Add the archive --purge flag to clean up table files

go/cmd/dolt/commands/archive.go

@@ -54,6 +54,7 @@ table files into archives. Currently, for safety, table files are left in place.
 
 const groupChunksFlag = "group-chunks"
 const revertFlag = "revert"
+const purgeFlag = "purge"
 
 // Description returns a description of the command
 func (cmd ArchiveCmd) Description() string {
@@ -71,9 +72,7 @@ func (cmd ArchiveCmd) ArgParser() *argparser.ArgParser {
 	ap := argparser.NewArgParserWithMaxArgs(cmd.Name(), 0)
 	ap.SupportsFlag(groupChunksFlag, "", "Attempt to group chunks. This will produce smaller archives, but can take much longer to build.")
 	ap.SupportsFlag(revertFlag, "", "Return to unpurged table files, or rebuilt table files from archives")
-	/* TODO: Implement these flags
-	ap.SupportsFlag("purge", "", "remove table files after archiving")
-	*/
+	ap.SupportsFlag(purgeFlag, "", "remove table files after archiving")
 	return ap
 }
 func (cmd ArchiveCmd) Hidden() bool {
@@ -137,7 +136,9 @@ func (cmd ArchiveCmd) Exec(ctx context.Context, commandStr string, args []string
 			}
 		}
 
-		err = nbs.BuildArchive(ctx, cs, &groupings, progress)
+		purge := apr.Contains(purgeFlag)
+
+		err = nbs.BuildArchive(ctx, cs, &groupings, purge, progress)
 		if err != nil {
 			cli.PrintErrln(err)
 			return 1

go/store/nbs/archive_build.go

@@ -114,7 +114,7 @@ func UnArchive(ctx context.Context, cs chunks.ChunkStore, smd StorageMetadata, p
 	return nil
 }
 
-func BuildArchive(ctx context.Context, cs chunks.ChunkStore, dagGroups *ChunkRelations, progress chan interface{}) (err error) {
+func BuildArchive(ctx context.Context, cs chunks.ChunkStore, dagGroups *ChunkRelations, purge bool, progress chan interface{}) (err error) {
 	// Currently, we don't have any stats to report. Required for calls to the lower layers tho.
 	var stats Stats
 
@@ -165,12 +165,15 @@ func BuildArchive(ctx context.Context, cs chunks.ChunkStore, dagGroups *ChunkRel
 			return fmt.Errorf("No tables found to archive. Run 'dolt gc' first")
 		}
 
+		cleanup := make([]hash.Hash, 0, len(swapMap))
+
 		//NM4 TODO: This code path must only be run on an offline database. We should add a check for that.
 		specs, err := gs.oldGen.tables.toSpecs()
 		newSpecs := make([]tableSpec, 0, len(specs))
 		for _, spec := range specs {
 			if newSpec, exists := swapMap[spec.name]; exists {
 				newSpecs = append(newSpecs, tableSpec{newSpec, spec.chunkCount})
+				cleanup = append(cleanup, spec.name)
 			} else {
 				newSpecs = append(newSpecs, spec)
 			}
@@ -179,6 +182,17 @@ func BuildArchive(ctx context.Context, cs chunks.ChunkStore, dagGroups *ChunkRel
 		if err != nil {
 			return err
 		}
+
+		if purge && len(cleanup) > 0 {
+			for _, h := range cleanup {
+				tf := filepath.Join(outPath, h.String())
+				err = os.Remove(tf)
+				if err != nil {
+					return err
+				}
+			}
+		}
+
 	} else {
 		return errors.New("Modern DB Expected")
 	}

integration-tests/bats/archive.bats

@@ -128,7 +128,6 @@ mutations_and_gc_statement() {
 @test "archive: archive --revert (rebuild)" {
   dolt sql -q "$(mutations_and_gc_statement)"
   dolt archive
-  dolt gc                         # This will delete the unused table files.
   dolt archive --revert
 
   # dolt log --stat will load every single chunk. 66 manually verified.
@@ -136,6 +135,18 @@ mutations_and_gc_statement() {
   [ "$commits" -eq "66" ]
 }
 
+@test "archive: archive --purge" {
+  dolt sql -q "$(mutations_and_gc_statement)"
+
+  tablefile=$(find ./.dolt/noms/oldgen -type f -regex '.*/[a-v0-9]\{32\}')
+
+  [ -e "$tablefile" ] # extreme paranoia. make sure it exists before.
+  dolt archive --purge
+  # Ensure the table file is gone.
+  [ ! -e "$tablefile" ]
+}
+
+
 @test "archive: can clone archived repository" {
     mkdir -p remote/.dolt
     mkdir cloned