Files
hackingtool/tests/test_tui_launch.py
Hardik Zinzu 4a9c3d0075 feat: interactive TUI, self-update & native settings; apply community bug fixes
Interactive terminal UI (Textual, default on a TTY)
- Category browser with live tool counts; type to search EVERY tool across all
  categories; up/down move, Enter open, left/Esc back; `/` opens a command palette.
- Tool screen gates Run/Install/Uninstall behind a confirm, dropping to a real
  TTY so a tool can't crash the UI; Open-repo is http(s)-guarded.
- The engine is unchanged; `--classic` keeps the full prompt_toolkit REPL and the
  headless `--engagement` orchestrator is untouched.

Update awareness + self-update helper
- Cached-daily GitHub Releases check (offline-safe, opt-out via
  HACKINGTOOL_NO_UPDATE_CHECK / update_check config); shows an "update available"
  line on the main screen via a background worker (never blocks first render).
- `/update` detects the install method (pipx/pip/apt/docker/source) and shows the
  exact upgrade command to copy - it never runs it.

Native settings + config hygiene
- `/update`/`/config` run as in-TUI screens instead of suspending to
  prompt_toolkit (which raised "event loop already running" nested under Textual).
- Sponsor feed URL moved from user config to a fixed code constant (SPONSORS_URL);
  update_check is now a proper bool toggle.

Docker / README
- Docker image references aligned to hardikzinzu/hackingtool.
- README refreshed for the interactive UI; first screenshot replaced with the new UI.

Community-reported bug fixes (verified still present in current code)
- Data safety: atomic JSON/YAML writes (#773, #765); malformed config.json no
  longer crashes load/import (#775); UTF-8 for catalog/settings IO (#766);
  findings parsers skip non-object/typed records instead of aborting (#768);
  pipeline decodes invalid bytes and salvages timed-out evidence (#771, #770).
- AI layer: reject null/non-text provider responses (#778); auto-mode connection
  test honors the Ollama fallback (#779); settings stay responsive during the AI
  probe (#780); one findings snapshot per AI report (#772).
- CLI/REPL: `@nmap`/`@sqlmap` resolve canonically instead of fuzzy mis-hits (#767);
  bare `/` no longer crashes the REPL (#761); bad/incomplete args exit non-zero
  instead of opening the menu (#763); search/tag result #99 is reachable (#762);
  background pane creation failures are reported, not shown as "started" (#777);
  config/API-key save failures surface without closing settings (#776).
- Misc: .dockerignore keeps README.md for the build (#781); PortScan prompt markup
  fixed (#698); uninstall command lines use [warning] not [error] (#666, #663);
  untrusted-output fence can't be broken out of + script import fix (#757).
2026-10-15 19:01:16 +05:30

31 lines
1.3 KiB
Python

import sys
from unittest import mock
import hackingtool.cli as cli
# NOTE: cli imports get_tools_dir via `from hackingtool.config import get_tools_dir`,
# so the live name is hackingtool.cli.get_tools_dir (patching hackingtool.config.*
# would not intercept it). ensure_user_files is called as config.ensure_user_files,
# so it is patched on the config module.
def test_default_launches_tui(monkeypatch):
monkeypatch.setattr(sys, "argv", ["hackingtool"])
monkeypatch.setattr("sys.stdin.isatty", lambda: True)
monkeypatch.setattr("sys.stdout.isatty", lambda: True)
with mock.patch("hackingtool.tui.app.run") as tui, \
mock.patch("hackingtool.config.ensure_user_files"), \
mock.patch("hackingtool.cli.get_tools_dir"):
cli.main()
assert tui.called
def test_classic_flag_skips_tui(monkeypatch):
monkeypatch.setattr(sys, "argv", ["hackingtool", "--classic"])
with mock.patch("hackingtool.tui.app.run") as tui, \
mock.patch("hackingtool.cli.interact_menu") as classic, \
mock.patch("hackingtool.repl._use_repl", return_value=False), \
mock.patch("hackingtool.config.ensure_user_files"), \
mock.patch("hackingtool.cli.get_tools_dir"):
cli.main()
assert not tui.called
assert classic.called