pre-commit: add a hook for checking for duplicate empty lines

It's a bit annoying to only find those during a CI pipeline run, so
let's add a hook for it. Pre-commit doesn't seem to have something
useful for duplicate line detection so let's hack up a quick script
that can do this for us, together with the other whitespace checks in
the CI so we're consistent.

Excluded from the duplicate line checks are anything "Not C" and the
"include/" directory since we don't control those files.

Part-of: <https://gitlab.freedesktop.org/libinput/libinput/-/merge_requests/1154>

.gitlab-ci.yml

@@ -204,17 +204,7 @@ check-whitespace:
     - .fdo.ci-fairy
   stage: sanity check
   script:
-    # remove trailing whitespaces
+    - .gitlab-ci/whitespace-check.py $(git ls-files)
-    - sed -i 's/ *$//' $(git ls-files)
-    - git diff --exit-code || (echo "ERROR - Trailing whitespaces in patchset, please fix" && false)
-    - sed -i 's/\t*$//' $(git ls-files)
-    - git diff --exit-code || (echo "ERROR - Trailing tabs in patchset, please fix" && false)
-    # search for tab after space
-    - sed -i 's/ \t//' $(git ls-files)
-    - git diff --exit-code || (echo "ERROR - Tab after space in patchset, please fix" && false)
-    # search for duplicated empty lines
-    - sed -i '/^$/N;/^\n$/D' $(git ls-files src/*.{h,c} tools/*.{h,c} udev/*.c test/*.c)
-    - git diff --exit-code || (echo "ERROR - Duplicated empty lines, please fix" && false)
 
 #
 # pre-commit hooks

.gitlab-ci/ci.template

@@ -192,17 +192,7 @@ check-whitespace:
     - .fdo.ci-fairy
   stage: sanity check
   script:
-    # remove trailing whitespaces
+    - .gitlab-ci/whitespace-check.py $(git ls-files)
-    - sed -i 's/ *$//' $(git ls-files)
-    - git diff --exit-code || (echo "ERROR - Trailing whitespaces in patchset, please fix" && false)
-    - sed -i 's/\t*$//' $(git ls-files)
-    - git diff --exit-code || (echo "ERROR - Trailing tabs in patchset, please fix" && false)
-    # search for tab after space
-    - sed -i 's/ \t//' $(git ls-files)
-    - git diff --exit-code || (echo "ERROR - Tab after space in patchset, please fix" && false)
-    # search for duplicated empty lines
-    - sed -i '/^$/N;/^\n$/D' $(git ls-files src/*.{h,c} tools/*.{h,c} udev/*.c test/*.c)
-    - git diff --exit-code || (echo "ERROR - Duplicated empty lines, please fix" && false)
 
 #
 # pre-commit hooks

.gitlab-ci/whitespace-check.py

@@ -0,0 +1,107 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: MIT
+
+from pathlib import Path
+from dataclasses import dataclass
+
+import argparse
+import itertools
+import os
+import sys
+
+
+@dataclass
+class WhitespaceError:
+    message: str
+    lineno: int
+    nlines: int = 1
+    column: None | int = None
+    ncolumns: int = 1
+
+
+def test_duplicate_empty_lines(lines: list[str]) -> list[WhitespaceError]:
+    errors = []
+    for idx, (l1, l2) in enumerate(itertools.pairwise(lines)):
+        if not l1 and not l2:
+            errors.append(WhitespaceError("Duplicated empty lines", idx, nlines=2))
+    return errors
+
+
+def test_tab_after_space(lines: list[str]) -> list[WhitespaceError]:
+    errors = []
+    for idx, l in enumerate(lines):
+        index = l.find(" \t")
+        if index > -1:
+            errors.append(
+                WhitespaceError(
+                    "Tab after space", idx, nlines=index, column=index, ncolumns=2
+                )
+            )
+    return errors
+
+
+def test_trailing_whitespace(lines: list[str]) -> list[WhitespaceError]:
+    errors = []
+    for idx, l in enumerate(lines):
+        if l.rstrip() != l:
+            errors.append(WhitespaceError("Trailing whitespace", idx))
+    return errors
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Whitespace checker script")
+    parser.add_argument(
+        "files",
+        metavar="FILES",
+        type=Path,
+        nargs="+",
+        help="The files to check",
+    )
+
+    args = parser.parse_args()
+
+    have_errors: bool = False
+
+    if os.isatty(sys.stderr.fileno()):
+        red = "\x1b[0;31m"
+        reset = "\x1b[0m"
+    else:
+        red = ""
+        reset = ""
+
+    for file in args.files:
+        lines = [l.rstrip("\n") for l in file.open().readlines()]
+
+        errors = []
+        errors.extend(test_tab_after_space(lines))
+        errors.extend(test_trailing_whitespace(lines))
+        if any(file.name.endswith(suffix) for suffix in [".c", ".h"]):
+            if not file.parts[0] == "include":
+                errors.extend(test_duplicate_empty_lines(lines))
+
+        for e in errors:
+            print(f"{red}ERROR: {e.message} in {file}:{reset}", file=sys.stderr)
+            print(f"{'-' * 72}", file=sys.stderr)
+            lineno = max(0, e.lineno - 5)
+            for idx, l in enumerate(lines[lineno : lineno + 10]):
+                if e.lineno <= lineno + idx < e.lineno + e.nlines:
+                    prefix = "->"
+                    hl = red
+                    nohl = reset
+                else:
+                    prefix = "  "
+                    hl = ""
+                    nohl = ""
+                print(f"{hl}{lineno + idx:3d}: {prefix} {l.rstrip()}{nohl}")
+
+            print(f"{'-' * 72}", file=sys.stderr)
+
+        if errors:
+            have_errors = True
+
+    if have_errors:
+        sys.exit(1)
+
+
+if __name__ == "__main__":
+    main()

.pre-commit-config.yaml

@@ -18,3 +18,9 @@ repos:
   hooks:
   - id: check-commits
   - id: generate-template
+- repo: local
+  hooks:
+  - id: run-sed-script
+    name: Check for whitespace errors
+    entry: ./.gitlab-ci/whitespace-check.py
+    language: system