feat: [dnslytics] add domain dataset searches

documentation/expansion/dnslytics.md

@@ -0,0 +1,11 @@
+# DNSlytics Domain Search
+
+Run an explicit DNSlytics domain dataset query supplied as a text attribute. One request retrieves page 1; at most max_results domains are imported. Domain metadata preserves active or dropped status when supplied. The premium endpoint consumes credits; a local cap does not reduce the provider charge. This initial module does not perform additional DNS or IP2ASN requests.
+
+Input: text.
+
+Configuration: api_key, max_results. Required: api_key.
+
+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://dnslytics.com/api

misp_modules/modules/expansion/dnslytics.py

@@ -0,0 +1,190 @@
+import ipaddress
+import json
+import re
+import zipfile
+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 _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 _pivot(event, attribute, kind, value):
+    value = _normalise(kind, value)
+    if kind == attribute["type"] and value == _normalise(attribute["type"], attribute["value"]):
+        return
+    obj = _object(
+        event, attribute, "file" if kind in ("md5", "sha1", "sha256") else ("url" if kind == "url" else "domain-ip")
+    )
+    relation = "ip" if kind in ("ip-src", "ip-dst") else kind
+    obj.add_attribute(relation, value=value, to_ids=False)
+
+
+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": "DNSlytics Domain Search",
+    "description": (
+        "Run an explicit DNSlytics domain dataset query supplied as a text attribute. One request retrieves page 1; at"
+        " most max_results domains are imported. Domain metadata preserves active or dropped status when supplied. The"
+        " premium endpoint consumes credits; a local cap does not reduce the provider charge. This initial module does"
+        " not perform additional DNS or IP2ASN requests."
+    ),
+    "module-type": ["expansion"],
+    "data-source": "external",
+    "logo": "",
+    "requirements": ["requests", "pymisp"],
+    "features": (
+        "Run an explicit DNSlytics domain dataset query supplied as a text attribute. One request retrieves page 1; at"
+        " most max_results domains are imported. Domain metadata preserves active or dropped status when supplied. The"
+        " premium endpoint consumes credits; a local cap does not reduce the provider charge. This initial module does"
+        " not perform additional DNS or IP2ASN requests."
+    ),
+    "references": ["https://dnslytics.com/api"],
+    "input": "text",
+    "output": "MISP objects linked to the original attribute.",
+}
+mispattributes = {"input": ["text"], "format": "misp_standard"}
+moduleconfig = ["api_key", "max_results"]
+REQUIRED = ["api_key"]
+
+
+def _expand(event, attribute, value, config, limit):
+    data = _call(
+        "GET",
+        "https://api.dnslytics.net/v2/dataset/domains",
+        params={"apikey": config["api_key"], "q": value, "page": 1},
+    )
+    if data is None:
+        return
+    if not isinstance(data, dict):
+        raise ModuleError("DNSlytics returned an invalid search response.")
+    if data.get("status") not in (None, "succeed", "success", "ok", 200):
+        raise ModuleError("DNSlytics rejected the search.")
+    payload = data.get("data", data)
+    rows = payload.get("domains", payload.get("results", []))
+    if not isinstance(rows, list):
+        raise ModuleError("DNSlytics returned invalid domain results.")
+    total = payload.get("ndomains", len(rows))
+    for row in rows[:limit]:
+        domain = row if isinstance(row, str) else row["domain"]
+        _pivot(event, attribute, "domain", domain)
+        if isinstance(row, dict):
+            _annotation(event, attribute, "DNSlytics domain observation", row)
+    if len(rows) > limit or int(total) > len(rows):
+        _annotation(event, attribute, "Search limit", {"returned": min(len(rows), limit), "total": total, "page": 1})

tests/test_dnslytics.py

@@ -0,0 +1,121 @@
+import json
+from unittest.mock import Mock, patch
+
+import pytest
+import requests
+
+from misp_modules.modules.expansion import dnslytics 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_search_preserves_query_and_caps_domains():
+    data = {
+        "status": "succeed",
+        "data": {"domains": [{"domain": "a.example", "active": False}, {"domain": "b.example"}], "ndomains": 90},
+    }
+    with patch.object(module.requests, "request", return_value=reply(data)) as request:
+        result = module.handler(
+            query(kind="text", value="name:*example*", config={"api_key": SECRET, "max_results": 1})
+        )
+    assert_source(result)
+    assert request.call_count == 1
+    assert request.call_args.kwargs["params"]["q"] == "name:*example*"
+    assert len(objects(result, "domain-ip")) == 1
+    assert "false" in values(objects(result, "annotation")[0], "text")[0]
+    assert json.loads(values(objects(result, "annotation")[-1], "text")[0])["total"] == 90
+
+
+def test_empty_search_keeps_only_source():
+    with patch.object(module.requests, "request", return_value=reply({"data": {"domains": []}})):
+        result = module.handler(query(kind="text"))
+    assert_source(result)
+    assert not result["results"]["Object"]
+
+
+def test_pivots_normalize_source_and_use_the_url_template():
+    from pymisp import MISPEvent
+
+    event = MISPEvent()
+    attribute = {"type": "sha256", "value": "A" * 64, "uuid": UUID}
+    module._pivot(event, attribute, "sha256", "a" * 64)
+    assert not event.objects
+    module._pivot(event, attribute, "url", "https://example.com/path")
+    assert event.objects[0].name == "url"
+    assert event.objects[0].attributes[0].object_relation == "url"