scripts/wrong-filename: use Python for speedup + compatibility (#18381)

scripts/README.md

@@ -19,7 +19,7 @@ This section contains a summary of the scripts available in this directory. For
 - [set-more-info-link.py](set-more-info-link.py) is a Python script to generate or update more information links across pages.
 - [set-page-title.py](set-page-title.py) is a Python script to update the title across pages.
 - [test.sh](test.sh) script runs some basic tests on every PR/commit to ensure the pages are valid and the code is formatted correctly.
-- [wrong-filename.sh](wrong-filename.sh) script checks the consistency between the filenames and the page title.
+- [wrong-filename.py](wrong-filename.py) script checks the consistency between the filenames and the page title.
 - [update-command.py](update-command.py) is a Python script to update the common contents of a command example across all languages.
 
 ## Compatibility
@@ -34,5 +34,5 @@ The table below shows the compatibility of user-executable scripts with differen
 | [set-alias-pages.py](set-alias-pages.py) | ✅ | ✅ | ✅ |
 | [set-more-info-link.py](set-more-info-link.py) | ✅ | ✅ | ✅ |
 | [set-page-title.py](set-page-title.py) | ✅ | ✅ | ✅ |
-| [wrong-filename.sh](wrong-filename.sh) | ✅ | ❌ | ❌ (WSL ✅)|
+| [wrong-filename.py](wrong-filename.py) | ✅ | ✅ | ✅ |
 | [update-command.py](update-command.py) | ✅ | ✅ | ✅ |

scripts/wrong-filename.py

@@ -0,0 +1,104 @@
+#!/usr/bin/env python3
+# SPDX-License-Identifier: MIT
+
+import sys
+from pathlib import Path
+import re
+
+OUTPUT_FILE = Path("inconsistent-filenames.txt")
+
+IGNORE_SET = {
+    ">",
+    "<",
+    "<>",
+    ":",
+    "?",
+    "|",
+    "jc.json",
+    "lid.libuser",
+    "mc.cli",
+    "mc.fm",
+    "pacman d",
+    "pacman f",
+    "pacman q",
+    "pacman r",
+    "pacman s",
+    "pacman t",
+    "pacman u",
+    "parted",
+    "print.runmailcap",
+    "print.win",
+    "print.zsh",
+    "python m json.tool",
+    "rename",
+    "snap.esa",
+    "snap.pkg",
+}
+
+
+def normalize(text: str) -> str:
+    """
+    Normalize a string:
+    - replace '-' with spaces
+    - lowercase
+    - collapse multiple spaces into one
+    - strip leading/trailing whitespace
+    """
+    text = text.replace("-", " ").lower().strip()
+    text = re.sub(r"\s+", " ", text)
+    return text
+
+
+def check_file(path: Path) -> str | None:
+    """Check a single markdown file for name/title consistency."""
+    filename = path.name
+
+    # Remove known suffixes
+    command_file = filename
+    for suffix in (".md", ".fish", ".js", ".1", ".2", ".3"):
+        if command_file.endswith(suffix):
+            command_file = command_file[: -len(suffix)]
+    command_file = normalize(command_file)
+
+    try:
+        with path.open("r", encoding="utf-8") as f:
+            firstline = f.readline().strip()
+    except Exception as exc:
+        return f"Error reading {path}: {exc}"
+
+    if not firstline.startswith("#"):
+        return f"Inconsistency found in file: {path} has no title"
+
+    command_page = normalize(firstline[2:])
+
+    # Skip if either filename or title is in the ignore set
+    if command_file in IGNORE_SET or command_page in IGNORE_SET:
+        return None
+
+    if command_file != command_page:
+        return (
+            f"Inconsistency found in file: {path}: "
+            f"{command_page} should be {command_file}"
+        )
+    return None
+
+
+def main() -> int:
+    """Run the filename consistency check."""
+    base_dirs = [p for p in Path(".").glob("pages*") if p.is_dir()]
+    files = [f for base in base_dirs for f in base.rglob("*.md")]
+
+    # Ensure OUTPUT_FILE is always empty at the start
+    OUTPUT_FILE.write_text("", encoding="utf-8")
+
+    with OUTPUT_FILE.open("a", encoding="utf-8") as out:
+        for path in files:
+            result = check_file(path)
+            if result:
+                out.write(result + "\n")
+
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())

scripts/wrong-filename.sh

@@ -1,29 +0,0 @@
-#!/usr/bin/env bash
-# SPDX-License-Identifier: MIT
-
-# This script checks consistency between the filenames and the page title.
-# Usage: ./scripts/wrong-filename.sh
-
-# Output file for recording inconsistencies
-OUTPUT_FILE="inconsistent-filenames.txt"
-# Remove existing output file (if any)
-rm -f "$OUTPUT_FILE"
-touch "$OUTPUT_FILE"
-
-IGNORE_LIST=(">" "<" "<>" ":" "?" "jc.json" "lid.libuser" "mc.cli" "mc.fm" "pacman  d" "pacman  f" "pacman  q" "pacman  r" "pacman  s" "pacman  t" "pacman  u" "parted" "print.runmailcap" "print.win" "python  m json.tool" "rename" "snap.esa" "snap.pkg")
-
-set -e
-
-# Iterate through all Markdown files in the 'pages' directories
-find pages* -name '*.md' -type f | while read -r path; do
-  # Extract the expected command name from the filename
-  COMMAND_NAME_FILE=$(basename "$path" | head -c-4 | sed 's/\.fish//' | sed 's/\.js//' | sed 's/\.1//' | sed 's/\.2//' | sed 's/\.3//' | tr '-' ' ' | tr '[:upper:]' '[:lower:]')
-
-  # Extract the command name from the first line of the Markdown file
-  COMMAND_NAME_PAGE=$(head -n1 "$path" | tail -c+3 | sed 's/--//' | tr '-' ' ' | tr '[:upper:]' '[:lower:]')
-
-  # Check if there is a mismatch between filename and content command names
-  if [[ "$COMMAND_NAME_FILE" != "$COMMAND_NAME_PAGE" && ! ${IGNORE_LIST[*]} =~ $COMMAND_NAME_PAGE ]]; then
-    echo "Inconsistency found in file: $path: $COMMAND_NAME_PAGE should be $COMMAND_NAME_FILE" >> "$OUTPUT_FILE"
-  fi
-done