Initial commit

m365_launcher.py

@@ -0,0 +1,446 @@
+"""
+gdpr_launcher.py — entry point for the packaged GDPRScanner app.
+
+Responsibilities:
+  1. Find a free port (default 5100)
+  2. Start Flask in a background thread
+  3. Open the UI in a native webview window (pywebview)
+     — falls back to the system browser if pywebview is unavailable
+
+Generated by build_gdpr.py — do not edit manually.
+"""
+import os
+os.environ.setdefault("OBJC_DISABLE_INITIALIZE_FORK_SAFETY", "YES")
+
+import subprocess
+import sys
+import socket
+import threading
+import time
+import webbrowser
+from pathlib import Path
+
+if getattr(sys, "frozen", False):
+    BASE_DIR = Path(sys._MEIPASS)
+else:
+    BASE_DIR = Path(__file__).parent
+
+
+def _setup_external_tools():
+    """
+    Locate Tesseract and Poppler regardless of how the app was launched.
+    GDPRScanner calls document_scanner for file content extraction, which
+    may need OCR for scanned PDFs — same setup as Document Scanner.
+    """
+    extra_paths = []
+
+    if sys.platform == "darwin":
+        brew_prefix = None
+        for brew_candidate in ["/opt/homebrew/bin/brew", "/usr/local/bin/brew"]:
+            if Path(brew_candidate).exists():
+                try:
+                    result = subprocess.run(
+                        [brew_candidate, "--prefix"],
+                        capture_output=True, text=True, timeout=5
+                    )
+                    if result.returncode == 0:
+                        brew_prefix = result.stdout.strip()
+                        break
+                except Exception:
+                    pass
+
+        brew_candidates = []
+        if brew_prefix:
+            brew_candidates.append(brew_prefix)
+        brew_candidates += ["/opt/homebrew", "/usr/local", "/home/linuxbrew/.linuxbrew"]
+
+        for prefix in brew_candidates:
+            bin_dir = Path(prefix) / "bin"
+            if bin_dir.exists():
+                extra_paths.append(str(bin_dir))
+            tessdata = Path(prefix) / "share" / "tessdata"
+            if tessdata.exists():
+                os.environ.setdefault("TESSDATA_PREFIX", str(tessdata))
+
+        for t in ["/opt/homebrew/bin/tesseract", "/usr/local/bin/tesseract"]:
+            if Path(t).exists():
+                os.environ.setdefault("TESSERACT_CMD", t)
+                break
+
+        for p in ["/opt/homebrew/bin", "/usr/local/bin",
+                  "/opt/homebrew/opt/poppler/bin", "/usr/local/opt/poppler/bin"]:
+            if (Path(p) / "pdftoppm").exists():
+                os.environ.setdefault("POPPLER_PATH", p)
+                extra_paths.insert(0, p)
+                break
+
+    elif sys.platform == "win32":
+        import winreg
+        tess_dir = None
+        try:
+            key = winreg.OpenKey(winreg.HKEY_LOCAL_MACHINE, r"SOFTWARE\Tesseract-OCR")
+            tess_dir, _ = winreg.QueryValueEx(key, "InstallDir")
+            winreg.CloseKey(key)
+        except Exception:
+            pass
+
+        for d in ([tess_dir] if tess_dir else []) + [
+            r"C:\Program Files\Tesseract-OCR",
+            r"C:\Program Files (x86)\Tesseract-OCR",
+            r"C:\Tesseract-OCR",
+        ]:
+            if d and Path(d, "tesseract.exe").exists():
+                os.environ.setdefault("TESSERACT_CMD", str(Path(d) / "tesseract.exe"))
+                extra_paths.append(d)
+                tessdata = Path(d) / "tessdata"
+                if tessdata.exists():
+                    os.environ.setdefault("TESSDATA_PREFIX", str(tessdata))
+                break
+
+        for d in [
+            r"C:\poppler\Library\bin", r"C:\poppler\bin",
+            r"C:\Program Files\poppler\Library\bin",
+            r"C:\Program Files\poppler\bin",
+            r"C:\tools\poppler\Library\bin",
+        ]:
+            if (Path(d) / "pdftoppm.exe").exists():
+                os.environ.setdefault("POPPLER_PATH", d)
+                extra_paths.insert(0, d)
+                break
+
+    if getattr(sys, "frozen", False):
+        tess_bin = BASE_DIR / ("tesseract.exe" if sys.platform == "win32" else "tesseract")
+        if tess_bin.exists():
+            os.environ.setdefault("TESSERACT_CMD", str(tess_bin))
+        for sub in ["poppler/bin", "poppler/Library/bin", "."]:
+            pdftoppm = BASE_DIR / sub / ("pdftoppm.exe" if sys.platform == "win32" else "pdftoppm")
+            if pdftoppm.exists():
+                os.environ.setdefault("POPPLER_PATH", str(pdftoppm.parent))
+                extra_paths.insert(0, str(pdftoppm.parent))
+                break
+        extra_paths.insert(0, str(BASE_DIR))
+
+    if extra_paths:
+        current = os.environ.get("PATH", "")
+        additions = os.pathsep.join(p for p in extra_paths if p not in current)
+        if additions:
+            os.environ["PATH"] = additions + os.pathsep + current
+
+    cmd = os.environ.get("TESSERACT_CMD")
+    if cmd and Path(cmd).exists():
+        try:
+            import pytesseract
+            pytesseract.pytesseract.tesseract_cmd = cmd
+        except ImportError:
+            pass
+
+    poppler = os.environ.get("POPPLER_PATH")
+    if poppler:
+        try:
+            import pdf2image.pdf2image as _p2i
+            _orig = _p2i.convert_from_path
+            def _patched(pdf_path, *a, poppler_path=None, **kw):
+                return _orig(pdf_path, *a, poppler_path=poppler_path or poppler, **kw)
+            _p2i.convert_from_path = _patched
+        except Exception:
+            pass
+
+
+_setup_external_tools()
+
+
+def find_free_port(start: int = 5100) -> int:
+    for port in range(start, start + 100):
+        with socket.socket(socket.AF_INET, socket.SOCK_STREAM) as s:
+            try:
+                s.bind(("127.0.0.1", port))
+                return port
+            except OSError:
+                continue
+    raise RuntimeError("No free port found in range 5100-5200")
+
+
+# ── Single-instance lock ──────────────────────────────────────────────────────
+_LOCK_FH = None
+
+def acquire_instance_lock() -> bool:
+    """
+    Acquire an exclusive process lock so only one instance runs at a time.
+    Returns True if the lock was acquired, False if another instance holds it.
+    The lock is released automatically when the process exits.
+    """
+    global _LOCK_FH
+    lock_dir = Path.home() / ".gdprscanner"
+    lock_dir.mkdir(parents=True, exist_ok=True)
+    lock_path = lock_dir / "app.lock"
+    try:
+        _LOCK_FH = open(lock_path, "w")
+        if sys.platform == "win32":
+            import msvcrt
+            msvcrt.locking(_LOCK_FH.fileno(), msvcrt.LK_NBLCK, 1)
+        else:
+            import fcntl
+            fcntl.flock(_LOCK_FH, fcntl.LOCK_EX | fcntl.LOCK_NB)
+        _LOCK_FH.write(str(os.getpid()))
+        _LOCK_FH.flush()
+        return True
+    except (IOError, OSError):
+        if _LOCK_FH:
+            _LOCK_FH.close()
+            _LOCK_FH = None
+        return False
+
+
+def _activate_venv():
+    if getattr(sys, "frozen", False):
+        return
+    for candidate in [BASE_DIR / "venv", Path(__file__).parent / "venv"]:
+        if sys.platform == "win32":
+            site_pkg = candidate / "Lib" / "site-packages"
+        else:
+            lib = candidate / "lib"
+            site_pkg = None
+            if lib.exists():
+                for d in lib.iterdir():
+                    sp = d / "site-packages"
+                    if sp.exists():
+                        site_pkg = sp
+                        break
+        if site_pkg and site_pkg.exists():
+            sys.path.insert(0, str(site_pkg))
+            os.environ["VIRTUAL_ENV"] = str(candidate)
+            os.environ.pop("PYTHONHOME", None)
+            break
+
+
+_activate_venv()
+
+
+def start_flask(port: int):
+    import gdpr_scanner as _app
+    _app.app.run(host="127.0.0.1", port=port, debug=False,
+                 threaded=True, use_reloader=False)
+
+
+def wait_for_flask(port: int, timeout: float = 20.0) -> bool:
+    deadline = time.monotonic() + timeout
+    while time.monotonic() < deadline:
+        try:
+            with socket.create_connection(("127.0.0.1", port), timeout=0.2):
+                return True
+        except OSError:
+            time.sleep(0.1)
+    return False
+
+
+def _load_icon_image():
+    try:
+        from PIL import Image as PILImage
+        for name in ["icon_gdpr.ico", "icon_gdpr.icns", "icon_gdpr.png",
+                     "icon.ico", "icon.icns", "icon.png",
+                     "icon_m365.ico", "icon_m365.icns", "icon_m365.png"]:  # legacy fallback
+            p = BASE_DIR / name
+            if p.exists():
+                return PILImage.open(p).convert("RGBA").resize((64, 64))
+        # Minimal fallback — blue square
+        img = PILImage.new("RGBA", (64, 64), (0, 114, 206, 255))
+        return img
+    except Exception:
+        return None
+
+
+def run_webview(port: int):
+    """
+    Open the app in a native webview window.
+    Returns True on success, False if pywebview is unavailable.
+    """
+    try:
+        import webview
+    except ImportError:
+        return False
+
+    class Api:
+        def quit(self):
+            import webview as _wv
+            for w in _wv.windows:
+                w.destroy()
+
+        def save_excel(self):
+            """Fetch the Excel export from Flask and save via native dialog."""
+            import urllib.request, datetime, os, webview as _wv
+            try:
+                url = f"http://127.0.0.1:{port}/api/export_excel"
+                with urllib.request.urlopen(url) as resp:
+                    data = resp.read()
+                fname = f"gdpr_scan_{datetime.datetime.now().strftime('%Y%m%d_%H%M%S')}.xlsx"
+                win = _wv.windows[0] if _wv.windows else None
+                if win:
+                    paths = win.create_file_dialog(
+                        _wv.SAVE_DIALOG,
+                        save_filename=fname,
+                        file_types=("Excel Files (*.xlsx)",),
+                    )
+                    if paths:
+                        dest = paths[0] if isinstance(paths, (list, tuple)) else paths
+                        if not dest.endswith(".xlsx"):
+                            dest += ".xlsx"
+                        with open(dest, "wb") as f:
+                            f.write(data)
+                        return {"ok": True, "path": dest}
+                return {"ok": False, "error": "cancelled"}
+            except Exception as e:
+                return {"ok": False, "error": str(e)}
+
+        def save_db_export(self):
+            """Fetch the DB export ZIP from Flask and save via native dialog."""
+            import urllib.request, datetime, webview as _wv
+            try:
+                url = f"http://127.0.0.1:{port}/api/db/export"
+                with urllib.request.urlopen(url) as resp:
+                    data = resp.read()
+                fname = f"gdpr_export_{datetime.datetime.now().strftime('%Y%m%d_%H%M%S')}.zip"
+                win = _wv.windows[0] if _wv.windows else None
+                if win:
+                    paths = win.create_file_dialog(
+                        _wv.SAVE_DIALOG,
+                        save_filename=fname,
+                        file_types=("ZIP Archive (*.zip)",),
+                    )
+                    if paths:
+                        dest = paths[0] if isinstance(paths, (list, tuple)) else paths
+                        if not dest.endswith(".zip"):
+                            dest += ".zip"
+                        with open(dest, "wb") as f:
+                            f.write(data)
+                        return {"ok": True, "path": dest}
+                return {"ok": False, "error": "cancelled"}
+            except Exception as e:
+                return {"ok": False, "error": str(e)}
+
+        def save_article30(self):
+            """Fetch the Article 30 Word doc from Flask and save via native dialog."""
+            import urllib.request, datetime, webview as _wv
+            try:
+                url = f"http://127.0.0.1:{port}/api/export_article30"
+                with urllib.request.urlopen(url) as resp:
+                    data = resp.read()
+                fname = f"article30_{datetime.datetime.now().strftime('%Y%m%d_%H%M%S')}.docx"
+                win = _wv.windows[0] if _wv.windows else None
+                if win:
+                    paths = win.create_file_dialog(
+                        _wv.SAVE_DIALOG,
+                        save_filename=fname,
+                        file_types=("Word Document (*.docx)",),
+                    )
+                    if paths:
+                        dest = paths[0] if isinstance(paths, (list, tuple)) else paths
+                        if not dest.endswith(".docx"):
+                            dest += ".docx"
+                        with open(dest, "wb") as f:
+                            f.write(data)
+                        return {"ok": True, "path": dest}
+                return {"ok": False, "error": "cancelled"}
+            except Exception as e:
+                return {"ok": False, "error": str(e)}
+
+        def open_manual(self, lang: str):
+            """Open the user manual in a new native webview window."""
+            import webview as _wv
+            url = f"http://127.0.0.1:{port}/manual?lang={lang}"
+            existing = next((w for w in _wv.windows if getattr(w, "_is_manual", False)), None)
+            if existing:
+                existing.load_url(url)
+            else:
+                mw = _wv.create_window(
+                    title="GDPRScanner — Manual",
+                    url=url,
+                    width=960,
+                    height=800,
+                    resizable=True,
+                )
+                mw._is_manual = True
+
+    w = webview.create_window(
+        title="GDPRScanner",
+        url=f"http://127.0.0.1:{port}/",
+        width=1400,
+        height=900,
+        min_size=(900, 600),
+        js_api=Api(),
+    )
+
+    def _on_closed():
+        os._exit(0)
+
+    w.events.closed += _on_closed
+    webview.start(debug=False)
+    return True
+
+
+def _run_browser_fallback(port: int):
+    """Open in system browser + optional tray icon."""
+    url = f"http://127.0.0.1:{port}/"
+    webbrowser.open(url)
+
+    try:
+        import pystray
+        from PIL import Image as PILImage
+
+        img = _load_icon_image()
+        if img is None:
+            return
+
+        def _quit(icon, item):
+            icon.stop()
+            os._exit(0)
+
+        def _open(icon, item):
+            webbrowser.open(url)
+
+        menu = pystray.Menu(
+            pystray.MenuItem("Open GDPRScanner", _open, default=True),
+            pystray.MenuItem("Quit", _quit),
+        )
+        icon = pystray.Icon("GDPRScanner", img, "GDPRScanner", menu)
+        icon.run()
+    except ImportError:
+        # No pystray — just keep the process alive
+        try:
+            while True:
+                time.sleep(60)
+        except KeyboardInterrupt:
+            pass
+
+
+if __name__ == "__main__":
+    if not acquire_instance_lock():
+        print("GDPRScanner is already running.", file=sys.stderr)
+        sys.exit(1)
+
+    # On macOS, multiprocessing uses "fork" which is unsafe with some
+    # frameworks — use "spawn" to match PyInstaller's behaviour.
+    if sys.platform == "darwin":
+        import multiprocessing
+        multiprocessing.set_start_method("spawn", force=True)
+
+    port = find_free_port()
+    # Machine-readable port line — stdout pipe for any parent process.
+    print(f"GDPR_PORT={port}", flush=True)
+
+    # Pre-import on main thread so cv2 / numpy initialise safely
+    try:
+        import gdpr_scanner  # noqa: F401 — side effect: loads Flask app
+    except Exception as e:
+        print(f"[!] Failed to import gdpr_scanner: {e}", file=sys.stderr)
+        sys.exit(1)
+
+    flask_thread = threading.Thread(target=start_flask, args=(port,), daemon=True)
+    flask_thread.start()
+
+    if not wait_for_flask(port):
+        print("[!] Flask did not start in time", file=sys.stderr)
+        sys.exit(1)
+
+    webview_ok = run_webview(port)
+    if not webview_ok:
+        _run_browser_fallback(port)