feat: add ForceReindex flag to VersionSyncWorkflow

Allows re-triggering asset indexing for all existing versions, not just
newly discovered ones. Useful for backfilling data after schema changes
like the new checksum columns.

Also trims the Versions slice on continue-as-new in
VersionBatchIndexWorkflow to avoid re-serializing already-processed
entries on every cycle.

Signed-off-by: Gabriel Harris-Rouquette <gabizou@me.com>

Author: gabizou

docs/WORKFLOWS.md

@@ -12,7 +12,7 @@ HTTP API (RegisterArtifact → triggers VersionSyncWorkflow)
 VersionSyncWorkflow ─────────────────────────────────────────────────┐
  ├── FetchVersions           (Sonatype maven-metadata.xml)           |
  ├── StoreNewVersions        (DB: compare + insert new versions)     |
- ├── VersionBatchIndexWorkflow (child, if new versions found)        |
+ ├── VersionBatchIndexWorkflow (child, if new versions or ForceReindex)|
  │    └── VersionIndexWorkflow (per version, sliding window of 5)    |
  │         ├── FetchVersionAssets   (Sonatype REST API)              |
  │         ├── StoreVersionAssets   (DB)                             |
@@ -57,13 +57,15 @@ The top-level orchestrator, triggered when an artifact is registered via the HTT
 
 **Trigger:** `POST /groups/{groupID}/artifacts` (fire-and-forget)
 
+**Input:** `VersionSyncInput{GroupID, ArtifactID, ForceReindex}`
+
 **Steps:**
 1. Fetch all version strings from Sonatype Nexus (`maven-metadata.xml`)
 2. Compare against DB and store only new versions
-3. Launch batch indexing for new versions (assets + commits)
+3. Launch batch indexing for new versions (assets + commits) — or **all** versions when `ForceReindex` is true
 4. Compute version ordering and extract schema-driven tags for **all** versions
 
-The ordering step runs for all versions (not just new ones) because `sort_order` is a relative ranking that must be recomputed when new versions are inserted.
+The ordering step runs for all versions (not just new ones) because `sort_order` is a relative ranking that must be recomputed when new versions are inserted. The `ForceReindex` flag is useful for backfilling data after schema changes (e.g. new asset columns).
 
 ### VersionBatchIndexWorkflow
 

internal/workflow/version_batch_index.go

@@ -118,9 +118,9 @@ func (s *versionBatchState) continueOrComplete(ctx workflow.Context) (int, error
 		s.drainCompletionSignals(ctx)
 
 		return 0, workflow.NewContinueAsNewError(ctx, VersionBatchIndexWorkflow, VersionBatchIndexInput{
-			Versions:       s.input.Versions,
+			Versions:       s.input.Versions[s.offset:],
 			WindowSize:     s.input.WindowSize,
-			Offset:         s.offset,
+			Offset:         0,
 			Progress:       s.progress,
 			CurrentRecords: s.currentRecords,
 		})

internal/workflow/version_sync.go

@@ -17,8 +17,9 @@ const (
 
 // VersionSyncInput is the input payload for the VersionSyncWorkflow.
 type VersionSyncInput struct {
-	GroupID    string
-	ArtifactID string
+	GroupID      string
+	ArtifactID   string
+	ForceReindex bool
 }
 
 // VersionSyncOutput is the result of the VersionSyncWorkflow.
@@ -95,15 +96,19 @@ func VersionSyncWorkflow(ctx workflow.Context, input VersionSyncInput) (*Version
 	// Ordering needs version rows but not assets or commits.
 	// Both can run concurrently.
 
-	// Start indexing (only for new versions)
+	// Start indexing — for new versions, or all versions when force-reindexing.
+	versionsToIndex := storeResult.NewVersions
+	if input.ForceReindex {
+		versionsToIndex = fetchResult.Versions
+	}
 	var indexFuture workflow.ChildWorkflowFuture
-	if len(storeResult.NewVersions) > 0 {
+	if len(versionsToIndex) > 0 {
 		indexOpts := workflow.ChildWorkflowOptions{
 			WorkflowID: fmt.Sprintf("version-batch-index-%s-%s", input.GroupID, input.ArtifactID),
 		}
 		indexCtx := workflow.WithChildOptions(ctx, indexOpts)
 		indexFuture = workflow.ExecuteChildWorkflow(indexCtx, VersionBatchIndexWorkflow, VersionBatchIndexInput{
-			Versions:   storeResult.NewVersions,
+			Versions:   versionsToIndex,
 			WindowSize: versionBatchDefaultWindowSize,
 		})
 	}
@@ -138,7 +143,10 @@ func VersionSyncWorkflow(ctx workflow.Context, input VersionSyncInput) (*Version
 	enrichCtx := workflow.WithChildOptions(ctx, enrichOpts)
 
 	var enrichResult CommitEnrichmentOutput
-	err = workflow.ExecuteChildWorkflow(enrichCtx, CommitEnrichmentWorkflow, CommitEnrichmentInput(input)).Get(ctx, &enrichResult)
+	err = workflow.ExecuteChildWorkflow(enrichCtx, CommitEnrichmentWorkflow, CommitEnrichmentInput{
+		GroupID:    input.GroupID,
+		ArtifactID: input.ArtifactID,
+	}).Get(ctx, &enrichResult)
 	if err != nil {
 		return nil, fmt.Errorf("enriching commits: %w", err)
 	}