Grayhat Warfare module (smicallef#1136)

* Grayhat Warfare Module

modules/sfp_grayhatwarfare.py

@@ -0,0 +1,175 @@
+# -------------------------------------------------------------------------------
+# Name:        sfp_grayhatwarfare
+# Purpose:     Find bucket names matching the keyword extracted from a domain from Grayhat API.
+#
+# Author:      <krishnasis@hotmail.com>
+#
+# Created:     24-01-2021
+# Copyright:   (c) Steve Micallef
+# Licence:     GPL
+# -------------------------------------------------------------------------------
+
+import json
+import time
+import urllib
+from spiderfoot import SpiderFootEvent, SpiderFootPlugin
+
+
+class sfp_grayhatwarfare(SpiderFootPlugin):
+
+    meta = {
+        'name': "Grayhat Warfare",
+        'summary': "Find bucket names matching the keyword extracted from a domain from Grayhat API.",
+        'flags': ["apikey"],
+        'useCases': ["Footprint", "Investigate", "Passive"],
+        'categories': ["Reputation Systems"],
+        'dataSource': {
+            'website': "https://buckets.grayhatwarfare.com/",
+            'model': "FREE_AUTH_LIMITED",
+            'references': [
+                "https://buckets.grayhatwarfare.com/docs/api/v1"
+            ],
+            'favIcon': "https://buckets.grayhatwarfare.com/assets/template/images/favicon.png",
+            'logo': "https://buckets.grayhatwarfare.com/assets/images/logo/logo-sm.png",
+            'description': "It is a searchable database of open buckets."
+            "Has up to million results of each bucket."
+            "Full text search with binary logic (can search for keywords and also stopwords)",
+        }
+    }
+
+    # Default options
+    opts = {
+        'api_key': '',
+        'per_page': 1000,
+        'max_pages': 2,
+        'pause': 1
+    }
+
+    # Option descriptions
+    optdescs = {
+        'api_key': 'Grayhat Warfare API key.',
+        'per_page': 'Maximum number of results per page (Max: 1000).',
+        'max_pages': 'Maximum number of pages to fetch.',
+        'pause': 'Number of seconds to wait between each API call.'
+    }
+
+    results = None
+    errorState = False
+
+    def setup(self, sfc, userOpts=dict()):
+        self.sf = sfc
+        self.results = self.tempStorage()
+
+        for opt in list(userOpts.keys()):
+            self.opts[opt] = userOpts[opt]
+
+    # What events is this module interested in for input
+    def watchedEvents(self):
+        return [
+            "DOMAIN_NAME",
+        ]
+
+    # What events this module produces
+    def producedEvents(self):
+        return [
+            'CLOUD_STORAGE_BUCKET',
+            'CLOUD_STORAGE_BUCKET_OPEN',
+            'RAW_RIR_DATA'
+        ]
+
+    # Query Grayhat Warfare
+    def query(self, keyword, start):
+        params = urllib.parse.urlencode({
+            'keywords': keyword.encode('raw_unicode_escape'),
+            'access_token': self.opts['api_key']
+        })
+
+        headers = {
+            'Accept': 'application/json',
+        }
+
+        res = self.sf.fetchUrl(
+            f"https://buckets.grayhatwarfare.com/api/v1/buckets/{start}/{self.opts['per_page']}?{params}",
+            headers=headers,
+            timeout=15,
+            useragent=self.opts['_useragent'],
+            verify=True
+        )
+
+        time.sleep(self.opts['pause'])
+
+        if res['code'] != "200":
+            self.sf.error("Unable to fetch data from Grayhat Warfare API.")
+            self.errorState = True
+            return None
+
+        if res['content'] is None:
+            self.sf.debug('No response from Grayhat Warfare API.')
+            return None
+
+        try:
+            return json.loads(res['content'])
+        except Exception as e:
+            self.sf.debug(f"Error processing JSON response: {e}")
+            return None
+
+    # Handle events sent to this module
+    def handleEvent(self, event):
+        eventName = event.eventType
+        srcModuleName = event.module
+        eventData = event.data
+
+        if eventData in self.results:
+            return None
+
+        if self.errorState:
+            return None
+
+        self.results[eventData] = True
+
+        self.sf.debug(f"Received event, {eventName}, from {srcModuleName}")
+
+        if self.opts['api_key'] == "":
+            self.sf.error("You enabled sfp_grayhatwarfare but did not set an API key!")
+            self.errorState = True
+            return
+
+        currentIndex = 0
+        currentPage = 0
+        maxPages = self.opts['max_pages']
+        perPage = self.opts['per_page']
+
+        keyword = self.sf.domainKeyword(eventData, self.opts['_internettlds'])
+
+        while currentPage < maxPages:
+            currentIndex = currentPage * perPage
+            if self.checkForStop():
+                return None
+
+            if self.errorState:
+                break
+
+            data = self.query(keyword=keyword, start=currentIndex)
+
+            if not data:
+                return None
+
+            for row in data.get('buckets'):
+                bucketName = row.get('bucket')
+                bucketKeyword = bucketName.split('.')[0]
+                self.sf.debug(bucketKeyword)
+                if bucketKeyword.startswith(keyword) or bucketKeyword.endswith(keyword):
+                    evt = SpiderFootEvent('CLOUD_STORAGE_BUCKET', bucketName, self.__name__, event)
+                    self.notifyListeners(evt)
+
+                    evt = SpiderFootEvent('CLOUD_STORAGE_BUCKET_OPEN', f"{bucketName}: {row.get('fileCount')} files found.", self.__name__, event)
+                    self.notifyListeners(evt)
+
+                    evt = SpiderFootEvent('RAW_RIR_DATA', str(row), self.__name__, event)
+                    self.notifyListeners(evt)
+
+            currentPage += 1
+            if data.get('buckets_count') < perPage:
+                break
+
+# End of sfp_grayhatwarfare class

test/unit/modules/test_sfp_grayhatwarfare.py

@@ -0,0 +1,80 @@
+# test_sfp_grayhatwarfare.py
+import unittest
+
+from modules.sfp_grayhatwarfare import sfp_grayhatwarfare
+from sflib import SpiderFoot
+from spiderfoot import SpiderFootEvent, SpiderFootTarget
+
+
+class TestModuleGrayhatWarfare(unittest.TestCase):
+    """
+    Test modules.sfp_grayhatwarfare
+    """
+
+    default_options = {
+        '_debug': False,  # Debug
+        '__logging': True,  # Logging in general
+        '__outputfilter': None,  # Event types to filter from modules' output
+        '_useragent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:62.0) Gecko/20100101 Firefox/62.0',  # User-Agent to use for HTTP requests
+        '_dnsserver': '',  # Override the default resolver
+        '_fetchtimeout': 5,  # number of seconds before giving up on a fetch
+        '_internettlds': 'https://publicsuffix.org/list/effective_tld_names.dat',
+        '_internettlds_cache': 72,
+        '_genericusers': "abuse,admin,billing,compliance,devnull,dns,ftp,hostmaster,inoc,ispfeedback,ispsupport,list-request,list,maildaemon,marketing,noc,no-reply,noreply,null,peering,peering-notify,peering-request,phish,phishing,postmaster,privacy,registrar,registry,root,routing-registry,rr,sales,security,spam,support,sysadmin,tech,undisclosed-recipients,unsubscribe,usenet,uucp,webmaster,www",
+        '__version__': '3.3-DEV',
+        '__database': 'spiderfoot.test.db',  # note: test database file
+        '__modules__': None,  # List of modules. Will be set after start-up.
+        '_socks1type': '',
+        '_socks2addr': '',
+        '_socks3port': '',
+        '_socks4user': '',
+        '_socks5pwd': '',
+        '_torctlport': 9051,
+        '__logstdout': False
+    }
+
+    def test_opts(self):
+        module = sfp_grayhatwarfare()
+        self.assertEqual(len(module.opts), len(module.optdescs))
+
+    def test_setup(self):
+        """
+        Test setup(self, sfc, userOpts=dict())
+        """
+        sf = SpiderFoot(self.default_options)
+
+        module = sfp_grayhatwarfare()
+        module.setup(sf, dict())
+
+    def test_watchedEvents_should_return_list(self):
+        module = sfp_grayhatwarfare()
+        self.assertIsInstance(module.watchedEvents(), list)
+
+    def test_producedEvents_should_return_list(self):
+        module = sfp_grayhatwarfare()
+        self.assertIsInstance(module.producedEvents(), list)
+
+    @unittest.skip("todo")
+    def test_handleEvent(self):
+        """
+        Test handleEvent(self, event)
+        """
+        sf = SpiderFoot(self.default_options)
+
+        module = sfp_grayhatwarfare()
+        module.setup(sf, dict())
+
+        target_value = 'example target value'
+        target_type = 'DOMAIN_NAME'
+        target = SpiderFootTarget(target_value, target_type)
+        module.setTarget(target)
+
+        event_type = 'ROOT'
+        event_data = 'example data'
+        event_module = ''
+        source_event = ''
+        evt = SpiderFootEvent(event_type, event_data, event_module, source_event)
+
+        result = module.handleEvent(evt)
+
+        self.assertIsNone(result)