Added handling for artifacts with classifiers and added debug mode which will list files for deletion

Author: [email protected]

custom/conf/app.example.ini

@@ -2633,6 +2633,8 @@ LEVEL = Info
 ;; Cleanup expired packages/data then targets the files within all maven snapshots versions 
 ;RETAIN_MAVEN_SNAPSHOT_BUILDS = -1
 ;; Maximum size of a npm upload (`-1` means no limits, format `1000`, `1 MB`, `1 GiB`)
+; Enable debug logging for Maven cleanup. Enabling debug will stop snapshot version artifacts from being deleted but will log the files which were meant for deletion.
+; DEBUG_MAVEN_CLEANUP = true
 ;LIMIT_SIZE_NPM = -1
 ;; Maximum size of a NuGet upload (`-1` means no limits, format `1000`, `1 MB`, `1 GiB`)
 ;LIMIT_SIZE_NUGET = -1

models/packages/package_file.go

@@ -12,7 +12,6 @@ import (
 	"time"
 
 	"code.gitea.io/gitea/models/db"
-	"code.gitea.io/gitea/modules/log"
 	"code.gitea.io/gitea/modules/timeutil"
 	"code.gitea.io/gitea/modules/util"
 
@@ -24,6 +23,8 @@ func init() {
 }
 
 var (
+	// ErrMetadataFile indicated a metadata file
+	ErrMetadataFile = errors.New("metadata file")
 	// ErrDuplicatePackageFile indicates a duplicated package file error
 	ErrDuplicatePackageFile = util.NewAlreadyExistErrorf("package file already exists")
 	// ErrPackageFileNotExist indicates a package file not exist error
@@ -229,57 +230,69 @@ func HasFiles(ctx context.Context, opts *PackageFileSearchOptions) (bool, error)
 	return db.Exist[PackageFile](ctx, opts.toConds())
 }
 
-// GetFilesByBuildNumber retrieves all files for a package version with build numbers <= maxBuildNumber.
-func GetFilesByBuildNumber(ctx context.Context, versionID int64, maxBuildNumber int) ([]*PackageFile, error) {
-	if maxBuildNumber < 0 {
-		return nil, errors.New("maxBuildNumber must be a non-negative integer")
+// GetFilesBelowBuildNumber retrieves all files for maven snapshot version where the build number is <= maxBuildNumber.
+// Returns two slices: one for filtered files and one for skipped files.
+func GetFilesBelowBuildNumber(ctx context.Context, versionID int64, maxBuildNumber int, classifiers ...string) ([]*PackageFile, []*PackageFile, error) {
+	if maxBuildNumber <= 0 {
+		return nil, nil, errors.New("maxBuildNumber must be a positive integer")
 	}
 
 	files, err := GetFilesByVersionID(ctx, versionID)
 	if err != nil {
-		return nil, fmt.Errorf("failed to retrieve files: %w", err)
+		return nil, nil, fmt.Errorf("failed to retrieve files: %w", err)
 	}
 
-	var filteredFiles []*PackageFile
+	var filteredFiles, skippedFiles []*PackageFile
 	for _, file := range files {
-		buildNumber, err := extractBuildNumberFromFileName(file.Name)
+		buildNumber, err := extractBuildNumberFromFileName(file.Name, classifiers...)
 		if err != nil {
-			if err.Error() == "metadata file" {
-				continue
+			if !errors.Is(err, ErrMetadataFile) {
+				skippedFiles = append(skippedFiles, file)
 			}
-			log.Warn("Failed to extract build number from file name '%s': %v", file.Name, err)
 			continue
 		}
-
 		if buildNumber <= maxBuildNumber {
 			filteredFiles = append(filteredFiles, file)
 		}
 	}
 
-	log.Info("Filtered %d files out of %d total files for version ID %d with maxBuildNumber %d", len(filteredFiles), len(files), versionID, maxBuildNumber)
-	return filteredFiles, nil
+	return filteredFiles, skippedFiles, nil
 }
 
-// extractBuildNumberFromFileName extracts the build number from the file name.
-func extractBuildNumberFromFileName(filename string) (int, error) {
-	// Skip metadata files
+// extractBuildNumberFromFileName extracts the build number from a Maven snapshot file name.
+// Expected formats:
+//
+//	"artifact-1.0.0-20250311.083409-9.tgz" returns 9
+//	"artifact-to-test-2.0.0-20250311.083409-10-sources.tgz" returns 10
+func extractBuildNumberFromFileName(filename string, classifiers ...string) (int, error) {
 	if strings.Contains(filename, "maven-metadata.xml") {
-		return 0, errors.New("metadata file")
+		return 0, ErrMetadataFile
 	}
 
-	// Split filename by hyphens to extract the build number
-	parts := strings.Split(filename, "-")
-	if len(parts) < 3 {
-		return 0, fmt.Errorf("invalid file name format: '%s'", filename)
+	dotIdx := strings.LastIndex(filename, ".")
+	if dotIdx == -1 {
+		return 0, fmt.Errorf("extract build number from filename: no file extension found in '%s'", filename)
+	}
+	base := filename[:dotIdx]
+
+	// Remove classifier suffix if present.
+	for _, classifier := range classifiers {
+		suffix := "-" + classifier
+		if strings.HasSuffix(base, suffix) {
+			base = base[:len(base)-len(suffix)]
+			break
+		}
 	}
 
-	// Extract the last part before the extension
-	buildNumberWithExt := parts[len(parts)-1]
-	buildNumberStr := strings.Split(buildNumberWithExt, ".")[0]
-
+	// The build number should be the token after the last dash.
+	lastDash := strings.LastIndex(base, "-")
+	if lastDash == -1 {
+		return 0, fmt.Errorf("extract build number from filename: invalid file name format in '%s'", filename)
+	}
+	buildNumberStr := base[lastDash+1:]
 	buildNumber, err := strconv.Atoi(buildNumberStr)
 	if err != nil {
-		return 0, fmt.Errorf("failed to convert maven package build number to integer: '%s'", buildNumberStr)
+		return 0, fmt.Errorf("extract build number from filename: failed to convert build number '%s' to integer in '%s': %v", buildNumberStr, filename, err)
 	}
 
 	return buildNumber, nil

modules/packages/maven/metadata.go

@@ -7,6 +7,7 @@ import (
 	"encoding/xml"
 	"errors"
 	"io"
+	"strconv"
 
 	"code.gitea.io/gitea/modules/util"
 	"code.gitea.io/gitea/modules/validation"
@@ -32,6 +33,12 @@ type Dependency struct {
 	Version    string `json:"version,omitempty"`
 }
 
+// SnapshotMetadata struct holds the build number and the list of classifiers for a snapshot version
+type SnapshotMetadata struct {
+	BuildNumber int      `json:"build_number,omitempty"`
+	Classifiers []string `json:"classifiers,omitempty"`
+}
+
 type pomStruct struct {
 	XMLName xml.Name `xml:"project"`
 
@@ -62,7 +69,7 @@ type pomStruct struct {
 	} `xml:"dependencies>dependency"`
 }
 
-type MavenMetadata struct {
+type snapshotMetadataStruct struct {
 	XMLName    xml.Name `xml:"metadata"`
 	GroupID    string   `xml:"groupId"`
 	ArtifactID string   `xml:"artifactId"`
@@ -74,11 +81,10 @@ type MavenMetadata struct {
 			BuildNumber string `xml:"buildNumber"`
 		} `xml:"snapshot"`
 		SnapshotVersions []struct {
-			SnapshotVersion struct {
-				Extension string `xml:"extension"`
-				Value     string `xml:"value"`
-				Updated   string `xml:"updated"`
-			} `xml:"snapshotVersion"`
+			Extension  string `xml:"extension"`
+			Classifier string `xml:"classifier"`
+			Value      string `xml:"value"`
+			Updated    string `xml:"updated"`
 		} `xml:"snapshotVersions>snapshotVersion"`
 	} `xml:"versioning"`
 }
@@ -132,19 +138,30 @@ func ParsePackageMetaData(r io.Reader) (*Metadata, error) {
 	}, nil
 }
 
-// ParseMavenMetadata parses the Maven metadata XML to extract the build number.
-func ParseMavenMetaData(r io.Reader) (string, error) {
-	var metadata MavenMetadata
+// ParseSnapshotVersionMetadata parses the Maven Snapshot Version metadata to extract the build number and list of available classifiers.
+func ParseSnapshotVersionMetaData(r io.Reader) (*SnapshotMetadata, error) {
+	var metadata snapshotMetadataStruct
 
 	dec := xml.NewDecoder(r)
-	dec.CharsetReader = charset.NewReaderLabel // Assuming charset.NewReaderLabel is a function you've set up to handle character encoding.
+	dec.CharsetReader = charset.NewReaderLabel
 	if err := dec.Decode(&metadata); err != nil {
-		return "", err
+		return nil, err
 	}
 
-	if metadata.Versioning.Snapshot.BuildNumber == "" {
-		return "", errors.New("no build number in snapshot metadata found")
+	buildNumber, err := strconv.Atoi(metadata.Versioning.Snapshot.BuildNumber)
+	if err != nil {
+		return nil, errors.New("invalid or missing build number in snapshot metadata")
 	}
 
-	return metadata.Versioning.Snapshot.BuildNumber, nil
+	var classifiers []string
+	for _, snapshotVersion := range metadata.Versioning.SnapshotVersions {
+		if snapshotVersion.Classifier != "" {
+			classifiers = append(classifiers, snapshotVersion.Classifier)
+		}
+	}
+
+	return &SnapshotMetadata{
+		BuildNumber: buildNumber,
+		Classifiers: classifiers,
+	}, nil
 }

modules/setting/packages.go

@@ -43,6 +43,7 @@ var (
 
 		DefaultRPMSignEnabled     bool
 		RetainMavenSnapshotBuilds int
+		DebugMavenCleanup         bool
 	}{
 		Enabled:                   true,
 		LimitTotalOwnerCount:      -1,
@@ -91,6 +92,7 @@ func loadPackagesFrom(rootCfg ConfigProvider) (err error) {
 	Packages.LimitSizeVagrant = mustBytes(sec, "LIMIT_SIZE_VAGRANT")
 	Packages.DefaultRPMSignEnabled = sec.Key("DEFAULT_RPM_SIGN_ENABLED").MustBool(false)
 	Packages.RetainMavenSnapshotBuilds = sec.Key("RETAIN_MAVEN_SNAPSHOT_BUILDS").MustInt(Packages.RetainMavenSnapshotBuilds)
+	Packages.DebugMavenCleanup = sec.Key("DEBUG_MAVEN_CLEANUP").MustBool(true)
 	return nil
 }
 

services/packages/maven/cleanup.go

@@ -3,7 +3,6 @@ package maven
 import (
 	"context"
 	"fmt"
-	"strconv"
 	"strings"
 
 	"code.gitea.io/gitea/models/packages"
@@ -13,105 +12,122 @@ import (
 	packages_service "code.gitea.io/gitea/services/packages"
 )
 
-// CleanupSnapshotVersion removes outdated files for SNAPHOT versions for all Maven packages.
+// CleanupSnapshotVersions removes outdated files for SNAPHOT versions for all Maven packages.
 func CleanupSnapshotVersions(ctx context.Context) error {
 	retainBuilds := setting.Packages.RetainMavenSnapshotBuilds
-	log.Info("Starting CleanupSnapshotVersion with retainBuilds: %d", retainBuilds)
+	debugSession := setting.Packages.DebugMavenCleanup
+	log.Debug("Starting Maven CleanupSnapshotVersions with retainBuilds: %d, debugSession: %t", retainBuilds, debugSession)
 
 	if retainBuilds == -1 {
-		log.Info("CleanupSnapshotVersion skipped because retainBuilds is set to -1")
+		log.Info("Maven CleanupSnapshotVersions skipped because retainBuilds is set to -1")
 		return nil
 	}
 
 	if retainBuilds < 1 {
-		return fmt.Errorf("forbidden value for retainBuilds: %d. Minimum 1 build should be retained", retainBuilds)
+		return fmt.Errorf("Maven CleanupSnapshotVersions: forbidden value for retainBuilds: %d. Minimum 1 build should be retained", retainBuilds)
 	}
 
 	versions, err := packages.GetVersionsByPackageType(ctx, 0, packages.TypeMaven)
 	if err != nil {
-		return fmt.Errorf("failed to retrieve Maven package versions: %w", err)
+		return fmt.Errorf("Maven CleanupSnapshotVersions: failed to retrieve Maven package versions: %w", err)
 	}
 
-	for _, version := range versions {
-		log.Info("Processing version: %s (ID: %d)", version.Version, version.ID)
+	var errors []error
 
+	for _, version := range versions {
 		if !isSnapshotVersion(version.Version) {
-			log.Info("Skipping non-SNAPSHOT version: %s (ID: %d)", version.Version, version.ID)
 			continue
 		}
 
-		if err := cleanSnapshotFiles(ctx, version.ID, retainBuilds); err != nil {
-			log.Error("Failed to clean up snapshot files for version '%s' (ID: %d): %v", version.Version, version.ID, err)
-			return err
+		if err := cleanSnapshotFiles(ctx, version.ID, retainBuilds, debugSession); err != nil {
+			errors = append(errors, fmt.Errorf("Maven CleanupSnapshotVersions: version '%s' (ID: %d): %w", version.Version, version.ID, err))
+		}
+	}
+
+	if len(errors) > 0 {
+		for _, err := range errors {
+			log.Warn("Maven CleanupSnapshotVersions: Error during cleanup: %v", err)
 		}
+		return fmt.Errorf("Maven CleanupSnapshotVersions: cleanup completed with errors: %v", errors)
 	}
 
-	log.Info("Completed CleanupSnapshotVersion")
+	log.Debug("Completed Maven CleanupSnapshotVersions")
 	return nil
 }
 
 func isSnapshotVersion(version string) bool {
 	return strings.HasSuffix(version, "-SNAPSHOT")
 }
 
-func cleanSnapshotFiles(ctx context.Context, versionID int64, retainBuilds int) error {
-	log.Info("Starting cleanSnapshotFiles for versionID: %d with retainBuilds: %d", versionID, retainBuilds)
+func cleanSnapshotFiles(ctx context.Context, versionID int64, retainBuilds int, debugSession bool) error {
+	log.Debug("Starting Maven cleanSnapshotFiles for versionID: %d with retainBuilds: %d, debugSession: %t", versionID, retainBuilds, debugSession)
 
 	metadataFile, err := packages.GetFileForVersionByName(ctx, versionID, "maven-metadata.xml", packages.EmptyFileKey)
 	if err != nil {
-		return fmt.Errorf("failed to retrieve Maven metadata file for version ID %d: %w", versionID, err)
+		return fmt.Errorf("cleanSnapshotFiles: failed to retrieve Maven metadata file for version ID %d: %w", versionID, err)
 	}
 
-	maxBuildNumber, err := extractMaxBuildNumberFromMetadata(ctx, metadataFile)
+	maxBuildNumber, classifiers, err := extractMaxBuildNumber(ctx, metadataFile)
 	if err != nil {
-		return fmt.Errorf("failed to extract max build number from maven-metadata.xml for version ID %d: %w", versionID, err)
+		return fmt.Errorf("cleanSnapshotFiles: failed to extract max build number from maven-metadata.xml for version ID %d: %w", versionID, err)
 	}
 
-	log.Info("Max build number for versionID %d: %d", versionID, maxBuildNumber)
-
 	thresholdBuildNumber := maxBuildNumber - retainBuilds
 	if thresholdBuildNumber <= 0 {
-		log.Info("No files to clean up, as the threshold build number is less than or equal to zero for versionID %d", versionID)
+		log.Debug("cleanSnapshotFiles: No files to clean up, as the threshold build number is less than or equal to zero for versionID %d", versionID)
 		return nil
 	}
 
-	filesToRemove, err := packages.GetFilesByBuildNumber(ctx, versionID, thresholdBuildNumber)
+	filesToRemove, skippedFiles, err := packages.GetFilesBelowBuildNumber(ctx, versionID, thresholdBuildNumber, classifiers...)
 	if err != nil {
-		return fmt.Errorf("failed to retrieve files for version ID %d: %w", versionID, err)
+		return fmt.Errorf("cleanSnapshotFiles: failed to retrieve files for version ID %d: %w", versionID, err)
+	}
+
+	if debugSession {
+		var fileNamesToRemove, skippedFileNames []string
+
+		for _, file := range filesToRemove {
+			fileNamesToRemove = append(fileNamesToRemove, file.Name)
+		}
+
+		for _, file := range skippedFiles {
+			skippedFileNames = append(skippedFileNames, file.Name)
+		}
+
+		log.Info("cleanSnapshotFiles: Debug session active. Files to remove: %v, Skipped files: %v", fileNamesToRemove, skippedFileNames)
+		return nil
 	}
 
 	for _, file := range filesToRemove {
 		log.Debug("Removing file '%s' below threshold %d", file.Name, thresholdBuildNumber)
 		if err := packages_service.DeletePackageFile(ctx, file); err != nil {
-			return fmt.Errorf("failed to delete file '%s': %w", file.Name, err)
+			return fmt.Errorf("Maven cleanSnapshotFiles: failed to delete file '%s': %w", file.Name, err)
 		}
 	}
 
-	log.Info("Completed cleanSnapshotFiles for versionID: %d", versionID)
+	log.Debug("Completed Maven cleanSnapshotFiles for versionID: %d", versionID)
 	return nil
 }
 
-func extractMaxBuildNumberFromMetadata(ctx context.Context, metadataFile *packages.PackageFile) (int, error) {
+func extractMaxBuildNumber(ctx context.Context, metadataFile *packages.PackageFile) (int, []string, error) {
 	pb, err := packages.GetBlobByID(ctx, metadataFile.BlobID)
 	if err != nil {
-		return 0, fmt.Errorf("failed to get package blob: %w", err)
+		return 0, nil, fmt.Errorf("extractMaxBuildNumber: failed to get package blob: %w", err)
 	}
 
 	content, _, _, err := packages_service.GetPackageBlobStream(ctx, metadataFile, pb, nil, true)
 	if err != nil {
-		return 0, fmt.Errorf("failed to get package file stream: %w", err)
+		return 0, nil, fmt.Errorf("extractMaxBuildNumber: failed to get package file stream: %w", err)
 	}
 	defer content.Close()
 
-	buildNumberStr, err := maven.ParseMavenMetaData(content)
+	snapshotMetadata, err := maven.ParseSnapshotVersionMetaData(content)
 	if err != nil {
-		return 0, fmt.Errorf("failed to parse maven-metadata.xml: %w", err)
+		return 0, nil, fmt.Errorf("extractMaxBuildNumber: failed to parse maven-metadata.xml: %w", err)
 	}
 
-	buildNumber, err := strconv.Atoi(buildNumberStr)
-	if err != nil {
-		return 0, fmt.Errorf("invalid build number format: %w", err)
-	}
+	buildNumber := snapshotMetadata.BuildNumber
+	classifiers := snapshotMetadata.Classifiers
 
-	return buildNumber, nil
+	return buildNumber, classifiers, nil
 }