Merged in feature/mutable-metadata1 (pull request #221)

M1, M2 and M3 complete * M1, M2 and M3 complete * review changes * docs * docs
2026-04-16 23:11:26 +00:00
parent ad7b21f3a2
commit 17fc813823
164 changed files with 44700 additions and 371 deletions
@@ -82,15 +82,17 @@ func (s *Service) GetEnriched(ctx context.Context, id uuid.UUID, includeTextReco
 	}

 	result := &DocumentEnriched{
-		ID:            docExternal.ID,
-		ClientID:      docExternal.Clientid,
-		Hash:          docExternal.Hash,
-		HasTextRecord: hasTextRecord,
-		FolderID:      docEnriched.Folderid,
-		Filename:      docEnriched.Filename,
-		OriginalPath:  docEnriched.Originalpath,
-		Labels:        labels,
-		FileSizeBytes: docEnriched.FileSizeBytes,
+		ID:                docExternal.ID,
+		ClientID:          docExternal.Clientid,
+		Hash:              docExternal.Hash,
+		HasTextRecord:     hasTextRecord,
+		FolderID:          docEnriched.Folderid,
+		Filename:          docEnriched.Filename,
+		OriginalPath:      docEnriched.Originalpath,
+		Labels:            labels,
+		FileSizeBytes:     docEnriched.FileSizeBytes,
+		CustomSchemaID:    docEnriched.CustomSchemaID,
+		HasCustomMetadata: docEnriched.HasCustomMetadata,
 	}

 	// 5. Optionally fetch full text record if requested and exists
@@ -1,6 +1,7 @@
 package document_test

 import (
+	"context"
 	"testing"

 	"queryorchestration/internal/database/repository"
@@ -210,3 +211,110 @@ func TestGetEnriched(t *testing.T) {
 		require.Error(t, err)
 	})
 }
+
+// TestGetEnriched_CustomSchemaFields covers Milestone 2.3a: the
+// DocumentEnriched struct surfaces CustomSchemaID and HasCustomMetadata
+// populated from the extended GetDocumentEnriched query in a single
+// round-trip. Three fixture shapes exercise every branch of the two new
+// fields.
+func TestGetEnriched_CustomSchemaFields(t *testing.T) {
+	ctx := t.Context()
+	cfg := &TestConfig{}
+	test.CreateDB(t, cfg)
+	svc := document.New(cfg)
+
+	clientID := "custom-schema-enriched"
+	// testcontainer pg instances are reused across runs; wipe the client's
+	// rows before re-seeding so the test is idempotent.
+	resetClientForEnrichedTest(t, ctx, cfg, clientID)
+	require.NoError(t, cfg.GetDBQueries().CreateClient(ctx, &repository.CreateClientParams{
+		Clientid: clientID,
+		Name:     "Custom Schema Enriched",
+	}))
+
+	// Seed one active schema for this client via direct SQL (avoids a
+	// hard dependency on the customschema package in the document tests).
+	var schemaID uuid.UUID
+	require.NoError(t, cfg.GetDBPool().QueryRow(ctx, `
+		INSERT INTO client_metadata_schemas (client_id, name, schema_def, version, created_by)
+		VALUES ($1, $2, $3::jsonb, 1, 'tester')
+		RETURNING id
+	`, clientID, "enriched-fields", `{"type":"object","additionalProperties":false}`).Scan(&schemaID))
+
+	t.Run("document with no schema bound", func(t *testing.T) {
+		docID, err := cfg.GetDBQueries().CreateDocument(ctx, &repository.CreateDocumentParams{
+			Clientid: clientID,
+			Hash:     "custom-no-schema",
+		})
+		require.NoError(t, err)
+
+		result, err := svc.GetEnriched(ctx, docID, false)
+		require.NoError(t, err)
+		assert.Nil(t, result.CustomSchemaID, "unbound document must have CustomSchemaID = nil")
+		assert.False(t, result.HasCustomMetadata, "unbound document must have HasCustomMetadata = false")
+	})
+
+	t.Run("document with schema bound but no metadata", func(t *testing.T) {
+		docID, err := cfg.GetDBQueries().CreateDocument(ctx, &repository.CreateDocumentParams{
+			Clientid: clientID,
+			Hash:     "custom-schema-no-metadata",
+		})
+		require.NoError(t, err)
+		_, err = cfg.GetDBPool().Exec(ctx,
+			`UPDATE documents SET custom_schema_id = $1 WHERE id = $2`, schemaID, docID)
+		require.NoError(t, err)
+
+		result, err := svc.GetEnriched(ctx, docID, false)
+		require.NoError(t, err)
+		require.NotNil(t, result.CustomSchemaID, "bound document must carry CustomSchemaID")
+		assert.Equal(t, schemaID, *result.CustomSchemaID)
+		assert.False(t, result.HasCustomMetadata, "bound document with no metadata must be false")
+	})
+
+	t.Run("document with schema and at least one metadata row", func(t *testing.T) {
+		docID, err := cfg.GetDBQueries().CreateDocument(ctx, &repository.CreateDocumentParams{
+			Clientid: clientID,
+			Hash:     "custom-schema-with-metadata",
+		})
+		require.NoError(t, err)
+		_, err = cfg.GetDBPool().Exec(ctx,
+			`UPDATE documents SET custom_schema_id = $1 WHERE id = $2`, schemaID, docID)
+		require.NoError(t, err)
+		_, err = cfg.GetDBPool().Exec(ctx, `
+			INSERT INTO document_custom_metadata (document_id, metadata, version, created_by)
+			VALUES ($1, '{"k":"v"}'::jsonb, 1, 'tester')
+		`, docID)
+		require.NoError(t, err)
+
+		result, err := svc.GetEnriched(ctx, docID, false)
+		require.NoError(t, err)
+		require.NotNil(t, result.CustomSchemaID)
+		assert.Equal(t, schemaID, *result.CustomSchemaID)
+		assert.True(t, result.HasCustomMetadata, "document with metadata row must be true")
+	})
+}
+
+// resetClientForEnrichedTest wipes every row owned by clientID in the order
+// required by the FK graph so the test starts clean even when the
+// testcontainer Postgres volume is reused across runs. Mirrors the pattern
+// from internal/database/repository/milestone2_migrations_test.go.
+func resetClientForEnrichedTest(t testing.TB, ctx context.Context, cfg serviceconfig.ConfigProvider, clientID string) {
+	t.Helper()
+	pool := cfg.GetDBPool()
+	stmts := []string{
+		`DELETE FROM document_custom_metadata
+		  WHERE document_id IN (SELECT id FROM documents WHERE clientId = $1)`,
+		`UPDATE documents SET custom_schema_id = NULL WHERE clientId = $1`,
+		`DELETE FROM client_metadata_schemas WHERE client_id = $1`,
+		`DELETE FROM documentEntries
+		  WHERE documentId IN (SELECT id FROM documents WHERE clientId = $1)`,
+		`DELETE FROM documents WHERE clientId = $1`,
+		`DELETE FROM folders WHERE clientId = $1`,
+		`DELETE FROM clientCanSync WHERE clientId = $1`,
+		`DELETE FROM clients WHERE clientId = $1`,
+	}
+	for _, stmt := range stmts {
+		_, err := pool.Exec(ctx, stmt, clientID)
+		require.NoErrorf(t, err, "reset stmt failed: %s", stmt)
+	}
+}
@@ -26,17 +26,26 @@ type DocumentExternal struct {
 // DocumentEnriched contains full document details including metadata, labels, and optionally text record.
 // Used by GET /document/{id} endpoint.
 // Note: Query functionality (Fields) has been removed. See remove_query_plan.md for details.
+//
+// Milestone 2.3a adds two fields for the mutable-metadata feature:
+//   - CustomSchemaID is the bound schema id (nil when the document has no schema).
+//   - HasCustomMetadata is true iff document_custom_metadata has >=1 row for the document.
+//
+// Both values are populated from the extended GetDocumentEnriched query in
+// one round-trip; no follow-up DB call is made.
 type DocumentEnriched struct {
-	ID            uuid.UUID
-	ClientID      string
-	Hash          string
-	HasTextRecord bool
-	FolderID      *uuid.UUID
-	Filename      *string
-	OriginalPath  *string
-	Labels        []LabelRecord
-	TextRecord    *TextRecord // Only populated when includeTextRecord=true
-	FileSizeBytes *int64      // File size in bytes (nil for legacy documents)
+	ID                uuid.UUID
+	ClientID          string
+	Hash              string
+	HasTextRecord     bool
+	FolderID          *uuid.UUID
+	Filename          *string
+	OriginalPath      *string
+	Labels            []LabelRecord
+	TextRecord        *TextRecord // Only populated when includeTextRecord=true
+	FileSizeBytes     *int64      // File size in bytes (nil for legacy documents)
+	CustomSchemaID    *uuid.UUID  // Bound schema id (nil = no schema)
+	HasCustomMetadata bool        // True iff any document_custom_metadata row exists
 }

 // LabelRecord represents a label applied to a document.