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

BACKUP_EXT=".bak"
DRY_RUN=0
WITH_CPP=0
WITH_BASH=0
WITH_PYTHON=0
WITH_CMAKE=0
TARGET_DIR="."

usage() {
  cat <<'EOF'
Usage:
  strip_c_comments_safe.sh [OPTIONEN] [VERZEICHNIS]

Optionen:
  --dry-run           Nur Vorschau anzeigen, keine Dateien ändern
  --backup-ext EXT    Backup-Endung setzen, Standard: .bak
  --cpp               Zusätzlich auch C++-Dateien verarbeiten
  --bash              Zusätzlich auch Bash-Skripte (*.sh) verarbeiten
  --python            Zusätzlich auch Python-Skripte (*.py) verarbeiten
  --cmake             Zusätzlich auch CMake-Skripte (*.cmake) verarbeiten
  -h, --help          Hilfe anzeigen

Argumente:
  VERZEICHNIS         Zielverzeichnis, Standard: aktuelles Verzeichnis

Beispiele:
  ./strip_c_comments_safe.sh
  ./strip_c_comments_safe.sh src
  ./strip_c_comments_safe.sh --dry-run include
  ./strip_c_comments_safe.sh --cpp src
  ./strip_c_comments_safe.sh --bash scripts
  ./strip_c_comments_safe.sh --python tools
  ./strip_c_comments_safe.sh --cmake cmake
  ./strip_c_comments_safe.sh --cpp --backup-ext .orig ./project
EOF
}

while [[ $# -gt 0 ]]; do
  case "$1" in
    --dry-run)
      DRY_RUN=1
      shift
      ;;
    --backup-ext)
      BACKUP_EXT="${2:-}"
      if [[ -z "$BACKUP_EXT" ]]; then
        echo "Fehler: --backup-ext braucht ein Argument" >&2
        exit 1
      fi
      shift 2
      ;;
    --cpp)
      WITH_CPP=1
      shift
      ;;
    --bash)
      WITH_BASH=1
      shift
      ;;
    --python)
      WITH_PYTHON=1
      shift
      ;;
    --cmake)
      WITH_CMAKE=1
      shift
      ;;
    -h|--help)
      usage
      exit 0
      ;;
    --)
      shift
      break
      ;;
    -*)
      echo "Unbekannte Option: $1" >&2
      usage >&2
      exit 1
      ;;
    *)
      if [[ "$TARGET_DIR" != "." ]]; then
        echo "Fehler: Es darf nur ein Verzeichnis angegeben werden" >&2
        usage >&2
        exit 1
      fi
      TARGET_DIR="$1"
      shift
      ;;
  esac
done

if [[ $# -gt 0 ]]; then
  if [[ "$TARGET_DIR" != "." ]]; then
    echo "Fehler: Es darf nur ein Verzeichnis angegeben werden" >&2
    usage >&2
    exit 1
  fi
  TARGET_DIR="$1"
  shift
fi

if [[ ! -d "$TARGET_DIR" ]]; then
  echo "Fehler: Verzeichnis nicht gefunden: $TARGET_DIR" >&2
  exit 1
fi

TARGET_DIR=$(cd -- "$TARGET_DIR" && pwd)

process_file() {
  local mode="$1"
  local file="$2"
  local tmp
  tmp="$(mktemp)"

  if [[ "$mode" == "python" ]]; then
    python3 - "$file" > "$tmp" <<'PY'
import io
import re
import sys
import tokenize

path = sys.argv[1]
with open(path, "rb") as f:
    data = f.read()

tokens = []
for tok in tokenize.tokenize(io.BytesIO(data).readline):
    if tok.type == tokenize.ENCODING:
        continue
    if tok.type == tokenize.COMMENT:
        line = tok.start[0]
        text = tok.string
        keep = (line == 1 and text.startswith("#!")) or (
            line <= 2 and re.search(r"coding[:=]\s*[-\w.]+", text)
        )
        if not keep:
            continue
    tokens.append(tok)

result = tokenize.untokenize(tokens)
if isinstance(result, str):
    sys.stdout.write(result)
else:
    sys.stdout.buffer.write(result)
PY
  else
    COMMENT_MODE="$mode" perl -0777 -pe '
      my $mode = $ENV{COMMENT_MODE} // "c";
      my $out = "";
      my $i = 0;
      my $len = length($_);
      my $state = "code";
      my $line = 1;

      sub is_ident {
        my ($ch) = @_;
        return defined($ch) && $ch =~ /[A-Za-z0-9_]/;
      }

      sub skip_line_comment {
        my ($s, $idx, $total) = @_;
        while ($idx < $total && substr($s, $idx, 1) ne "\n") {
          $idx++;
        }
        return $idx;
      }

      while ($i < $len) {
        my $c  = substr($_, $i, 1);
        my $c2 = $i + 1 < $len ? substr($_, $i, 2) : "";

        if ($mode eq "c") {
          if ($state eq "code") {
            if ($c2 eq "/*") {
              my $prev = length($out) ? substr($out, -1, 1) : "";
              $i += 2;

              my $saw_nl = 0;
              while ($i < $len) {
                my $d  = substr($_, $i, 1);
                my $d2 = $i + 1 < $len ? substr($_, $i, 2) : "";
                $saw_nl = 1 if $d eq "\n";
                if ($d2 eq "*/") {
                  $i += 2;
                  last;
                }
                $i++;
              }

              my $next = $i < $len ? substr($_, $i, 1) : "";

              if ($saw_nl) {
                $out .= "\n";
                $line++;
              } elsif (is_ident($prev) && is_ident($next)) {
                $out .= " ";
              }

              next;
            }
            elsif ($c2 eq "//") {
              my $prev = length($out) ? substr($out, -1, 1) : "";
              $i += 2;
              $i = skip_line_comment($_, $i, $len);
              my $next = $i < $len ? substr($_, $i, 1) : "";

              if ($i < $len && $next eq "\n") {
                $out .= "\n";
                $line++;
                $i++;
              } elsif (is_ident($prev)) {
                $out .= " ";
              }

              next;
            }
            elsif ($c eq "\"") {
              $out .= $c;
              $i++;
              $state = "string";
              next;
            }
            elsif ($c eq "'\''") {
              $out .= $c;
              $i++;
              $state = "char";
              next;
            }
            else {
              $out .= $c;
              $line++ if $c eq "\n";
              $i++;
              next;
            }
          }
          elsif ($state eq "string") {
            if ($c eq "\\") {
              if ($i + 1 < $len) {
                my $pair = substr($_, $i, 2);
                $out .= $pair;
                $line++ if $pair =~ /\n/;
                $i += 2;
              } else {
                $out .= $c;
                $i++;
              }
              next;
            }
            elsif ($c eq "\"") {
              $out .= $c;
              $i++;
              $state = "code";
              next;
            }
            else {
              $out .= $c;
              $line++ if $c eq "\n";
              $i++;
              next;
            }
          }
          elsif ($state eq "char") {
            if ($c eq "\\") {
              if ($i + 1 < $len) {
                my $pair = substr($_, $i, 2);
                $out .= $pair;
                $line++ if $pair =~ /\n/;
                $i += 2;
              } else {
                $out .= $c;
                $i++;
              }
              next;
            }
            elsif ($c eq "'\''") {
              $out .= $c;
              $i++;
              $state = "code";
              next;
            }
            else {
              $out .= $c;
              $line++ if $c eq "\n";
              $i++;
              next;
            }
          }
        } else {
          if ($state eq "code") {
            if ($c eq "\n") {
              $out .= $c;
              $line++;
              $i++;
              next;
            }

            if ($c eq "#") {
              my $bol = ($i == 0 || substr($_, $i - 1, 1) eq "\n") ? 1 : 0;
              my $prev = length($out) ? substr($out, -1, 1) : "";
              my $starts_comment = 0;

              if ($line == 1 && $bol && $i + 1 < $len && substr($_, $i, 2) eq "#!") {
                $starts_comment = 0;
              } elsif ($bol || $prev =~ /[\s;]/) {
                $starts_comment = 1;
              }

              if ($starts_comment) {
                $i = skip_line_comment($_, $i, $len);
                next;
              }
            }

            if ($c eq "'\''") {
              $out .= $c;
              $i++;
              $state = "single";
              next;
            }
            elsif ($c eq "\"") {
              $out .= $c;
              $i++;
              $state = "double";
              next;
            }
            elsif ($c eq "\\") {
              if ($i + 1 < $len) {
                my $pair = substr($_, $i, 2);
                $out .= $pair;
                $line++ if $pair =~ /\n/;
                $i += 2;
              } else {
                $out .= $c;
                $i++;
              }
              next;
            }
            else {
              $out .= $c;
              $i++;
              next;
            }
          }
          elsif ($state eq "single") {
            $out .= $c;
            $i++;
            $state = "code" if $c eq "'\''";
            next;
          }
          elsif ($state eq "double") {
            if ($c eq "\\") {
              if ($i + 1 < $len) {
                my $pair = substr($_, $i, 2);
                $out .= $pair;
                $line++ if $pair =~ /\n/;
                $i += 2;
              } else {
                $out .= $c;
                $i++;
              }
              next;
            }
            $out .= $c;
            $line++ if $c eq "\n";
            $i++;
            $state = "code" if $c eq "\"";
            next;
          }
        }
      }

      $_ = $out;
    ' "$file" > "$tmp"
  fi

  if cmp -s "$file" "$tmp"; then
    rm -f "$tmp"
    return 0
  fi

  if [[ "$DRY_RUN" -eq 1 ]]; then
    echo "=== $file ==="
    diff -u -- "$file" "$tmp" || true
    rm -f "$tmp"
  else
    cp -p -- "$file" "$file$BACKUP_EXT"
    cat -- "$tmp" > "$file"
    rm -f "$tmp"
    echo "Geändert: $file (Backup: $file$BACKUP_EXT)"
  fi
}

detect_mode() {
  local file="$1"
  case "$file" in
    *.c|*.h|*.cpp|*.cxx|*.cc|*.hpp|*.hxx|*.hh)
      echo "c"
      ;;
    *.sh)
      echo "bash"
      ;;
    *.py)
      echo "python"
      ;;
    *.cmake)
      echo "bash"
      ;;
    *)
      echo ""
      ;;
  esac
}

find_expr=()
find_expr+=( -name "*.c" -o -name "*.h" )

if [[ "$WITH_CPP" -eq 1 ]]; then
  find_expr+=( -o -name "*.cpp" -o -name "*.cxx" -o -name "*.cc" -o -name "*.hpp" -o -name "*.hxx" -o -name "*.hh" )
fi

if [[ "$WITH_BASH" -eq 1 ]]; then
  find_expr+=( -o -name "*.sh" )
fi

if [[ "$WITH_PYTHON" -eq 1 ]]; then
  find_expr+=( -o -name "*.py" )
fi

if [[ "$WITH_CMAKE" -eq 1 ]]; then
  find_expr+=( -o -name "*.cmake" )
fi

declare -A seen_files=()

while IFS= read -r -d '' file; do
  real_file="$file"
  if [[ -L "$file" ]]; then
    target=$(readlink "$file")
    if [[ "$target" != /* ]]; then
      dir="$(cd "$(dirname "$file")" && pwd -P)" || continue
      target="$dir/$target"
    fi
    [[ -L "$target" ]] && continue
    [[ ! -f "$target" ]] && continue
    real_file="$target"
  fi
  [[ -n "${seen_files[$real_file]+x}" ]] && continue
  seen_files[$real_file]=1
  mode="$(detect_mode "$real_file")"
  [[ -n "$mode" ]] && process_file "$mode" "$real_file"
done < <(find "$TARGET_DIR" \( -type f -o -type l \) \( "${find_expr[@]}" \) -print0)