[Tools] (scripts): format exactly the files the format check covers

- Skip MG_Impl/GLImpl/Exporting/Definitions.cpp, whose table clang-format cannot wrap

scripts/check_text_hygiene.py

@@ -89,6 +89,10 @@ SOURCE_DIRS = ("MobileGL/", "scripts/", "tools/")
 # Files whose exact bytes are pinned elsewhere: changing them, even a comment, means a protocol
 # revision bump (scripts/ci/protocol_revision_pins.json), so only the UTF-8 rule applies to them.
 FROZEN = {"MobileGL/MG_Remote/Protocol/protocol.fbs"}
+# Left unformatted: its export table's HEAD/END macro pairs end no statement, so clang-format reads
+# the whole table as one expression and needs minutes and gigabytes to wrap it, even under
+# `clang-format off`.
+FORMAT_SKIP = {"MobileGL/MG_Impl/GLImpl/Exporting/Definitions.cpp"}
 NOTES_DIR = "docs/Disaggregated/notes/"
 # The archived planning notes and agent contracts. The commit that adds ARCHIVE_MARKER fills the
 # archive; every later commit leaves it alone.
@@ -112,6 +116,12 @@ def is_generated(path):
     return "/generated/" in path or "/gen/" in path or os.path.basename(path).startswith("generated_")
 
 
+def is_formatted(path):
+    """Whether clang-format owns the file's layout: the format check and format_code.sh cover it."""
+    return (path.startswith(SOURCE_DIRS) and path.endswith(FORMAT_EXT) and not is_generated(path)
+            and path not in FORMAT_SKIP)
+
+
 def submodule_paths(repo=ROOT):
     out = git("config", "--file", ".gitmodules", "--get-regexp", r"submodule\..*\.path", check=False, repo=repo)
     return {line.split(" ", 1)[1].strip() for line in out.splitlines() if " " in line}
@@ -595,7 +605,7 @@ def check_format(rev_range):
         if is_merge(sha) or not git("rev-parse", "--verify", "--quiet", sha + "^", check=False).strip():
             continue
         for path, ranges in unmoved_ranges(git("diff", "-U0", "-M", "--no-color", sha + "^", sha)).items():
-            if not (path.endswith(FORMAT_EXT) and path.startswith(SOURCE_DIRS)) or is_generated(path):
+            if not is_formatted(path):
                 continue
             checked += 1
             content = subprocess.run(["git", "-C", ROOT, "show", "%s:%s" % (sha, path)], stdout=subprocess.PIPE).stdout

scripts/format_code.sh

@@ -7,156 +7,36 @@
 # SPDX-License-Identifier: LGPL-3.0-only
 # End of Source File Header
 
-set -uo pipefail
+# Reformat in place every file the CI format check covers (is_formatted in check_text_hygiene.py:
-if [ -t 1 ]; then
+# tracked C/C++ sources, minus generated files, .inc fragments and FORMAT_SKIP). The tree is
-    CYAN=$'\033[36m'
+# already formatted, so this only rewrites lines that drifted. CLANG_FORMAT overrides the binary.
-    GREEN=$'\033[32m'
+set -euo pipefail
-    YELLOW=$'\033[33m'
+cd "$(dirname "$0")/.."
-    RED=$'\033[31m'
+exec python3 - <<'EOF'
-    BOLD=$'\033[1m'
+import importlib.util
-    RESET=$'\033[0m'
+import os
-else
+import shutil
-    CYAN='' ; GREEN='' ; YELLOW='' ; RED='' ; BOLD='' ; RESET=''
+import subprocess
-fi
+import sys
+from concurrent.futures import ThreadPoolExecutor
 
-OK="✔"; SKIP="⤼"; FAIL="✖"
+spec = importlib.util.spec_from_file_location("hygiene", os.path.join("scripts", "check_text_hygiene.py"))
+hygiene = importlib.util.module_from_spec(spec)
+spec.loader.exec_module(hygiene)
+binary = os.environ.get("CLANG_FORMAT") or shutil.which("clang-format-23") or shutil.which("clang-format")
+if not binary:
+    sys.exit("clang-format not found; install it with: pip install 'clang-format==23.1.3'")
+print(subprocess.run([binary, "--version"], stdout=subprocess.PIPE).stdout.decode().strip())
+files = [f for f in hygiene.git("ls-files", *hygiene.SOURCE_DIRS).splitlines() if hygiene.is_formatted(f)]
 
-cd "$(dirname "$0")" || { printf "%s\n" "Cannot cd to script dir"; exit 1; }
 
-DIR="../MobileGL"
+def format_one(path):
-CLANG_FORMAT_CONFIG="../.clang-format"
+    before = open(path, "rb").read()
+    subprocess.run([binary, "-i", "--style=file", path], check=True)
+    return open(path, "rb").read() != before
 
-CLANG_FORMAT_PATH="$(command -v clang-format-23 2>/dev/null || true)"
-if [ -z "$CLANG_FORMAT_PATH" ]; then
-    CLANG_FORMAT_PATH="$(command -v clang-format 2>/dev/null || true)"
-fi
-if [ -z "$CLANG_FORMAT_PATH" ]; then
-    printf '%s\n' "${RED}Error:${RESET} no clang-format found in PATH (tried clang-format-21 and clang-format)."
-    exit 1
-fi
 
-if [ ! -f "$CLANG_FORMAT_CONFIG" ]; then
+with ThreadPoolExecutor(os.cpu_count() or 4) as pool:
-    printf '%s\n' "${RED}Error:${RESET} clang-format config not found at ${CLANG_FORMAT_CONFIG}"
+    changed = sum(pool.map(format_one, files))
-    exit 1
+print("%d of %d files reformatted" % (changed, len(files)))
-fi
+EOF
-
-printf '%s %s\n' "${CYAN}${BOLD}Using clang-format:${RESET}" "${CLANG_FORMAT_PATH}"
-printf '%s %s\n' "${CYAN}${BOLD}Config file:${RESET}" "${CLANG_FORMAT_CONFIG}"
-DIR_ABS="(missing)"
-if [ -d "$DIR" ]; then
-    DIR_ABS="$(cd "$DIR" && pwd)"
-fi
-printf '%s %s -> %s\n\n' "${CYAN}${BOLD}Target dir:${RESET}" "${DIR}" "${DIR_ABS}"
-
-if [ "$DIR_ABS" = "(missing)" ]; then
-    printf '%s\n' "${YELLOW}Target directory '${DIR}' does not exist relative to script dir '$(pwd)'.${RESET}"
-    printf '%s\n' "ls -la of current dir:"
-    ls -la
-    exit 1
-fi
-
-candidates=()
-while IFS= read -r -d '' f; do
-    candidates+=("$f")
-done < <(
-    find "$DIR" \
-        \( -type d \( -name out -o -name build \) -prune \) -o \
-        -type f \( -name "*.h" -o -name "*.hpp" -o -name "*.cpp" -o -name "*.cc" -o -name "*.c" \) -print0 2>/dev/null
-)
-
-printf '%s\n' "${YELLOW}find results:${RESET} ${#candidates[@]} candidate(s) found."
-if [ "${#candidates[@]}" -eq 0 ]; then
-    printf '%s\n' "${YELLOW}No files matched.${RESET}"
-    ls -la "$DIR" | sed -n '1,40p'
-    exit 0
-fi
-
-todo=()
-skipped_def=0
-skipped_big=0
-skipped_err=0
-i=0
-total_candidates=${#candidates[@]}
-for f in "${candidates[@]}"; do
-    i=$((i+1))
-    if (( i % 50 == 0 )); then
-        printf '%s\n' "${CYAN}Filtering progress:${RESET} ${i}/${total_candidates}"
-    fi
-
-    base=$(basename "$f" 2>/dev/null) || base=""
-    if [ "$base" = "Definitions.cpp" ]; then
-        skipped_def=$((skipped_def+1))
-        continue
-    fi
-
-    line_count=$(awk 'END{print NR+0}' "$f" 2>/dev/null) || { line_count=0; skipped_err=$((skipped_err+1)); printf '%s\n' "${YELLOW}Warn:${RESET} awk failed on '$f' (skipping)"; continue; }
-    if ! [[ "$line_count" =~ ^[0-9]+$ ]]; then
-        skipped_err=$((skipped_err+1))
-        printf '%s\n' "${YELLOW}Warn:${RESET} got non-numeric line count for '$f' -> '${line_count}', skipping"
-        continue
-    fi
-    if [ "$line_count" -gt 10000 ]; then
-        skipped_big=$((skipped_big+1))
-        continue
-    fi
-    todo+=("$f")
-done
-
-printf '\n%s\n\n' "${YELLOW}Filtering result:${RESET} total candidates=${total_candidates}, to-format=${#todo[@]}, skipped Definitions.cpp=${skipped_def}, skipped >10000=${skipped_big}, awk-errors=${skipped_err}"
-
-total=${#todo[@]}
-if [ "$total" -eq 0 ]; then
-    printf '%s\n' "${YELLOW}Nothing to format.${RESET}"
-    exit 0
-fi
-
-OK="✔"; SKIP="⤼"; FAIL="✖"
-ARROW="╰─>"; SIMPLE_ARROW="─>"
-
-draw_progress() {
-    local current=$1 total=$2 width=36
-
-    local percent=0
-    if [ "$total" -gt 0 ]; then
-        percent=$(( (current * 100 + total/2) / total ))
-        if [ "$percent" -gt 100 ]; then percent=100; fi
-    fi
-
-    local filled=0
-    if [ "$total" -gt 0 ]; then
-        filled=$(( (current * width + total - 1) / total ))
-    fi
-    if [ "$filled" -gt "$width" ]; then filled=$width; fi
-    if [ "$filled" -lt 0 ]; then filled=0; fi
-
-    local empty=$(( width - filled ))
-
-    local bar_filled=""
-    local bar_empty=""
-    if [ "$filled" -gt 0 ]; then
-        bar_filled="$(printf '%0.s█' $(seq 1 "$filled") 2>/dev/null || true)"
-    fi
-    if [ "$empty" -gt 0 ]; then
-        bar_empty="$(printf '%0.s░' $(seq 1 "$empty") 2>/dev/null || true)"
-    fi
-
-    printf "\r%s [%3d/%3d] %3d%% |%s%s| " "${CYAN}${BOLD}Progress${RESET}" "$current" "$total" "$percent" "$bar_filled" "$bar_empty"
-}
-
-processed=0
-trap 'printf "\nInterrupted\n"; exit' INT TERM
-
-for f in "${todo[@]}"; do
-    processed=$((processed+1))
-    draw_progress "$processed" "$total"
-
-    printf '\n%s %s %s\n' "${CYAN}Formatting${RESET}" "$SIMPLE_ARROW" "$f"
-    if "$CLANG_FORMAT_PATH" -i -style="file:${CLANG_FORMAT_CONFIG}" "$f"; then
-        printf '   %s %s %s\n' "$ARROW" "${GREEN}${OK}${RESET}" "done"
-    else
-        printf '   %s %s %s\n' "$ARROW" "${RED}${FAIL}${RESET}" "failed"
-    fi
-    draw_progress "$processed" "$total"
-done
-
-printf '\n\n%s %s\n' "${GREEN}All done!${RESET}" "Formatted: ${processed}, Skipped: $((skipped_def + skipped_big + skipped_err))"