query-orchestration/api/queryAPI/custommetadata_integration_test.go

// Package queryapi_test — Milestone 2 §2.8 end-to-end integration tests.
//
// These tests compose the Milestone 2 HTTP handlers, the customschema
// service, the fieldextraction service, and the real Postgres testcontainer
// produced by internal/test.CreateDB to prove the mutual-exclusivity
// invariant, the full happy-path round-trip, and the edge cases the plan
// calls out in §2.8.
//
// No mocks. Every request goes through a real handler receiver. The
// helpers below reuse the Milestone 1 testutils (newCustomSchemaContext,
// seedCustomSchema, resetClientForSchemaTests) so the two files share the
// same fixture shape.
package queryapi_test

import (
	"bytes"
	"encoding/json"
	"errors"
	"net/http"
	"net/http/httptest"
	"testing"

	queryapi "queryorchestration/api/queryAPI"
	"queryorchestration/internal/cognitoauth"
	"queryorchestration/internal/database/repository"
	"queryorchestration/internal/fieldextraction"
	"queryorchestration/internal/test"

	"github.com/google/uuid"
	"github.com/labstack/echo/v4"
	"github.com/oapi-codegen/nullable"
	openapi_types "github.com/oapi-codegen/runtime/types"
	"github.com/stretchr/testify/assert"
	"github.com/stretchr/testify/require"
)

// newCustomMetadataContext is the Milestone 2 sibling of
// newCustomSchemaContext — it exists only so the tests in this file can
// read clearly at a glance; the implementation is identical.
func newCustomMetadataContext(t testing.TB, method string, body interface{}) (echo.Context, *httptest.ResponseRecorder) {
	t.Helper()
	e := echo.New()
	var reqBody []byte
	if body != nil {
		var err error
		reqBody, err = json.Marshal(body)
		require.NoError(t, err)
	}
	req := httptest.NewRequest(method, "/", bytes.NewReader(reqBody))
	if reqBody != nil {
		req.Header.Set(echo.HeaderContentType, echo.MIMEApplicationJSON)
	}
	rec := httptest.NewRecorder()
	ctx := e.NewContext(req, rec)
	ctx.Set("user_claims", map[string]interface{}{"sub": testActorSubject})
	ctx.Set("user_info", cognitoauth.UserInfo{Username: "admin"})
	return ctx, rec
}

// setupM2Fixture boots a fresh DB, wipes any rows belonging to clientID,
// creates the client and returns the controllers. Mirrors the Milestone 1
// pattern but in one helper so the §2.8 tests stay short.
func setupM2Fixture(t *testing.T, clientID string) (*queryapi.Controllers, *ControllerConfig) {
	t.Helper()
	cfg := &ControllerConfig{}
	test.CreateDB(t, cfg)
	initializeTestConfig(t, cfg)
	svc := createControllerServices(cfg)
	cons := queryapi.NewControllers(svc, cfg)

	resetClientForSchemaTests(t, cfg, clientID)
	test.CreateTestClient(t, cfg, clientID, "M2 integration "+clientID)
	return cons, cfg
}

// seedDoc creates a document for clientID with a unique hash.
func seedDoc(t *testing.T, cfg *ControllerConfig, clientID, hash string) uuid.UUID {
	t.Helper()
	id, err := cfg.GetDBQueries().CreateDocument(t.Context(), &repository.CreateDocumentParams{
		Clientid: clientID,
		Hash:     hash,
	})
	require.NoError(t, err)
	return id
}

// assignSchemaForTest wraps AssignDocumentSchema against a freshly built
// echo context so the §2.8 tests can bind a document in one line.
func assignSchemaForTest(t *testing.T, cons *queryapi.Controllers, docID, schemaID uuid.UUID) {
	t.Helper()
	body := queryapi.DocumentSchemaAssignRequest{
		CustomSchemaId: nullable.NewNullableWithValue(openapi_types.UUID(schemaID)),
	}
	ctx, rec := newCustomMetadataContext(t, http.MethodPatch, body)
	require.NoError(t, cons.AssignDocumentSchema(ctx, openapi_types.UUID(docID)))
	require.Equal(t, http.StatusOK, rec.Code, "assign must return 200")
}

// ---------- end-to-end happy round-trip ----------

// TestM2_HappyRoundTrip covers the §2.8 full round-trip:
// create schema → assign to document → POST valid metadata → GET returns
// it with populated schemaId / schemaName / schemaVersion / createdBy.
func TestM2_HappyRoundTrip(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_HAPPY"
	cons, cfg := setupM2Fixture(t, clientID)

	schema := seedCustomSchema(t, cons, clientID, "m2-happy")
	docID := seedDoc(t, cfg, clientID, "m2-happy-doc")
	assignSchemaForTest(t, cons, docID, uuid.UUID(schema.Id))

	// POST valid metadata.
	postBody := queryapi.CustomMetadataRequest{
		DocumentId: openapi_types.UUID(docID),
		Metadata:   map[string]interface{}{"name": "contract-42"},
	}
	ctxPost, recPost := newCustomMetadataContext(t, http.MethodPost, postBody)
	require.NoError(t, cons.SetCustomMetadata(ctxPost))
	require.Equal(t, http.StatusCreated, recPost.Code)

	var posted queryapi.CustomMetadataResponse
	require.NoError(t, json.Unmarshal(recPost.Body.Bytes(), &posted))
	assert.Equal(t, int32(1), posted.Version)
	assert.Equal(t, testActorSubject, posted.CreatedBy)

	// GET /custom-metadata?documentId=...
	ctxGet, recGet := newCustomMetadataContext(t, http.MethodGet, nil)
	err := cons.GetCurrentCustomMetadata(ctxGet, queryapi.GetCurrentCustomMetadataParams{
		DocumentId: openapi_types.UUID(docID),
	})
	require.NoError(t, err)
	require.Equal(t, http.StatusOK, recGet.Code)

	var got queryapi.CustomMetadataResponse
	require.NoError(t, json.Unmarshal(recGet.Body.Bytes(), &got))
	assert.Equal(t, posted.Version, got.Version)
	assert.Equal(t, posted.Id, got.Id)
	require.True(t, got.SchemaId.IsSpecified() && !got.SchemaId.IsNull())
	sid, _ := got.SchemaId.Get()
	assert.Equal(t, schema.Id, sid)
	schemaName, _ := got.SchemaName.Get()
	assert.Equal(t, "m2-happy", schemaName)
	schemaVersion, _ := got.SchemaVersion.Get()
	assert.Equal(t, int32(1), schemaVersion)
}

// TestM2_PostWithoutSchemaReturns400 proves POST /custom-metadata refuses
// writes to documents that have no schema binding.
func TestM2_PostWithoutSchemaReturns400(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_NOSCH"
	cons, cfg := setupM2Fixture(t, clientID)

	docID := seedDoc(t, cfg, clientID, "m2-nosch-doc")

	postBody := queryapi.CustomMetadataRequest{
		DocumentId: openapi_types.UUID(docID),
		Metadata:   map[string]interface{}{"name": "nope"},
	}
	ctxPost, recPost := newCustomMetadataContext(t, http.MethodPost, postBody)
	err := cons.SetCustomMetadata(ctxPost)
	// The handler propagates echo.HTTPError for mapped sentinels; the
	// real HTTP server flushes this as a response. For the handler unit
	// test we inspect the error directly.
	requireHTTPStatus(t, err, recPost, http.StatusBadRequest)
}

// TestM2_LegacyWriteOnCustomDoc covers the §2.8 mutual-exclusivity end:
// schema bound → legacy field extraction write → 409.
func TestM2_LegacyWriteOnCustomDoc(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_LEGACY"
	cons, cfg := setupM2Fixture(t, clientID)

	schema := seedCustomSchema(t, cons, clientID, "m2-legacy")
	docID := seedDoc(t, cfg, clientID, "m2-legacy-doc")
	assignSchemaForTest(t, cons, docID, uuid.UUID(schema.Id))

	// Attempt legacy write via the fieldextraction service directly
	// (the REST handler would do the same mapping — we test service +
	// handler mapping separately in the fieldextraction package).
	svc := createControllerServices(cfg)
	_, err := svc.FieldExtraction.CreateFieldExtraction(t.Context(), &fieldextraction.CreateFieldExtractionInput{
		DocumentID: docID,
		SingleFields: &repository.AddFieldExtractionParams{
			Documentid: docID,
			Createdby:  "tester",
		},
		ArrayFields: nil,
	})
	require.Error(t, err)
	assert.ErrorIs(t, err, fieldextraction.ErrMutualExclusivityViolation)
}

// TestM2_GetByVersion covers §2.8 GET /custom-metadata/version happy
// path: post 3 versions, retrieve v2 exactly.
func TestM2_GetByVersion(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_VERSION"
	cons, cfg := setupM2Fixture(t, clientID)

	schema := seedCustomSchema(t, cons, clientID, "m2-version")
	docID := seedDoc(t, cfg, clientID, "m2-version-doc")
	assignSchemaForTest(t, cons, docID, uuid.UUID(schema.Id))

	for _, payload := range []string{`{"name":"v1"}`, `{"name":"v2"}`, `{"name":"v3"}`} {
		var m map[string]interface{}
		require.NoError(t, json.Unmarshal([]byte(payload), &m))
		body := queryapi.CustomMetadataRequest{
			DocumentId: openapi_types.UUID(docID),
			Metadata:   m,
		}
		ctx, rec := newCustomMetadataContext(t, http.MethodPost, body)
		require.NoError(t, cons.SetCustomMetadata(ctx))
		require.Equal(t, http.StatusCreated, rec.Code)
	}

	ctxVer, recVer := newCustomMetadataContext(t, http.MethodGet, nil)
	err := cons.GetCustomMetadataByVersion(ctxVer, queryapi.GetCustomMetadataByVersionParams{
		DocumentId: openapi_types.UUID(docID),
		Version:    2,
	})
	require.NoError(t, err)
	require.Equal(t, http.StatusOK, recVer.Code)

	var got queryapi.CustomMetadataResponse
	require.NoError(t, json.Unmarshal(recVer.Body.Bytes(), &got))
	assert.Equal(t, int32(2), got.Version)
	// metadata must be the v2 payload, not v1 or v3
	v, ok := got.Metadata["name"]
	require.True(t, ok)
	assert.Equal(t, "v2", v)
}

// TestM2_History covers the §2.8 history GET happy path with pagination.
func TestM2_History(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_HIST"
	cons, cfg := setupM2Fixture(t, clientID)

	schema := seedCustomSchema(t, cons, clientID, "m2-hist")
	docID := seedDoc(t, cfg, clientID, "m2-hist-doc")
	assignSchemaForTest(t, cons, docID, uuid.UUID(schema.Id))

	for i := 0; i < 5; i++ {
		body := queryapi.CustomMetadataRequest{
			DocumentId: openapi_types.UUID(docID),
			Metadata:   map[string]interface{}{"name": "x"},
		}
		ctx, rec := newCustomMetadataContext(t, http.MethodPost, body)
		require.NoError(t, cons.SetCustomMetadata(ctx))
		require.Equal(t, http.StatusCreated, rec.Code)
	}

	limit := int32(2)
	offset := int32(0)
	ctx1, rec1 := newCustomMetadataContext(t, http.MethodGet, nil)
	require.NoError(t, cons.GetCustomMetadataHistory(ctx1, queryapi.GetCustomMetadataHistoryParams{
		DocumentId: openapi_types.UUID(docID),
		Limit:      &limit,
		Offset:     &offset,
	}))
	require.Equal(t, http.StatusOK, rec1.Code)
	var page1 queryapi.CustomMetadataHistoryResponse
	require.NoError(t, json.Unmarshal(rec1.Body.Bytes(), &page1))
	require.Len(t, page1.Versions, 2)
	assert.Equal(t, int32(5), page1.Versions[0].Version)
	assert.Equal(t, int32(4), page1.Versions[1].Version)

	offset2 := int32(2)
	ctx2, rec2 := newCustomMetadataContext(t, http.MethodGet, nil)
	require.NoError(t, cons.GetCustomMetadataHistory(ctx2, queryapi.GetCustomMetadataHistoryParams{
		DocumentId: openapi_types.UUID(docID),
		Limit:      &limit,
		Offset:     &offset2,
	}))
	var page2 queryapi.CustomMetadataHistoryResponse
	require.NoError(t, json.Unmarshal(rec2.Body.Bytes(), &page2))
	require.Len(t, page2.Versions, 2)
	assert.Equal(t, int32(3), page2.Versions[0].Version)
	assert.Equal(t, int32(2), page2.Versions[1].Version)
}

// TestM2_AssignToDocWithLegacyExtractionsReturns409 covers the reverse
// direction of the mutual-exclusivity invariant: a doc with legacy
// extractions cannot be bound to a schema.
func TestM2_AssignToDocWithLegacyExtractionsReturns409(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_LEGACYASSIGN"
	cons, cfg := setupM2Fixture(t, clientID)

	// Create a document and seed a legacy extraction via the service.
	svc := createControllerServices(cfg)
	docID := seedDoc(t, cfg, clientID, "m2-legacy-assign-doc")
	_, err := svc.FieldExtraction.CreateFieldExtraction(t.Context(), &fieldextraction.CreateFieldExtractionInput{
		DocumentID: docID,
		SingleFields: &repository.AddFieldExtractionParams{
			Documentid: docID,
			Createdby:  "tester",
		},
		ArrayFields: nil,
	})
	require.NoError(t, err)

	schema := seedCustomSchema(t, cons, clientID, "m2-legacy-assign")

	// Attempt to bind — must 409.
	body := queryapi.DocumentSchemaAssignRequest{
		CustomSchemaId: nullable.NewNullableWithValue(schema.Id),
	}
	ctx, rec := newCustomMetadataContext(t, http.MethodPatch, body)
	err = cons.AssignDocumentSchema(ctx, openapi_types.UUID(docID))
	requireHTTPStatus(t, err, rec, http.StatusConflict)
}

// TestM2_FKCascadeDeletesMetadata proves the FK cascade from documents
// removes document_custom_metadata rows on document delete, covering the
// §2.8 "FK-cascade delete" row.
func TestM2_FKCascadeDeletesMetadata(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_FKCASCADE"
	cons, cfg := setupM2Fixture(t, clientID)

	schema := seedCustomSchema(t, cons, clientID, "m2-fkcascade")
	docID := seedDoc(t, cfg, clientID, "m2-fkcascade-doc")
	assignSchemaForTest(t, cons, docID, uuid.UUID(schema.Id))

	// Write two metadata versions via the handler.
	for i := 0; i < 2; i++ {
		body := queryapi.CustomMetadataRequest{
			DocumentId: openapi_types.UUID(docID),
			Metadata:   map[string]interface{}{"name": "x"},
		}
		ctx, rec := newCustomMetadataContext(t, http.MethodPost, body)
		require.NoError(t, cons.SetCustomMetadata(ctx))
		require.Equal(t, http.StatusCreated, rec.Code)
	}

	var before int
	require.NoError(t, cfg.GetDBPool().QueryRow(t.Context(),
		`SELECT COUNT(*) FROM document_custom_metadata WHERE document_id = $1`,
		docID).Scan(&before))
	require.Equal(t, 2, before)

	// Delete the document row directly (the feature relies on the FK
	// cascade, not on an application-level delete path).
	_, err := cfg.GetDBPool().Exec(t.Context(),
		`DELETE FROM documents WHERE id = $1`, docID)
	require.NoError(t, err)

	var after int
	require.NoError(t, cfg.GetDBPool().QueryRow(t.Context(),
		`SELECT COUNT(*) FROM document_custom_metadata WHERE document_id = $1`,
		docID).Scan(&after))
	assert.Equal(t, 0, after, "FK cascade must remove every metadata row")
}

// TestM2_GetByVersionNotFound exercises the ErrMetadataVersionNotFound (404)
// branch in mapCustomMetadataError by requesting a version that does not exist.
func TestM2_GetByVersionNotFound(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_VERNOTFOUND"
	cons, cfg := setupM2Fixture(t, clientID)

	schema := seedCustomSchema(t, cons, clientID, "m2-ver-notfound")
	docID := seedDoc(t, cfg, clientID, "m2-ver-notfound-doc")
	assignSchemaForTest(t, cons, docID, uuid.UUID(schema.Id))

	// POST one version so the document exists and has metadata.
	body := queryapi.CustomMetadataRequest{
		DocumentId: openapi_types.UUID(docID),
		Metadata:   map[string]interface{}{"name": "x"},
	}
	ctx, rec := newCustomMetadataContext(t, http.MethodPost, body)
	require.NoError(t, cons.SetCustomMetadata(ctx))
	require.Equal(t, http.StatusCreated, rec.Code)

	// Request a version that does not exist → 404.
	ctxVer, recVer := newCustomMetadataContext(t, http.MethodGet, nil)
	err := cons.GetCustomMetadataByVersion(ctxVer, queryapi.GetCustomMetadataByVersionParams{
		DocumentId: openapi_types.UUID(docID),
		Version:    99,
	})
	requireHTTPStatus(t, err, recVer, http.StatusNotFound)
}

// TestM2_CreateFieldExtractionOnCustomDocReturns409 exercises the
// ErrMutualExclusivityViolation branch in the CreateFieldExtraction handler.
// A document bound to a custom schema must reject legacy field extraction writes
// with 409.
func TestM2_CreateFieldExtractionOnCustomDocReturns409(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_LEGFX409"
	cons, cfg := setupM2Fixture(t, clientID)

	schema := seedCustomSchema(t, cons, clientID, "m2-legfx-409")
	docID := seedDoc(t, cfg, clientID, "m2-legfx-409-doc")
	assignSchemaForTest(t, cons, docID, uuid.UUID(schema.Id))

	// Attempt a legacy field extraction write — must 409.
	reqBody := queryapi.FieldExtractionRequest{
		DocumentId: queryapi.DocumentID(docID),
		CreatedBy:  "test@example.com",
		SingleFields: queryapi.SingleFields{
			FileName: func() *string { s := "test.pdf"; return &s }(),
		},
	}
	b, err := json.Marshal(reqBody)
	require.NoError(t, err)

	e := echo.New()
	req := httptest.NewRequest(http.MethodPost, "/field-extractions", bytes.NewReader(b))
	req.Header.Set(echo.HeaderContentType, echo.MIMEApplicationJSON)
	rec := httptest.NewRecorder()
	ctx := e.NewContext(req, rec)
	ctx.Set("user_claims", map[string]interface{}{"sub": testActorSubject})

	fxErr := cons.CreateFieldExtraction(ctx)
	requireHTTPStatus(t, fxErr, rec, http.StatusConflict)
}

// TestM2_CreateFieldExtractionDocNotFound exercises the ErrDocumentNotFound (404)
// branch in the CreateFieldExtraction handler when the document does not exist.
func TestM2_CreateFieldExtractionDocNotFound(t *testing.T) {
	if testing.Short() {
		t.SkipNow()
	}
	const clientID = "M2_FXDOCNOTFOUND"
	cons, _ := setupM2Fixture(t, clientID)

	// Use a UUID that will not match any document.
	nonExistentDocID := uuid.New()
	reqBody := queryapi.FieldExtractionRequest{
		DocumentId: queryapi.DocumentID(nonExistentDocID),
		CreatedBy:  "test@example.com",
		SingleFields: queryapi.SingleFields{
			FileName: func() *string { s := "test.pdf"; return &s }(),
		},
	}
	b, err := json.Marshal(reqBody)
	require.NoError(t, err)

	e := echo.New()
	req := httptest.NewRequest(http.MethodPost, "/field-extractions", bytes.NewReader(b))
	req.Header.Set(echo.HeaderContentType, echo.MIMEApplicationJSON)
	rec := httptest.NewRecorder()
	ctx := e.NewContext(req, rec)
	ctx.Set("user_claims", map[string]interface{}{"sub": testActorSubject})

	fxErr := cons.CreateFieldExtraction(ctx)
	requireHTTPStatus(t, fxErr, rec, http.StatusNotFound)
}

// requireHTTPStatus asserts the handler produced the expected HTTP code,
// handling both the "echo.HTTPError returned" and "recorder has code
// already" shapes the handlers can emit. Milestone 2 handlers always
// return an echo.HTTPError for mapped sentinels rather than writing to
// the recorder, so the error branch is the common path.
func requireHTTPStatus(t *testing.T, err error, rec *httptest.ResponseRecorder, want int) {
	t.Helper()
	if err != nil {
		var httpErr *echo.HTTPError
		if errors.As(err, &httpErr) {
			assert.Equalf(t, want, httpErr.Code, "expected status %d, got %d: %v", want, httpErr.Code, httpErr.Message)
			return
		}
		t.Fatalf("expected echo.HTTPError with status %d, got %T: %v", want, err, err)
	}
	assert.Equal(t, want, rec.Code)
}