Merge bed592c075 into a440116a16

custom/conf/app.example.ini

@@ -2676,7 +2676,12 @@ LEVEL = Info
 ;LIMIT_SIZE_HELM = -1
 ;; Maximum size of a Maven upload (`-1` means no limits, format `1000`, `1 MB`, `1 GiB`)
 ;LIMIT_SIZE_MAVEN = -1
+;; Specifies the number of most recent Maven snapshot builds to retain. `-1` retains all builds, while `1` retains only the latest build. Value should be -1 or positive.
+;; 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/fixtures/package.yml

@@ -0,0 +1,6 @@
+-
+  id: 1
+  owner_id: 2
+  type: maven
+  name: com.gitea:test-project
+  lower_name: com.gitea:test-project

models/fixtures/package_blob.yml

@@ -0,0 +1,88 @@
+-
+  id: 1
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b1
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c01
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a01
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a801
+  created_unix: 1672531200
+-
+  id: 2
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b2
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c02
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a02
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a802
+  created_unix: 1672531200
+-
+  id: 3
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b3
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c03
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a03
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a803
+  created_unix: 1672531200
+-
+  id: 4
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b4
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c04
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a04
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a804
+  created_unix: 1672531200
+-
+  id: 5
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b5
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c05
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a05
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a805
+  created_unix: 1672531200
+-
+  id: 6
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b6
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c06
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a06
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a806
+  created_unix: 1672531200
+-
+  id: 7
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b7
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c07
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a07
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a807
+  created_unix: 1672531200
+-
+  id: 8
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b8
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c08
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a08
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a808
+  created_unix: 1672531200
+-
+  id: 9
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0b9
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c09
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a09
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a809
+  created_unix: 1672531200
+-
+  id: 10
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0ba
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c10
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a10
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a810
+  created_unix: 1672531200
+-
+  id: 11
+  size: 14
+  hash_md5: 27224a672372115e5a1d125ed7b2a0bb
+  hash_sha1: 9854582a2958b2d31541ce4a6a1a36201b80c11
+  hash_sha256: 9f86d081884c7d659a2feaa0c55ad015a3bf4f1b2b0b822cd15d6c15b0f00a11
+  hash_sha512: ee26b0dd4af7e749aa1a8ee3c10ae9923f618980772e473f8819a5d4940e0db27ac185f8a0e1d5f84f88bc887fd67b143732c304cc5fa9ad8e6f57f50028a811
+  created_unix: 1672531200

models/fixtures/package_file.yml

@@ -0,0 +1,99 @@
+-
+  id: 1
+  version_id: 1
+  blob_id: 1
+  name: gitea-test-1.0-20230101.000000-1.jar
+  lower_name: gitea-test-1.0-20230101.000000-1.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 2
+  version_id: 1
+  blob_id: 2
+  name: gitea-test-1.0-20230101.000000-2.jar
+  lower_name: gitea-test-1.0-20230101.000000-2.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 3
+  version_id: 1
+  blob_id: 3
+  name: gitea-test-1.0-20230101.000000-3.jar
+  lower_name: gitea-test-1.0-20230101.000000-3.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 4
+  version_id: 1
+  blob_id: 4
+  name: gitea-test-1.0-20230101.000000-4.jar
+  lower_name: gitea-test-1.0-20230101.000000-4.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 5
+  version_id: 1
+  blob_id: 5
+  name: gitea-test-1.0-20230101.000000-5.jar
+  lower_name: gitea-test-1.0-20230101.000000-5.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 6
+  version_id: 1
+  blob_id: 6
+  name: gitea-test-1.0-20230101.000000-3-sources.jar
+  lower_name: gitea-test-1.0-20230101.000000-3-sources.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 7
+  version_id: 1
+  blob_id: 7
+  name: gitea-test-1.0-20230101.000000-3-javadoc.jar
+  lower_name: gitea-test-1.0-20230101.000000-3-javadoc.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 8
+  version_id: 1
+  blob_id: 8
+  name: gitea-test-1.0-20230101.000000-4-sources.jar
+  lower_name: gitea-test-1.0-20230101.000000-4-sources.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 9
+  version_id: 1
+  blob_id: 9
+  name: gitea-test-1.0-20230101.000000-4-javadoc.jar
+  lower_name: gitea-test-1.0-20230101.000000-4-javadoc.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 10
+  version_id: 1
+  blob_id: 10
+  name: gitea-test-1.0-20230101.000000-5-sources.jar
+  lower_name: gitea-test-1.0-20230101.000000-5-sources.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200
+-
+  id: 11
+  version_id: 1
+  blob_id: 11
+  name: gitea-test-1.0-20230101.000000-5-javadoc.jar
+  lower_name: gitea-test-1.0-20230101.000000-5-javadoc.jar
+  composite_key: ""
+  is_lead: false
+  created_unix: 1672531200

models/fixtures/package_version.yml

@@ -0,0 +1,8 @@
+-
+  id: 1
+  package_id: 1
+  creator_id: 2
+  version: 1.0-SNAPSHOT
+  lower_version: 1.0-snapshot
+  metadata_json: '{"artifact_id":"test-project","group_id":"com.gitea"}'
+  created_unix: 1672531200

models/packages/package_file.go

@@ -5,6 +5,9 @@ package packages
 
 import (
 	"context"
+	"errors"
+	"fmt"
+	"sort"
 	"strconv"
 	"strings"
 	"time"
@@ -21,6 +24,8 @@ 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
@@ -231,6 +236,80 @@ func HasFiles(ctx context.Context, opts *PackageFileSearchOptions) (bool, error)
 	return db.Exist[PackageFile](ctx, opts.toConds())
 }
 
+// GetFilesBelowBuildNumber retrieves all files for a 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, nil, fmt.Errorf("failed to retrieve files: %w", err)
+	}
+
+	// Sort classifiers by length (longest first) once per call
+	sort.SliceStable(classifiers, func(i, j int) bool {
+		return len(classifiers[i]) > len(classifiers[j])
+	})
+
+	var filteredFiles, skippedFiles []*PackageFile
+	for _, file := range files {
+		buildNumber, err := extractBuildNumberFromFileName(file.Name, classifiers...)
+		if err != nil {
+			if !errors.Is(err, ErrMetadataFile) {
+				skippedFiles = append(skippedFiles, file)
+			}
+			continue
+		}
+
+		if buildNumber <= maxBuildNumber {
+			filteredFiles = append(filteredFiles, file)
+		}
+	}
+
+	return filteredFiles, skippedFiles, nil
+}
+
+// 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, ErrMetadataFile
+	}
+
+	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
+		}
+	}
+
+	// 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("extract build number from filename: failed to convert build number '%s' to integer in '%s': %v", buildNumberStr, filename, err)
+	}
+
+	return buildNumber, nil
+}
+
 // CalculateFileSize sums up all blob sizes matching the search options.
 // It does NOT respect the deduplication of blobs.
 func CalculateFileSize(ctx context.Context, opts *PackageFileSearchOptions) (int64, error) {

models/packages/package_version.go

@@ -129,11 +129,16 @@ func getVersionByNameAndVersion(ctx context.Context, ownerID int64, packageType
 
 // GetVersionsByPackageType gets all versions of a specific type
 func GetVersionsByPackageType(ctx context.Context, ownerID int64, packageType Type) ([]*PackageVersion, error) {
-	pvs, _, err := SearchVersions(ctx, &PackageSearchOptions{
-		OwnerID:    ownerID,
+	opts := &PackageSearchOptions{
 		Type:       packageType,
 		IsInternal: optional.Some(false),
-	})
+	}
+
+	if ownerID != 0 {
+		opts.OwnerID = ownerID
+	}
+
+	pvs, _, err := SearchVersions(ctx, opts)
 	return pvs, err
 }
 

modules/packages/maven/metadata.go

@@ -5,7 +5,9 @@ package maven
 
 import (
 	"encoding/xml"
+	"errors"
 	"io"
+	"strconv"
 
 	"code.gitea.io/gitea/modules/util"
 	"code.gitea.io/gitea/modules/validation"
@@ -31,6 +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"`
 
@@ -61,6 +69,26 @@ type pomStruct struct {
 	} `xml:"dependencies>dependency"`
 }
 
+type snapshotMetadataStruct struct {
+	XMLName    xml.Name `xml:"metadata"`
+	GroupID    string   `xml:"groupId"`
+	ArtifactID string   `xml:"artifactId"`
+	Version    string   `xml:"version"`
+	Versioning struct {
+		LastUpdated string `xml:"lastUpdated"`
+		Snapshot    struct {
+			Timestamp   string `xml:"timestamp"`
+			BuildNumber string `xml:"buildNumber"`
+		} `xml:"snapshot"`
+		SnapshotVersions []struct {
+			Extension  string `xml:"extension"`
+			Classifier string `xml:"classifier"`
+			Value      string `xml:"value"`
+			Updated    string `xml:"updated"`
+		} `xml:"snapshotVersions>snapshotVersion"`
+	} `xml:"versioning"`
+}
+
 // ParsePackageMetaData parses the metadata of a pom file
 func ParsePackageMetaData(r io.Reader) (*Metadata, error) {
 	var pom pomStruct
@@ -109,3 +137,31 @@ func ParsePackageMetaData(r io.Reader) (*Metadata, error) {
 		Dependencies: dependencies,
 	}, nil
 }
+
+// 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
+	if err := dec.Decode(&metadata); err != nil {
+		return nil, err
+	}
+
+	buildNumber, err := strconv.Atoi(metadata.Versioning.Snapshot.BuildNumber)
+	if err != nil {
+		return nil, errors.New("invalid or missing build number in snapshot metadata")
+	}
+
+	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

@@ -41,10 +41,13 @@ var (
 		LimitSizeSwift       int64
 		LimitSizeVagrant     int64
 
-		DefaultRPMSignEnabled bool
+		DefaultRPMSignEnabled     bool
+		RetainMavenSnapshotBuilds int
+		DebugMavenCleanup         bool
 	}{
-		Enabled:              true,
-		LimitTotalOwnerCount: -1,
+		Enabled:                   true,
+		LimitTotalOwnerCount:      -1,
+		RetainMavenSnapshotBuilds: -1,
 	}
 )
 
@@ -88,6 +91,8 @@ func loadPackagesFrom(rootCfg ConfigProvider) (err error) {
 	Packages.LimitSizeSwift = mustBytes(sec, "LIMIT_SIZE_SWIFT")
 	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/cleanup/cleanup.go

@@ -1,7 +1,7 @@
 // Copyright 2022 The Gitea Authors. All rights reserved.
 // SPDX-License-Identifier: MIT
 
-package container
+package cleanup
 
 import (
 	"context"
@@ -20,6 +20,7 @@ import (
 	cargo_service "code.gitea.io/gitea/services/packages/cargo"
 	container_service "code.gitea.io/gitea/services/packages/container"
 	debian_service "code.gitea.io/gitea/services/packages/debian"
+	maven_service "code.gitea.io/gitea/services/packages/maven"
 	rpm_service "code.gitea.io/gitea/services/packages/rpm"
 )
 
@@ -171,6 +172,10 @@ func CleanupExpiredData(ctx context.Context, olderThan time.Duration) error {
 			return err
 		}
 
+		if err := maven_service.CleanupSnapshotVersions(ctx); err != nil {
+			log.Error("Error during maven snapshot versions cleanup: %v", err)
+		}
+
 		ps, err := packages_model.FindUnreferencedPackages(ctx)
 		if err != nil {
 			return err

services/packages/maven/cleanup.go

@@ -0,0 +1,157 @@
+// Copyright 2025 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package maven
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"strings"
+
+	"code.gitea.io/gitea/models/packages"
+	"code.gitea.io/gitea/modules/json"
+	"code.gitea.io/gitea/modules/log"
+	"code.gitea.io/gitea/modules/packages/maven"
+	"code.gitea.io/gitea/modules/setting"
+	packages_service "code.gitea.io/gitea/services/packages"
+)
+
+// CleanupSnapshotVersions removes outdated files for SNAPHOT versions for all Maven packages.
+func CleanupSnapshotVersions(ctx context.Context) error {
+	retainBuilds := setting.Packages.RetainMavenSnapshotBuilds
+	debugSession := setting.Packages.DebugMavenCleanup
+	log.Debug("Maven Cleanup: starting with retainBuilds: %d, debugSession: %t", retainBuilds, debugSession)
+
+	if retainBuilds < 1 {
+		log.Warn("Maven Cleanup: skipped as value for retainBuilds less than 1: %d. Minimum 1 build should be retained", retainBuilds)
+		return nil
+	}
+
+	versions, err := packages.GetVersionsByPackageType(ctx, 0, packages.TypeMaven)
+	if err != nil {
+		return fmt.Errorf("maven Cleanup: failed to retrieve Maven package versions: %w", err)
+	}
+
+	var errs []error
+	var metadataErrors []error
+
+	for _, version := range versions {
+		if !isSnapshotVersion(version.Version) {
+			continue
+		}
+
+		var artifactID, groupID string
+		if version.MetadataJSON != "" {
+			var metadata map[string]any
+			if err := json.Unmarshal([]byte(version.MetadataJSON), &metadata); err != nil {
+				log.Warn("Maven Cleanup: error during cleanup: failed to unmarshal metadataJSON for package version ID: %d: %w", version.ID, err)
+			} else {
+				artifactID, _ = metadata["artifact_id"].(string)
+				groupID, _ = metadata["group_id"].(string)
+				log.Debug("Maven Cleanup: processing package version with ID: %s, Group ID: %s, Artifact ID: %s, Version: %s", version.ID, groupID, artifactID, version.Version)
+			}
+		}
+
+		if err := cleanSnapshotFiles(ctx, version.ID, retainBuilds, debugSession); err != nil {
+			formattedErr := fmt.Errorf("version '%s' (ID: %d, Group ID: %s, Artifact ID: %s): %w",
+				version.Version, version.ID, groupID, artifactID, err)
+
+			if errors.Is(err, packages.ErrMetadataFile) {
+				metadataErrors = append(metadataErrors, formattedErr)
+			} else {
+				errs = append(errs, formattedErr)
+			}
+		}
+	}
+
+	for _, err := range metadataErrors {
+		log.Warn("Maven Cleanup: error during cleanup: %v", err)
+	}
+
+	if len(errs) > 0 {
+		for _, err := range errs {
+			log.Error("Maven Cleanup: error during cleanup: %v", err)
+		}
+		return fmt.Errorf("maven Cleanup: completed with errors: %v", errs)
+	}
+
+	log.Debug("Completed Maven Cleanup")
+	return nil
+}
+
+func isSnapshotVersion(version string) bool {
+	return strings.HasSuffix(version, "-SNAPSHOT")
+}
+
+func cleanSnapshotFiles(ctx context.Context, versionID int64, retainBuilds int, debugSession bool) error {
+	log.Debug("Maven Cleanup: starting 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("%w: failed to retrieve maven-metadata.xml: %w", packages.ErrMetadataFile, err)
+	}
+
+	maxBuildNumber, classifiers, err := extractMaxBuildNumber(ctx, metadataFile)
+	if err != nil {
+		return fmt.Errorf("%w: failed to extract max build number from maven-metadata.xml: %w", packages.ErrMetadataFile, err)
+	}
+
+	thresholdBuildNumber := maxBuildNumber - retainBuilds
+	if thresholdBuildNumber <= 0 {
+		log.Debug("Maven Cleanup: no files to clean up, as the threshold build number is less than or equal to zero for versionID %d", versionID)
+		return nil
+	}
+
+	filesToRemove, skippedFiles, err := packages.GetFilesBelowBuildNumber(ctx, versionID, thresholdBuildNumber, classifiers...)
+	if err != nil {
+		return fmt.Errorf("cleanSnapshotFiles: failed to retrieve files for version: %w", 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.Debug("Maven Cleanup: debug session active. Files to remove: %v, Skipped files: %v", fileNamesToRemove, skippedFileNames)
+		return nil
+	}
+
+	for _, file := range filesToRemove {
+		log.Debug("Maven Cleanup: removing file '%s' below threshold %d", file.Name, thresholdBuildNumber)
+		if err := packages_service.DeletePackageFile(ctx, file); err != nil {
+			return fmt.Errorf("cleanSnapshotFiles: failed to delete file '%s': %w", file.Name, err)
+		}
+	}
+
+	return nil
+}
+
+func extractMaxBuildNumber(ctx context.Context, metadataFile *packages.PackageFile) (int, []string, error) {
+	pb, err := packages.GetBlobByID(ctx, metadataFile.BlobID)
+	if err != nil {
+		return 0, nil, fmt.Errorf("failed to get package blob: %w", err)
+	}
+
+	content, _, _, err := packages_service.OpenBlobForDownload(ctx, metadataFile, pb, "", nil, true)
+	if err != nil {
+		return 0, nil, fmt.Errorf("failed to get package file stream: %w", err)
+	}
+	defer content.Close()
+
+	snapshotMetadata, err := maven.ParseSnapshotVersionMetaData(content)
+	if err != nil {
+		return 0, nil, fmt.Errorf("failed to parse maven-metadata.xml: %w", err)
+	}
+
+	buildNumber := snapshotMetadata.BuildNumber
+	classifiers := snapshotMetadata.Classifiers
+
+	return buildNumber, classifiers, nil
+}

services/packages/maven/cleanup_test.go

@@ -0,0 +1,250 @@
+// Copyright 2023 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package maven
+
+import (
+	"bytes"
+	"testing"
+
+	"code.gitea.io/gitea/models/packages"
+	"code.gitea.io/gitea/models/unittest"
+	packages_module "code.gitea.io/gitea/modules/packages"
+	"code.gitea.io/gitea/modules/setting"
+	packages_service "code.gitea.io/gitea/services/packages"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestMain(m *testing.M) {
+	unittest.MainTest(m)
+}
+
+func addMavenMetadataToPackageVersion(t *testing.T, pv *packages.PackageVersion) {
+	// Create maven-metadata.xml content with build number 5 (matching the fixtures)
+	// Maven metadata structure explanation:
+	// - <snapshot>: Contains the latest snapshot timestamp and build number
+	// - <snapshotVersions>: Lists all available files for each build number
+	//   - <extension>: File extension (jar, pom, etc.)
+	//   - <classifier>: Optional classifier (sources, javadoc, tests, etc.)
+	//   - <value>: The actual version string with timestamp and build number
+	//   - <updated>: Timestamp when the artifact was deployed
+	metadataXML := `<?xml version="1.0" encoding="UTF-8"?>
+<metadata>
+  <groupId>com.gitea</groupId>
+  <artifactId>test-project</artifactId>
+  <version>1.0-SNAPSHOT</version>
+  <versioning>
+    <snapshot>
+      <timestamp>20230101.000000</timestamp>
+      <buildNumber>5</buildNumber>
+    </snapshot>
+    <lastUpdated>20230101000000</lastUpdated>
+    <snapshotVersions>
+      <snapshotVersion>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-1</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-2</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-3</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <classifier>sources</classifier>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-3</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <classifier>javadoc</classifier>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-3</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-4</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <classifier>sources</classifier>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-4</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <classifier>javadoc</classifier>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-4</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-5</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <classifier>sources</classifier>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-5</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+      <snapshotVersion>
+        <classifier>javadoc</classifier>
+        <extension>jar</extension>
+        <value>1.0-20230101.000000-5</value>
+        <updated>20230101000000</updated>
+      </snapshotVersion>
+    </snapshotVersions>
+  </versioning>
+</metadata>`
+
+	// Add metadata file to the existing package version using service method
+	metadataReader := bytes.NewReader([]byte(metadataXML))
+	hsr, err := packages_module.CreateHashedBufferFromReader(metadataReader)
+	assert.NoError(t, err)
+
+	pfci := &packages_service.PackageFileCreationInfo{
+		PackageFileInfo: packages_service.PackageFileInfo{
+			Filename: "maven-metadata.xml",
+		},
+		Data: hsr,
+	}
+
+	_, err = packages_service.AddFileToPackageVersionInternal(t.Context(), pv, pfci)
+	assert.NoError(t, err)
+}
+
+func TestCleanupSnapshotVersions(t *testing.T) {
+	assert.NoError(t, unittest.PrepareTestDatabase())
+
+	t.Run("Should skip when retainBuilds is negative", func(t *testing.T) {
+		setting.Packages.RetainMavenSnapshotBuilds = -1
+		setting.Packages.DebugMavenCleanup = false
+		t.Logf("Test settings: retainBuilds=%d, debug=%t", setting.Packages.RetainMavenSnapshotBuilds, setting.Packages.DebugMavenCleanup)
+		err := CleanupSnapshotVersions(t.Context())
+		assert.NoError(t, err)
+	})
+
+	t.Run("Should skip when retainBuilds is zero", func(t *testing.T) {
+		setting.Packages.RetainMavenSnapshotBuilds = 0
+		setting.Packages.DebugMavenCleanup = false
+		t.Logf("Test settings: retainBuilds=%d, debug=%t", setting.Packages.RetainMavenSnapshotBuilds, setting.Packages.DebugMavenCleanup)
+		err := CleanupSnapshotVersions(t.Context())
+		assert.NoError(t, err)
+	})
+
+	t.Run("Should handle missing metadata file gracefully", func(t *testing.T) {
+		assert.NoError(t, unittest.PrepareTestDatabase())
+
+		setting.Packages.RetainMavenSnapshotBuilds = 2
+		setting.Packages.DebugMavenCleanup = false
+
+		// Get the existing package version from fixtures (ID 1)
+		pv, err := packages.GetVersionByID(t.Context(), 1)
+		assert.NoError(t, err)
+
+		// Verify all 11 files exist before cleanup (5 base jars + 6 classifier jars)
+		filesBefore, err := packages.GetFilesByVersionID(t.Context(), pv.ID)
+		assert.NoError(t, err)
+		assert.Len(t, filesBefore, 11) // 5 base jars + 6 classifier jars (sources + javadoc for builds 3,4,5)
+
+		// No metadata file exists in fixtures - should handle gracefully
+		err = CleanupSnapshotVersions(t.Context())
+		assert.NoError(t, err)
+
+		// Verify all 11 files still exist after cleanup (no cleanup should occur without metadata)
+		filesAfter, err := packages.GetFilesByVersionID(t.Context(), pv.ID)
+		assert.NoError(t, err)
+		assert.Len(t, filesAfter, 11, "All files should remain when metadata is missing")
+	})
+
+	t.Run("Should work with debug mode", func(t *testing.T) {
+		assert.NoError(t, unittest.PrepareTestDatabase())
+
+		setting.Packages.RetainMavenSnapshotBuilds = 2
+		setting.Packages.DebugMavenCleanup = true
+
+		pv, err := packages.GetVersionByID(t.Context(), 1)
+		assert.NoError(t, err)
+
+		addMavenMetadataToPackageVersion(t, pv)
+
+		filesBefore, err := packages.GetFilesByVersionID(t.Context(), pv.ID)
+		assert.NoError(t, err)
+		assert.Len(t, filesBefore, 12) // 11 jar files + 1 metadata file
+
+		err = CleanupSnapshotVersions(t.Context())
+		assert.NoError(t, err)
+
+		// Verify all files still exist after cleanup (debug mode should not delete anything)
+		filesAfter, err := packages.GetFilesByVersionID(t.Context(), pv.ID)
+		assert.NoError(t, err)
+		assert.Len(t, filesAfter, 12, "All files should remain in debug mode")
+	})
+
+	t.Run("Should test actual cleanup with metadata", func(t *testing.T) {
+		assert.NoError(t, unittest.PrepareTestDatabase())
+
+		setting.Packages.DebugMavenCleanup = false
+		setting.Packages.RetainMavenSnapshotBuilds = 2
+		t.Logf("Test settings: retainBuilds=%d, debug=%t", setting.Packages.RetainMavenSnapshotBuilds, setting.Packages.DebugMavenCleanup)
+
+		// Get the existing package version from fixtures (ID 1)
+		pv, err := packages.GetVersionByID(t.Context(), 1)
+		assert.NoError(t, err)
+		assert.Equal(t, "1.0-SNAPSHOT", pv.Version)
+
+		addMavenMetadataToPackageVersion(t, pv)
+
+		filesBefore, err := packages.GetFilesByVersionID(t.Context(), pv.ID)
+		assert.NoError(t, err)
+		assert.Len(t, filesBefore, 12) // 11 jar files + 1 metadata file
+
+		// Test cleanup with retainBuilds = 2 (should keep builds 4 and 5, remove 1, 2, 3)
+		// Build 4: base jar + sources + javadoc = 3 files
+		// Build 5: base jar + sources + javadoc = 3 files
+		// Total retained: 6 files + 1 metadata = 7 files
+		err = CleanupSnapshotVersions(t.Context())
+		assert.NoError(t, err)
+
+		filesAfter, err := packages.GetFilesByVersionID(t.Context(), pv.ID)
+		assert.NoError(t, err)
+
+		// Should have metadata file + 6 retained build artifacts (2 builds × 3 files each)
+		assert.Len(t, filesAfter, 7)
+
+		// Check that metadata file is still there
+		var hasMetadata bool
+		var retainedBuilds []string
+		for _, file := range filesAfter {
+			if file.Name == "maven-metadata.xml" {
+				hasMetadata = true
+			} else {
+				retainedBuilds = append(retainedBuilds, file.Name)
+			}
+		}
+
+		assert.True(t, hasMetadata, "maven-metadata.xml should be retained")
+		assert.Len(t, retainedBuilds, 6, "Should retain exactly 6 files (2 builds with 3 artifacts each)")
+
+		t.Logf("Retained builds: %v", retainedBuilds)
+
+		// Verify build 4 artifacts are retained
+		assert.Contains(t, retainedBuilds, "gitea-test-1.0-20230101.000000-4.jar")
+		assert.Contains(t, retainedBuilds, "gitea-test-1.0-20230101.000000-4-sources.jar")
+		assert.Contains(t, retainedBuilds, "gitea-test-1.0-20230101.000000-4-javadoc.jar")
+
+		// Verify build 5 artifacts are retained
+		assert.Contains(t, retainedBuilds, "gitea-test-1.0-20230101.000000-5.jar")
+		assert.Contains(t, retainedBuilds, "gitea-test-1.0-20230101.000000-5-sources.jar")
+		assert.Contains(t, retainedBuilds, "gitea-test-1.0-20230101.000000-5-javadoc.jar")
+	})
+}

services/packages/packages.go

@@ -598,8 +598,8 @@ func OpenBlobStream(pb *packages_model.PackageBlob) (io.ReadSeekCloser, error) {
 }
 
 // OpenBlobForDownload returns the content of the specific package blob and increases the download counter.
-// If the storage supports direct serving and it's enabled, only the direct serving url is returned.
-func OpenBlobForDownload(ctx context.Context, pf *packages_model.PackageFile, pb *packages_model.PackageBlob, method string, serveDirectReqParams url.Values) (io.ReadSeekCloser, *url.URL, *packages_model.PackageFile, error) {
+// If the storage supports direct serving and it's enabled, only the direct serving url is returned; otherwise, forceInternalServe should be set to true.
+func OpenBlobForDownload(ctx context.Context, pf *packages_model.PackageFile, pb *packages_model.PackageBlob, method string, serveDirectReqParams url.Values, forceInternalServe ...bool) (io.ReadSeekCloser, *url.URL, *packages_model.PackageFile, error) {
 	key := packages_module.BlobHash256Key(pb.HashSHA256)
 
 	cs := packages_module.NewContentStore()
@@ -608,7 +608,9 @@ func OpenBlobForDownload(ctx context.Context, pf *packages_model.PackageFile, pb
 	var u *url.URL
 	var err error
 
-	if cs.ShouldServeDirect() {
+	internalServe := len(forceInternalServe) > 0 && forceInternalServe[0]
+
+	if !internalServe && cs.ShouldServeDirect() {
 		u, err = cs.GetServeDirectURL(key, pf.Name, method, serveDirectReqParams)
 		if err != nil && !errors.Is(err, storage.ErrURLNotSupported) {
 			log.Error("Error getting serve direct url (fallback to local reader): %v", err)