feat: add repo finder and commit finder outcomes to database

Signed-off-by: Ben Selwyn-Smith <[email protected]>

docs/source/pages/developers_guide/apidoc/macaron.parsers.rst

@@ -40,3 +40,11 @@ macaron.parsers.github\_workflow\_model module
    :members:
    :undoc-members:
    :show-inheritance:
+
+macaron.parsers.pomparser module
+--------------------------------
+
+.. automodule:: macaron.parsers.pomparser
+   :members:
+   :undoc-members:
+   :show-inheritance:

docs/source/pages/developers_guide/apidoc/macaron.repo_finder.rst

@@ -57,6 +57,14 @@ macaron.repo\_finder.repo\_finder\_deps\_dev module
    :undoc-members:
    :show-inheritance:
 
+macaron.repo\_finder.repo\_finder\_enums module
+-----------------------------------------------
+
+.. automodule:: macaron.repo_finder.repo_finder_enums
+   :members:
+   :undoc-members:
+   :show-inheritance:
+
 macaron.repo\_finder.repo\_finder\_java module
 ----------------------------------------------
 

src/macaron/database/table_definitions.py

@@ -36,6 +36,7 @@
 from macaron.database.database_manager import ORMBase
 from macaron.database.db_custom_types import RFC3339DateTime
 from macaron.errors import InvalidPURLError
+from macaron.repo_finder.repo_finder_enums import CommitFinderOutcome, RepoFinderOutcome
 from macaron.slsa_analyzer.provenance.intoto import InTotoPayload, ProvenanceSubjectPURLMatcher
 from macaron.slsa_analyzer.slsa_req import ReqName
 
@@ -177,7 +178,14 @@ class Component(PackageURLMixin, ORMBase):
         lazy="immediate",
     )
 
-    def __init__(self, purl: str, analysis: Analysis, repository: "Repository | None"):
+    #: The one-to-one relationship with Repo Finder metadata.
+    repo_finder_metadata: Mapped["RepoFinderMetadata"] = relationship(
+        uselist=False, back_populates="component", lazy="immediate"
+    )
+
+    def __init__(
+        self, purl: str, analysis: Analysis, repository: "Repository | None", repo_finder_metadata: "RepoFinderMetadata"
+    ):
         """
         Instantiate the software component using PURL identifier.
 
@@ -204,7 +212,13 @@ def __init__(self, purl: str, analysis: Analysis, repository: "Repository | None
         # TODO: Explore the ``dbm`` or ``shelve`` packages to support dict type, which are part of the Python standard library.
         purl_kwargs = purl_parts.to_dict(encode=True)
 
-        super().__init__(purl=purl, analysis=analysis, repository=repository, **purl_kwargs)
+        super().__init__(
+            purl=purl,
+            analysis=analysis,
+            repository=repository,
+            repo_finder_metadata=repo_finder_metadata,
+            **purl_kwargs,
+        )
 
     @property
     def report_file_name(self) -> str:
@@ -605,3 +619,34 @@ def from_purl_and_provenance(
             return cls(sha256=sha256)
 
         return None
+
+
+class RepoFinderMetadata(ORMBase):
+    """Metadata from the Repo Finder and Commit Finder runs for an associated Component."""
+
+    __tablename__ = "_repo_finder_metadata"
+
+    #: The primary key.
+    id: Mapped[int] = mapped_column(Integer, primary_key=True, autoincrement=True)  # noqa: A003
+
+    #: The foreign key to the software component.
+    component_id: Mapped[int] = mapped_column(Integer, ForeignKey(Component.id), nullable=False)
+
+    #: A one-to-one relationship with software components.
+    component: Mapped["Component"] = relationship(back_populates="repo_finder_metadata")
+
+    #: The outcome of the Repo Finder.
+    repo_finder_outcome: Mapped[Enum] = mapped_column(
+        Enum(RepoFinderOutcome), nullable=False  # pylint: disable=protected-access,no-member
+    )
+
+    #: The outcome of the Commit Finder.
+    commit_finder_outcome: Mapped[Enum] = mapped_column(
+        Enum(CommitFinderOutcome), nullable=False  # pylint: disable=protected-access,no-member
+    )
+
+    #: The URL found by the Repo Finder (if applicable).
+    found_url: Mapped[str] = mapped_column(String)
+
+    #: The commit of the tag matched by the Commit Finder.
+    found_commit: Mapped[str] = mapped_column(String)

src/macaron/dependency_analyzer/cyclonedx.py

@@ -30,6 +30,7 @@
 from macaron.errors import CycloneDXParserError, DependencyAnalyzerError
 from macaron.output_reporter.scm import SCMStatus
 from macaron.repo_finder.repo_finder import find_repo
+from macaron.repo_finder.repo_finder_enums import RepoFinderOutcome
 from macaron.repo_finder.repo_validator import find_valid_repository_url
 
 logger: logging.Logger = logging.getLogger(__name__)
@@ -465,12 +466,12 @@ def _resolve_more_dependencies(dependencies: dict[str, DependencyInfo]) -> None:
         for item in dependencies.values():
             if item["available"] != SCMStatus.MISSING_SCM:
                 continue
-
-            item["url"] = find_repo(item["purl"])
-            if item["url"] == "":
+            url, outcome = find_repo(item["purl"])
+            if outcome not in {RepoFinderOutcome.FOUND, RepoFinderOutcome.FOUND_FROM_PARENT}:
                 logger.debug("Failed to find url for purl: %s", item["purl"])
             else:
                 # TODO decide how to handle possible duplicates here
+                item["url"] = url
                 item["available"] = SCMStatus.AVAILABLE
                 item["note"] = ""
 

src/macaron/repo_finder/commit_finder.py

@@ -14,6 +14,7 @@
 
 from macaron.repo_finder import repo_finder_deps_dev
 from macaron.repo_finder.repo_finder import to_domain_from_known_purl_types
+from macaron.repo_finder.repo_finder_enums import CommitFinderOutcome
 from macaron.slsa_analyzer.git_service import GIT_SERVICES
 
 logger: logging.Logger = logging.getLogger(__name__)
@@ -122,7 +123,7 @@ class AbstractPurlType(Enum):
     UNSUPPORTED = (2,)
 
 
-def find_commit(git_obj: Git, purl: PackageURL) -> str | None:
+def find_commit(git_obj: Git, purl: PackageURL) -> tuple[str | None, CommitFinderOutcome]:
     """Try to find the commit matching the passed PURL.
 
     The PURL may be a repository type, e.g. GitHub, in which case the commit might be in its version part.
@@ -138,21 +139,21 @@ def find_commit(git_obj: Git, purl: PackageURL) -> str | None:
 
     Returns
     -------
-    str | None
-        The digest, or None if the commit cannot be correctly retrieved.
+    tuple[str | None, CommitFinderOutcome]
+        The digest, or None if the commit cannot be correctly retrieved, and the outcome to report.
     """
     version = purl.version
     if not version:
         logger.debug("Missing version for analysis target: %s", purl.name)
-        return None
+        return None, CommitFinderOutcome.NO_VERSION_PROVIDED
 
     repo_type = determine_abstract_purl_type(purl)
     if repo_type == AbstractPurlType.REPOSITORY:
         return extract_commit_from_version(git_obj, version)
     if repo_type == AbstractPurlType.ARTIFACT:
         return find_commit_from_version_and_name(git_obj, purl.name, version)
     logger.debug("Type of PURL is not supported for commit finding: %s", purl.type)
-    return None
+    return None, CommitFinderOutcome.UNSUPPORTED_PURL_TYPE
 
 
 def determine_abstract_purl_type(purl: PackageURL) -> AbstractPurlType:
@@ -182,7 +183,7 @@ def determine_abstract_purl_type(purl: PackageURL) -> AbstractPurlType:
         return AbstractPurlType.UNSUPPORTED
 
 
-def extract_commit_from_version(git_obj: Git, version: str) -> str | None:
+def extract_commit_from_version(git_obj: Git, version: str) -> tuple[str | None, CommitFinderOutcome]:
     """Try to extract the commit from the PURL's version parameter.
 
     E.g.
@@ -198,8 +199,8 @@ def extract_commit_from_version(git_obj: Git, version: str) -> str | None:
 
     Returns
     -------
-    str | None
-        The digest, or None if the commit cannot be correctly retrieved.
+    tuple[str | None, CommitFinderOutcome]
+        The digest, or None if the commit cannot be correctly retrieved, and the outcome to report.
     """
     # A commit hash is 40 characters in length, but commits are often referenced using only some of those.
     commit: Commit | None = None
@@ -219,12 +220,12 @@ def extract_commit_from_version(git_obj: Git, version: str) -> str | None:
             logger.debug("Failed to retrieve commit: %s", error)
 
     if not commit:
-        return None
+        return None, CommitFinderOutcome.REPO_PURL_FAILURE
 
-    return commit.hash if commit else None
+    return commit.hash if commit else None, CommitFinderOutcome.MATCHED
 
 
-def find_commit_from_version_and_name(git_obj: Git, name: str, version: str) -> str | None:
+def find_commit_from_version_and_name(git_obj: Git, name: str, version: str) -> tuple[str | None, CommitFinderOutcome]:
     """Try to find the matching commit in a repository of a given version (and name) via tags.
 
     The passed version is used to match with the tags in the target repository. The passed name is used in cases where
@@ -241,14 +242,19 @@ def find_commit_from_version_and_name(git_obj: Git, name: str, version: str) ->
 
     Returns
     -------
-    str | None
-        The digest, or None if the commit cannot be correctly retrieved.
+    tuple[str | None, CommitFinderOutcome]
+        The digest, or None if the commit cannot be correctly retrieved, and the outcome to report.
     """
     logger.debug("Searching for commit of artifact version using tags: %s@%s", name, version)
 
     # Only consider tags that have a commit.
+    repo_tags = git_obj.repo.tags
+    if not repo_tags:
+        logger.debug("No tags found for %s", name)
+        return None, CommitFinderOutcome.NO_TAGS
+
     valid_tags = {}
-    for tag in git_obj.repo.tags:
+    for tag in repo_tags:
         commit = _get_tag_commit(tag)
         if not commit:
             logger.debug("No commit found for tag: %s", tag)
@@ -259,14 +265,14 @@ def find_commit_from_version_and_name(git_obj: Git, name: str, version: str) ->
 
     if not valid_tags:
         logger.debug("No tags with commits found for %s", name)
-        return None
+        return None, CommitFinderOutcome.NO_TAGS_WITH_COMMITS
 
     # Match tags.
-    matched_tags = match_tags(list(valid_tags.keys()), name, version)
+    matched_tags, outcome = match_tags(list(valid_tags.keys()), name, version)
 
     if not matched_tags:
         logger.debug("No tags matched for %s", name)
-        return None
+        return None, outcome
 
     if len(matched_tags) > 1:
         logger.debug("Tags found for %s: %s", name, len(matched_tags))
@@ -283,7 +289,7 @@ def find_commit_from_version_and_name(git_obj: Git, name: str, version: str) ->
         hexsha = tag.commit.hexsha
     except ValueError:
         logger.debug("Error trying to retrieve digest of commit: %s", tag.commit)
-        return None
+        return None, CommitFinderOutcome.NO_TAG_COMMIT
 
     logger.debug(
         "Found tag %s with commit %s for artifact version %s@%s",
@@ -292,7 +298,7 @@ def find_commit_from_version_and_name(git_obj: Git, name: str, version: str) ->
         name,
         version,
     )
-    return hexsha if hexsha else None
+    return hexsha if hexsha else None, CommitFinderOutcome.MATCHED
 
 
 def _split_name(name: str) -> list[str]:
@@ -350,7 +356,7 @@ def _split_separators(version: str) -> list[str]:
     return [item for item in split if item]
 
 
-def _build_version_pattern(name: str, version: str) -> tuple[Pattern | None, list[str]]:
+def _build_version_pattern(name: str, version: str) -> tuple[Pattern | None, list[str], CommitFinderOutcome]:
     """Build a version pattern to match the passed version string.
 
     Parameters
@@ -363,12 +369,12 @@ def _build_version_pattern(name: str, version: str) -> tuple[Pattern | None, lis
     Returns
     -------
     tuple[Pattern | None, list[str]]
-        The tuple of the regex pattern that will match the version, and the list of version parts that were extracted.
-        If an exception occurs from any regex operation, the pattern will be returned as None.
+        The tuple of the regex pattern that will match the version, the list of version parts that were extracted, and
+        the outcome to report. If an exception occurs from any regex operation, the pattern will be returned as None.
 
     """
     if not version:
-        return None, []
+        return None, [], CommitFinderOutcome.NO_VERSION_PROVIDED
 
     # Escape input to prevent it being treated as regex.
     name = re.escape(name)
@@ -377,7 +383,7 @@ def _build_version_pattern(name: str, version: str) -> tuple[Pattern | None, lis
 
     if not parts:
         logger.debug("Version contained no valid parts: %s", version)
-        return None, []
+        return None, [], CommitFinderOutcome.INVALID_PURL
 
     logger.debug("Final version parts: %s", parts)
 
@@ -471,14 +477,14 @@ def _build_version_pattern(name: str, version: str) -> tuple[Pattern | None, lis
 
     # Compile the pattern.
     try:
-        return re.compile(this_version_pattern, flags=re.IGNORECASE), parts
+        return re.compile(this_version_pattern, flags=re.IGNORECASE), parts, CommitFinderOutcome.MATCHED
     except Exception as error:  # pylint: disable=broad-exception-caught
         # The regex library uses an internal error that cannot be used here to satisfy pylint.
         logger.debug("Error while compiling version regex: %s", error)
-        return None, []
+        return None, [], CommitFinderOutcome.REGEX_COMPILE_FAILURE
 
 
-def match_tags(tag_list: list[str], name: str, version: str) -> list[str]:
+def match_tags(tag_list: list[str], name: str, version: str) -> tuple[list[str], CommitFinderOutcome]:
     """Return items of the passed tag list that match the passed artifact name and version.
 
     Parameters
@@ -492,8 +498,8 @@ def match_tags(tag_list: list[str], name: str, version: str) -> list[str]:
 
     Returns
     -------
-    list[str]
-        The list of tags that matched the pattern.
+    tuple[list[str], CommitFinderOutcome]
+        The list of tags that matched the pattern, if any, and the outcome to report.
     """
     logger.debug("Tag Sample: %s", tag_list[:5])
 
@@ -519,14 +525,14 @@ def match_tags(tag_list: list[str], name: str, version: str) -> list[str]:
             if match.group(1):
                 prefix_match = tag
     if prefix_match:
-        return [prefix_match]
+        return [prefix_match], CommitFinderOutcome.MATCHED
     if last_match:
-        return [last_match]
+        return [last_match], CommitFinderOutcome.MATCHED
 
     # Create the more complicated pattern for the passed version.
-    pattern, parts = _build_version_pattern(name, version)
+    pattern, parts, outcome = _build_version_pattern(name, version)
     if not pattern:
-        return []
+        return [], outcome
 
     # Match the tags.
     matched_tags = []
@@ -547,8 +553,12 @@ def match_tags(tag_list: list[str], name: str, version: str) -> list[str]:
 
     matched_tags = _fix_misaligned_tag_matches(matched_tags, version)
 
-    if len(matched_tags) <= 1:
-        return [_["tag"] for _ in matched_tags]
+    if not matched_tags:
+        logger.debug("Failed to match any tags.")
+        return [], CommitFinderOutcome.NO_TAGS_MATCHED
+
+    if len(matched_tags) == 1:
+        return [_["tag"] for _ in matched_tags], CommitFinderOutcome.MATCHED
 
     # In the case of multiple matches, further work must be done.
 
@@ -589,7 +599,7 @@ def match_tags(tag_list: list[str], name: str, version: str) -> list[str]:
             )
         )
 
-    return [_["tag"] for _ in matched_tags]
+    return [_["tag"] for _ in matched_tags], CommitFinderOutcome.MATCHED
 
 
 def _fix_misaligned_tag_matches(matched_tags: list[dict[str, str]], version: str) -> list[dict[str, str]]:

src/macaron/repo_finder/provenance_extractor.py

@@ -320,7 +320,7 @@ def check_if_input_purl_provenance_conflict(
 
     # Check the PURL commit against the provenance.
     if not digest_input and provenance_commit_digest and purl.version:
-        purl_commit = extract_commit_from_version(git_obj, purl.version)
+        purl_commit, _ = extract_commit_from_version(git_obj, purl.version)
         if purl_commit and purl_commit != provenance_commit_digest:
             logger.debug(
                 "The commit digest passed via purl input does not match what exists in the "