Add schema validation for safety_db importer

Signed-off-by: Shivam Sandbhor <shivam.sandbhor@gmail.com>

Author: sbs2001

requirements.txt

@@ -30,4 +30,5 @@ tqdm==4.41.1
 wcwidth==0.1.7
 whitenoise==5.0.1
 zipp==0.6.0
-pytoml==0.1.21
+pytoml==0.1.21
+schema==0.7.1

vulnerabilities/scraper/safety_db.py

@@ -1,5 +1,6 @@
 # Data Imported from https://github.com/pyupio/safety-db
 import json
+from schema import Regex, Or, Schema, SchemaError
 from urllib.error import HTTPError
 from urllib.request import urlopen
 from dephell_specifier import RangeSpecifier
@@ -13,6 +14,24 @@ def get_data_file():
     return json_file
 
 
+def validate_schema(advisory_dict):
+
+    scheme = {
+        str:
+            [
+                {
+                    "advisory": str,
+                    "cve": Or(None, Regex(r"CVE-\d+-\d+")),
+                    "id": Regex(r"^pyup.io-\d"),
+                    "specs": list,
+                    "v": str
+                }
+            ]
+    }
+
+    Schema(scheme).validate(advisory_dict)
+
+
 def get_all_versions_of_package(package_name):
     url = "https://pypi.org/pypi/{}/json".format(package_name)
     releases = set()
@@ -29,6 +48,7 @@ def get_all_versions_of_package(package_name):
 def import_vulnerabilities():
     vulnerability_package_dicts = []
     data = get_data_file()
+    validate_schema(data)
     cves = set()
     for package_name in data:
         all_package_versions = set(get_all_versions_of_package(package_name))