Swap out those pesty SMB illegals with benign characters!

rmSMB_Illegals.sh

#!/bin/bash

# USAGE: ./rmSMB_Illegals.sh path

# Find searches through each file on a filesystem and passes the filenames to a
# function that tests for the presence of illegal characters and remediates.
# Bad characters are either removed or replaced with a ``-''.

# This has been tested on APFS on macOS 10.13.5. It'd be pretty safe to run this
# on a mounted volume, but it'd be nicer to test on Synology Linux and have it
# run directly.

LOG_FILE="$(pwd)/$0.log"
REGEX='^[][a-zA-Z0-9_ .-]*$'
TRAILING_SPACE=' $'
TRAILING_PERIOD='\.$'

writeToLog(){
  local LOG_DIRNAME
  LOG_DIRNAME=$(dirname "${LOG_FILE}")
  if [[ ! -d "${LOG_DIRNAME}" ]]; then
    mkdir -p "${LOG_DIRNAME}"
    touch "${LOG_FILE}"
    printf "%s Log and log parent directory created.\n" \
      "$(date "+%Y/%m/%d %H:%M:%S")" >> "${LOG_FILE}"
    fi
    if [[ ! -e "${LOG_FILE}" ]]; then
      touch "$LOG_FILE"
      printf "%s Log created.\n" "$(/bin/date "+%Y/%m/%d %H:%M:%S")" \
        >> "${LOG_FILE}"
    fi
    LOGGED_MESSAGE="$*"
    printf "%s $LOGGED_MESSAGE\n" "$(/bin/date "+%Y/%m/%d %H:%M:%S")" \
      >> "$LOG_FILE"
}

testForIllegalCharacters(){
  STRING="$1"
  # Check that file name string only contains white-listed characters.
  if [[ $STRING =~ $REGEX ]]; then
    # Check that white-listed characters are not in bad places
    if [[ ! $STRING =~ $TRAILING_SPACE ]]; then
      : # No trailing space... moving along
    else
      return 1 # Trailing space found
    fi
    if [[ ! $STRING =~ $TRAILING_PERIOD ]]; then
      : # Not trailing period found... moving along
    else
      return 1 # Trailing period found
    fi
    return 0 # YAY! This file is clean!
  else
    return 1 # Non-whitelisted character found. We'll take some action.
  fi
}

sanitizeFilename(){
  STRING="$1"
  STRING_REPLACEMENT="$STRING" # Set variable outside of for loop
  writeToLog "WORKING: \"$TARGET_FILE\""
  # Let's iterate through this string and replace anything not on the whitelist
  for (( INDEX = 0; INDEX < ${#STRING}; INDEX++ )); do
    CHAR=${STRING:$INDEX:1}
    if [[ ! $CHAR =~ $REGEX ]]; then # We found one!
      # writeToLog "\t$STRING: '$CHAR' at index $INDEX"
      STRING_REPLACEMENT=$(printf $STRING_REPLACEMENT \
        | sed "s,.,-,$(expr $INDEX + 1)")
    fi
  done

  if [[ $STRING_REPLACEMENT =~ $TRAILING_SPACE ]]; then # return without trailing space
    # writeToLog "\t$STRING_REPLACEMENT: ' ' at index $(expr ${#STRING_REPLACEMENT} - 1)"
    printf "$STRING_REPLACEMENT" | sed "s, $,,"
    return 0
  elif [[ $STRING_REPLACEMENT =~ $TRAILING_PERIOD ]]; then # return without trailing period
    # writeToLog "\t$STRING_REPLACEMENT: '.' at index $(expr ${#STRING_REPLACEMENT} - 1)"
    printf "$STRING_REPLACEMENT" | sed "s,.$,,"
    return 0
  else # We're good to go.
    printf "$STRING_REPLACEMENT"
    return 0
  fi
}

if [[ $EUID -ne 0 ]]; then
   writeToLog "This must be run as root."
   exit 1
fi

OIFS=$IFS
IFS=$'\n'
# find will crawl the filesystem performing some tests on each filename
for TARGET_FILE in $(find "$1" -name "*" | grep -vE "^\.*$"); do
  TARGET_BASE="$(basename $TARGET_FILE)"
  if [[ $TARGET_BASE == ".DS_Store" ]]; then
    rm "$TARGET_FILE"
  fi
  testForIllegalCharacters "$TARGET_BASE" # Run initial test. Clean returns 0.
  if [[ "$?" -eq "1" ]]; then # We found a dirty filename
    # Find a decent, sanitized name for the file.
    SANITIZED_NAME="$(sanitizeFilename "$TARGET_BASE")"
    # Let's make sure we don't overwrite anything.
    if [[ -e "$(dirname $TARGET_FILE)/$SANITIZED_NAME" ]]; then
      # The sanitized name would overwrite something, so we'll add the date
      # to the beginning of the filename.
      mv "$TARGET_FILE" \
        "$(dirname $TARGET_FILE)/$(date +%Y-%m-%d)_$SANITIZED_NAME"
      writeToLog "\t$(dirname $TARGET_FILE)/$SANITIZED_NAME already exists!"
      writeToLog "CHANGE: $TARGET_FILE to $(dirname $TARGET_FILE)/$(date +%Y-%m-%d)_$SANITIZED_NAME"
    else
      # No conflict. Going to go ahead and overwrite.
      mv "$TARGET_FILE" "$(dirname $TARGET_FILE)/$SANITIZED_NAME"
      writeToLog "CHANGE: $TARGET_FILE to $(dirname $TARGET_FILE)/$SANITIZED_NAME"
    fi
  fi
done
IFS=$OIFS