extract erofs file to a folder

erofs.extract

#!/usr/bin/env bash
set -euo pipefail

if [[ $# -lt 2 || $# -gt 3 ]]; then
  echo "Usage: $0 <image.erofs> <output_dir> [offset_bytes]" >&2
  exit 1
fi

IMAGE="$1"
OUTDIR="$2"
OFFSET="${3:-}"

command -v dump.erofs >/dev/null 2>&1 || { echo "dump.erofs not found in PATH" >&2; exit 1; }
mkdir -p "$OUTDIR"

DUMP=(dump.erofs)
[[ -n "$OFFSET" ]] && DUMP+=(--offset="$OFFSET")

abs_child() {
  local dir="$1" name="$2"
  if [[ "$dir" == "/" ]]; then printf "/%s" "$name"; else printf "%s/%s" "$dir" "$name"; fi
}

parse_ls_table() {
  local dir="$1"
  "${DUMP[@]}" --path "$dir" --ls "$IMAGE" | awk '
    BEGIN { intable=0 }
    /^[[:space:]]*NID[[:space:]]+TYPE[[:space:]]+FILENAME[[:space:]]*$/ { intable=1; next }
    intable && $0 ~ /^[[:space:]]*[0-9]+[[:space:]]+[0-9]+[[:space:]]+/ {
      # Capture first two numeric fields and the rest as filename (may contain spaces)
      nid=$1; type=$2;
      sub(/^[[:space:]]*[0-9]+[[:space:]]+[0-9]+[[:space:]]+/, "", $0);
      fn=$0;
      if (fn != "." && fn != "..") {
        print nid "\t" type "\t" fn
      }
      next
    }
  '
}

probe_path_meta() {
  local img_path="$1"
  "${DUMP[@]}" --path "$img_path" "$IMAGE" 2>/dev/null || true
}

extract_file() {
  local img_path="$1" out_path="$2"
  mkdir -p "$(dirname "$out_path")"
  if ! "${DUMP[@]}" --path "$img_path" --cat "$IMAGE" > "$out_path"; then
    echo "warn: failed to cat $img_path (not a regular file?)" >&2
    rm -f "$out_path" || true
  fi
}

extract_symlink() {
  local img_path="$1" out_path="$2"
  local meta target
  meta="$(probe_path_meta "$img_path")"
  target="$(sed -n 's/.*[[:space:]]target:[[:space:]]\([^[:cntrl:]]*\).*/\1/p' <<<"$meta" | head -n1)"
  if [[ -n "$target" ]]; then
    mkdir -p "$(dirname "$out_path")"
    ln -sf "$target" "$out_path"
    return 0
  fi
  return 1
}

queue=("/")

while ((${#queue[@]})); do
  dir="${queue[0]}"; queue=("${queue[@]:1}")

  if [[ "$dir" == "/" ]]; then
    mkdir -p "$OUTDIR"
  else
    mkdir -p "$OUTDIR/${dir#/}"
  fi

  while IFS=$'\t' read -r nid typecode name; do
    [[ -z "${name:-}" ]] && continue

    child="$(abs_child "$dir" "$name")"
    out="$OUTDIR/${child#/}"

    case "$typecode" in
      2)  # directory
          mkdir -p "$out"
          queue+=("$child")
          ;;
      1)  # regular file
          extract_file "$child" "$out"
          ;;
      *)  # unknown type -> probe
          meta="$(probe_path_meta "$child")"
          if grep -qi '\<symlink\>' <<<"$meta"; then
            extract_symlink "$child" "$out" || echo "warn: could not reconstruct symlink $child" >&2
          elif grep -qi '\bdirectory\b' <<<"$meta"; then
            mkdir -p "$out"; queue+=("$child")
          else
            # fall back to trying as a file
            extract_file "$child" "$out"
          fi
          ;;
    esac
  done < <(parse_ls_table "$dir")
done

echo "Done: extracted to $OUTDIR"