT FILENAME...". For other source files,
+# any shell syntax is permitted (including e.g. command substitution).
+
 check scripts/generate_errors.pl library/error.c
 check scripts/generate_query_config.pl programs/test/query_config.c
 check scripts/generate_driver_wrappers.py library/psa_crypto_driver_wrappers.c