From a900b97939546b2f6144c709254770cf5abfbb99 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jos=C3=A9=20Raddaoui=20Mar=C3=ADn?=
 <raddaouimarin@gmail.com>
Date: Fri, 31 May 2024 14:45:41 +0200
Subject: [PATCH] Keep PREMIS files in SIP

---
 cmd/worker/workercmd/cmd.go             |  5 -----
 internal/activities/transform_sip.go    |  3 +--
 internal/workflow/preprocessing.go      | 18 ------------------
 internal/workflow/preprocessing_test.go | 18 ------------------
 4 files changed, 1 insertion(+), 43 deletions(-)

diff --git a/cmd/worker/workercmd/cmd.go b/cmd/worker/workercmd/cmd.go
index 2219ba8d..fd73eef7 100644
--- a/cmd/worker/workercmd/cmd.go
+++ b/cmd/worker/workercmd/cmd.go
@@ -4,7 +4,6 @@ import (
 	"context"
 
 	"github.com/artefactual-sdps/temporal-activities/bagit"
-	"github.com/artefactual-sdps/temporal-activities/removefiles"
 	"github.com/go-logr/logr"
 	"go.artefactual.dev/tools/temporal"
 	temporalsdk_activity "go.temporal.io/sdk/activity"
@@ -80,10 +79,6 @@ func (m *Main) Run(ctx context.Context) error {
 		activities.NewTransformSIP().Execute,
 		temporalsdk_activity.RegisterOptions{Name: activities.TransformSIPName},
 	)
-	w.RegisterActivityWithOptions(
-		removefiles.NewActivity().Execute,
-		temporalsdk_activity.RegisterOptions{Name: removefiles.ActivityName},
-	)
 	w.RegisterActivityWithOptions(
 		bagit.NewCreateBagActivity(m.cfg.Bagit).Execute,
 		temporalsdk_activity.RegisterOptions{Name: bagit.CreateBagActivityName},
diff --git a/internal/activities/transform_sip.go b/internal/activities/transform_sip.go
index e495bf78..5cec4711 100644
--- a/internal/activities/transform_sip.go
+++ b/internal/activities/transform_sip.go
@@ -46,8 +46,7 @@ func (a *TransformSIP) Execute(ctx context.Context, params *TransformSIPParams)
 			return err
 		}
 		if d.Name() == "Prozess_Digitalisierung_PREMIS.xml" {
-			// Adding the parent dir to the filename reduces the likelihood of filename
-			// conflicts and prevents deletion when all PREMIS files are deleted after.
+			// Adding the parent dir to the filename reduces the likelihood of filename conflicts.
 			dir := filepath.Base(filepath.Dir(p))
 			dest := filepath.Join(mdPath, fmt.Sprintf("Prozess_Digitalisierung_PREMIS_%s.xml", dir))
 			err := fsutil.Move(p, dest)
diff --git a/internal/workflow/preprocessing.go b/internal/workflow/preprocessing.go
index 0a917496..bbb3aca5 100644
--- a/internal/workflow/preprocessing.go
+++ b/internal/workflow/preprocessing.go
@@ -4,11 +4,9 @@ import (
 	"errors"
 	"fmt"
 	"path/filepath"
-	"regexp"
 	"time"
 
 	"github.com/artefactual-sdps/temporal-activities/bagit"
-	"github.com/artefactual-sdps/temporal-activities/removefiles"
 	"go.artefactual.dev/tools/temporal"
 	temporalsdk_temporal "go.temporal.io/sdk/temporal"
 	temporalsdk_workflow "go.temporal.io/sdk/workflow"
@@ -16,8 +14,6 @@ import (
 	"github.com/artefactual-sdps/preprocessing-sfa/internal/activities"
 )
 
-var premisRe *regexp.Regexp = regexp.MustCompile("(?i)_PREMIS.xml$")
-
 type PreprocessingWorkflowParams struct {
 	RelativePath string
 }
@@ -106,20 +102,6 @@ func (w *PreprocessingWorkflow) Execute(
 		return nil, e
 	}
 
-	// Remove PREMIS files.
-	var removeFiles removefiles.ActivityResult
-	e = temporalsdk_workflow.ExecuteActivity(
-		withLocalActOpts(ctx),
-		removefiles.ActivityName,
-		&removefiles.ActivityParams{
-			Path:           localPath,
-			RemovePatterns: []*regexp.Regexp{premisRe},
-		},
-	).Get(ctx, &removeFiles)
-	if e != nil {
-		return nil, e
-	}
-
 	// Bag the SIP for Enduro processing.
 	var createBag bagit.CreateBagActivityResult
 	e = temporalsdk_workflow.ExecuteActivity(
diff --git a/internal/workflow/preprocessing_test.go b/internal/workflow/preprocessing_test.go
index 4683cf81..4def859d 100644
--- a/internal/workflow/preprocessing_test.go
+++ b/internal/workflow/preprocessing_test.go
@@ -2,11 +2,9 @@ package workflow_test
 
 import (
 	"path/filepath"
-	"regexp"
 	"testing"
 
 	"github.com/artefactual-sdps/temporal-activities/bagit"
-	"github.com/artefactual-sdps/temporal-activities/removefiles"
 	"github.com/stretchr/testify/mock"
 	"github.com/stretchr/testify/suite"
 	temporalsdk_activity "go.temporal.io/sdk/activity"
@@ -22,8 +20,6 @@ import (
 
 const sharedPath = "/shared/path/"
 
-var premisRe *regexp.Regexp = regexp.MustCompile("(?i)_PREMIS.xml$")
-
 type PreprocessingTestSuite struct {
 	suite.Suite
 	temporalsdk_testsuite.WorkflowTestSuite
@@ -57,10 +53,6 @@ func (s *PreprocessingTestSuite) SetupTest(cfg config.Configuration) {
 		activities.NewTransformSIP().Execute,
 		temporalsdk_activity.RegisterOptions{Name: activities.TransformSIPName},
 	)
-	s.env.RegisterActivityWithOptions(
-		removefiles.NewActivity().Execute,
-		temporalsdk_activity.RegisterOptions{Name: removefiles.ActivityName},
-	)
 	s.env.RegisterActivityWithOptions(
 		bagit.NewCreateBagActivity(cfg.Bagit).Execute,
 		temporalsdk_activity.RegisterOptions{Name: bagit.CreateBagActivityName},
@@ -127,16 +119,6 @@ func (s *PreprocessingTestSuite) TestPreprocessingWorkflowSuccess() {
 	).Return(
 		&activities.TransformSIPResult{}, nil,
 	)
-	s.env.OnActivity(
-		removefiles.ActivityName,
-		sessionCtx,
-		&removefiles.ActivityParams{
-			Path:           sipPath,
-			RemovePatterns: []*regexp.Regexp{premisRe},
-		},
-	).Return(
-		&removefiles.ActivityResult{}, nil,
-	)
 	s.env.OnActivity(
 		bagit.CreateBagActivityName,
 		sessionCtx,