test(engine): add unit tests for commit metadata extraction

internal/engine/commitinfo/commitinfo.go

@@ -0,0 +1,48 @@
+// SPDX-FileCopyrightText: Copyright 2023 The Minder Authors
+// SPDX-License-Identifier: Apache-2.0
+
+// Package commitinfo provides utilities for extracting and normalizing
+// pull request commit metadata into a provider-agnostic structure.
+//
+// It is designed to decouple commit data handling from provider-specific
+// implementations (e.g., GitHub) and avoid import cycles between engine
+// components.
+package commitinfo
+
+import (
+	"strings"
+
+	"github.com/google/go-github/v63/github"
+)
+
+// CommitInfo represents normalized pull request commit metadata.
+type CommitInfo struct {
+	SHA     string
+	Message string
+	Author  string
+}
+
+// Extract normalizes a GitHub commit into a CommitInfo struct.
+func Extract(c *github.RepositoryCommit) CommitInfo {
+	msg := ""
+	if c.GetCommit() != nil {
+		msg = c.GetCommit().GetMessage()
+	}
+
+	firstLine := msg
+	if idx := strings.Index(msg, "\n"); idx != -1 {
+		firstLine = msg[:idx]
+	}
+	firstLine = strings.TrimSpace(firstLine)
+
+	author := ""
+	if c.GetCommit() != nil && c.GetCommit().Author != nil {
+		author = c.GetCommit().Author.GetName()
+	}
+
+	return CommitInfo{
+		SHA:     c.GetSHA(),
+		Message: firstLine,
+		Author:  author,
+	}
+}

internal/engine/commitinfo/commitinfo_test.go

@@ -0,0 +1,41 @@
+// SPDX-FileCopyrightText: Copyright 2023 The Minder Authors
+// SPDX-License-Identifier: Apache-2.0
+
+package commitinfo
+
+import (
+	"testing"
+
+	"github.com/google/go-github/v63/github"
+)
+
+func TestExtract_Basic(t *testing.T) {
+	t.Parallel()
+
+	msg := "feat: add new feature\n\nmore details"
+	author := "Sachin"
+
+	c := &github.RepositoryCommit{
+		SHA: github.String("abc123"),
+		Commit: &github.Commit{
+			Message: github.String(msg),
+			Author: &github.CommitAuthor{
+				Name: github.String(author),
+			},
+		},
+	}
+
+	info := Extract(c)
+
+	if info.SHA != "abc123" {
+		t.Fatalf("expected SHA abc123, got %s", info.SHA)
+	}
+
+	if info.Message != "feat: add new feature" {
+		t.Fatalf("unexpected message: %s", info.Message)
+	}
+
+	if info.Author != author {
+		t.Fatalf("expected author %s, got %s", author, info.Author)
+	}
+}

internal/engine/ingester/diff/diff.go

@@ -18,6 +18,7 @@ import (
 
 	"github.com/go-git/go-billy/v5"
 	"github.com/go-git/go-billy/v5/helper/iofs"
+	"github.com/google/go-github/v63/github"
 	scalibr "github.com/google/osv-scalibr"
 	"github.com/google/osv-scalibr/extractor"
 	scalibr_fs "github.com/google/osv-scalibr/fs"
@@ -27,6 +28,7 @@ import (
 	"github.com/rs/zerolog"
 	"google.golang.org/protobuf/reflect/protoreflect"
 
+	"github.com/mindersec/minder/internal/engine/commitinfo"
 	pbinternal "github.com/mindersec/minder/internal/proto"
 	pb "github.com/mindersec/minder/pkg/api/protobuf/go/minder/v1"
 	"github.com/mindersec/minder/pkg/engine/v1/interfaces"
@@ -37,9 +39,21 @@ const (
 	// DiffRuleDataIngestType is the type of the diff rule data ingest engine
 	DiffRuleDataIngestType = "diff"
 	prFilesPerPage         = 30
+	prCommitsPerPage       = 30
 	wildcard               = "*"
 )
 
+type commitLister interface {
+	ListPullRequestCommits(
+		ctx context.Context,
+		owner string,
+		repo string,
+		prNumber int,
+		perPage int,
+		pageNumber int,
+	) ([]*github.RepositoryCommit, *github.Response, error)
+}
+
 // Diff is the diff rule data ingest engine
 type Diff struct {
 	cli interfaces.GitHubListAndClone
@@ -92,6 +106,10 @@ func (di *Diff) Ingest(
 	}
 	prNumber := int(pr.Number)
 
+	if err := di.logPullRequestCommits(ctx, pr, prNumber); err != nil {
+		return nil, fmt.Errorf("error listing pull request commits: %w", err)
+	}
+
 	switch di.cfg.GetType() {
 	case "", pb.DiffTypeDep:
 		return di.getDepTypeDiff(ctx, prNumber, pr)
@@ -108,6 +126,46 @@ func (di *Diff) Ingest(
 	}
 }
 
+func (di *Diff) logPullRequestCommits(ctx context.Context, pr *pbinternal.PullRequest, prNumber int) error {
+	lister, ok := di.cli.(commitLister)
+	if !ok {
+		return nil
+	}
+
+	logger := zerolog.Ctx(ctx)
+	page := 0
+	for {
+		commits, resp, err := lister.ListPullRequestCommits(
+			ctx,
+			pr.RepoOwner,
+			pr.RepoName,
+			prNumber,
+			prCommitsPerPage,
+			page,
+		)
+		if err != nil {
+			return fmt.Errorf("failed to list pull request commits: %w", err)
+		}
+
+		for _, c := range commits {
+			info := commitinfo.Extract(c)
+			logger.Debug().
+				Str("commit_sha", info.SHA).
+				Str("commit_msg", info.Message).
+				Str("commit_author", info.Author).
+				Msg("pull request commit")
+		}
+
+		if resp == nil || resp.NextPage == 0 {
+			break
+		}
+
+		page = resp.NextPage
+	}
+
+	return nil
+}
+
 func (di *Diff) getDepTypeDiff(ctx context.Context, prNumber int, pr *pbinternal.PullRequest) (*interfaces.Ingested, error) {
 	deps := pbinternal.PrDependencies{Pr: pr}
 	page := 0

internal/providers/github/common.go

@@ -422,6 +422,47 @@ func (c *GitHub) ListFiles(
 	return resp.files, resp.resp, err
 }
 
+// ListPullRequestCommits is a wrapper for the GitHub API to list commits in a pull request.
+func (c *GitHub) ListPullRequestCommits(
+	ctx context.Context,
+	owner string,
+	repo string,
+	prNumber int,
+	perPage int,
+	pageNumber int,
+) ([]*github.RepositoryCommit, *github.Response, error) {
+	type listPullRequestCommitsRespWrapper struct {
+		commits []*github.RepositoryCommit
+		resp    *github.Response
+	}
+
+	op := func() (listPullRequestCommitsRespWrapper, error) {
+		opt := &github.ListOptions{
+			Page:    pageNumber,
+			PerPage: perPage,
+		}
+		commits, resp, err := c.client.PullRequests.ListCommits(ctx, owner, repo, prNumber, opt)
+
+		listCommitsResp := listPullRequestCommitsRespWrapper{
+			commits: commits,
+			resp:    resp,
+		}
+
+		if isRateLimitError(err) {
+			waitErr := c.waitForRateLimitReset(ctx, err)
+			if waitErr == nil {
+				return listCommitsResp, err
+			}
+			return listCommitsResp, backoffv4.Permanent(err)
+		}
+
+		return listCommitsResp, backoffv4.Permanent(err)
+	}
+
+	resp, err := performWithRetry(ctx, op)
+	return resp.commits, resp.resp, err
+}
+
 // CreateReview is a wrapper for the GitHub API to create a review
 func (c *GitHub) CreateReview(
 	ctx context.Context, owner, repo string, number int, reviewRequest *github.PullRequestReviewRequest,