test: offline suite + committed fixtures for every stage

recon-triage/tests/test_exploitdb.py

@@ -0,0 +1,68 @@
+"""Exploit-DB matcher: parses real searchsploit JSON, never fabricates, fails soft."""
+
+from __future__ import annotations
+
+import json
+from pathlib import Path
+
+from recon_triage.grounding.exploitdb import ground_report
+from recon_triage.schema import Host, Port, ReconReport, Service
+from recon_triage.tools import searchsploit
+
+FIXTURES = Path(__file__).parent / "fixtures"
+
+
+def _index() -> dict:
+    return json.loads((FIXTURES / "searchsploit" / "index.json").read_text())
+
+
+def test_normalize_parses_results():
+    doc = _index()["Apache httpd 2.4.49"]
+    matches = searchsploit.normalize(json.dumps(doc), query="Apache httpd 2.4.49")
+    assert {m.edb_id for m in matches} == {"50383", "50406"}
+    assert all(m.verified is False for m in matches)
+    assert all(m.source == "exploit-db" for m in matches)
+    assert matches[0].url.endswith("/50383")
+
+
+def test_normalize_empty_results():
+    doc = _index()["nginx 1.18.0"]
+    assert searchsploit.normalize(json.dumps(doc)) == []
+
+
+def test_normalize_handles_garbage():
+    assert searchsploit.normalize("") == []
+    assert searchsploit.normalize("not json") == []
+
+
+def test_ground_report_attaches_only_returned(fixture_search_fn):
+    report = ReconReport(
+        run_id="t",
+        hosts=[
+            Host(
+                hostname="api.example.com",
+                ports=[
+                    Port(number=443, service=Service(product="Apache httpd", version="2.4.49")),
+                    Port(number=22, service=Service(product="OpenSSH", version="8.2p1 Ubuntu 4ubuntu0.5")),
+                ],
+            )
+        ],
+    )
+    attached = ground_report(report, fixture_search_fn)
+    assert attached == 2  # only the Apache service yields 2 candidates
+    apache = report.hosts[0].ports[0].service
+    ssh = report.hosts[0].ports[1].service
+    assert {m.edb_id for m in apache.exploitdb_candidates} == {"50383", "50406"}
+    assert ssh.exploitdb_candidates == []  # nothing returned -> empty, not fabricated
+
+
+def test_ground_report_failsoft_on_search_error():
+    def boom(term: str):
+        raise RuntimeError("searchsploit blew up")
+
+    report = ReconReport(
+        run_id="t",
+        hosts=[Host(hostname="h", ports=[Port(number=443, service=Service(product="X", version="1"))])],
+    )
+    attached = ground_report(report, boom)  # must not raise
+    assert attached == 0