Merged in feature/batch-status (pull request #215)

Implement and test the batch status feature * working
2026-03-12 18:53:42 +00:00
parent 8a4029058b
commit aafe7d5b5f
39 changed files with 4252 additions and 516 deletions
@@ -6,7 +6,10 @@ import (
 	"bytes"
 	"fmt"
 	"testing"
+	"time"

+	"queryorchestration/internal/database/repository"
+	"queryorchestration/internal/serviceconfig/objectstore"
 	"queryorchestration/internal/test"

 	"github.com/aws/aws-sdk-go-v2/aws"
@@ -105,3 +108,130 @@ func TestMeasureFileSize_Integration_NonExistentKey(t *testing.T) {
 	assert.Error(t, err, "measureFileSize should return an error for non-existent key")
 	assert.Contains(t, err.Error(), "HeadObject failed", "error should indicate HeadObject failure")
 }
+
+// TestDocumentInitCreate_ReturnsCreateResult verifies that Create() returns a
+// *CreateResult with correct IsDuplicate, BatchID, and Filename fields for both
+// new and duplicate documents.
+func TestDocumentInitCreate_ReturnsCreateResult(t *testing.T) {
+	ctx := t.Context()
+
+	// Set up DB, S3, and SQS
+	cfg := &DocInitConfig{}
+	test.CreateDB(t, cfg)
+	acfg := test.CreateAWSContainer(t, cfg)
+
+	test.SetQueueClient(t, ctx, cfg, acfg.ExternalEndpoint)
+	cfg.DocumentSyncURL = test.CreateQueue(t, cfg, test.DocSyncRunnerName)
+
+	test.SetStoreClient(t, ctx, cfg, acfg.ExternalEndpoint)
+	test.CreateBucket(t, cfg)
+
+	svc := New(cfg)
+
+	clientID := "create_result_" + uuid.New().String()[:8]
+	err := cfg.GetDBQueries().CreateClient(ctx, &repository.CreateClientParams{
+		Clientid: clientID,
+		Name:     "Test CreateResult",
+	})
+	require.NoError(t, err)
+
+	// Create a batch
+	batchID, err := cfg.GetDBQueries().CreateBatchUpload(ctx, &repository.CreateBatchUploadParams{
+		ClientID:         clientID,
+		OriginalFilename: "test.zip",
+		TotalDocuments:   2,
+	})
+	require.NoError(t, err)
+
+	// Upload a file to S3
+	location := objectstore.BucketKey{
+		Location:  objectstore.Import,
+		ClientID:  clientID,
+		CreatedAt: time.Now().UTC(),
+		EntityID:  uuid.New(),
+		BatchID:   &batchID,
+	}
+	testContent := []byte("test content for create result")
+	_, err = cfg.GetStoreClient().PutObject(ctx, &s3.PutObjectInput{
+		Bucket: aws.String(cfg.GetBucket()),
+		Key:    aws.String(location.String()),
+		Body:   bytes.NewReader(testContent),
+	})
+	require.NoError(t, err)
+
+	filename := "invoice.pdf"
+	uniqueHash := "create_result_hash_" + uuid.New().String()[:8]
+
+	// Test 1: New document returns IsDuplicate=false with correct fields
+	result, err := svc.Create(ctx, &Create{
+		Bucket:   cfg.GetBucket(),
+		Key:      location,
+		Hash:     uniqueHash,
+		Filename: &filename,
+	})
+	require.NoError(t, err)
+	require.NotNil(t, result)
+
+	assert.NotEqual(t, uuid.Nil, result.ID, "new document should have a valid ID")
+	assert.False(t, result.IsDuplicate, "new document should not be a duplicate")
+	assert.NotNil(t, result.BatchID, "batch document should have BatchID set")
+	assert.Equal(t, batchID, *result.BatchID, "BatchID should match the batch")
+	assert.NotNil(t, result.Filename, "Filename should be set")
+	assert.Equal(t, filename, *result.Filename, "Filename should match")
+
+	// Test 2: Duplicate document (same hash) returns IsDuplicate=true
+	location2 := objectstore.BucketKey{
+		Location:  objectstore.Import,
+		ClientID:  clientID,
+		CreatedAt: time.Now().UTC(),
+		EntityID:  uuid.New(),
+		BatchID:   &batchID,
+	}
+	_, err = cfg.GetStoreClient().PutObject(ctx, &s3.PutObjectInput{
+		Bucket: aws.String(cfg.GetBucket()),
+		Key:    aws.String(location2.String()),
+		Body:   bytes.NewReader(testContent),
+	})
+	require.NoError(t, err)
+
+	dupResult, err := svc.Create(ctx, &Create{
+		Bucket:   cfg.GetBucket(),
+		Key:      location2,
+		Hash:     uniqueHash, // Same hash triggers duplicate
+		Filename: &filename,
+	})
+	require.NoError(t, err)
+	require.NotNil(t, dupResult)
+
+	assert.Equal(t, result.ID, dupResult.ID, "duplicate should reference the original document ID")
+	assert.True(t, dupResult.IsDuplicate, "second document with same hash should be duplicate")
+	assert.NotNil(t, dupResult.BatchID, "duplicate should still have BatchID")
+	assert.Equal(t, batchID, *dupResult.BatchID, "duplicate BatchID should match")
+
+	// Test 3: Non-batch document has nil BatchID
+	location3 := objectstore.BucketKey{
+		Location:  objectstore.Import,
+		ClientID:  clientID,
+		CreatedAt: time.Now().UTC(),
+		EntityID:  uuid.New(),
+		// No BatchID
+	}
+	_, err = cfg.GetStoreClient().PutObject(ctx, &s3.PutObjectInput{
+		Bucket: aws.String(cfg.GetBucket()),
+		Key:    aws.String(location3.String()),
+		Body:   bytes.NewReader(testContent),
+	})
+	require.NoError(t, err)
+
+	nonBatchResult, err := svc.Create(ctx, &Create{
+		Bucket: cfg.GetBucket(),
+		Key:    location3,
+		Hash:   "nonbatch_hash_" + uuid.New().String()[:8],
+	})
+	require.NoError(t, err)
+	require.NotNil(t, nonBatchResult)
+
+	assert.False(t, nonBatchResult.IsDuplicate, "non-batch new document should not be duplicate")
+	assert.Nil(t, nonBatchResult.BatchID, "non-batch document should have nil BatchID")
+	assert.Nil(t, nonBatchResult.Filename, "non-batch document should have nil Filename")
+}