Add dates and changelog for packages and vulnerabilities (#1310)

* Add chaneglog Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Add importer name to advisory importers Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Fix views Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * clean-up models.py Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Add importer name and importing authority for istio Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Fix models Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Make code more readable Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Cleanup models Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Add tests for changelogs Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Adjust package changelog according to new requirements Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Remove importing_authority Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Add vulnerability changelog Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Address review comments Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Make changes according to recent PR merging Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Format package_details.html Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Add tests for checking unicity of importer names Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Add tooltip for package and vulnerability views Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Typo fixes Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Fix failing tests Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Address review comments Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Fix typos Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Address review comments Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Fix pypa importer changelog url Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> * Refactor ubuntu usn Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com> --------- Signed-off-by: Tushar Goel <tushar.goel.dav@gmail.com>
aboutcode-org · Dec 26, 2023 · e2b60c9 · e2b60c9
1 parent d21d2c1
commit e2b60c9
Show file tree

Hide file tree

Showing 118 changed files with 4,942 additions and 2,290 deletions.
diff --git a/vulnerabilities/import_runner.py b/vulnerabilities/import_runner.py
@@ -18,17 +18,20 @@
 
 from vulnerabilities.importer import AdvisoryData
 from vulnerabilities.importer import Importer
+from vulnerabilities.importers import IMPORTERS_REGISTRY
 from vulnerabilities.improver import Inference
 from vulnerabilities.improvers.default import DefaultImporter
 from vulnerabilities.models import Advisory
 from vulnerabilities.models import Alias
 from vulnerabilities.models import Package
 from vulnerabilities.models import PackageRelatedVulnerability
 from vulnerabilities.models import Vulnerability
+from vulnerabilities.models import VulnerabilityChangeLog
 from vulnerabilities.models import VulnerabilityReference
 from vulnerabilities.models import VulnerabilityRelatedReference
 from vulnerabilities.models import VulnerabilitySeverity
 from vulnerabilities.models import Weakness
+from vulnerabilities.utils import get_importer_name
 
 logger = logging.getLogger(__name__)
 
@@ -102,6 +105,7 @@ def process_advisories(
                         "created_by": importer_name,
                         "date_collected": datetime.datetime.now(tz=datetime.timezone.utc),
                     },
+                    url=data.url,
                 )
                 if not obj.date_imported:
                     advisories.append(obj)
@@ -151,6 +155,7 @@ def process_inferences(inferences: List[Inference], advisory: Advisory, improver
             vulnerability_id=inference.vulnerability_id,
             aliases=inference.aliases,
             summary=inference.summary,
+            advisory=advisory,
         )
 
         if not vulnerability:
@@ -193,24 +198,24 @@ def process_inferences(inferences: List[Inference], advisory: Advisory, improver
                     )
 
         for affected_purl in inference.affected_purls or []:
-            vulnerable_package = Package.objects.get_or_create_from_purl(purl=affected_purl)
+            vulnerable_package, _ = Package.objects.get_or_create_from_purl(purl=affected_purl)
             PackageRelatedVulnerability(
                 vulnerability=vulnerability,
                 package=vulnerable_package,
                 created_by=improver_name,
                 confidence=inference.confidence,
                 fix=False,
-            ).update_or_create()
+            ).update_or_create(advisory=advisory)
 
         if inference.fixed_purl:
-            fixed_package = Package.objects.get_or_create_from_purl(purl=inference.fixed_purl)
+            fixed_package, _ = Package.objects.get_or_create_from_purl(purl=inference.fixed_purl)
             PackageRelatedVulnerability(
                 vulnerability=vulnerability,
                 package=fixed_package,
                 created_by=improver_name,
                 confidence=inference.confidence,
                 fix=True,
-            ).update_or_create()
+            ).update_or_create(advisory=advisory)
 
         if inference.weaknesses and vulnerability:
             for cwe_id in inference.weaknesses:
@@ -246,7 +251,7 @@ def create_valid_vulnerability_reference(url, reference_id=None):
 
 
 def get_or_create_vulnerability_and_aliases(
-    aliases: List[str], vulnerability_id=None, summary=None
+    aliases: List[str], vulnerability_id=None, summary=None, advisory=None
 ):
     """
     Get or create vulnerabilitiy and aliases such that all existing and new
@@ -297,6 +302,12 @@ def get_or_create_vulnerability_and_aliases(
             vulnerability = create_vulnerability_and_add_aliases(
                 aliases=new_alias_names, summary=summary
             )
+            importer_name = get_importer_name(advisory)
+            VulnerabilityChangeLog.log_import(
+                importer=importer_name,
+                source_url=advisory.url,
+                vulnerability=vulnerability,
+            )
         except Exception as e:
             logger.error(
                 f"Cannot create vulnerability with summary {summary!r} and {new_alias_names!r} {e!r}.\n{traceback_format_exc()}."

diff --git a/vulnerabilities/importer.py b/vulnerabilities/importer.py
@@ -248,6 +248,7 @@ class AdvisoryData:
     references: List[Reference] = dataclasses.field(default_factory=list)
     date_published: Optional[datetime.datetime] = None
     weaknesses: List[int] = dataclasses.field(default_factory=list)
+    url: Optional[str] = None
 
     def __post_init__(self):
         if self.date_published and not self.date_published.tzinfo:
@@ -271,6 +272,7 @@ def to_dict(self):
             "references": [ref.to_dict() for ref in self.references],
             "date_published": self.date_published.isoformat() if self.date_published else None,
             "weaknesses": self.weaknesses,
+            "url": self.url if self.url else "",
         }
 
     @classmethod
@@ -287,6 +289,7 @@ def from_dict(cls, advisory_data):
             if date_published
             else None,
             "weaknesses": advisory_data["weaknesses"],
+            "url": advisory_data.get("url") or None,
         }
         return cls(**transformed)
 
@@ -313,6 +316,8 @@ class Importer:
     license_url = ""
     notice = ""
     vcs_response: VCSResponse = None
+    # It needs to be unique and immutable
+    importer_name = ""
 
     def __init__(self):
         if not self.spdx_license_expression:
@@ -358,6 +363,9 @@ class OvalImporter(Importer):
     `OvalDataSource` class. Subclasses must implement the methods `_fetch` and `set_api`.
     """
 
+    data_url: str = ""
+    importer_name = "Oval Importer"
+
     @staticmethod
     def create_purl(pkg_name: str, pkg_data: Mapping) -> PackageURL:
         """
@@ -472,4 +480,5 @@ def get_data_from_xml_doc(
                 affected_packages=affected_packages,
                 references=sorted(references),
                 date_published=date_published,
+                url=self.data_url,
             )
diff --git a/vulnerabilities/importers/alpine_linux.py b/vulnerabilities/importers/alpine_linux.py
@@ -35,6 +35,7 @@
 class AlpineImporter(Importer):
     spdx_license_expression = "CC-BY-SA-4.0"
     license_url = "https://secdb.alpinelinux.org/license.txt"
+    importer_name = "Alpine Linux Importer"
 
     def advisory_data(self) -> Iterable[AdvisoryData]:
         page_response_content = fetch_response(BASE_URL).content
@@ -50,7 +51,7 @@ def advisory_data(self) -> Iterable[AdvisoryData]:
             if not record["packages"]:
                 LOGGER.error(f'"packages" not found in {link!r}')
                 continue
-            yield from process_record(record)
+            yield from process_record(record=record, url=link)
 
 
 def fetch_advisory_directory_links(page_response_content: str) -> List[str]:
@@ -98,12 +99,12 @@ def check_for_attributes(record) -> bool:
     return True
 
 
-def process_record(record: dict) -> Iterable[AdvisoryData]:
+def process_record(record: dict, url: str) -> Iterable[AdvisoryData]:
     """
     Return a list of AdvisoryData objects by processing data
     present in that `record`
     """
-    if not record["packages"]:
+    if not record.get("packages"):
         LOGGER.error(f'"packages" not found in this record {record!r}')
         return []
 
@@ -114,10 +115,11 @@ def process_record(record: dict) -> Iterable[AdvisoryData]:
         if not check_for_attributes(record):
             continue
         yield from load_advisories(
-            package["pkg"],
-            record["distroversion"],
-            record["reponame"],
-            record["archs"],
+            pkg_infos=package["pkg"],
+            distroversion=record["distroversion"],
+            reponame=record["reponame"],
+            archs=record["archs"],
+            url=url,
         )
 
 
@@ -126,6 +128,7 @@ def load_advisories(
     distroversion: str,
     reponame: str,
     archs: List[str],
+    url: str,
 ) -> Iterable[AdvisoryData]:
     """
     Yield AdvisoryData by mapping data from `pkg_infos`
@@ -211,4 +214,5 @@ def load_advisories(
                 references=references,
                 affected_packages=affected_packages,
                 aliases=aliases,
+                url=url,
             )
diff --git a/vulnerabilities/importers/apache_httpd.py b/vulnerabilities/importers/apache_httpd.py
@@ -33,6 +33,7 @@ class ApacheHTTPDImporter(Importer):
     base_url = "https://httpd.apache.org/security/json/"
     spdx_license_expression = "Apache-2.0"
     license_url = "https://www.apache.org/licenses/LICENSE-2.0"
+    importer_name = "Apache HTTPD Importer"
 
     def advisory_data(self):
         links = fetch_links(self.base_url)
@@ -106,6 +107,7 @@ def to_advisory(self, data):
             summary=description or "",
             affected_packages=affected_packages,
             references=[reference],
+            url=reference.url,
         )
 
     def to_version_ranges(self, versions_data, fixed_versions):

diff --git a/vulnerabilities/importers/apache_kafka.py b/vulnerabilities/importers/apache_kafka.py
@@ -95,6 +95,7 @@ class ApacheKafkaImporter(Importer):
     ASF_PAGE_URL = "https://kafka.apache.org/cve-list"
     spdx_license_expression = "Apache-2.0"
     license_url = "https://www.apache.org/licenses/"
+    importer_name = "Apache Kafka Importer"
 
     @staticmethod
     def fetch_advisory_page(self):
@@ -191,6 +192,7 @@ def to_advisory(self, advisory_page):
                         affected_packages=affected_packages,
                         references=references,
                         date_published=date_published,
+                        url=f"{self.ASF_PAGE_URL}#{cve_id}",
                     )
                 )
 

diff --git a/vulnerabilities/importers/apache_tomcat.py b/vulnerabilities/importers/apache_tomcat.py
@@ -118,14 +118,15 @@ class ApacheTomcatImporter(Importer):
 
     spdx_license_expression = "Apache-2.0"
     license_url = "https://www.apache.org/licenses/LICENSE-2.0"
+    importer_name = "Apache Tomcat Importer"
 
     def fetch_advisory_pages(self):
         """
         Yield the content of each HTML page containing version-related security data.
         """
         links = self.fetch_advisory_links("https://tomcat.apache.org/security")
         for page_url in links:
-            yield requests.get(page_url).content
+            yield page_url, requests.get(page_url).content
 
     def fetch_advisory_links(self, url):
         """
@@ -147,12 +148,12 @@ def advisory_data(self):
         """
         advisories = []
 
-        for advisory_page in self.fetch_advisory_pages():
-            advisories.extend(self.extract_advisories_from_page(advisory_page))
+        for url, advisory_page in self.fetch_advisory_pages():
+            advisories.extend(self.extract_advisories_from_page(url, advisory_page))
 
         return advisories
 
-    def extract_advisories_from_page(self, apache_tomcat_advisory_html):
+    def extract_advisories_from_page(self, url, apache_tomcat_advisory_html):
         """
         Yield AdvisoryData objects extracted from the HTML text ``apache_tomcat_advisory_html``.
         """
@@ -162,7 +163,7 @@ def extract_advisories_from_page(self, apache_tomcat_advisory_html):
         )
 
         for advisory_group in fixed_version_advisory_groups:
-            yield from generate_advisory_data_objects(advisory_group)
+            yield from generate_advisory_data_objects(url, advisory_group)
 
 
 @dataclasses.dataclass(order=True)
@@ -265,7 +266,7 @@ def extract_tomcat_advisory_data_from_page(apache_tomcat_advisory_html):
         )
 
 
-def generate_advisory_data_objects(tomcat_advisory_data_object):
+def generate_advisory_data_objects(url, tomcat_advisory_data_object):
     fixed_versions = tomcat_advisory_data_object.fixed_versions
     severity_scores = ("Low:", "Moderate:", "Important:", "High:", "Critical:")
 
@@ -361,6 +362,7 @@ def generate_advisory_data_objects(tomcat_advisory_data_object):
                 summary="",
                 affected_packages=affected_packages,
                 references=references,
+                url=url,
             )
 
 

diff --git a/vulnerabilities/importers/archlinux.py b/vulnerabilities/importers/archlinux.py
@@ -28,6 +28,7 @@ class ArchlinuxImporter(Importer):
     url = "https://security.archlinux.org/json"
     spdx_license_expression = "MIT"
     license_url = "https://github.com/archlinux/arch-security-tracker/blob/master/LICENSE"
+    importer_name = "Arch Linux Importer"
 
     def fetch(self) -> Iterable[Mapping]:
         response = fetch_response(self.url)
@@ -91,6 +92,7 @@ def parse_advisory(self, record) -> List[AdvisoryData]:
                     summary=summary,
                     affected_packages=affected_packages,
                     references=references,
+                    url=f"https://security.archlinux.org/{record['name']}",
                 )
             )
 

diff --git a/vulnerabilities/importers/debian.py b/vulnerabilities/importers/debian.py
@@ -77,6 +77,7 @@ class DebianImporter(Importer):
     """
 
     api_url = "https://security-tracker.debian.org/tracker/data/json"
+    importer_name = "Debian Importer"
 
     def get_response(self):
         response = requests.get(self.api_url)
@@ -154,4 +155,5 @@ def parse(self, pkg_name: str, records: Mapping[str, Any]) -> Iterable[AdvisoryD
                 summary=record.get("description", ""),
                 affected_packages=affected_packages,
                 references=references,
+                url=self.api_url,
             )
diff --git a/vulnerabilities/importers/debian_oval.py b/vulnerabilities/importers/debian_oval.py
@@ -54,6 +54,7 @@ class DebianOvalImporter(OvalImporter):
     Cheers,
         Moritz
     """
+    importer_name = "Debian Oval Importer"
 
     def __init__(self, *args, **kwargs):
         super().__init__(*args, **kwargs)
@@ -66,6 +67,7 @@ def _fetch(self):
         releases = ["wheezy", "stretch", "jessie", "buster", "bullseye"]
         for release in releases:
             file_url = f"https://www.debian.org/security/oval/oval-definitions-{release}.xml.bz2"
+            self.data_url = file_url
             resp = requests.get(file_url).content
             extracted = bz2.decompress(resp)
             yield (

diff --git a/vulnerabilities/importers/elixir_security.py b/vulnerabilities/importers/elixir_security.py
@@ -6,9 +6,11 @@
 # See https://github.com/nexB/vulnerablecode for support or download.
 # See https://aboutcode.org for more information about nexB OSS projects.
 #
+import urllib.parse as urlparse
 from pathlib import Path
 from typing import Set
 
+from dateutil import parser as dateparser
 from packageurl import PackageURL
 from univers.version_constraint import VersionConstraint
 from univers.version_range import HexVersionRange
@@ -26,21 +28,26 @@ class ElixirSecurityImporter(Importer):
     repo_url = "git+https://github.com/dependabot/elixir-security-advisories"
     license_url = "https://github.com/dependabot/elixir-security-advisories/blob/master/LICENSE.txt"
     spdx_license_expression = "CC0-1.0"
+    importer_name = "Elixir Security Importer"
 
     def advisory_data(self) -> Set[AdvisoryData]:
         try:
-            self.clone(repo_url=self.repo_url)
-            path = Path(self.vcs_response.dest_dir)
-            vuln = path / "packages"
+            self.clone(self.repo_url)
+            base_path = Path(self.vcs_response.dest_dir)
+            vuln = base_path / "packages"
             for file in vuln.glob("**/*.yml"):
-                yield from self.process_file(file)
+                yield from self.process_file(file, base_path)
         finally:
             if self.vcs_response:
                 self.vcs_response.delete()
 
-    def process_file(self, path):
-        path = str(path)
-        yaml_file = load_yaml(path)
+    def process_file(self, file, base_path):
+        relative_path = str(file.relative_to(base_path)).strip("/")
+        advisory_url = (
+            f"https://github.com/dependabot/elixir-security-advisories/blob/master/{relative_path}"
+        )
+        file = str(file)
+        yaml_file = load_yaml(file)
         cve_id = ""
         summary = yaml_file.get("description") or ""
         pkg_name = yaml_file.get("package") or ""
@@ -94,9 +101,15 @@ def process_file(self, path):
                 )
             )
 
+        date_published = None
+        if yaml_file.get("disclosure_date"):
+            date_published = dateparser.parse(yaml_file.get("disclosure_date"))
+
         yield AdvisoryData(
             aliases=[cve_id],
             summary=summary,
             references=references,
             affected_packages=affected_packages,
+            url=advisory_url,
+            date_published=date_published,
         )