ailtyposquatting (#2341)

* ailtyposquatting

* restore a file that was deleted

* fix

* fix

* changes

* tests

* no files

* logs

* files

* variables

* test

* test

* enum

* tests

* tests

* dns_resolve

* migration

* a log :p

---------

Co-authored-by: g4ze <[email protected]>

api_app/analyzers_manager/migrations/0093_analyzer_config_ailtyposquatting.py

@@ -0,0 +1,151 @@
+from django.db import migrations
+from django.db.models.fields.related_descriptors import (
+    ForwardManyToOneDescriptor,
+    ForwardOneToOneDescriptor,
+    ManyToManyDescriptor,
+)
+
+plugin = {
+    "python_module": {
+        "health_check_schedule": None,
+        "update_schedule": None,
+        "module": "ailtyposquatting.AilTypoSquatting",
+        "base_path": "api_app.analyzers_manager.observable_analyzers",
+    },
+    "name": "AILTypoSquatting",
+    "description": "[AILTypoSquatting](https://github.com/typosquatter/ail-typo-squatting) is a Python library to generate list of potential typo squatting domains with domain name permutation engine to feed AIL and other systems.",
+    "disabled": False,
+    "soft_time_limit": 60,
+    "routing_key": "default",
+    "health_check_status": True,
+    "type": "observable",
+    "docker_based": False,
+    "maximum_tlp": "RED",
+    "observable_supported": ["domain"],
+    "supported_filetypes": [],
+    "run_hash": False,
+    "run_hash_type": "",
+    "not_supported_filetypes": [],
+    "model": "analyzers_manager.AnalyzerConfig",
+}
+
+params = [
+    {
+        "python_module": {
+            "module": "ailtyposquatting.AilTypoSquatting",
+            "base_path": "api_app.analyzers_manager.observable_analyzers",
+        },
+        "name": "dns_resolving",
+        "type": "bool",
+        "description": "dns_resolving for AilTypoSquatting; only works for TLP CLEAR",
+        "is_secret": False,
+        "required": False,
+    },
+]
+values = [
+    {
+        "parameter": {
+            "python_module": {
+                "module": "ailtyposquatting.AilTypoSquatting",
+                "base_path": "api_app.analyzers_manager.observable_analyzers",
+            },
+            "name": "dns_resolving",
+            "type": "bool",
+            "description": "dns_resolving for AilTypoSquatting; only works for TLP CLEAR",
+            "is_secret": False,
+            "required": False,
+        },
+        "analyzer_config": "AILTypoSquatting",
+        "connector_config": None,
+        "visualizer_config": None,
+        "ingestor_config": None,
+        "pivot_config": None,
+        "for_organization": False,
+        "value": False,
+        "updated_at": "2024-05-26T00:10:15.236358Z",
+        "owner": None,
+    },
+]
+
+
+def _get_real_obj(Model, field, value):
+    def _get_obj(Model, other_model, value):
+        if isinstance(value, dict):
+            real_vals = {}
+            for key, real_val in value.items():
+                real_vals[key] = _get_real_obj(other_model, key, real_val)
+            value = other_model.objects.get_or_create(**real_vals)[0]
+        # it is just the primary key serialized
+        else:
+            if isinstance(value, int):
+                if Model.__name__ == "PluginConfig":
+                    value = other_model.objects.get(name=plugin["name"])
+                else:
+                    value = other_model.objects.get(pk=value)
+            else:
+                value = other_model.objects.get(name=value)
+        return value
+
+    if (
+        type(getattr(Model, field))
+        in [ForwardManyToOneDescriptor, ForwardOneToOneDescriptor]
+        and value
+    ):
+        other_model = getattr(Model, field).get_queryset().model
+        value = _get_obj(Model, other_model, value)
+    elif type(getattr(Model, field)) in [ManyToManyDescriptor] and value:
+        other_model = getattr(Model, field).rel.model
+        value = [_get_obj(Model, other_model, val) for val in value]
+    return value
+
+
+def _create_object(Model, data):
+    mtm, no_mtm = {}, {}
+    for field, value in data.items():
+        value = _get_real_obj(Model, field, value)
+        if type(getattr(Model, field)) is ManyToManyDescriptor:
+            mtm[field] = value
+        else:
+            no_mtm[field] = value
+    try:
+        o = Model.objects.get(**no_mtm)
+    except Model.DoesNotExist:
+        o = Model(**no_mtm)
+        o.full_clean()
+        o.save()
+        for field, value in mtm.items():
+            attribute = getattr(o, field)
+            if value is not None:
+                attribute.set(value)
+        return False
+    return True
+
+
+def migrate(apps, schema_editor):
+    Parameter = apps.get_model("api_app", "Parameter")
+    PluginConfig = apps.get_model("api_app", "PluginConfig")
+    python_path = plugin.pop("model")
+    Model = apps.get_model(*python_path.split("."))
+    if not Model.objects.filter(name=plugin["name"]).exists():
+        exists = _create_object(Model, plugin)
+        if not exists:
+            for param in params:
+                _create_object(Parameter, param)
+            for value in values:
+                _create_object(PluginConfig, value)
+
+
+def reverse_migrate(apps, schema_editor):
+    python_path = plugin.pop("model")
+    Model = apps.get_model(*python_path.split("."))
+    Model.objects.get(name=plugin["name"]).delete()
+
+
+class Migration(migrations.Migration):
+    atomic = False
+    dependencies = [
+        ("api_app", "0062_alter_parameter_python_module"),
+        ("analyzers_manager", "0092_alter_validin_desc"),
+    ]
+
+    operations = [migrations.RunPython(migrate, reverse_migrate)]

api_app/analyzers_manager/observable_analyzers/ailtyposquatting.py

@@ -0,0 +1,67 @@
+import logging
+import math
+
+from ail_typo_squatting import typo
+from ail_typo_squatting.dns_local import resolving
+
+from api_app.analyzers_manager import classes
+from tests.mock_utils import if_mock_connections, patch
+
+logger = logging.getLogger(__name__)
+
+
+class AilTypoSquatting(classes.ObservableAnalyzer):
+    """
+    wrapper for https://github.com/typosquatter/ail-typo-squatting
+    """
+
+    dns_resolving: bool = False
+
+    def update(self) -> bool:
+        pass
+
+    def run(self):
+        response = {}
+        logger.info(
+            f"""running AilTypoSquatting on {self.observable_name}
+                    with tlp {self._job.tlp}
+                    and dns resolving {self.dns_resolving}"""
+        )
+
+        response["algorithms"] = typo.runAll(
+            domain=self.observable_name,
+            limit=math.inf,
+            formatoutput="text",
+            pathOutput=None,
+        )
+        if self._job.tlp == self._job.TLP.CLEAR.value and self.dns_resolving:
+            # for "x.com", response["algorithms"][0]=".com"
+            # which is not valid for look up
+            if len(self.observable_name.split(".")[0]) == 1:
+                logger.info(
+                    f"""running dns resolving on {self.observable_name}
+                     excluding {response['algorithms'][0]}"""
+                )
+                response["dnsResolving"] = resolving.dnsResolving(
+                    resultList=response["algorithms"][1:],
+                    domain=self.observable_name,
+                    pathOutput=None,
+                )
+            else:
+                response["dnsResolving"] = resolving.dnsResolving(
+                    resultList=response["algorithms"],
+                    domain=self.observable_name,
+                    pathOutput=None,
+                )
+
+        return response
+
+    @classmethod
+    def _monkeypatch(cls):
+        patches = [
+            if_mock_connections(
+                patch.object(typo, "runAll", return_value=None),
+                patch.object(resolving, "dnsResolving", return_value=None),
+            )
+        ]
+        return super()._monkeypatch(patches=patches)

docs/source/Usage.md

@@ -256,6 +256,7 @@ The following is the list of the available analyzers you can run out-of-the-box.
 * `HudsonRock`: [Hudson Rock](https://cavalier.hudsonrock.com/docs) provides its clients the ability to query a database of over 27,541,128 computers which were compromised through global info-stealer campaigns performed by threat actors.
 * `CyCat`: [CyCat](https://cycat.org/) or the CYbersecurity Resource CATalogue aims at mapping and documenting, in a single formalism and catalogue available cybersecurity tools, rules, playbooks, processes and controls. 
 * `Vulners`: [Vulners](vulners.com) is the most complete and the only fully correlated security intelligence database, which goes through constant updates and links 200+ data sources in a unified machine-readable format. It contains 8 mln+ entries, including CVEs, advisories, exploits, and IoCs — everything you need to stay abreast on the latest security threats.
+* `AILTypoSquatting`:[AILTypoSquatting](https://github.com/typosquatter/ail-typo-squatting) is a Python library to generate list of potential typo squatting domains with domain name permutation engine to feed AIL and other systems.
 
 ##### Generic analyzers (email, phone number, etc.; anything really)
 

requirements/project-requirements.txt

@@ -78,7 +78,7 @@ querycontacts==2.0.0
 blint==2.1.5
 hfinger==0.2.2
 permhash==0.1.4
-
+ail_typo_squatting==2.7.4
 # this is required because XLMMacroDeobfuscator does not pin the following packages
 pyxlsb2==0.0.8
 xlrd2==1.3.4