🔍 add scanner_signature_sources config option

This adds the option to configure which set of signatures `fdroid
scanner` should use, by configuring it in `config.yml`. It allows
fetching signatures in our custom json format. It also adds 3 additional
sources: 'suss', 'exodus', 'etip'

examples/config.yml

@@ -355,3 +355,19 @@
 # lint_licenses:
 #   - Custom-License-A
 #   - Another-License
+
+# `fdroid scanner` can scan for signatures from various sources. By default
+# it's configured to only use F-Droids official SUSS collection. We have
+# support for these special collections:
+#  * 'exodus' - official exodus-privacy.org signatures
+#  * 'etip' - exodus privacy investigation platfrom community contributed
+#           signatures
+#  * 'suss' - official F-Droid: Suspicious or Unwanted Software Signatures
+# You can also configure scanner to use custom collections of signatures here.
+# They have to follow the format specified in the SUSS readme.
+# (https://gitlab.com/fdroid/fdroid-suss/#cache-file-data-format)
+#
+# scanner_signature_sources:
+#   - suss
+#   - exodus
+#   - https://example.com/signatures.json

fdroidserver/common.py

@@ -163,6 +163,7 @@ default_config = {
     'archive_older': 0,
     'lint_licenses': fdroidserver.lint.APPROVED_LICENSES,  # type: ignore
     'git_mirror_size_limit': 10000000000,
+    'scanner_signature_sources': ['suss'],
 }
 
 

fdroidserver/scanner.py

@@ -25,6 +25,7 @@ import logging
 import zipfile
 import itertools
 import traceback
+import urllib.parse
 import urllib.request
 from argparse import ArgumentParser
 from copy import deepcopy
@@ -141,6 +142,10 @@ class SignatureDataCacheMissException(Exception):
     pass
 
 
+class SignatureDataNoDefaultsException(Exception):
+    pass
+
+
 class SignatureDataVersionMismatchException(Exception):
     pass
 
@@ -198,7 +203,7 @@ class SignatureDataController:
                 self.check_last_updated()
             except SignatureDataCacheMissException:
                 self.load_from_defaults()
-        except SignatureDataOutdatedException:
+        except (SignatureDataOutdatedException, SignatureDataNoDefaultsException):
             self.fetch_signatures_from_web()
             self.write_to_cache()
         except (SignatureDataMalformedException, SignatureDataVersionMismatchException) as e:
@@ -208,9 +213,7 @@ class SignatureDataController:
             raise e
 
     def load_from_defaults(self):
-        sig_file = (Path(__file__).parent / 'data' / 'scanner' / self.filename).resolve()
+        raise SignatureDataNoDefaultsException()
-        with open(sig_file) as f:
-            self.set_data(json.load(f))
 
     def load_from_cache(self):
         sig_file = scanner._scanner_cachedir() / self.filename
@@ -254,8 +257,9 @@ class SignatureDataController:
 
 class ExodusSignatureDataController(SignatureDataController):
     def __init__(self):
-        super().__init__('Exodus signatures', 'exodus.yml', 'https://reports.exodus-privacy.eu.org/api/trackers')
+        super().__init__('Exodus signatures', 'exodus.json', 'https://reports.exodus-privacy.eu.org/api/trackers')
         self.cache_duration = timedelta(days=1)  # refresh exodus cache after one day
+        self.has_trackers_json_key = True
 
     def fetch_signatures_from_web(self):
         logging.debug(_("downloading '{}'").format(self.url))
@@ -270,8 +274,10 @@ class ExodusSignatureDataController(SignatureDataController):
         if not self.url.startswith("https://"):
             raise Exception(_("can't open non-https url: '{};".format(self.url)))
         with urllib.request.urlopen(self.url) as f:  # nosec B310 scheme filtered above
-            d = json.load(f)
+            trackerlist = json.load(f)
-            for tracker in d["trackers"].values():
+            if self.has_trackers_json_key:
+                trackerlist = trackerlist["trackers"].values()
+            for tracker in trackerlist:
                 if tracker.get('code_signature'):
                     data["signatures"][tracker["name"]] = {
                         "name": tracker["name"],
@@ -288,6 +294,15 @@ class ExodusSignatureDataController(SignatureDataController):
         self.set_data(data)
 
 
+class EtipSignatureDataController(ExodusSignatureDataController):
+    def __init__(self):
+        super().__init__()
+        self.name = 'ETIP signatures'
+        self.filename = 'etip.json'
+        self.url = 'https://etip.exodus-privacy.eu.org/api/trackers/?format=json'
+        self.has_trackers_json_key = False
+
+
 class SUSSDataController(SignatureDataController):
     def __init__(self):
         super().__init__(
@@ -302,16 +317,42 @@ class SUSSDataController(SignatureDataController):
 
 class ScannerTool():
     def __init__(self):
-        self.sdcs = [
-            SUSSDataController(),
-        ]
 
         # we could add support for loading additional signature source
         # definitions from config.yml here
 
+        self.scanner_data_lookup()
         self.load()
         self.compile_regexes()
 
+    def scanner_data_lookup(self):
+        sigsources = common.get_config().get('scanner_signature_sources', [])
+        logging.debug(
+            "scanner is configured to use signature data from: '{}'"
+            .format("', '".join(sigsources))
+        )
+        self.sdcs = []
+        for i, source_url in enumerate(sigsources):
+            if source_url.lower() == 'suss':
+                self.sdcs.append(SUSSDataController())
+            elif source_url.lower() == 'exodus':
+                self.sdcs.append(ExodusSignatureDataController())
+            elif source_url.lower() == 'etip':
+                self.sdcs.append(EtipSignatureDataController())
+            else:
+                u = urllib.parse.urlparse(source_url)
+                if u.scheme != 'https' or u.path == "":
+                    raise ConfigurationException(
+                        "Invalid 'scanner_signature_sources' configuration: '{}'. "
+                        "Has to be a valid HTTPS-URL or match a predefined "
+                        "constants: 'suss', 'exodus'".format(source_url)
+                    )
+                self.sdcs.append(SignatureDataController(
+                    source_url,
+                    '{}_{}'.format(i, os.path.basename(u.path)),
+                    source_url,
+                ))
+
     def load(self):
         for sdc in self.sdcs:
             sdc.load()
@@ -697,15 +738,11 @@ def main():
     # initialize/load configuration values
     common.get_config(opts=options)
 
+    if options.exodus:
+        if "exodus" not in common.get_config()['scanner_signature_sources']:
+            common.get_config()['scanner_signature_sources'].append('exodus')
     if options.refresh:
         scanner._get_tool().refresh()
-    if options.exodus:
-        c = ExodusSignatureDataController()
-        if options.refresh:
-            c.fetch_signatures_from_web()
-        else:
-            c.fetch()
-        scanner._get_tool().add(c)
 
     probcount = 0
 

tests/scanner.TestCase

@@ -656,6 +656,37 @@ class Test_SignatureDataController(unittest.TestCase):
         func_fsfw.assert_called_once_with()
         func_wtc.assert_called_once_with()
 
+    def test_load_try_web_when_no_defaults(self):
+        sdc = fdroidserver.scanner.SignatureDataController(
+            'nnn', 'fff.yml', 'https://example.com/test.json'
+        )
+        func_lfc = mock.Mock(
+            side_effect=fdroidserver.scanner.SignatureDataCacheMissException()
+        )
+        func_lfd = mock.Mock(
+            side_effect=fdroidserver.scanner.SignatureDataNoDefaultsException()
+        )
+        func_fsfw = mock.Mock()
+        func_wtc = mock.Mock()
+        with mock.patch(
+            'fdroidserver.scanner.SignatureDataController.load_from_cache',
+            func_lfc,
+        ), mock.patch(
+            'fdroidserver.scanner.SignatureDataController.load_from_defaults',
+            func_lfd,
+        ), mock.patch(
+            'fdroidserver.scanner.SignatureDataController.fetch_signatures_from_web',
+            func_fsfw,
+        ), mock.patch(
+            'fdroidserver.scanner.SignatureDataController.write_to_cache',
+            func_wtc,
+        ):
+            sdc.load()
+        func_lfc.assert_called_once_with()
+        func_lfd.assert_called_once_with()
+        func_fsfw.assert_called_once_with()
+        func_wtc.assert_called_once_with()
+
     @unittest.skipIf(
         sys.version_info < (3, 9, 0),
         "mock_open doesn't allow easy access to written data in older python versions",