Pre-publish privacy audit: scrub the simulator serial, harden the sdist, add licences

Audited the package against the two-stage procedure before a first PyPI
publish. The sdist and wheel were already tight, but three things needed
fixing and the controls needed to become real rather than documented.

The simulator volunteered high="PI04" as its product serial, which is the
real Bebop 2 serial prefix. Nothing unique to one aircraft, but a realistic
prefix invites being quoted into a bug report as a specimen, so it now reads
"N0TAREAL" / "0000000SIM" with a comment saying why it is nonsense on
purpose. The low half and the 500.0 no-fix GPS sentinel were already fake.

Hardened [tool.uv.build-backend] source-exclude well past the directories
that exist today: captures at any depth, log dumps, recorded media by
extension, caches, and anything credential-shaped. .gitignore governs git
and source-exclude governs the sdist; a capture can sit in one and not the
other, which is how this kind of data reaches an immutable index. Verified
the broad patterns do not over-reach: arsdk-xml/ with PROVENANCE.md and
tools/logs.py both still ship.

Added LICENSE (MIT) and LICENSE-arsdk-xml (Parrot SA's BSD-3-Clause), and
corrected the declared licence to "MIT AND BSD-3-Clause". The vendored XML
ships in both artifacts because nothing here decodes a command without it,
so MIT alone understated what is in the box. Both texts now appear in the
artifacts and in the metadata.

test_packaging.py grows privacy guards that fail on a serial prefix, a P7
CPU id, any MAC, a high-precision coordinate, an absolute home path, or any
private address other than the drone's own documented 192.168.42.0/24. Each
pattern was checked against the real identifiers to confirm it bites, since
a guard that passes on an empty tree proves nothing.

Example address in test_arsdk_session.py moved to RFC 5737 space.

504 tests pass, ruff clean.

tests/test_arsdk_session.py

@@ -404,10 +404,12 @@ def test_probe_gives_up_on_a_closed_port_quickly():
 def test_an_explicit_address_is_trusted_without_probing():
     # A caller who names an address wants the connection error, not a
     # discovery verdict, so find() must not quietly return None here.
-    found = discovery.find("10.1.2.3")
+    # 192.0.2.0/24 is RFC 5737 documentation space, so this cannot name a host
+    # on whatever network the suite happens to run on.
+    found = discovery.find("192.0.2.7")
     assert found is not None
-    assert (found.ip, found.via) == ("10.1.2.3", "given")
-    assert found.address == ("10.1.2.3", 44444)
+    assert (found.ip, found.via) == ("192.0.2.7", "given")
+    assert found.address == ("192.0.2.7", 44444)
 
 
 # -- an event we cannot name ---------------------------------------------

tests/test_packaging.py

@@ -1,14 +1,61 @@
-"""Guards that the package stays coherent as the streams land."""
+"""Guards that the package stays coherent as the streams land.
 
+The second half of this file is a privacy guard rather than a coherence one.
+This package was written against a real aircraft parked at a real address, so
+the things that identify it -- the serial, the SoC CPU id, the Wi-Fi BSSID, the
+first GPS fix -- all passed through this working tree at some point. PyPI is
+immutable per version and sdists are mirrored within minutes, so the audit that
+catches a stray identifier has to run before every publish, not after one.
+"""
+
+import re
 import tomllib
 from pathlib import Path
 
 import mcbebop
 from mcbebop.server import build_server
 
+ROOT = Path(__file__).parent.parent
+
+# Every pattern here is a shape that identifies the owner's aircraft or
+# network. The comment on each says what it is, so a future hit is diagnosable
+# rather than mysterious.
+_FORBIDDEN = (
+    # A Bebop 2 serial's high half. The full serial also appears in the
+    # aircraft's SSID, which makes it a locator and not just a label.
+    (re.compile(r"PI04"), "a real Bebop 2 serial prefix"),
+    # The P7 SoC CPU id: 'P7' then a long uppercase run. 'P7ID', the command
+    # name in ardrone3.xml, is too short to match and is meant to be here.
+    (re.compile(r"P7[0-9A-Z]{8,}"), "a P7 SoC CPU id"),
+    # Any MAC address at all, not just a Parrot OUI.
+    (re.compile(r"\b[0-9A-Fa-f]{2}(?::[0-9A-Fa-f]{2}){5}\b"), "a MAC address"),
+    # A plausible residential coordinate. The simulator reports ARSDK's 500.0
+    # no-fix sentinel instead, which has no decimal places to match.
+    (re.compile(r"\b[0-9]{1,3}\.[0-9]{6,}"), "a high-precision coordinate"),
+    # An absolute build path leaks a username and the local layout.
+    (re.compile(r"/home/[a-z]"), "an absolute home-directory path"),
+)
+
+# 192.168.42.0/24 is the drone's own fixed network and is in Parrot's public
+# documentation, so it is expected. Any other private address is somebody's
+# real LAN; use RFC 5737 documentation space in examples instead.
+_PRIVATE_IP = re.compile(
+    r"\b(?:10\.\d{1,3}\.\d{1,3}\.\d{1,3}"
+    r"|172\.(?:1[6-9]|2\d|3[01])\.\d{1,3}\.\d{1,3}"
+    r"|192\.168\.\d{1,3}\.\d{1,3})\b"
+)
+_ALLOWED_IP = re.compile(r"\b192\.168\.42\.\d{1,3}\b")
+
+
+def _shipped_files() -> list[Path]:
+    """Every text file that reaches an artifact, which is src/ plus the docs."""
+    files = [ROOT / "README.md", ROOT / "pyproject.toml"]
+    files += [p for p in (ROOT / "src").rglob("*") if p.is_file() and "__pycache__" not in p.parts]
+    return files
+
 
 def _pyproject() -> dict:
-    return tomllib.loads((Path(__file__).parent.parent / "pyproject.toml").read_text())
+    return tomllib.loads((ROOT / "pyproject.toml").read_text())
 
 
 def test_version_matches_pyproject():
@@ -38,3 +85,54 @@ def test_vendored_xml_present_and_parses():
         assert proj.find("class") is not None, f"{name} has no <class> elements"
         total += sum(len(c.findall("cmd")) for c in proj.iter("class"))
     assert total == 264, f"expected 264 commands, found {total}"
+
+
+# -- privacy guards ------------------------------------------------------
+def test_no_shipped_file_carries_an_aircraft_identifier():
+    hits = []
+    for path in _shipped_files():
+        text = path.read_text(encoding="utf-8", errors="replace")
+        for lineno, line in enumerate(text.splitlines(), 1):
+            for pattern, what in _FORBIDDEN:
+                if pattern.search(line):
+                    hits.append(f"{path.relative_to(ROOT)}:{lineno} looks like {what}: {line.strip()[:90]}")
+    assert not hits, "identifying data in a file that ships:\n" + "\n".join(hits)
+
+
+def test_no_shipped_file_names_a_private_network_but_the_drones_own():
+    hits = []
+    for path in _shipped_files():
+        text = path.read_text(encoding="utf-8", errors="replace")
+        for lineno, line in enumerate(text.splitlines(), 1):
+            for found in _PRIVATE_IP.finditer(line):
+                if not _ALLOWED_IP.fullmatch(found.group()):
+                    hits.append(f"{path.relative_to(ROOT)}:{lineno} names {found.group()}")
+    assert not hits, (
+        "a private address other than the drone's own 192.168.42.0/24; "
+        "use RFC 5737 documentation space:\n" + "\n".join(hits)
+    )
+
+
+def test_sdist_excludes_captures_and_dev_trees():
+    """The sdist exclusions are a separate mechanism from .gitignore.
+
+    A file can be gitignored and still swept into an sdist, which is how
+    captures reach PyPI. Assert the directories that hold real data off the
+    aircraft are named in both places.
+    """
+    excluded = set(_pyproject()["tool"]["uv"]["build-backend"]["source-exclude"])
+    ignored = (ROOT / ".gitignore").read_text().split()
+    for name in ("captures", "tests"):
+        assert name in excluded or f"{name}/**" in excluded, f"{name} is not excluded from the sdist"
+    for name in ("captures",):
+        assert f"{name}/" in ignored or name in ignored, f"{name} is not gitignored"
+
+
+def test_declared_licence_files_exist():
+    """A licence named in metadata but absent from the tree ships a lie."""
+    declared = _pyproject()["project"]["license-files"]
+    assert declared, "license-files must name the licence texts so they ship"
+    for name in declared:
+        assert (ROOT / name).is_file(), f"{name} is declared in license-files but missing"
+    # The vendored XML is Parrot's and the expression has to say so.
+    assert "BSD-3-Clause" in _pyproject()["project"]["license"]