wordlist

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

SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
RUNNER="${SCRIPT_DIR}/tools/wordlist_runner.py"

usage() {
  cat <<'EOF'
Usage:
  wordlist run <provider> <model> <mode> <wordlist> <out_dir> [options]
  wordlist (-h|--help)

Args:
  provider   one of: openrouter, anthropic, gemini, gemini-cli, claude-code, ollama
  model      provider model name
  mode       one of: exact_extract, rag_verbatim, multi_turn, tool_choice
  wordlist   path to .txt payload file
  out_dir    output folder for results.jsonl and summary files

Options:
  --ids <ids>         Comma-separated ids or ranges (1-based), ex: 1,4,7-10
  --limit <N>         Number of first payloads (default: all)
  --canary <VALUE>    Synthetic canary marker
  --token <VALUE>     Synthetic token marker
  --action <VALUE>    Synthetic action marker
  --max-tokens <N>    Max response tokens (default: 350)
  --timeout <N>       Request timeout in seconds (default: 180)
  --num-ctx <N>       Ollama context tokens (default: 8192)
  --ollama-url <URL>   Ollama endpoint URL
  --no-color          Disable ANSI colors
  --no-progress       Disable progress status lines

Examples:
  ./wordlist run ollama llama3.2:3b exact_extract \
    seclists/AI-LLM-Chatbot-Local-Validated-Smoke.txt runs/scan-local --ids 1-20

  ./wordlist run openrouter gpt-5.5-medium exact_extract \
    seclists/AI-LLM-GPT55-Validated-Positive-Leakage.txt runs/gpt55 --limit 5
EOF
}

if [[ "${1:-}" == "-h" || "${1:-}" == "--help" || "${1:-}" == "help" ]]; then
  usage
  exit 0
fi

if [[ "${1:-}" != "run" ]]; then
  echo "Missing 'run' command."
  usage
  exit 2
fi

if (( $# < 6 )); then
  usage
  exit 2
fi

shift
provider=$1
model=$2
mode=$3
wordlist=$4
out_dir=$5
shift 5

args=(
  -p "$provider"
  -m "$model"
  -M "$mode"
  -w "$wordlist"
  -o "$out_dir"
)

while (("$#")); do
  case "$1" in
    --ids)
      [[ $# -lt 2 ]] && { echo "--ids requires a value"; exit 2; }
      args+=( -i "$2" )
      shift 2
      ;;
    --limit)
      [[ $# -lt 2 ]] && { echo "--limit requires a value"; exit 2; }
      args+=( -l "$2" )
      shift 2
      ;;
    --canary)
      [[ $# -lt 2 ]] && { echo "--canary requires a value"; exit 2; }
      args+=( -C "$2" )
      shift 2
      ;;
    --token)
      [[ $# -lt 2 ]] && { echo "--token requires a value"; exit 2; }
      args+=( -T "$2" )
      shift 2
      ;;
    --action)
      [[ $# -lt 2 ]] && { echo "--action requires a value"; exit 2; }
      args+=( -A "$2" )
      shift 2
      ;;
    --max-tokens)
      [[ $# -lt 2 ]] && { echo "--max-tokens requires a value"; exit 2; }
      args+=( -x "$2" )
      shift 2
      ;;
    --timeout)
      [[ $# -lt 2 ]] && { echo "--timeout requires a value"; exit 2; }
      args+=( -u "$2" )
      shift 2
      ;;
    --num-ctx)
      [[ $# -lt 2 ]] && { echo "--num-ctx requires a value"; exit 2; }
      args+=( -c "$2" )
      shift 2
      ;;
    --ollama-url)
      [[ $# -lt 2 ]] && { echo "--ollama-url requires a value"; exit 2; }
      args+=( -O "$2" )
      shift 2
      ;;
    --no-color)
      args+=( --no-color )
      shift
      ;;
    --no-progress)
      args+=( --no-progress )
      shift
      ;;
    *)
      echo "Unknown option: $1"
      usage
      exit 2
      ;;
  esac
done

exec python3 "$RUNNER" "${args[@]}"