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

tests/profiler/test_identity_rollup.py

@@ -29,11 +29,12 @@ class TestExtractFpSummaries:
 
     def test_empty_input_returns_all_none(self):
         result = extract_fp_summaries([])
-        assert result == {
-            "ja3_hashes": None,
-            "hassh_hashes": None,
-            "tls_cert_sha256": None,
-        }
+        assert all(v is None for v in result.values())
+        assert "ja3_hashes" in result
+        assert "hassh_hashes" in result
+        assert "tls_cert_sha256" in result
+        assert "ja4h_hashes" in result
+        assert "ja4_quic_hashes" in result
 
     def test_single_row_single_cert(self):
         row = _row_with(_bounty("tls_certificate", cert_sha256="ab" * 32))
@@ -139,3 +140,50 @@ class TestExtractFpSummaries:
         assert json.loads(result["ja3_hashes"]) == sorted(
             ["ja3-shared", "ja3-second", "ja3-third"]
         )
+
+    # ── ja4h + ja4_quic (PR2 columns) ────────────────────────────────
+
+    def test_ja4h_single_value(self):
+        row = _row_with(_bounty("ja4h", ja4h="GE11nn0000_02_abc_000"))
+        result = extract_fp_summaries([row])
+        assert json.loads(result["ja4h_hashes"]) == ["GE11nn0000_02_abc_000"]
+
+    def test_ja4_quic_single_value(self):
+        row = _row_with(_bounty("ja4_quic", ja4_quic="q13d0310h2_002f_0403_h3"))
+        result = extract_fp_summaries([row])
+        assert json.loads(result["ja4_quic_hashes"]) == ["q13d0310h2_002f_0403_h3"]
+
+    def test_ja4h_dedup_across_rows(self):
+        a = _row_with(_bounty("ja4h", ja4h="GE11nn0000_02_abc_000"))
+        b = _row_with(_bounty("ja4h", ja4h="GE11nn0000_02_abc_000"))
+        c = _row_with(_bounty("ja4h", ja4h="GE20nn0000_04_def_000"))
+        result = extract_fp_summaries([a, b, c])
+        hashes = json.loads(result["ja4h_hashes"])
+        assert len(hashes) == 2
+        assert "GE11nn0000_02_abc_000" in hashes
+        assert "GE20nn0000_04_def_000" in hashes
+
+    def test_ja4h_and_ja4_quic_coexist(self):
+        row = _row_with(
+            _bounty("ja4h", ja4h="GE11nn0000_02_abc_000"),
+            _bounty("ja4_quic", ja4_quic="q13d0310h2_002f_0403_h3"),
+        )
+        result = extract_fp_summaries([row])
+        assert json.loads(result["ja4h_hashes"]) == ["GE11nn0000_02_abc_000"]
+        assert json.loads(result["ja4_quic_hashes"]) == ["q13d0310h2_002f_0403_h3"]
+
+    def test_ja4h_missing_payload_key_skipped(self):
+        # bounty shaped like a fingerprint but missing the 'ja4h' key
+        row = _row_with({
+            "bounty_type": "fingerprint",
+            "payload": {"fingerprint_type": "ja4h", "protocol": "h1"},
+        })
+        result = extract_fp_summaries([row])
+        assert result["ja4h_hashes"] is None
+
+    def test_empty_returns_none_for_new_columns(self):
+        result = extract_fp_summaries([])
+        assert "ja4h_hashes" in result
+        assert result["ja4h_hashes"] is None
+        assert "ja4_quic_hashes" in result
+        assert result["ja4_quic_hashes"] is None