feat(pr2): HTTP/2+HTTP/3 fingerprint extractors — JA4H, H2 SETTINGS, JA4-QUIC

2026-05-10 00:47:19 -04:00
parent 0653e500b5
commit 92632d7afd
25 changed files with 1885 additions and 48 deletions
--- a/tests/sniffer/test_ja4h.py
+++ b/tests/sniffer/test_ja4h.py
@@ -0,0 +1,153 @@
+"""Tests for _ja4h computation and QUIC helpers in decnet.sniffer.fingerprint."""
+from __future__ import annotations
+
+import pytest
+
+from decnet.sniffer.fingerprint import _ja4h, _quic_varint, _extract_crypto_frames
+
+
+class TestJA4H:
+    def test_basic_get_h11(self):
+        result = _ja4h(
+            method="GET",
+            version="HTTP/1.1",
+            headers_ordered=["Host", "User-Agent", "Accept"],
+        )
+        parts = result.split("_")
+        assert len(parts) == 4
+        assert parts[0].startswith("GE11")  # method + version
+        assert parts[0][4] == "n"           # no cookie
+        assert parts[0][5] == "n"           # no referer
+        assert parts[0][6:10] == "0000"     # no Accept-Language
+
+    def test_cookie_flag(self):
+        result = _ja4h(
+            method="POST",
+            version="HTTP/1.1",
+            headers_ordered=["Host", "Cookie", "Content-Type"],
+            cookie_val="session=abc",
+        )
+        parts = result.split("_")
+        assert parts[0][4] == "c"           # has cookie
+        assert parts[0][5] == "n"           # no referer
+
+    def test_referer_flag(self):
+        result = _ja4h(
+            method="GET",
+            version="HTTP/1.1",
+            headers_ordered=["Host", "Referer"],
+        )
+        parts = result.split("_")
+        assert parts[0][5] == "r"           # has referer
+
+    def test_h2_version_tag(self):
+        result = _ja4h(
+            method="GET",
+            version="HTTP/2.0",
+            headers_ordered=["Host", "User-Agent"],
+        )
+        assert result.startswith("GE20")
+
+    def test_h3_version_tag(self):
+        result = _ja4h(
+            method="GET",
+            version="HTTP/3.0",
+            headers_ordered=["Host", "User-Agent"],
+        )
+        assert result.startswith("GE30")
+
+    def test_cookie_and_referer_excluded_from_header_hash(self):
+        result_with = _ja4h(
+            method="GET",
+            version="HTTP/1.1",
+            headers_ordered=["Host", "User-Agent", "Cookie", "Referer"],
+            cookie_val="x=1",
+        )
+        result_without = _ja4h(
+            method="GET",
+            version="HTTP/1.1",
+            headers_ordered=["Host", "User-Agent"],
+        )
+        # Header hash (parts[2]) must be identical — cookie/referer excluded from it
+        assert result_with.split("_")[2] == result_without.split("_")[2]
+
+    def test_header_count_excludes_cookie_and_referer(self):
+        result = _ja4h(
+            method="GET",
+            version="HTTP/1.1",
+            headers_ordered=["Host", "Cookie", "Accept", "Referer"],
+        )
+        parts = result.split("_")
+        # 2 headers after dropping Cookie and Referer (Host + Accept)
+        assert parts[1] == "02"
+
+    def test_cookie_hash_alphabetical_sort(self):
+        r1 = _ja4h("GET", "HTTP/1.1", [], cookie_val="z=3; a=1; m=2")
+        r2 = _ja4h("GET", "HTTP/1.1", [], cookie_val="a=1; m=2; z=3")
+        # Both should produce the same cookie hash regardless of original order
+        assert r1.split("_")[3] == r2.split("_")[3]
+
+    def test_no_cookie_produces_12_zeros(self):
+        result = _ja4h("GET", "HTTP/1.1", ["Host"])
+        assert result.split("_")[3] == "000000000000"
+
+    def test_accept_lang_truncated_to_4_chars(self):
+        result = _ja4h("GET", "HTTP/1.1", [], accept_lang="en-US,en;q=0.9")
+        parts = result.split("_")
+        lang_tag = parts[0][6:10]
+        assert lang_tag == "en-U"
+
+    def test_deterministic(self):
+        kwargs = dict(
+            method="POST",
+            version="HTTP/1.1",
+            headers_ordered=["Host", "Content-Type", "Accept"],
+        )
+        assert _ja4h(**kwargs) == _ja4h(**kwargs)
+
+
+class TestQuicVarint:
+    def test_1_byte(self):
+        assert _quic_varint(b"\x3f", 0) == (63, 1)
+
+    def test_2_byte(self):
+        # 0x4000 → big 2-byte form: 01 + 14 bits = 0x4000 = 16384
+        data = bytes([0x40, 0x00])
+        assert _quic_varint(data, 0) == (0, 2)
+
+    def test_4_byte(self):
+        # 0x80000000 → 2 MSB = 10, value = 0
+        data = bytes([0x80, 0x00, 0x00, 0x00])
+        assert _quic_varint(data, 0) == (0, 4)
+
+    def test_small_values(self):
+        assert _quic_varint(b"\x00", 0) == (0, 1)
+        assert _quic_varint(b"\x01", 0) == (1, 1)
+        assert _quic_varint(b"\x25", 0) == (37, 1)
+
+
+class TestExtractCryptoFrames:
+    def test_single_crypto_frame(self):
+        # CRYPTO frame: type=0x06, offset=0x00 (varint), length=5 (varint), data
+        data_bytes = b"hello"
+        frame = bytes([0x06, 0x00, 0x05]) + data_bytes
+        result = _extract_crypto_frames(frame)
+        assert result == b"hello"
+
+    def test_empty_payload(self):
+        result = _extract_crypto_frames(b"")
+        assert result == b""
+
+    def test_padding_skipped(self):
+        # PADDING (0x00) + CRYPTO frame
+        data_bytes = b"world"
+        frame = bytes([0x00, 0x00, 0x06, 0x00, 0x05]) + data_bytes
+        result = _extract_crypto_frames(frame)
+        assert result == b"world"
+
+    def test_non_crypto_frame_stops_parsing(self):
+        # Unknown frame type (0x10) after CRYPTO — should stop and return what we have
+        data = b"hello"
+        frame = bytes([0x06, 0x00, 0x05]) + data + bytes([0x10, 0x00])
+        result = _extract_crypto_frames(frame)
+        assert result == b"hello"
--- a/tests/sniffer/test_quic_initial.py
+++ b/tests/sniffer/test_quic_initial.py
@@ -0,0 +1,123 @@
+"""Tests for QUIC v1 Initial packet key derivation (RFC 9001 Appendix A vectors)."""
+from __future__ import annotations
+
+import pytest
+
+from decnet.sniffer.fingerprint import (
+    _hkdf_extract,
+    _hkdf_expand_label,
+    _quic_initial_keys,
+    _QUIC_V1_INITIAL_SALT,
+    _ja4_quic,
+    _parse_quic_initial,
+)
+
+
+# RFC 9001 Appendix A.1 key derivation test vectors
+_RFC9001_DCID = bytes.fromhex("8394c8f03e515708")
+_RFC9001_CLIENT_KEY = bytes.fromhex("1f369613dd76d5467730efcbe3b1a22d")
+_RFC9001_CLIENT_IV = bytes.fromhex("fa044b2f42a3fd3b46fb255c")
+_RFC9001_CLIENT_HP = bytes.fromhex("9f50449e04a0e810283a1e9933adedd2")
+
+
+class TestHKDF:
+    def test_extract_sha256(self):
+        # HKDF-Extract is HMAC-SHA256(salt, IKM). Cross-check with a known value.
+        result = _hkdf_extract(b"salt", b"ikm")
+        import hmac, hashlib
+        expected = hmac.new(b"salt", b"ikm", hashlib.sha256).digest()
+        assert result == expected
+
+    def test_expand_label_length(self):
+        secret = _hkdf_extract(_QUIC_V1_INITIAL_SALT, _RFC9001_DCID)
+        # "client in" expand should be 32 bytes
+        client_secret = _hkdf_expand_label(secret, "client in", b"", 32)
+        assert len(client_secret) == 32
+
+    def test_expand_label_key_length(self):
+        secret = _hkdf_extract(_QUIC_V1_INITIAL_SALT, _RFC9001_DCID)
+        client_secret = _hkdf_expand_label(secret, "client in", b"", 32)
+        key = _hkdf_expand_label(client_secret, "quic key", b"", 16)
+        assert len(key) == 16
+
+    def test_expand_label_iv_length(self):
+        secret = _hkdf_extract(_QUIC_V1_INITIAL_SALT, _RFC9001_DCID)
+        client_secret = _hkdf_expand_label(secret, "client in", b"", 32)
+        iv = _hkdf_expand_label(client_secret, "quic iv", b"", 12)
+        assert len(iv) == 12
+
+
+class TestQuicInitialKeys:
+    def test_rfc9001_appendix_a_vectors(self):
+        """Key derivation must match RFC 9001 Appendix A.1 test vectors exactly."""
+        key, iv, hp = _quic_initial_keys(_RFC9001_DCID)
+        assert key == _RFC9001_CLIENT_KEY, f"key mismatch: {key.hex()}"
+        assert iv == _RFC9001_CLIENT_IV, f"iv mismatch: {iv.hex()}"
+        assert hp == _RFC9001_CLIENT_HP, f"hp mismatch: {hp.hex()}"
+
+
+class TestJA4Quic:
+    def test_proto_prefix_is_q(self):
+        ch = {
+            "cipher_suites": [0x1301, 0x1302],
+            "extensions": [0x000a, 0x000d, 0x002b],
+            "signature_algorithms": [0x0403, 0x0804],
+            "supported_versions": [0x0304],
+            "sni": "example.com",
+            "alpn": ["h3"],
+            "tls_version": 0x0303,
+        }
+        result = _ja4_quic(ch)
+        assert result.startswith("q"), f"expected 'q' prefix: {result}"
+
+    def test_structure(self):
+        ch = {
+            "cipher_suites": [0x1301],
+            "extensions": [0x000a],
+            "signature_algorithms": [],
+            "supported_versions": [0x0304],
+            "sni": "",
+            "alpn": [],
+            "tls_version": 0x0303,
+        }
+        result = _ja4_quic(ch)
+        parts = result.split("_")
+        assert len(parts) == 3
+
+    def test_deterministic(self):
+        ch = {
+            "cipher_suites": [0x1301, 0x1302, 0x1303],
+            "extensions": [0x000a, 0x000d],
+            "signature_algorithms": [0x0403],
+            "supported_versions": [0x0304],
+            "sni": "host.example",
+            "alpn": ["h3"],
+            "tls_version": 0x0303,
+        }
+        assert _ja4_quic(ch) == _ja4_quic(ch)
+
+
+class TestParseQuicInitial:
+    def test_short_header_rejected(self):
+        # Short header: bit 7 clear
+        assert _parse_quic_initial(b"\x40" + b"\x00" * 20) is None
+
+    def test_wrong_version_rejected(self):
+        # Long header, Initial type, version = 0x00000002
+        pkt = bytearray(30)
+        pkt[0] = 0xC0  # long header + Initial
+        pkt[1:5] = b"\x00\x00\x00\x02"  # version 2
+        assert _parse_quic_initial(bytes(pkt)) is None
+
+    def test_non_initial_type_rejected(self):
+        # Long header, Handshake type (0x20 set)
+        pkt = bytearray(30)
+        pkt[0] = 0xE0  # long header + Handshake
+        pkt[1:5] = b"\x00\x00\x00\x01"
+        assert _parse_quic_initial(bytes(pkt)) is None
+
+    def test_garbage_returns_none(self):
+        assert _parse_quic_initial(b"garbage bytes that are not QUIC") is None
+
+    def test_too_short_returns_none(self):
+        assert _parse_quic_initial(b"\xc0\x00") is None