feat: [censys_enrichmentapis] add enrichment API support

documentation/expansion/censys_enrichmentapis.md

@@ -0,0 +1,11 @@
+# Censys Enrichment APIs
+
+Retrieve Censys host enrichment, web properties on ports 80 and 443, or certificate metadata. Configure a platform API token and organization ID. Host service results are bounded; the annotation retains provider fields including scores and labels when present. Web property and certificate requests may consume standard credits under the account plan.
+
+Input: ip-src, ip-dst, domain, hostname, x509-fingerprint-sha256.
+
+Configuration: api_token, organization_id, max_results. Required: api_token, organization_id.
+
+max_results defaults to 25 and accepts 1–100. New attributes use `to_ids=false`; the source UUID, detection flag and metadata are preserved. Objects reference the source attribute. Requests verify TLS, reject redirects and use connection/read timeouts of 5/30 seconds. API errors do not echo credentials. No automatic submission retries or background polling occur.
+
+Reference: https://docs.censys.com/reference/v3-globaldata-asset-host-enrichment

misp_modules/modules/expansion/censys_enrichmentapis.py

@@ -0,0 +1,200 @@
+import ipaddress
+import json
+import re
+import zipfile
+from urllib.parse import quote
+from uuid import UUID
+import requests
+from pymisp import MISPEvent, MISPObject
+from pymisp.exceptions import PyMISPError
+
+
+class ModuleError(ValueError):
+    """A diagnostic safe to show to the analyst."""
+
+
+def _integer(config, key, default, maximum=100):
+    value = config.get(key)
+    if value is None or value == "":
+        return default
+    if isinstance(value, bool) or not str(value).isdigit() or (not 1 <= int(value) <= maximum):
+        raise ModuleError(f"{key} must be an integer between 1 and {maximum}.")
+    return int(value)
+
+
+def _segment(value):
+    return quote(str(value), safe="")
+
+
+def _normalise(kind, value):
+    if not isinstance(value, str) or not value.strip() or any(ord(c) < 32 for c in value):
+        raise ModuleError("A nonempty string value is required.")
+    value = value.strip()
+    if kind in ("md5", "sha1", "sha256", "x509-fingerprint-sha256"):
+        length = {"md5": 32, "sha1": 40}.get(kind, 64)
+        if not re.fullmatch("[a-fA-F0-9]{%d}" % length, value):
+            raise ModuleError("Invalid hash.")
+        return value.lower()
+    if kind in ("ip-src", "ip-dst"):
+        return str(ipaddress.ip_address(value))
+    if kind in ("domain", "hostname"):
+        value = value.rstrip(".").encode("idna").decode("ascii").lower()
+        if len(value) > 253 or not all(
+            re.fullmatch("[a-z0-9](?:[a-z0-9-]{0,61}[a-z0-9])?", p) for p in value.split(".")
+        ):
+            raise ModuleError("Invalid DNS name.")
+    if kind == "vulnerability":
+        value = value.upper()
+        if not re.fullmatch("CVE-[0-9]{4}-[0-9]{4,}", value):
+            raise ModuleError("A CVE identifier is required.")
+    return value
+
+
+def _call(method, url, **kwargs):
+    try:
+        response = requests.request(method, url, timeout=(5, 30), verify=True, allow_redirects=False, **kwargs)
+    except requests.RequestException:
+        raise ModuleError(
+            "The service request failed or timed out; check the service before repeating a submission."
+        ) from None
+    if response.status_code in (204, 404) and method == "GET":
+        return None
+    if response.status_code not in (200, 201, 202):
+        raise ModuleError(f"The service returned HTTP {response.status_code}.")
+    try:
+        return response.json()
+    except ValueError:
+        raise ModuleError("The service returned invalid JSON.") from None
+
+
+def _object(event, attribute, name):
+    obj = MISPObject(name)
+    obj.add_reference(attribute["uuid"], "describes")
+    event.add_object(obj)
+    return obj
+
+
+def _annotation(event, attribute, title, data):
+    obj = _object(event, attribute, "annotation")
+    obj.comment = title
+    obj.add_attribute("text", value=json.dumps(data, sort_keys=True), to_ids=False, disable_correlation=True)
+    return obj
+
+
+def handler(q=False):
+    if q is False:
+        return False
+    try:
+        request = json.loads(q)
+        if not isinstance(request, dict) or not isinstance(request.get("attribute"), dict):
+            raise ModuleError("An attribute containing type, value and uuid is required.")
+        attribute = request["attribute"]
+        UUID(attribute["uuid"])
+        if attribute["type"] not in mispattributes["input"]:
+            raise ModuleError("Unsupported attribute type.")
+        value = _normalise(attribute["type"], attribute["value"])
+        config = request.get("config") or {}
+        if not isinstance(config, dict):
+            raise ModuleError("The config must be an object.")
+        for key in REQUIRED:
+            if not isinstance(config.get(key), str) or not config[key].strip():
+                raise ModuleError(f"Missing {key} configuration.")
+        limit = _integer(config, "max_results", 25)
+        event = MISPEvent()
+        event.add_attribute(**attribute)
+        _expand(event, attribute, value, config, limit)
+        result = json.loads(event.to_json())
+        return {"results": {key: result.get(key, []) for key in ("Attribute", "Object")}}
+    except ModuleError as exc:
+        return {"error": str(exc)}
+    except (
+        ValueError,
+        TypeError,
+        KeyError,
+        AttributeError,
+        IndexError,
+        OSError,
+        RuntimeError,
+        zipfile.BadZipFile,
+        PyMISPError,
+    ):
+        return {"error": "Invalid request or service response."}
+
+
+def introspection():
+    return mispattributes
+
+
+def version():
+    return dict(moduleinfo, config=moduleconfig)
+
+
+moduleinfo = {
+    "version": "1.0",
+    "author": "MISP Project",
+    "name": "Censys Enrichment APIs",
+    "description": (
+        "Retrieve Censys host enrichment, web properties on ports 80 and 443, or certificate metadata. Configure a"
+        " platform API token and organization ID. Host service results are bounded; the annotation retains provider"
+        " fields including scores and labels when present. Web property and certificate requests may consume standard"
+        " credits under the account plan."
+    ),
+    "module-type": ["expansion"],
+    "data-source": "external",
+    "logo": "",
+    "requirements": ["requests", "pymisp"],
+    "features": (
+        "Retrieve Censys host enrichment, web properties on ports 80 and 443, or certificate metadata. Configure a"
+        " platform API token and organization ID. Host service results are bounded; the annotation retains provider"
+        " fields including scores and labels when present. Web property and certificate requests may consume standard"
+        " credits under the account plan."
+    ),
+    "references": ["https://docs.censys.com/reference/v3-globaldata-asset-host-enrichment"],
+    "input": "ip-src, ip-dst, domain, hostname, x509-fingerprint-sha256",
+    "output": "MISP objects linked to the original attribute.",
+}
+mispattributes = {
+    "input": ["ip-src", "ip-dst", "domain", "hostname", "x509-fingerprint-sha256"],
+    "format": "misp_standard",
+}
+moduleconfig = ["api_token", "organization_id", "max_results"]
+REQUIRED = ["api_token", "organization_id"]
+
+
+def _expand(event, attribute, value, config, limit):
+    headers = {"Authorization": "Bearer " + config["api_token"], "Accept": "application/json"}
+    params = {"organization_id": config["organization_id"]}
+    root = "https://api.platform.censys.io/v3/global/asset/"
+    kind = attribute["type"]
+    if kind in ("ip-src", "ip-dst"):
+        paths = ["enrichment/host/" + _segment(value)]
+    elif kind == "x509-fingerprint-sha256":
+        paths = ["certificate/" + value]
+    else:
+        paths = ["webproperty/" + _segment(value + ":" + str(port)) for port in (80, 443)]
+    for path in paths:
+        data = _call("GET", root + path, headers=headers, params=params)
+        if data is None:
+            continue
+        resource = data["result"]["resource"]
+        if not isinstance(resource, dict):
+            raise ModuleError("Censys returned an invalid resource.")
+        _annotation(event, attribute, moduleinfo["name"], resource)
+        if kind in ("ip-src", "ip-dst"):
+            for service in resource.get("services", [])[:limit]:
+                obj = _object(event, attribute, "network-connection")
+                obj.add_attribute(kind, value=value, to_ids=False)
+                if service.get("port") is not None:
+                    obj.add_attribute(
+                        "src-port" if kind == "ip-src" else "dst-port", value=int(service["port"]), to_ids=False
+                    )
+                if service.get("transport_protocol"):
+                    obj.add_attribute("layer4-protocol", value=service["transport_protocol"], to_ids=False)
+            if len(resource.get("services", [])) > limit:
+                _annotation(event, attribute, "Result limit", {"truncated": True, "max_results": limit})
+        elif kind == "x509-fingerprint-sha256":
+            obj = _object(event, attribute, "x509")
+            obj.add_attribute("x509-fingerprint-sha256", value=value, to_ids=False)
+            for key, relation in [("issuer_dn", "issuer"), ("subject_dn", "subject")]:
+                if resource.get("parsed", {}).get(key):
+                    obj.add_attribute(relation, value=resource["parsed"][key], to_ids=False)

tests/test_censys_enrichmentapis.py

@@ -0,0 +1,118 @@
+import json
+from unittest.mock import Mock, patch
+
+import pytest
+import requests
+
+from misp_modules.modules.expansion import censys_enrichmentapis as module
+
+UUID = "32876e4a-916d-4935-9e51-7b69a76718f8"
+SECRET = "test-service-secret"
+
+
+def query(kind=None, value="example.com", config=None, **attribute_fields):
+    attribute = {
+        "type": kind or module.mispattributes["input"][0],
+        "value": value,
+        "uuid": UUID,
+        "to_ids": True,
+        "comment": "source context",
+        **attribute_fields,
+    }
+    return json.dumps(
+        {"attribute": attribute, "config": config if config is not None else {k: SECRET for k in module.REQUIRED}}
+    )
+
+
+def reply(payload=None, status=200):
+    return Mock(status_code=status, headers={}, json=Mock(return_value=payload))
+
+
+def assert_source(result):
+    assert "results" in result, result
+    source = next(a for a in result["results"]["Attribute"] if a["uuid"] == UUID)
+    assert source["to_ids"] is True
+    assert source["comment"] == "source context"
+    for obj in result["results"]["Object"]:
+        assert any(r["referenced_uuid"] == UUID for r in obj["ObjectReference"])
+        assert all(a["to_ids"] is False for a in obj["Attribute"])
+
+
+def objects(result, name):
+    return [obj for obj in result["results"]["Object"] if obj["name"] == name]
+
+
+def values(obj, relation):
+    return [a["value"] for a in obj["Attribute"] if a["object_relation"] == relation]
+
+
+def test_discovery():
+    assert module.handler(False) is False
+    assert module.introspection()["format"] == "misp_standard"
+    assert module.version()["config"] == module.moduleconfig
+
+
+@pytest.mark.parametrize("raw", ["{", "[]", "null", "{}", '{"attribute": []}'])
+def test_invalid_envelope(raw):
+    with patch.object(module.requests, "request") as request:
+        assert "error" in module.handler(raw)
+    request.assert_not_called()
+
+
+@pytest.mark.parametrize("status", [301, 401, 403, 429, 503])
+def test_http_errors_are_sanitized(status):
+    with patch.object(module.requests, "request", return_value=reply({"error": SECRET}, status)):
+        with pytest.raises(module.ModuleError) as failure:
+            module._call("GET", "https://service.example/report")
+    assert str(status) in str(failure.value)
+    assert SECRET not in str(failure.value)
+
+
+def test_timeout_is_bounded_and_sanitized():
+    with patch.object(module.requests, "request", side_effect=requests.Timeout(SECRET)) as request:
+        with pytest.raises(module.ModuleError) as failure:
+            module._call("POST", "https://service.example/submit")
+    assert SECRET not in str(failure.value)
+    assert request.call_count == 1
+    assert request.call_args.kwargs["verify"] is True
+    assert request.call_args.kwargs["allow_redirects"] is False
+    assert request.call_args.kwargs["timeout"] == (5, 30)
+
+
+@pytest.mark.parametrize("raw", [0, -1, True, "bad", 101])
+def test_invalid_result_limit(raw):
+    with pytest.raises(module.ModuleError):
+        module._integer({"max_results": raw}, "max_results", 25)
+
+
+def test_host_uses_enrichment_endpoint_and_organization():
+    data = {
+        "result": {"resource": {"ip": "192.0.2.1", "services": [{"port": 443, "transport_protocol": "TCP"}], "risk": 0}}
+    }
+    with patch.object(module.requests, "request", return_value=reply(data)) as request:
+        result = module.handler(query(kind="ip-dst", value="192.0.2.1"))
+    assert_source(result)
+    assert request.call_args.args == ("GET", "https://api.platform.censys.io/v3/global/asset/enrichment/host/192.0.2.1")
+    assert request.call_args.kwargs["params"] == {"organization_id": SECRET}
+    assert values(objects(result, "network-connection")[0], "dst-port") == ["443"]
+
+
+def test_domain_requests_two_web_properties_and_keeps_partial_results():
+    with patch.object(
+        module.requests,
+        "request",
+        side_effect=[reply(status=404), reply({"result": {"resource": {"hostname": "example.com"}}})],
+    ) as request:
+        result = module.handler(query(kind="domain"))
+    assert_source(result)
+    assert request.call_count == 2
+    assert request.call_args.args[1].endswith("webproperty/example.com%3A443")
+    assert len(objects(result, "annotation")) == 1
+
+
+def test_ip_source_direction_is_preserved():
+    with patch.object(
+        module.requests, "request", return_value=reply({"result": {"resource": {"services": [{"port": 443}]}}})
+    ):
+        result = module.handler(query(kind="ip-src", value="192.0.2.1"))
+    assert values(objects(result, "network-connection")[0], "ip-src") == ["192.0.2.1"]