From 38c369e0db19f7416e14bac57490e083c0298ab7 Mon Sep 17 00:00:00 2001
From: Jonathan Ringer <jringer@anduril.com>
Date: Thu, 26 Sep 2024 14:33:50 -0700
Subject: [PATCH] CDX 1.4

---
 src/sbomnix/cdx.py     | 31 ++++++++++++++++++++++++++++++-
 src/sbomnix/sbomdb.py  | 41 ++++++++++++++++++++++++++++++++++++++---
 src/vulnxscan/utils.py | 10 +++++++++-
 3 files changed, 77 insertions(+), 5 deletions(-)

diff --git a/src/sbomnix/cdx.py b/src/sbomnix/cdx.py
index 3cdc3aa..1b9beb1 100644
--- a/src/sbomnix/cdx.py
+++ b/src/sbomnix/cdx.py
@@ -4,12 +4,17 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 
-""" CycloneDX utils """
+################################################################################
 
+# CycloneDX
+
+""" Module for generating cdx entries """
 
 import re
+
 from reuse._licenses import LICENSE_MAP as SPDX_LICENSES
 from common.utils import LOG, LOG_SPAM
+from vulnxscan.utils import _vuln_source, _vuln_url
 
 
 def _drv_to_cdx_licenses_entry(drv, column_name, cdx_license_type):
@@ -128,3 +133,27 @@ def _drv_to_cdx_dependency(drv, deps_list, uid="store_path"):
     if deps_list:
         dependency["dependsOn"] = deps_list
     return dependency
+
+
+def _vuln_to_cdx_vuln(vuln):
+    """Return cdx vulnerability entry from vulnix row"""
+    vulnerability = {}
+    vulnerability["bom-ref"] = vuln.store_path
+    vulnerability["id"] = vuln.vuln_id
+    source = {}
+    source["url"] = _vuln_url(vuln)
+    source["name"] = _vuln_source(vuln)
+    vulnerability["source"] = source
+    vulnerability["ratings"] = []
+    # If the vulnerability is still being assessed, it will be missing a valid number
+    if vuln.severity != "":
+        rating = {}
+        rating["source"] = source
+        rating["score"] = vuln.severity
+        vulnerability["ratings"].append(rating)
+    vulnerability["tools"] = []
+    for scanner in vuln.scanner:
+        tool = {}
+        tool["name"] = scanner
+        vulnerability["tools"].append(tool)
+    return vulnerability
diff --git a/src/sbomnix/sbomdb.py b/src/sbomnix/sbomdb.py
index abac19b..164d179 100644
--- a/src/sbomnix/sbomdb.py
+++ b/src/sbomnix/sbomdb.py
@@ -4,10 +4,11 @@
 #
 # SPDX-License-Identifier: Apache-2.0
 
-# pylint: disable=too-many-instance-attributes, too-many-arguments
+# pylint: disable=too-many-instance-attributes, too-many-arguments, too-many-locals
 
 """ Module for generating SBOMs in various formats """
 
+from tempfile import NamedTemporaryFile
 import uuid
 import logging
 import json
@@ -18,10 +19,11 @@
 import numpy as np
 from reuse._licenses import LICENSE_MAP as SPDX_LICENSES
 from nixgraph.graph import NixDependencies
-from sbomnix.cdx import _drv_to_cdx_component, _drv_to_cdx_dependency
+from sbomnix.cdx import _drv_to_cdx_component, _drv_to_cdx_dependency, _vuln_to_cdx_vuln
 from sbomnix.nix import Store, find_deriver
 from sbomnix.meta import Meta
 from common.utils import LOG, df_to_csv_file, get_py_pkg_version
+from vulnxscan.vulnscan import VulnScan
 
 ###############################################################################
 
@@ -168,7 +170,7 @@ def to_cdx(self, cdx_path, printinfo=True):
         """Export sbomdb to cyclonedx json file"""
         cdx = {}
         cdx["bomFormat"] = "CycloneDX"
-        cdx["specVersion"] = "1.3"
+        cdx["specVersion"] = "1.4"
         cdx["version"] = 1
         cdx["serialNumber"] = f"urn:uuid:{self.uuid}"
         cdx["metadata"] = {}
@@ -202,6 +204,39 @@ def to_cdx(self, cdx_path, printinfo=True):
             deps = self._lookup_dependencies(drv, uid=self.uid)
             dependency = _drv_to_cdx_dependency(drv, deps, uid=self.uid)
             cdx["dependencies"].append(dependency)
+        scanner = VulnScan()
+        scanner.scan_vulnix(self.target_deriver, self.buildtime)
+        # Write incomplete sbom to a temporary path, then perform a vulnerability scan
+        with NamedTemporaryFile(
+            delete=False, prefix="vulnxscan_", suffix=".json"
+        ) as fcdx:
+            self._write_json(fcdx.name, cdx, printinfo=False)
+            scanner.scan_grype(fcdx.name)
+            scanner.scan_osv(fcdx.name)
+        cdx["vulnerabilities"] = []
+        # Union all scans into a single dataframe
+        vulns = pd.concat(
+            [scanner.df_grype, scanner.df_osv, scanner.df_vulnix],
+            ignore_index=True,
+        )
+        # Concat adds a modified column, remove
+        vulns.drop("modified", axis=1, inplace=True)
+        # Deduplicate repeated vulnerabilities, making the scanner column into an array
+        vuln_grouped = vulns.groupby(
+            ["package", "version", "severity", "vuln_id"],
+            as_index=False,
+        ).agg({"scanner": pd.Series.unique})
+        # Do a join so we have access to bom-ref
+        vulnix_components = pd.merge(
+            left=vuln_grouped,
+            right=self.df_sbomdb,
+            how="left",
+            left_on=["package", "version"],
+            right_on=["pname", "version"],
+        )
+        for vuln in vulnix_components.itertuples():
+            vulnix_vuln = _vuln_to_cdx_vuln(vuln)
+            cdx["vulnerabilities"].append(vulnix_vuln)
         self._write_json(cdx_path, cdx, printinfo)
 
     def to_spdx(self, spdx_path, printinfo=True):
diff --git a/src/vulnxscan/utils.py b/src/vulnxscan/utils.py
index 7982d77..0619ab5 100644
--- a/src/vulnxscan/utils.py
+++ b/src/vulnxscan/utils.py
@@ -63,11 +63,19 @@ def _vuln_url(row):
     nvd_url = "https://nvd.nist.gov/vuln/detail/"
     if "cve" in row.vuln_id.lower():
         return f"{nvd_url}{row.vuln_id}"
-    if row.osv:
+    if getattr(row, "osv", False) or ("osv" in getattr(row, "scanner", [])):
         return f"{osv_url}{row.vuln_id}"
     return ""
 
 
+def _vuln_source(row):
+    if "cve" in row.vuln_id.lower():
+        return "NVD"
+    if getattr(row, "osv", False) or ("osv" in getattr(row, "scanner", [])):
+        return "OSV"
+    return ""
+
+
 def _is_patched(row):
     if row.vuln_id and str(row.vuln_id).lower() in str(row.patches).lower():
         patches = row.patches.split()