zeroshade commented on code in PR #1883: URL: https://github.com/apache/iceberg-go/pull/1883#discussion_r3937950868
########## table/incremental_changelog_scan.go: ########## @@ -0,0 +1,393 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +package table + +import ( + "cmp" + "context" + "fmt" + "maps" + "slices" + "time" + + "github.com/apache/iceberg-go" + "github.com/apache/iceberg-go/metrics" +) + +// IncrementalChangelogScan plans data-file changes between snapshots. It +// emits insert and delete tasks for data-manifest entries and skips replace +// snapshots. PlanFiles returns an error if any in-range snapshot's manifest +// list references a delete manifest, including ones carried forward from +// earlier snapshots. +type IncrementalChangelogScan struct { + scan *Scan + fromSnapshotID *int64 + fromInclusive bool + toSnapshotID *int64 +} + +type plannedChangelogTask struct { + task ChangelogScanTask + file FileScanTask +} + +// NewIncrementalChangelogScan creates an incremental changelog planner. +// Projection and row limits are not applied to returned tasks. Auto planning +// falls back to local planning, while remote planning is not supported. Use +// ChangelogScanTask.ScanTask with Scan.ReadTasks to read the returned files. +// Row filters are attached to each task as residuals without partition-specific +// simplification, matching the existing incremental append scan behavior. +func (t Table) NewIncrementalChangelogScan(opts ...ScanOption) *IncrementalChangelogScan { + return &IncrementalChangelogScan{scan: t.Scan(opts...)} +} + +// FromSnapshotInclusive includes changes committed by the starting snapshot. +// The snapshot is validated when files are planned. +func (s *IncrementalChangelogScan) FromSnapshotInclusive(snapshotID int64) *IncrementalChangelogScan { + out := *s + out.fromSnapshotID = &snapshotID + out.fromInclusive = true + + return &out +} + +// FromSnapshotExclusive starts after the given snapshot. The starting +// snapshot must be a parent ancestor of the ending snapshot when planning. +// The snapshot is validated when files are planned. +func (s *IncrementalChangelogScan) FromSnapshotExclusive(snapshotID int64) *IncrementalChangelogScan { + out := *s + out.fromSnapshotID = &snapshotID + out.fromInclusive = false + + return &out +} + +// ToSnapshot sets the inclusive ending snapshot. +// The snapshot is validated when files are planned. +func (s *IncrementalChangelogScan) ToSnapshot(snapshotID int64) *IncrementalChangelogScan { + out := *s + out.toSnapshotID = &snapshotID + + return &out +} + +// PlanFiles returns one task for each added or deleted data-file entry. A +// cancelled context returns its cancellation error before planning starts. +// Tasks are ordered by change ordinal, then by DELETE before INSERT within an +// ordinal, and finally by data-file path. When an ending snapshot is +// available, it emits a ScanReport through the configured reporter on +// successful planning. Changelog reports count every returned task in +// ResultDataFiles and TotalFileSizeInBytes, so a file inserted and deleted +// within the range is counted twice. +func (s *IncrementalChangelogScan) PlanFiles(ctx context.Context) ([]ChangelogScanTask, error) { + if s == nil || s.scan == nil { + return nil, fmt.Errorf("%w: incremental changelog scan is not initialized", ErrInvalidOperation) + } + if err := ctx.Err(); err != nil { + return nil, err + } + + switch s.scan.planningMode { + case ScanPlanningLocal, ScanPlanningAuto: + case ScanPlanningRemote: + return nil, fmt.Errorf("%w: incremental changelog scans do not support remote planning", ErrInvalidOperation) + default: + return nil, fmt.Errorf("%w: unknown scan planning mode %q", iceberg.ErrInvalidArgument, s.scan.planningMode) + } + start := time.Now() + + toSnapshot, err := s.toSnapshot() + if err != nil { + return nil, err + } + if toSnapshot == nil { + if s.fromSnapshotID != nil { + return nil, fmt.Errorf("%w: no ending snapshot found for incremental changelog scan from %d", + iceberg.ErrInvalidArgument, *s.fromSnapshotID) + } + + return nil, nil + } + + planningScan := *s.scan + planningScan.identifier = slices.Clone(s.scan.identifier) + planningScan.selectedFields = slices.Clone(s.scan.selectedFields) + planningScan.options = maps.Clone(s.scan.options) + if s.toSnapshotID != nil { + planningScan.snapshotID = &toSnapshot.SnapshotID + planningScan.asOfTimestamp = nil + } + schema, err := planningScan.effectiveSchema() + if err != nil { + return nil, err + } + residual, err := bindTaskFilter(schema, planningScan.rowFilter, planningScan.caseSensitive) + if err != nil { + return nil, fmt.Errorf("bind incremental changelog scan residual: %w", err) + } + var acc scanMetricsAccumulator + finish := func(plannedTasks []plannedChangelogTask) []ChangelogScanTask { + acc.resultDataFiles = int64(len(plannedTasks)) + tasks := make([]ChangelogScanTask, len(plannedTasks)) + fileTasks := make([]FileScanTask, len(plannedTasks)) + for i, planned := range plannedTasks { + tasks[i] = planned.task + fileTasks[i] = planned.file + acc.totalFileSize += planned.file.File.FileSizeBytes() + } + acc.applyResultDeleteMetrics(fileTasks) + planningDuration := time.Since(start) + + if rep := planningScan.Reporter(); !metrics.IsNop(rep) { + projected, _ := planningScan.Projection() + safeReport(ctx, rep, planningScan.buildScanReport(&acc, schema, projected, planningDuration)) + } + + return tasks + } + + snapshotRange, err := incrementalSnapshotsBetween( + s.scan.metadata, s.fromSnapshotID, s.fromInclusive, toSnapshot.SnapshotID) + if err != nil { + return nil, err + } + snapshots, err := changelogSnapshots(snapshotRange) + if err != nil { + return nil, err + } + if len(snapshots) == 0 { + return finish(nil), nil + } + + changelogSnapshotIDs := make(map[int64]struct{}, len(snapshots)) + snapshotOrdinals := make(map[int64]int, len(snapshots)) + for ordinal, snapshot := range snapshots { + changelogSnapshotIDs[snapshot.SnapshotID] = struct{}{} + snapshotOrdinals[snapshot.SnapshotID] = ordinal + } + + if s.scan.ioF == nil { + return nil, fmt.Errorf("%w: table file IO is not configured", ErrInvalidOperation) + } + fs, err := s.scan.ioF(ctx) + if err != nil { + return nil, err + } + + manifestsByPath := make(map[string]iceberg.ManifestFile) + for _, snapshot := range snapshots { + if err := ctx.Err(); err != nil { + return nil, err + } + manifests, err := snapshot.Manifests(fs) + if err != nil { + return nil, err + } + for _, manifest := range manifests { + if manifest.ManifestContent() == iceberg.ManifestContentDeletes { + return nil, fmt.Errorf("%w: incremental changelog scan range references a delete manifest originating in snapshot %d", + ErrInvalidOperation, manifest.SnapshotID()) + } + if manifest.ManifestContent() != iceberg.ManifestContentData { + continue + } + if _, ok := changelogSnapshotIDs[manifest.SnapshotID()]; !ok { + continue + } + manifestsByPath[manifest.FilePath()] = manifest + } + } + + paths := make([]string, 0, len(manifestsByPath)) + for path := range manifestsByPath { + paths = append(paths, path) + } + slices.Sort(paths) + manifestList := make([]iceberg.ManifestFile, 0, len(paths)) + for _, path := range paths { + manifestList = append(manifestList, manifestsByPath[path]) + } + + // Changelog metrics intentionally count only manifests that contain added or + // deleted data files; no-change manifests are removed before the scan metric + // accumulator sees them. + manifestList = slices.DeleteFunc(manifestList, func(manifest iceberg.ManifestFile) bool { + return !manifestHasChangelogEntries(manifest) + }) + partitionFilters := planningScan.partitionFiltersForSchema(schema) + manifestList, err = planningScan.filterManifestsWithSchemaOptions( + manifestList, schema, &acc, partitionFilters, + /* includeDeleted= */ true) + if err != nil { + return nil, err + } + if len(manifestList) == 0 { + return finish(nil), nil + } + entries, err := planningScan.collectManifestEntriesWithSchemaOptions( + ctx, manifestList, schema, + partitionFilters, + /* discardDeleted= */ false, + /* discardExisting= */ true, + ) + if err != nil { + return nil, err + } + + plannedTasks := make([]plannedChangelogTask, 0, len(entries.dataEntries)) + for _, entry := range entries.dataEntries { + ordinal, ok := snapshotOrdinals[entry.SnapshotID()] + if !ok { + continue + } + + task, err := newChangelogScanTask(entry, ordinal, residual) + if err != nil { + return nil, fmt.Errorf("incremental changelog scan snapshot %d: %w", entry.SnapshotID(), err) + } + plannedTasks = append(plannedTasks, plannedChangelogTask{ + task: task, + file: task.ScanTask(), + }) + } + slices.SortFunc(plannedTasks, func(left, right plannedChangelogTask) int { + if ordinal := cmp.Compare(left.task.ChangeOrdinal(), right.task.ChangeOrdinal()); ordinal != 0 { + return ordinal + } + if operation := cmp.Compare(changelogOperationOrder(left.task.Operation()), changelogOperationOrder(right.task.Operation())); operation != 0 { + return operation + } + + return cmp.Compare(left.file.File.FilePath(), right.file.File.FilePath()) + }) + return finish(plannedTasks), nil +} Review Comment: **blocking** — PR-added line trips the repo's enabled nlreturn linter, failing all four Go CI jobs `return finish(plannedTasks), nil` immediately follows the closing `})` of the slices.SortFunc block with no blank line. `nlreturn` is enabled at .golangci.yml:28, so this fails the lint step on ubuntu/macos x go1.25.9/go1.26.1 — the sole cause of the red CI. Fix is to insert a blank line before the return, matching the style used everywhere else in this same file (e.g. :63, :74, :83, :152). <details><summary>Evidence</summary> ```text CI (job 100728632052 ubuntu go1.25.9 and 100728632137 macos go1.25.9): '##[error]table/incremental_changelog_scan.go:278:2: return with no blank line before (nlreturn)' / '##[error]issues found'. Reproduced locally with golangci-lint 2.12.2 (same version CI pins) using a minimal nlreturn-only config: 'incremental_changelog_scan.go:278:2: return with no blank line before (nlreturn)\n\treturn finish(plannedTasks), nil\n\t^\n1 issues:\n* nlreturn: 1'. Still reproduces at clean head after all probes were reverted. ``` </details> ########## table/incremental_changelog_scan.go: ########## @@ -0,0 +1,393 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information +// regarding copyright ownership. The ASF licenses this file +// to you under the Apache License, Version 2.0 (the +// "License"); you may not use this file except in compliance +// with the License. You may obtain a copy of the License at +// +// http://www.apache.org/licenses/LICENSE-2.0 +// +// Unless required by applicable law or agreed to in writing, +// software distributed under the License is distributed on an +// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY +// KIND, either express or implied. See the License for the +// specific language governing permissions and limitations +// under the License. + +package table + +import ( + "cmp" + "context" + "fmt" + "maps" + "slices" + "time" + + "github.com/apache/iceberg-go" + "github.com/apache/iceberg-go/metrics" +) + +// IncrementalChangelogScan plans data-file changes between snapshots. It +// emits insert and delete tasks for data-manifest entries and skips replace +// snapshots. PlanFiles returns an error if any in-range snapshot's manifest +// list references a delete manifest, including ones carried forward from +// earlier snapshots. +type IncrementalChangelogScan struct { + scan *Scan + fromSnapshotID *int64 + fromInclusive bool + toSnapshotID *int64 +} + +type plannedChangelogTask struct { + task ChangelogScanTask + file FileScanTask +} + +// NewIncrementalChangelogScan creates an incremental changelog planner. +// Projection and row limits are not applied to returned tasks. Auto planning +// falls back to local planning, while remote planning is not supported. Use +// ChangelogScanTask.ScanTask with Scan.ReadTasks to read the returned files. +// Row filters are attached to each task as residuals without partition-specific +// simplification, matching the existing incremental append scan behavior. +func (t Table) NewIncrementalChangelogScan(opts ...ScanOption) *IncrementalChangelogScan { + return &IncrementalChangelogScan{scan: t.Scan(opts...)} +} + +// FromSnapshotInclusive includes changes committed by the starting snapshot. +// The snapshot is validated when files are planned. +func (s *IncrementalChangelogScan) FromSnapshotInclusive(snapshotID int64) *IncrementalChangelogScan { + out := *s + out.fromSnapshotID = &snapshotID + out.fromInclusive = true + + return &out +} + +// FromSnapshotExclusive starts after the given snapshot. The starting +// snapshot must be a parent ancestor of the ending snapshot when planning. +// The snapshot is validated when files are planned. +func (s *IncrementalChangelogScan) FromSnapshotExclusive(snapshotID int64) *IncrementalChangelogScan { + out := *s + out.fromSnapshotID = &snapshotID + out.fromInclusive = false + + return &out +} + +// ToSnapshot sets the inclusive ending snapshot. +// The snapshot is validated when files are planned. +func (s *IncrementalChangelogScan) ToSnapshot(snapshotID int64) *IncrementalChangelogScan { + out := *s + out.toSnapshotID = &snapshotID + + return &out +} + +// PlanFiles returns one task for each added or deleted data-file entry. A +// cancelled context returns its cancellation error before planning starts. +// Tasks are ordered by change ordinal, then by DELETE before INSERT within an +// ordinal, and finally by data-file path. When an ending snapshot is +// available, it emits a ScanReport through the configured reporter on +// successful planning. Changelog reports count every returned task in +// ResultDataFiles and TotalFileSizeInBytes, so a file inserted and deleted +// within the range is counted twice. +func (s *IncrementalChangelogScan) PlanFiles(ctx context.Context) ([]ChangelogScanTask, error) { + if s == nil || s.scan == nil { + return nil, fmt.Errorf("%w: incremental changelog scan is not initialized", ErrInvalidOperation) + } + if err := ctx.Err(); err != nil { + return nil, err + } + + switch s.scan.planningMode { + case ScanPlanningLocal, ScanPlanningAuto: + case ScanPlanningRemote: + return nil, fmt.Errorf("%w: incremental changelog scans do not support remote planning", ErrInvalidOperation) + default: + return nil, fmt.Errorf("%w: unknown scan planning mode %q", iceberg.ErrInvalidArgument, s.scan.planningMode) + } + start := time.Now() + + toSnapshot, err := s.toSnapshot() + if err != nil { + return nil, err + } + if toSnapshot == nil { + if s.fromSnapshotID != nil { + return nil, fmt.Errorf("%w: no ending snapshot found for incremental changelog scan from %d", + iceberg.ErrInvalidArgument, *s.fromSnapshotID) + } + + return nil, nil + } + + planningScan := *s.scan + planningScan.identifier = slices.Clone(s.scan.identifier) + planningScan.selectedFields = slices.Clone(s.scan.selectedFields) + planningScan.options = maps.Clone(s.scan.options) + if s.toSnapshotID != nil { + planningScan.snapshotID = &toSnapshot.SnapshotID + planningScan.asOfTimestamp = nil + } + schema, err := planningScan.effectiveSchema() + if err != nil { + return nil, err + } + residual, err := bindTaskFilter(schema, planningScan.rowFilter, planningScan.caseSensitive) + if err != nil { + return nil, fmt.Errorf("bind incremental changelog scan residual: %w", err) + } + var acc scanMetricsAccumulator + finish := func(plannedTasks []plannedChangelogTask) []ChangelogScanTask { + acc.resultDataFiles = int64(len(plannedTasks)) + tasks := make([]ChangelogScanTask, len(plannedTasks)) + fileTasks := make([]FileScanTask, len(plannedTasks)) + for i, planned := range plannedTasks { + tasks[i] = planned.task + fileTasks[i] = planned.file + acc.totalFileSize += planned.file.File.FileSizeBytes() + } + acc.applyResultDeleteMetrics(fileTasks) + planningDuration := time.Since(start) + + if rep := planningScan.Reporter(); !metrics.IsNop(rep) { + projected, _ := planningScan.Projection() + safeReport(ctx, rep, planningScan.buildScanReport(&acc, schema, projected, planningDuration)) + } + + return tasks + } + + snapshotRange, err := incrementalSnapshotsBetween( + s.scan.metadata, s.fromSnapshotID, s.fromInclusive, toSnapshot.SnapshotID) + if err != nil { + return nil, err + } + snapshots, err := changelogSnapshots(snapshotRange) + if err != nil { + return nil, err + } + if len(snapshots) == 0 { + return finish(nil), nil + } + + changelogSnapshotIDs := make(map[int64]struct{}, len(snapshots)) + snapshotOrdinals := make(map[int64]int, len(snapshots)) + for ordinal, snapshot := range snapshots { + changelogSnapshotIDs[snapshot.SnapshotID] = struct{}{} + snapshotOrdinals[snapshot.SnapshotID] = ordinal + } + + if s.scan.ioF == nil { + return nil, fmt.Errorf("%w: table file IO is not configured", ErrInvalidOperation) + } + fs, err := s.scan.ioF(ctx) + if err != nil { + return nil, err + } + + manifestsByPath := make(map[string]iceberg.ManifestFile) + for _, snapshot := range snapshots { + if err := ctx.Err(); err != nil { + return nil, err + } + manifests, err := snapshot.Manifests(fs) + if err != nil { + return nil, err + } + for _, manifest := range manifests { + if manifest.ManifestContent() == iceberg.ManifestContentDeletes { + return nil, fmt.Errorf("%w: incremental changelog scan range references a delete manifest originating in snapshot %d", + ErrInvalidOperation, manifest.SnapshotID()) + } + if manifest.ManifestContent() != iceberg.ManifestContentData { + continue + } + if _, ok := changelogSnapshotIDs[manifest.SnapshotID()]; !ok { + continue + } + manifestsByPath[manifest.FilePath()] = manifest + } + } + + paths := make([]string, 0, len(manifestsByPath)) + for path := range manifestsByPath { + paths = append(paths, path) + } + slices.Sort(paths) + manifestList := make([]iceberg.ManifestFile, 0, len(paths)) + for _, path := range paths { + manifestList = append(manifestList, manifestsByPath[path]) + } + + // Changelog metrics intentionally count only manifests that contain added or + // deleted data files; no-change manifests are removed before the scan metric + // accumulator sees them. + manifestList = slices.DeleteFunc(manifestList, func(manifest iceberg.ManifestFile) bool { + return !manifestHasChangelogEntries(manifest) + }) + partitionFilters := planningScan.partitionFiltersForSchema(schema) + manifestList, err = planningScan.filterManifestsWithSchemaOptions( + manifestList, schema, &acc, partitionFilters, + /* includeDeleted= */ true) + if err != nil { + return nil, err + } + if len(manifestList) == 0 { + return finish(nil), nil + } + entries, err := planningScan.collectManifestEntriesWithSchemaOptions( + ctx, manifestList, schema, + partitionFilters, + /* discardDeleted= */ false, + /* discardExisting= */ true, + ) + if err != nil { + return nil, err + } + + plannedTasks := make([]plannedChangelogTask, 0, len(entries.dataEntries)) + for _, entry := range entries.dataEntries { + ordinal, ok := snapshotOrdinals[entry.SnapshotID()] + if !ok { + continue + } + + task, err := newChangelogScanTask(entry, ordinal, residual) + if err != nil { + return nil, fmt.Errorf("incremental changelog scan snapshot %d: %w", entry.SnapshotID(), err) + } + plannedTasks = append(plannedTasks, plannedChangelogTask{ + task: task, + file: task.ScanTask(), + }) + } + slices.SortFunc(plannedTasks, func(left, right plannedChangelogTask) int { + if ordinal := cmp.Compare(left.task.ChangeOrdinal(), right.task.ChangeOrdinal()); ordinal != 0 { + return ordinal + } + if operation := cmp.Compare(changelogOperationOrder(left.task.Operation()), changelogOperationOrder(right.task.Operation())); operation != 0 { + return operation + } + + return cmp.Compare(left.file.File.FilePath(), right.file.File.FilePath()) + }) + return finish(plannedTasks), nil +} + +func manifestHasChangelogEntries(manifest iceberg.ManifestFile) bool { + // V1 manifest lists use -1 for unknown counts, so only zero means the + // manifest is known not to contain added or deleted entries. + return manifest.AddedDataFiles() != 0 || manifest.DeletedDataFiles() != 0 +} + +func changelogOperationOrder(operation ChangelogOperation) int { + switch operation { + // Deletes must be replayed before inserts within one change ordinal. Keep + // this explicit instead of relying on the string values' lexical order. + case ChangelogOpDelete: + return 0 + case ChangelogOpInsert: + return 1 + case ChangelogOpUpdateBefore: + return 2 + case ChangelogOpUpdateAfter: + return 3 + default: + return 4 + } +} + +func changelogOperation(status iceberg.ManifestEntryStatus) (ChangelogOperation, error) { + switch status { + case iceberg.EntryStatusADDED: + return ChangelogOpInsert, nil + case iceberg.EntryStatusDELETED: + return ChangelogOpDelete, nil + default: + return "", fmt.Errorf("%w: unknown manifest entry status %d", ErrInvalidMetadata, status) + } +} + +func newChangelogScanTask(entry iceberg.ManifestEntry, ordinal int, residual iceberg.BooleanExpression) (ChangelogScanTask, error) { + operation, err := changelogOperation(entry.Status()) + if err != nil { + return nil, err + } + + file := entry.DataFile() + configureFileScanTask := func(task *FileScanTask) { + task.Start = 0 + task.Length = file.FileSizeBytes() + task.Residual = residual + task.FirstRowID = file.FirstRowID() + if sequenceNumber := entry.SequenceNum(); sequenceNumber >= 0 { + task.DataSequenceNumber = &sequenceNumber + } + } + + switch operation { + case ChangelogOpInsert: + task, err := NewAddedRowsScanTask(file, nil, ordinal, entry.SnapshotID()) + if err != nil { + return nil, err + } + configureFileScanTask(&task.FileScanTask) + + return task, nil + case ChangelogOpDelete: + task, err := NewDeletedDataFileScanTask(file, nil, ordinal, entry.SnapshotID()) + if err != nil { + return nil, err + } + configureFileScanTask(&task.FileScanTask) + + return task, nil + default: + return nil, fmt.Errorf("%w: unsupported changelog operation %q", ErrInvalidOperation, operation) + } +} + +func changelogTaskFileScanTask(task ChangelogScanTask) (FileScanTask, error) { + if task == nil { Review Comment: **minor** — changelogTaskFileScanTask is now dead production code reachable only from tests The prior round moved finish() onto the stored plannedChangelogTask.file, which removed the only production caller of this helper. It now exists solely to serve two tests, and its error branch is unreachable in production: ChangelogScanTask is sealed by isChangelogScanTask() and every implementation returns a concrete FileScanTask from the public ScanTask() accessor, so the nil-task guard can only fire from a test passing an explicit nil. Either delete it and have the tests call task.ScanTask() directly, or move the helper into the test file. ########## table/incremental_changelog_scan.go: ########## @@ -0,0 +1,393 @@ +// Licensed to the Apache Software Foundation (ASF) under one +// or more contributor license agreements. See the NOTICE file +// distributed with this work for additional information Review Comment: **nit** — 3 of 13 branch commits still do not compile (ChangelogOperation redeclared) Down from the 5 flagged last round but not resolved. Commits 0ab13219, f9368181 and af32225c each carry two `type ChangelogOperation string` declarations in table/ (the rebase artifact against #1897's changelog_scan_task.go), so `go build ./table/` fails at those points and git bisect is broken across the branch. Cosmetic under squash-merge; a rebase would clear it. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
