Merged in feature/fix_test_timeouts (pull request #194)

text extractions tests

* increase timeout

for slow ci/cd systems

* add tests
This commit is contained in:
Jay Brown
2025-12-04 22:45:15 +00:00
parent 3bdc27f4de
commit a94637ab13
14 changed files with 2071 additions and 270 deletions
+503 -2
View File
@@ -251,7 +251,7 @@ func verifyFolderMetrics(t *testing.T, net test.Network, clientID string) {
t.Helper()
// List all folders for the client using the new endpoint
foldersRes, err := net.Client.ListClientFoldersWithResponse(t.Context(), queryapi.ClientID(clientID))
foldersRes, err := net.Client.ListClientFoldersWithResponse(t.Context(), queryapi.ClientID(clientID), nil)
require.NoError(t, err, "Failed to list client folders")
if foldersRes.JSON200 == nil {
t.Logf("Response status: %d, body: %s", foldersRes.StatusCode(), string(foldersRes.Body))
@@ -873,7 +873,7 @@ func TestFolderAPI(t *testing.T) {
// Get the root folder and verify its metrics separately
fmt.Println("\n=== Testing root folder metrics ===")
foldersRes, err := net.Client.ListClientFoldersWithResponse(t.Context(), queryapi.ClientID(clientID))
foldersRes, err := net.Client.ListClientFoldersWithResponse(t.Context(), queryapi.ClientID(clientID), nil)
require.NoError(t, err, "Failed to list client folders")
require.NotNil(t, foldersRes.JSON200, "Expected 200 for folder list")
@@ -912,5 +912,506 @@ func ptr(s string) *string {
return &s
}
// ptrFloat64 returns a pointer to a float64 value.
func ptrFloat64(f float64) *float64 {
return &f
}
// ptrBool returns a pointer to a bool value.
func ptrBool(b bool) *bool {
return &b
}
// TestDeepFolderHierarchy tests folder creation with 10+ levels of nesting
// (Phase 10 - Scenario 4: Folder hierarchy with recursive queries).
func TestDeepFolderHierarchy(t *testing.T) {
cfg := &TextExtractionConfig{}
// Create full network
net, cleanup := test.CreateFullNetwork(t, t.Context(), cfg)
defer cleanup()
// Create test client
clientName := fmt.Sprintf("DeepFolderTestClient_%d", time.Now().UnixNano())
clientExternalID := fmt.Sprintf("deep-fld-%d", time.Now().Unix())
createRes, err := net.Client.CreateClientWithResponse(t.Context(), queryapi.ClientCreate{
Name: clientName,
Id: clientExternalID,
})
require.NoError(t, err)
require.NotNil(t, createRes.JSON201, "Expected 201 response for client creation, got status %d", createRes.StatusCode())
clientID := createRes.JSON201.Id
fmt.Printf("Created client: %s\n", clientID)
// Enable sync
canSync := true
_, err = net.Client.UpdateClientWithResponse(t.Context(), clientID, queryapi.ClientUpdate{
CanSync: &canSync,
})
require.NoError(t, err)
// Create 12 levels of nested folders
fmt.Println("\n=== Creating 12 levels of nested folders ===")
const folderDepth = 12
var parentID *openapi_types.UUID
var folderIDs []openapi_types.UUID
var currentPath string
for i := 1; i <= folderDepth; i++ {
currentPath = fmt.Sprintf("%s/level%d", currentPath, i)
createFolderReq := queryapi.CreateFolderJSONRequestBody{
Path: currentPath,
ClientId: clientID,
CreatedBy: "test@example.com",
}
if parentID != nil {
createFolderReq.ParentId = parentID
}
folderRes, err := net.Client.CreateFolderWithResponse(t.Context(), createFolderReq)
require.NoError(t, err, "Failed to create folder at level %d", i)
require.NotNil(t, folderRes.JSON201, "Expected 201 for folder at level %d, got %d with body: %s",
i, folderRes.StatusCode(), string(folderRes.Body))
folderID := folderRes.JSON201.Id
folderIDs = append(folderIDs, folderID)
parentID = &folderID
fmt.Printf(" Level %d: ID=%s, Path=%s\n", i, folderID, currentPath)
}
assert.Len(t, folderIDs, folderDepth, "Should have created %d folders", folderDepth)
// List all folders and verify count
fmt.Println("\n=== Listing all folders for client ===")
foldersRes, err := net.Client.ListClientFoldersWithResponse(t.Context(), queryapi.ClientID(clientID), nil)
require.NoError(t, err, "Failed to list client folders")
require.NotNil(t, foldersRes.JSON200, "Expected 200 response for folder list")
// Should have at least 12 folders (plus root if created)
fmt.Printf(" Total folders found: %d\n", len(foldersRes.JSON200.Folders))
assert.GreaterOrEqual(t, len(foldersRes.JSON200.Folders), folderDepth,
"Expected at least %d folders", folderDepth)
// Verify we can get metrics for the deepest folder
fmt.Println("\n=== Testing metrics for deepest folder (level 12) ===")
deepestFolderID := folderIDs[len(folderIDs)-1]
metricsRes, err := net.Client.GetFolderMetricsWithResponse(t.Context(), deepestFolderID)
require.NoError(t, err, "Failed to get metrics for deepest folder")
require.NotNil(t, metricsRes.JSON200, "Expected 200 for deepest folder metrics")
fmt.Printf(" Deepest folder metrics: folderId=%s, totalDocuments=%d\n",
metricsRes.JSON200.FolderId, metricsRes.JSON200.TotalDocuments)
assert.Equal(t, deepestFolderID, metricsRes.JSON200.FolderId,
"Metrics should be for the requested folder")
// Verify we can get documents for the deepest folder (should be empty)
fmt.Println("\n=== Testing documents for deepest folder ===")
docsRes, err := net.Client.GetFolderDocumentsWithResponse(t.Context(), deepestFolderID)
require.NoError(t, err, "Failed to get documents for deepest folder")
require.NotNil(t, docsRes.JSON200, "Expected 200 for folder documents")
docCount := 0
if docsRes.JSON200.Documents != nil {
docCount = len(*docsRes.JSON200.Documents)
}
fmt.Printf(" Documents in deepest folder: %d\n", docCount)
fmt.Println("\n=== Deep folder hierarchy test complete ===")
}
// TestLargeArrayFieldExtraction tests field extraction creation with large array sizes (N=100)
// (Phase 10 - Performance test: Large array sizes).
func TestLargeArrayFieldExtraction(t *testing.T) {
cfg := &TextExtractionConfig{}
// Create full network
net, cleanup := test.CreateFullNetwork(t, t.Context(), cfg)
defer cleanup()
// Create test client
clientName := fmt.Sprintf("LargeArrayTestClient_%d", time.Now().UnixNano())
clientExternalID := fmt.Sprintf("lg-array-%d", time.Now().Unix())
createRes, err := net.Client.CreateClientWithResponse(t.Context(), queryapi.ClientCreate{
Name: clientName,
Id: clientExternalID,
})
require.NoError(t, err)
require.NotNil(t, createRes.JSON201, "Expected 201 response for client creation")
clientID := createRes.JSON201.Id
fmt.Printf("Created client: %s\n", clientID)
// Enable sync
canSync := true
_, err = net.Client.UpdateClientWithResponse(t.Context(), clientID, queryapi.ClientUpdate{
CanSync: &canSync,
})
require.NoError(t, err)
// Upload a single document
fmt.Println("\n=== Uploading document ===")
uploadSingleDocument(t, net.Client, clientID, "large_array_test.pdf")
// Wait for document to appear
var documents []queryapi.DocumentSummary
waitForDocumentCount(t, net.Client, clientID, 1, &documents)
require.Len(t, documents, 1, "Expected 1 document")
docID := documents[0].Id
fmt.Printf(" Document uploaded: ID=%s\n", docID)
// Create field extraction with 100 array field items
fmt.Println("\n=== Creating field extraction with 100 array items ===")
const arraySize = 100
arrayFields := make([]queryapi.ArrayFieldItem, arraySize)
for i := 0; i < arraySize; i++ {
arrayFields[i] = queryapi.ArrayFieldItem{
ExhibitTitle: ptr(fmt.Sprintf("Exhibit %d", i+1)),
ExhibitPage: ptr(fmt.Sprintf("%d", i+1)),
ReimbProvTin: ptr(fmt.Sprintf("%09d", i+100000000)),
ReimbProvNpi: ptr(fmt.Sprintf("%010d", i+1000000000)),
ReimbProvName: ptr(fmt.Sprintf("Provider %d", i+1)),
AareteDerivedClaimTypeCd: ptr("INPATIENT"),
ReimbPctRate: ptrFloat64(float64(80+i%20) + float64(i%100)/100.0),
ReimbFeeRate: ptrFloat64(float64(100 + i*10)),
CarveoutInd: ptrBool(i%2 == 0),
DefaultInd: ptrBool(i == 0),
}
}
startTime := time.Now()
createFieldExtRes, err := net.Client.CreateFieldExtractionWithResponse(t.Context(), queryapi.CreateFieldExtractionJSONRequestBody{
DocumentId: docID,
SingleFields: queryapi.SingleFields{
FileName: ptr("large_array_test.pdf"),
ContractTitle: ptr("Large Array Performance Test Contract"),
ClientName: ptr("Performance Test Client"),
},
ArrayFields: arrayFields,
CreatedBy: openapi_types.Email("test@example.com"),
})
createDuration := time.Since(startTime)
require.NoError(t, err, "Failed to create field extraction")
if createFieldExtRes.JSON400 != nil {
t.Fatalf("Got 400 error creating field extraction: %s", createFieldExtRes.JSON400.Message)
}
require.NotNil(t, createFieldExtRes.JSON201, "Expected 201 for field extraction, got %d with body: %s",
createFieldExtRes.StatusCode(), string(createFieldExtRes.Body))
fmt.Printf(" Field extraction created in %v\n", createDuration)
fmt.Printf(" Version: %d\n", createFieldExtRes.JSON201.Version)
// Verify the current field extraction
fmt.Println("\n=== Verifying current field extraction ===")
getFieldExtRes, err := net.Client.GetCurrentFieldExtractionWithResponse(t.Context(), &queryapi.GetCurrentFieldExtractionParams{
DocumentId: docID,
})
require.NoError(t, err, "Failed to get current field extraction")
require.NotNil(t, getFieldExtRes.JSON200, "Expected 200 for get field extraction")
// Verify array field count
actualArraySize := len(getFieldExtRes.JSON200.ArrayFields)
fmt.Printf(" Array fields returned: %d\n", actualArraySize)
assert.Equal(t, arraySize, actualArraySize, "Expected %d array fields", arraySize)
// Verify single fields
assert.NotNil(t, getFieldExtRes.JSON200.SingleFields.ContractTitle)
if getFieldExtRes.JSON200.SingleFields.ContractTitle != nil {
assert.Equal(t, "Large Array Performance Test Contract", *getFieldExtRes.JSON200.SingleFields.ContractTitle)
}
// Performance assertion: should complete reasonably quickly (under 30 seconds)
assert.Less(t, createDuration, 30*time.Second,
"Field extraction with 100 array items should complete in under 30 seconds")
fmt.Println("\n=== Large array field extraction test complete ===")
}
// TestLabelReapplicationHistory tests applying the same label multiple times to track history
// (Phase 10 - Performance test: Many labels per document history).
func TestLabelReapplicationHistory(t *testing.T) {
cfg := &TextExtractionConfig{}
// Create full network
net, cleanup := test.CreateFullNetwork(t, t.Context(), cfg)
defer cleanup()
// Create test client
clientName := fmt.Sprintf("LabelHistoryTestClient_%d", time.Now().UnixNano())
clientExternalID := fmt.Sprintf("lbl-hist-%d", time.Now().Unix())
createRes, err := net.Client.CreateClientWithResponse(t.Context(), queryapi.ClientCreate{
Name: clientName,
Id: clientExternalID,
})
require.NoError(t, err)
require.NotNil(t, createRes.JSON201, "Expected 201 response for client creation")
clientID := createRes.JSON201.Id
fmt.Printf("Created client: %s\n", clientID)
// Enable sync
canSync := true
_, err = net.Client.UpdateClientWithResponse(t.Context(), clientID, queryapi.ClientUpdate{
CanSync: &canSync,
})
require.NoError(t, err)
// Upload a single document
fmt.Println("\n=== Uploading document ===")
uploadSingleDocument(t, net.Client, clientID, "label_history_test.pdf")
// Wait for document to appear
var documents []queryapi.DocumentSummary
waitForDocumentCount(t, net.Client, clientID, 1, &documents)
require.Len(t, documents, 1, "Expected 1 document")
docID := documents[0].Id
fmt.Printf(" Document uploaded: ID=%s\n", docID)
// Apply each of the 5 available labels multiple times to build history
// Note: The system has 5 predefined labels: Ingested, OCR_Processed, GenAI_Processed, Doczy_AI_Completed, Dashboard_Ready
labels := []string{"Ingested", "OCR_Processed", "GenAI_Processed", "Doczy_AI_Completed", "Dashboard_Ready"}
const applicationsPerLabel = 4
fmt.Printf("\n=== Applying each label %d times (total %d label applications) ===\n",
applicationsPerLabel, len(labels)*applicationsPerLabel)
totalApplications := 0
for round := 1; round <= applicationsPerLabel; round++ {
for _, label := range labels {
applyLabelRes, err := net.Client.ApplyLabelWithResponse(t.Context(), docID, queryapi.ApplyLabelJSONRequestBody{
Label: label,
AppliedBy: openapi_types.Email(fmt.Sprintf("user%d@example.com", round)),
})
require.NoError(t, err, "Failed to apply label %s in round %d", label, round)
if applyLabelRes.JSON201 != nil {
totalApplications++
if round == 1 || round == applicationsPerLabel {
fmt.Printf(" Round %d: Applied '%s'\n", round, label)
}
} else if applyLabelRes.JSON400 != nil {
// This might happen if re-application isn't allowed - that's informative
t.Logf(" Round %d: Label '%s' returned 400: %s", round, label, applyLabelRes.JSON400.Message)
}
}
if round == 1 {
fmt.Println(" ...")
}
}
fmt.Printf(" Total successful label applications: %d\n", totalApplications)
// Get all labels for the document and verify history
fmt.Println("\n=== Verifying label history ===")
getLabelsRes, err := net.Client.GetDocumentLabelsWithResponse(t.Context(), docID)
require.NoError(t, err, "Failed to get document labels")
require.NotNil(t, getLabelsRes.JSON200, "Expected 200 for get labels")
labelCount := 0
if getLabelsRes.JSON200.Labels != nil {
labelCount = len(*getLabelsRes.JSON200.Labels)
}
fmt.Printf(" Total label entries in history: %d\n", labelCount)
// If re-application is allowed, we should have multiple entries
// If not, we should have exactly 5 (one per unique label)
if labelCount == len(labels) {
fmt.Println(" Note: System appears to keep only most recent application per label")
} else if labelCount == totalApplications {
fmt.Println(" Note: System preserves full history of all label applications")
}
// The key test: verify we have at least one entry per label
assert.GreaterOrEqual(t, labelCount, len(labels),
"Should have at least one entry per label type")
// Verify we can query documents by each label
fmt.Println("\n=== Verifying query by label ===")
for _, label := range labels {
getDocsByLabelRes, err := net.Client.GetDocumentsByLabelWithResponse(t.Context(), label, &queryapi.GetDocumentsByLabelParams{
ClientId: queryapi.ClientID(clientID),
})
require.NoError(t, err, "Failed to get documents by label %s", label)
require.NotNil(t, getDocsByLabelRes.JSON200, "Expected 200 for get documents by label %s", label)
docCount := 0
if getDocsByLabelRes.JSON200.Documents != nil {
docCount = len(*getDocsByLabelRes.JSON200.Documents)
}
assert.Equal(t, 1, docCount, "Expected 1 document with label '%s'", label)
}
fmt.Println(" All labels return the document correctly")
fmt.Println("\n=== Label reapplication history test complete ===")
}
// TestFieldExtractionByFolderAndLabel tests Scenario 2: Query field extractions by folder and label
// (Phase 10 - Scenario 2: Cross-feature integration).
// This test uses a simpler approach - single document upload to test cross-feature integration.
func TestFieldExtractionByFolderAndLabel(t *testing.T) {
cfg := &TextExtractionConfig{}
// Create full network
net, cleanup := test.CreateFullNetwork(t, t.Context(), cfg)
defer cleanup()
// Create test client
clientName := fmt.Sprintf("FolderLabelFETestClient_%d", time.Now().UnixNano())
clientExternalID := fmt.Sprintf("fld-lbl-fe-%d", time.Now().Unix())
createRes, err := net.Client.CreateClientWithResponse(t.Context(), queryapi.ClientCreate{
Name: clientName,
Id: clientExternalID,
})
require.NoError(t, err)
require.NotNil(t, createRes.JSON201, "Expected 201 response for client creation, got status: %d, body: %s",
createRes.StatusCode(), string(createRes.Body))
clientID := createRes.JSON201.Id
fmt.Printf("Created client: %s\n", clientID)
// Enable sync
canSync := true
_, err = net.Client.UpdateClientWithResponse(t.Context(), clientID, queryapi.ClientUpdate{
CanSync: &canSync,
})
require.NoError(t, err)
// Create two folders
fmt.Println("\n=== Creating folders ===")
folder1Res, err := net.Client.CreateFolderWithResponse(t.Context(), queryapi.CreateFolderJSONRequestBody{
Path: "/contracts",
ClientId: clientID,
CreatedBy: "test@example.com",
})
require.NoError(t, err, "Failed to create folder1")
require.NotNil(t, folder1Res.JSON201, "Expected 201 for folder1")
folder1ID := folder1Res.JSON201.Id
fmt.Printf(" Created /contracts folder: ID=%s\n", folder1ID)
folder2Res, err := net.Client.CreateFolderWithResponse(t.Context(), queryapi.CreateFolderJSONRequestBody{
Path: "/invoices",
ClientId: clientID,
CreatedBy: "test@example.com",
})
require.NoError(t, err, "Failed to create folder2")
require.NotNil(t, folder2Res.JSON201, "Expected 201 for folder2")
folder2ID := folder2Res.JSON201.Id
fmt.Printf(" Created /invoices folder: ID=%s\n", folder2ID)
// Upload single document and wait for it
fmt.Println("\n=== Uploading document ===")
uploadSingleDocument(t, net.Client, clientID, "contract_doc.pdf")
var documents []queryapi.DocumentSummary
waitForDocumentCount(t, net.Client, clientID, 1, &documents)
require.Len(t, documents, 1, "Expected 1 document")
docID := documents[0].Id
fmt.Printf(" Uploaded document: ID=%s\n", docID)
// Apply labels to document
fmt.Println("\n=== Applying labels ===")
for _, label := range []string{"Ingested", "OCR_Processed", "Dashboard_Ready"} {
_, err := net.Client.ApplyLabelWithResponse(t.Context(), docID, queryapi.ApplyLabelJSONRequestBody{
Label: label,
AppliedBy: openapi_types.Email("test@example.com"),
})
require.NoError(t, err, "Failed to apply label %s", label)
}
fmt.Printf(" Applied Ingested, OCR_Processed, Dashboard_Ready\n")
// Create field extraction
fmt.Println("\n=== Creating field extraction ===")
feRes, err := net.Client.CreateFieldExtractionWithResponse(t.Context(), queryapi.CreateFieldExtractionJSONRequestBody{
DocumentId: docID,
SingleFields: queryapi.SingleFields{
FileName: ptr("contract_doc.pdf"),
ContractTitle: ptr("Service Agreement"),
ClientName: ptr("Test Client"),
},
ArrayFields: []queryapi.ArrayFieldItem{
{
ExhibitTitle: ptr("Exhibit A"),
ReimbPctRate: ptrFloat64(85.0),
ReimbFeeRate: ptrFloat64(150.0),
},
},
CreatedBy: openapi_types.Email("test@example.com"),
})
require.NoError(t, err, "Failed to create field extraction")
require.NotNil(t, feRes.JSON201, "Expected 201 for field extraction")
fmt.Printf(" Field extraction created: version=%d\n", feRes.JSON201.Version)
// Query by label and verify field extraction exists
fmt.Println("\n=== Querying documents by label and verifying field extractions ===")
// Get documents with Dashboard_Ready label
dashboardReadyDocsRes, err := net.Client.GetDocumentsByLabelWithResponse(t.Context(), "Dashboard_Ready", &queryapi.GetDocumentsByLabelParams{
ClientId: queryapi.ClientID(clientID),
})
require.NoError(t, err, "Failed to get documents by Dashboard_Ready label")
require.NotNil(t, dashboardReadyDocsRes.JSON200, "Expected 200 for get documents by label")
dashboardReadyDocs := dashboardReadyDocsRes.JSON200.Documents
require.NotNil(t, dashboardReadyDocs, "Expected documents list")
assert.Len(t, *dashboardReadyDocs, 1, "Expected 1 document with Dashboard_Ready label")
fmt.Printf(" Documents with Dashboard_Ready: %d\n", len(*dashboardReadyDocs))
// Verify that document has a field extraction
for _, doc := range *dashboardReadyDocs {
feGetRes, err := net.Client.GetCurrentFieldExtractionWithResponse(t.Context(), &queryapi.GetCurrentFieldExtractionParams{
DocumentId: doc.Id,
})
require.NoError(t, err, "Failed to get field extraction for document %s", doc.Id)
require.NotNil(t, feGetRes.JSON200, "Expected field extraction for Dashboard_Ready document")
assert.NotNil(t, feGetRes.JSON200.SingleFields.ContractTitle, "Expected contract title")
fmt.Printf(" Document %s has field extraction with title: %s\n",
doc.Id, *feGetRes.JSON200.SingleFields.ContractTitle)
}
// Get documents with OCR_Processed label
ocrDocsRes, err := net.Client.GetDocumentsByLabelWithResponse(t.Context(), "OCR_Processed", &queryapi.GetDocumentsByLabelParams{
ClientId: queryapi.ClientID(clientID),
})
require.NoError(t, err, "Failed to get documents by OCR_Processed label")
require.NotNil(t, ocrDocsRes.JSON200, "Expected 200 for get documents by label")
ocrDocs := ocrDocsRes.JSON200.Documents
require.NotNil(t, ocrDocs, "Expected documents list")
assert.Len(t, *ocrDocs, 1, "Expected 1 document with OCR_Processed label")
fmt.Printf(" Documents with OCR_Processed: %d\n", len(*ocrDocs))
// Verify folder metrics
fmt.Println("\n=== Verifying folder metrics ===")
metrics1Res, err := net.Client.GetFolderMetricsWithResponse(t.Context(), folder1ID)
require.NoError(t, err, "Failed to get folder1 metrics")
require.NotNil(t, metrics1Res.JSON200, "Expected 200 for folder1 metrics")
fmt.Printf(" /contracts folder metrics: totalDocs=%d\n", metrics1Res.JSON200.TotalDocuments)
metrics2Res, err := net.Client.GetFolderMetricsWithResponse(t.Context(), folder2ID)
require.NoError(t, err, "Failed to get folder2 metrics")
require.NotNil(t, metrics2Res.JSON200, "Expected 200 for folder2 metrics")
fmt.Printf(" /invoices folder metrics: totalDocs=%d\n", metrics2Res.JSON200.TotalDocuments)
// Test: Get all documents with Ingested label
fmt.Println("\n=== Verifying Ingested documents ===")
allIngestedRes, err := net.Client.GetDocumentsByLabelWithResponse(t.Context(), "Ingested", &queryapi.GetDocumentsByLabelParams{
ClientId: queryapi.ClientID(clientID),
})
require.NoError(t, err, "Failed to get documents by Ingested label")
require.NotNil(t, allIngestedRes.JSON200, "Expected 200 for get documents by label")
ingestedDocs := allIngestedRes.JSON200.Documents
require.NotNil(t, ingestedDocs, "Expected documents list")
assert.Len(t, *ingestedDocs, 1, "Expected 1 document with Ingested label")
fmt.Printf(" Documents with Ingested label: %d\n", len(*ingestedDocs))
// Verify field extraction on the ingested document
fmt.Println("\n=== Checking field extraction on Ingested document ===")
for _, doc := range *ingestedDocs {
feGetRes, err := net.Client.GetCurrentFieldExtractionWithResponse(t.Context(), &queryapi.GetCurrentFieldExtractionParams{
DocumentId: doc.Id,
})
require.NoError(t, err, "Failed to query field extraction for document %s", doc.Id)
require.NotNil(t, feGetRes.JSON200, "Expected field extraction for Ingested document")
fmt.Printf(" Document %s has field extraction\n", doc.Id)
}
fmt.Println("\n=== Field extraction by folder and label test complete ===")
}
// Ensure io import is used
var _ = io.EOF