refactor: clean up linter scripts

Author: martinus

.github/copilot-instructions.md

@@ -99,7 +99,7 @@ All lint scripts are in `scripts/lint/`. They MUST pass before submitting change
 
 ```bash
 # Run ALL linters (recommended)
-./scripts/lint/lint-all.py
+./scripts/lint/all.py
 
 # Individual linters
 ./scripts/lint/lint-version.py        # Check version consistency across files
@@ -149,7 +149,7 @@ All lint scripts are in `scripts/lint/`. They MUST pass before submitting change
 
 ### Common CI Failures & Fixes
 
-1. **Linting failures**: Run `./scripts/lint/lint-all.py` locally first
+1. **Linting failures**: Run `./scripts/lint/all.py` locally first
 2. **32-bit failures**: Don't use `size_t` in hashes without consideration
 3. **Sanitizer failures**: Check for undefined behavior, use-after-free, data races
 4. **Windows/MSVC**: Check for MSVC-specific warnings (see `test/meson.build`)
@@ -216,7 +216,7 @@ unordered_dense/
 
 5. **Lint**: Verify all linters pass
    ```bash
-   ./scripts/lint/lint-all.py
+   ./scripts/lint/all.py
    ```
 
 6. **Multi-config** (optional): Test across configurations
@@ -300,4 +300,4 @@ unordered_dense/
 - You need information not covered here
 - The repository structure has changed significantly
 
-When in doubt, run `./scripts/lint/lint-all.py` and `meson test -C builddir/<config> -v` to verify your changes.
+When in doubt, run `./scripts/lint/all.py` and `meson test -C builddir/<config> -v` to verify your changes.

scripts/build.py

@@ -54,7 +54,7 @@ def run(cmd):
     if result.returncode != 0:
         exit(result.returncode)
 
-run('scripts/lint/lint-all.py')
+run('scripts/lint/all.py')
 
 for cmd_dir in cmd_and_dir:
     workdir = cmd_dir[-1]

scripts/lint/all.py

@@ -0,0 +1,27 @@
+#!/usr/bin/env python3
+
+from pathlib import Path
+from subprocess import run
+from time import perf_counter
+
+
+def main():
+    start = perf_counter()
+    linters_dir = Path(__file__).parent
+    linters = sorted(
+        p for p in linters_dir.iterdir() if p.is_file() and p.name.startswith("lint-")
+    )
+
+    rc = 0
+    for lint in linters:
+        res = run([str(lint)])
+        if res.returncode:
+            print(f"^---- failure(s) from {lint.name}\n")
+            rc |= res.returncode
+
+    print(f"{len(linters)} linters in {perf_counter() - start:0.2f}s")
+    raise SystemExit(rc)
+
+
+if __name__ == "__main__":
+    main()

scripts/lint/lint-all.py

@@ -1,50 +1,38 @@
 #!/usr/bin/env python3
-
-from glob import glob
 from pathlib import Path
+import re
+import shutil
 from subprocess import run
-from os import path
-import subprocess
 import sys
-from time import time
-import re
 
-root_path = path.abspath(Path(__file__).parent.parent.parent)
-
-globs = [
-    f"{root_path}/include/**/*.h",
-    f"{root_path}/test/**/*.h",
-    f"{root_path}/test/**/*.cpp",
-]
-exclusions = [
-    "nanobench\\.h",
-    "FuzzedDataProvider\\.h",
-    '/third-party/']
-
-files = []
-for g in globs:
-    r = glob(g, recursive=True)
-    files.extend(r)
-
-# filter out exclusions
-for exclusion in exclusions:
-    l = filter(lambda file: re.search(exclusion, file) == None, files)
-    files = list(l)
-
-if len(files) == 0:
-    print("could not find any files!")
-    sys.exit(1)
-
-command = ['clang-format', '--dry-run', '-Werror'] + files
-p = subprocess.Popen(command,
-                     stdout=subprocess.PIPE,
-                     stderr=None,
-                     stdin=subprocess.PIPE,
-                     universal_newlines=True)
-
-stdout, stderr = p.communicate()
-
-print(f"clang-format checked {len(files)} files")
-
-if p.returncode != 0:
-    sys.exit(p.returncode)
+ROOT = Path(__file__).resolve().parents[2]
+PATTERNS = ["include/**/*.h", "test/**/*.h", "test/**/*.cpp"]
+EXCLUDE_RE = re.compile(r"nanobench\.h|FuzzedDataProvider\.h|/third-party/")
+
+
+def collect_files(root: Path):
+    return [
+        f
+        for p in PATTERNS
+        for f in root.glob(p)
+        if f.is_file() and not EXCLUDE_RE.search(str(f))
+    ]
+
+
+def main():
+    files = collect_files(ROOT)
+    if not files:
+        print("could not find any files!")
+        raise SystemExit(1)
+
+    if not (clang_format := shutil.which("clang-format")):
+        print("clang-format not found in PATH")
+        raise SystemExit(2)
+
+    ec = run([clang_format, "--dry-run", "-Werror"] + files).returncode
+    print(f"clang-format checked {len(files)} files")
+    SystemExit(ec)
+
+
+if __name__ == "__main__":
+    main()

scripts/lint/lint-clang-format.py

@@ -1,44 +1,37 @@
 #!/usr/bin/env python3
-"""Run clang-tidy only on unordered_dense.h"""
-
 import argparse
-import subprocess
+import shutil
+from subprocess import run
 import sys
-from pathlib import Path
 
 
 def main():
-    parser = argparse.ArgumentParser(
-        description="Run clang-tidy on unordered_dense.h (via include_only.cpp)"
-    )
-    parser.add_argument(
-        "--std",
-        default="c++17",
-        help="C++ standard to use (default: c++17)",
+    p = argparse.ArgumentParser(
+        description=(
+            "Run clang-tidy on unordered_dense.h (via test/unit/include_only.cpp)"
+        )
     )
-    args = parser.parse_args()
-
-    # Define paths
-    source_file = Path("test/unit/include_only.cpp")
-
-    cmd = [
-        "clang-tidy",
-        str(source_file),
-        "--header-filter=.*unordered_dense\\.h",
-        "--warnings-as-errors=*",
-        "--",
-        f"-std={args.std}",
-        "-I",
-        "include",
-    ]
+    p.add_argument("--std", default="c++17", help="C++ standard (default: c++17)")
+    args = p.parse_args()
 
-    try:
-        subprocess.run(cmd, check=True)
-    except subprocess.CalledProcessError as e:
-        sys.exit(e.returncode)
-    except FileNotFoundError:
+    if not (clang_tidy := shutil.which("clang-tidy")):
         print("Error: clang-tidy not found in PATH", file=sys.stderr)
-        sys.exit(1)
+        raise SystemExit(1)
+
+    # Exit with clang-tidy's exit code.
+    ec = run(
+        [
+            clang_tidy,
+            "test/unit/include_only.cpp",
+            "--header-filter=.*unordered_dense\\.h",
+            "--warnings-as-errors=*",
+            "--",
+            f"-std={args.std}",
+            "-I",
+            "include",
+        ]
+    ).returncode
+    raise SystemExit(ec)
 
 
 if __name__ == "__main__":

scripts/lint/lint-clang-tidy.py

@@ -1,59 +1,45 @@
 #!/usr/bin/env python3
-
-import os
-import pathlib
+from pathlib import Path
 import re
 
-
-root = os.path.abspath(pathlib.Path(__file__).parent.parent.parent)
-
-# filename, pattern, number of occurrences
-file_pattern_count = [
-    (f"{root}/CMakeLists.txt", r"^\s+VERSION (\d+)\.(\d+)\.(\d+)\n", 1),
-    (f"{root}/include/ankerl/stl.h", r"Version (\d+)\.(\d+)\.(\d+)\n", 1),
-    (f"{root}/include/ankerl/unordered_dense.h", r"Version (\d+)\.(\d+)\.(\d+)\n", 1),
-    (f"{root}/meson.build", r"version: '(\d+)\.(\d+)\.(\d+)'", 1),
-    (f"{root}/test/unit/namespace.cpp", r"unordered_dense::v(\d+)_(\d+)_(\d+)", 1),
+# fmt: off
+ROOT = Path(__file__).resolve().parents[2]
+HEADER = ROOT / "include" / "ankerl" / "unordered_dense.h"
+CHECKS = [
+    (HEADER, r"Version (\d+)\.(\d+)\.(\d+)", 1),
+    (ROOT / "CMakeLists.txt", r"^\s+VERSION (\d+)\.(\d+)\.(\d+)", 1),
+    (ROOT / "include" / "ankerl" / "stl.h", r"Version (\d+)\.(\d+)\.(\d+)", 1),
+    (ROOT / "meson.build", r"version:\s*'?(\d+)\.(\d+)\.(\d+)'?", 1),
+    (ROOT / "test" / "unit" / "namespace.cpp", r"unordered_dense::v(\d+)_(\d+)_(\d+)", 1),
 ]
-
-# let's parse the reference from svector.h
-major = "??"
-minor = "??"
-patch = "??"
-with open(f"{root}/include/ankerl/unordered_dense.h", "r") as f:
-    for line in f:
-        r = re.search(r"#define ANKERL_UNORDERED_DENSE_VERSION_([A-Z]+) (\d+)", line)
-        if not r:
-            continue
-
-        if "MAJOR" == r.group(1):
-            major = r.group(2)
-        elif "MINOR" == r.group(1):
-            minor = r.group(2)
-        elif "PATCH" == r.group(1):
-            patch = r.group(2)
-        else:
-            "match but with something else!"
-            exit(1)
-
-is_ok = True
-for filename, pattern, count in file_pattern_count:
-    num_found = 0
-    with open(filename, "r") as f:
-        for line in f:
-            r = re.search(pattern, line)
-            if r:
-                num_found += 1
-                if major != r.group(1) or minor != r.group(2) or patch != r.group(3):
-                    is_ok = False
-                    print(
-                        f"ERROR in {filename}: got '{line.strip()}' but version should be '{major}.{minor}.{patch}'"
-                    )
-    if num_found != count:
-        is_ok = False
-        print(
-            f"ERROR in {filename}: expected {count} occurrences but found it {num_found} times"
+# fmt: on
+
+
+def read_version_from_header(p: Path) -> str:
+    m = re.findall(
+        r"#define\s+ANKERL_UNORDERED_DENSE_VERSION_(MAJOR|MINOR|PATCH)\s+(\d+)",
+        p.read_text(),
+    )
+    d = dict(m)
+    return f"{d['MAJOR']}.{d['MINOR']}.{d['PATCH']}"
+
+
+def main():
+    ref = read_version_from_header(HEADER)
+    errs = []
+    for path, pattern, count in CHECKS:
+        matches = list(re.finditer(pattern, path.read_text(), re.M))
+        if (n := len(matches)) != count:
+            errs.append(f"ERROR: {path}: expected {count} matches, found {n}")
+        errs.extend(
+            f"ERROR: {path}: found version {found}, expected {ref}"
+            for m in matches
+            if (found := ".".join(m.groups())) != ref
         )
 
-if not is_ok:
-    exit(1)
+    print("\n".join(errs))
+    raise SystemExit(1 if errs else 0)
+
+
+if __name__ == "__main__":
+    main()