From 0ce2e75560dde29a5f0eb8680e2123b36b8e1e68 Mon Sep 17 00:00:00 2001
From: Roman971 <32455037+Roman971@users.noreply.github.com>
Date: Sat, 30 Jul 2022 15:24:52 +0200
Subject: [PATCH] Rewrite format script in python to improve speed with
 multiprocessing (#1331)

* Rewrite format script in python to improve speed with multiprocessing

* Make format.py executable
---
 docs/tutorial/merging.md |   2 +-
 format.py                | 172 +++++++++++++++++++++++++++++++++++++++
 format.sh                |  84 -------------------
 3 files changed, 173 insertions(+), 85 deletions(-)
 create mode 100755 format.py
 delete mode 100755 format.sh

diff --git a/docs/tutorial/merging.md b/docs/tutorial/merging.md
index f2554d31e8..a5e7b66f69 100644
--- a/docs/tutorial/merging.md
+++ b/docs/tutorial/merging.md
@@ -59,7 +59,7 @@ If you can't match a function even with everyone's, don't worry overlong about i
 
 ### Format
 
-Run the formatting script `format.sh`, to format the C files in the standard way we use.
+Run the formatting script `format.py`, to format the C files in the standard way we use.
 
 ### Merge master
 
diff --git a/format.py b/format.py
new file mode 100755
index 0000000000..b1204aa511
--- /dev/null
+++ b/format.py
@@ -0,0 +1,172 @@
+#!/usr/bin/env python3
+
+import argparse
+import glob
+import multiprocessing
+import os
+import re
+import shutil
+import subprocess
+import sys
+import tempfile
+from functools import partial
+from typing import List
+
+
+# clang-format, clang-tidy and clang-apply-replacements default version
+# Version 11 is used when available for more consistency between contributors
+CLANG_VER = 11
+
+# Clang-Format options (see .clang-format for rules applied)
+FORMAT_OPTS = "-i -style=file"
+
+# Clang-Tidy options (see .clang-tidy for checks enabled)
+TIDY_OPTS = "-p . --fix --fix-errors"
+
+# Compiler options used with Clang-Tidy
+# Normal warnings are disabled with -Wno-everything to focus only on tidying
+INCLUDES = "-Iinclude -Isrc -Ibuild -I."
+DEFINES = "-D_LANGUAGE_C -DNON_MATCHING"
+COMPILER_OPTS = f"-fno-builtin -std=gnu90 -m32 -Wno-everything {INCLUDES} {DEFINES}"
+
+
+def get_clang_executable(allowed_executables: List[str]):
+    for executable in allowed_executables:
+        try:
+            subprocess.check_call([executable, "--version"], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+            return executable
+        except FileNotFoundError or subprocess.CalledProcessError:
+            pass
+    return None
+
+
+def get_tidy_version(tidy_executable: str):
+    tidy_version_run = subprocess.run([tidy_executable, "--version"], stdout=subprocess.PIPE, text=True)
+    match = re.search(r"LLVM version ([0-9]+)", tidy_version_run.stdout)
+    return int(match.group(1))
+
+
+CLANG_FORMAT = get_clang_executable([f"clang-format-{CLANG_VER}", "clang-format"])
+if CLANG_FORMAT is None:
+    sys.exit(f"Error: neither clang-format nor clang-format-{CLANG_VER} found")
+
+CLANG_TIDY = get_clang_executable([f"clang-tidy-{CLANG_VER}", "clang-tidy"])
+if CLANG_TIDY is None:
+    sys.exit(f"Error: neither clang-tidy nor clang-tidy-{CLANG_VER} found")
+
+CLANG_APPLY_REPLACEMENTS = get_clang_executable([f"clang-apply-replacements-{CLANG_VER}", "clang-apply-replacements"])
+
+# Try to detect the clang-tidy version and add --fix-notes for version 13+
+# This is used to ensure all fixes are applied properly in recent versions
+if get_tidy_version(CLANG_TIDY) >= 13:
+    TIDY_OPTS += " --fix-notes"
+
+
+def list_chunks(list: List, chunk_length: int):
+    for i in range(0, len(list), chunk_length):
+        yield list[i : i + chunk_length]
+
+
+def run_clang_format(files: List[str]):
+    exec_str = f"{CLANG_FORMAT} {FORMAT_OPTS} {' '.join(files)}"
+    subprocess.run(exec_str, shell=True)
+
+
+def run_clang_tidy(files: List[str]):
+    exec_str = f"{CLANG_TIDY} {TIDY_OPTS} {' '.join(files)} -- {COMPILER_OPTS}"
+    subprocess.run(exec_str, shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+
+
+def run_clang_tidy_with_export(tmp_dir: str, files: List[str]):
+    (handle, tmp_file) = tempfile.mkstemp(suffix=".yaml", dir=tmp_dir)
+    os.close(handle)
+
+    exec_str = f"{CLANG_TIDY} {TIDY_OPTS} --export-fixes={tmp_file} {' '.join(files)} -- {COMPILER_OPTS}"
+    subprocess.run(exec_str, shell=True, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+
+
+def run_clang_apply_replacements(tmp_dir: str):
+    exec_str = f"{CLANG_APPLY_REPLACEMENTS} --format --style=file --style-config=. {tmp_dir}"
+    subprocess.run(exec_str, shell=True, stderr=subprocess.DEVNULL)
+
+
+def add_final_new_line(file: str):
+    # https://backreference.org/2010/05/23/sanitizing-files-with-no-trailing-newline/index.html
+    # "gets the last character of the file pipes it into read, which will exit with a nonzero exit
+    # code if it encounters EOF before newline (so, if the last character of the file isn't a newline).
+    # If read exits nonzero, then append a newline onto the file using echo (if read exits 0,
+    # that satisfies the ||, so the echo command isn't run)." (https://stackoverflow.com/a/34865616)
+    exec_str = f"tail -c1 {file} | read -r _ || echo >> {file}"
+    subprocess.run(exec_str, shell=True)
+
+
+def format_files(src_files: List[str], extra_files: List[str], nb_jobs: int):
+    if nb_jobs != 1:
+        print(f"Formatting files with {nb_jobs} jobs")
+    else:
+        print(f"Formatting files with a single job (consider using -j to make this faster)")
+
+    # Format files in chunks to improve performance while still utilizing jobs
+    file_chunks = list(list_chunks(src_files, (len(src_files) // nb_jobs) + 1))
+
+    print("Running clang-format...")
+    # clang-format only applies changes in the given files, so it's safe to run in parallel
+    with multiprocessing.get_context("fork").Pool(nb_jobs) as pool:
+        pool.map(run_clang_format, file_chunks)
+
+    print("Running clang-tidy...")
+    if nb_jobs > 1:
+        # clang-tidy may apply changes in #included files, so when running it in parallel we use --export-fixes
+        # then we call clang-apply-replacements to apply all suggested fixes at the end
+        tmp_dir = tempfile.mkdtemp()
+
+        try:
+            with multiprocessing.get_context("fork").Pool(nb_jobs) as pool:
+                pool.map(partial(run_clang_tidy_with_export, tmp_dir), file_chunks)
+
+            run_clang_apply_replacements(tmp_dir)
+        finally:
+            shutil.rmtree(tmp_dir)
+    else:
+        run_clang_tidy(src_files)
+
+    print("Adding missing final new lines...")
+    # Adding final new lines is safe to do in parallel and can be applied to all types of files
+    with multiprocessing.get_context("fork").Pool(nb_jobs) as pool:
+        pool.map(add_final_new_line, src_files + extra_files)
+
+    print("Done formatting files.")
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Format files in the codebase to enforce most style rules")
+    parser.add_argument("files", metavar="file", nargs="*")
+    parser.add_argument(
+        "-j",
+        dest="jobs",
+        type=int,
+        nargs="?",
+        default=1,
+        help="number of jobs to run (default: 1 without -j, number of cpus with -j)",
+    )
+    args = parser.parse_args()
+
+    nb_jobs = args.jobs or multiprocessing.cpu_count()
+    if nb_jobs > 1:
+        if CLANG_APPLY_REPLACEMENTS is None:
+            sys.exit(
+                f"Error: neither clang-apply-replacements nor clang-apply-replacements-{CLANG_VER} found (required to use -j)"
+            )
+
+    if args.files:
+        files = args.files
+        extra_files = []
+    else:
+        files = glob.glob("src/**/*.c", recursive=True)
+        extra_files = glob.glob("assets/**/*.xml", recursive=True)
+
+    format_files(files, extra_files, nb_jobs)
+
+
+if __name__ == "__main__":
+    main()
diff --git a/format.sh b/format.sh
deleted file mode 100755
index 004274bfc8..0000000000
--- a/format.sh
+++ /dev/null
@@ -1,84 +0,0 @@
-#!/usr/bin/env bash
-
-# Clang-Format version and options (see .clang-format for rules applied)
-# Version 11 is used when available for more consistency between contributors
-FORMAT_VER="11"
-FORMAT_OPTS="-i -style=file"
-
-# Clang-Tidy options (see .clang-tidy for checks enabled)
-TIDY_OPTS="-p . --fix --fix-errors"
-
-# Compiler options used with Clang-Tidy
-# Normal warnings are disabled with -Wno-everything to focus only on tidying
-INCLUDES="-Iinclude -Isrc -Ibuild -I."
-DEFINES="-D_LANGUAGE_C -DNON_MATCHING"
-COMPILER_OPTS="-fno-builtin -std=gnu90 -m32 -Wno-everything ${INCLUDES} ${DEFINES}"
-
-TIDY_VERSION_REGEX="LLVM version ([0-9]+)"
-
-# https://backreference.org/2010/05/23/sanitizing-files-with-no-trailing-newline/index.html
-# "gets the last character of the file pipes it into read, which will exit with
-# a nonzero exit code if it encounters EOF before newline (so, if the last
-# character of the file isn't a newline). If read exits nonzero, then append a
-# newline onto the file using echo (if read exits 0, that satisfies the ||, so
-# the echo command isn't run)." (https://stackoverflow.com/a/34865616)
-function add_final_newline () {
-    for file in "$@"
-    do
-        tail -c1 $file | read -r _ || echo >> $file
-    done
-}
-export -f add_final_newline
-
-shopt -s globstar
-
-if [ $(command -v clang-format-${FORMAT_VER}) ]
-then
-    CLANG_FORMAT="clang-format-${FORMAT_VER}"
-else
-    if [ $(command -v clang-format) ]
-    then
-        CLANG_FORMAT="clang-format"
-    else
-        echo "Neither clang-format nor clang-format-${FORMAT_VER} found. Exiting."
-        exit 1
-    fi
-fi
-
-if [ $(command -v clang-tidy) ]
-then
-    CLANG_TIDY="clang-tidy"
-else
-    echo "clang-tidy not found. Exiting."
-    exit 1
-fi
-
-# Try to detect the clang-tidy version and add --fix-notes for version 13+
-# This is used to ensure all fixes are applied properly in recent versions
-if [[ $(${CLANG_TIDY} --version) =~ $TIDY_VERSION_REGEX ]]; then
-    if (( ${BASH_REMATCH[1]} >= 13 )); then
-        TIDY_OPTS="${TIDY_OPTS} --fix-notes"
-    fi
-fi
-
-if (( $# > 0 )); then
-    echo "Formatting file(s) $*"
-    echo "Running clang-format..."
-    ${CLANG_FORMAT} ${FORMAT_OPTS} "$@"
-    echo "Running clang-tidy..."
-    ${CLANG_TIDY} ${TIDY_OPTS} "$@" -- ${COMPILER_OPTS} &> /dev/null
-    echo "Adding missing final new lines..."
-    add_final_newline "$@"
-    echo "Done formatting file(s) $*"
-    exit
-fi
-
-echo "Formatting C files. This will take a bit"
-echo "Running clang-format..."
-${CLANG_FORMAT} ${FORMAT_OPTS} src/**/*.c
-echo "Running clang-tidy..."
-${CLANG_TIDY} ${TIDY_OPTS} src/**/*.c -- ${COMPILER_OPTS} &> /dev/null
-echo "Adding missing final new lines..."
-find src/ -type f -name "*.c" -exec bash -c 'add_final_newline "$@"' bash {} +
-find assets/xml/ -type f -name "*.xml" -exec bash -c 'add_final_newline "$@"' bash {} +
-echo "Done formatting all files."