Stream large database exports

src/export/csv.test.ts

@@ -1,13 +1,12 @@
 import { describe, it, expect, vi, beforeEach } from 'vitest'
 import { exportTableToCsvRoute } from './csv'
-import { getTableData, createExportResponse } from './index'
+import { executeOperation } from '.'
 import { createResponse } from '../utils'
 import type { DataSource } from '../types'
 import type { StarbaseDBConfiguration } from '../handler'
 
-vi.mock('./index', () => ({
-    getTableData: vi.fn(),
-    createExportResponse: vi.fn(),
+vi.mock('.', () => ({
+    executeOperation: vi.fn(),
 }))
 
 vi.mock('../utils', () => ({
@@ -42,51 +41,48 @@ beforeEach(() => {
 })
 
 describe('CSV Export Module', () => {
-    it('should return a CSV file when table data exists', async () => {
-        vi.mocked(getTableData).mockResolvedValue([
-            { id: 1, name: 'Alice', age: 30 },
-            { id: 2, name: 'Bob', age: 25 },
-        ])
-
-        vi.mocked(createExportResponse).mockReturnValue(
-            new Response('mocked-csv-content', {
-                headers: { 'Content-Type': 'text/csv' },
-            })
-        )
+    it('should stream a CSV file when table data exists', async () => {
+        vi.mocked(executeOperation)
+            .mockResolvedValueOnce([{ name: 'users' }])
+            .mockResolvedValueOnce([
+                { name: 'id' },
+                { name: 'name' },
+                { name: 'age' },
+            ])
+            .mockResolvedValueOnce([
+                {
+                    sql: 'CREATE TABLE users (id INTEGER, name TEXT, age INTEGER);',
+                },
+            ])
+            .mockResolvedValueOnce([
+                { id: 1, name: 'Alice', age: 30 },
+                { id: 2, name: 'Bob', age: 25 },
+            ])
 
         const response = await exportTableToCsvRoute(
             'users',
             mockDataSource,
             mockConfig
         )
 
-        expect(getTableData).toHaveBeenCalledWith(
-            'users',
-            mockDataSource,
-            mockConfig
+        expect(response.headers.get('Content-Type')).toBe('text/csv')
+        expect(response.headers.get('Content-Disposition')).toBe(
+            'attachment; filename="users_export.csv"'
         )
-        expect(createExportResponse).toHaveBeenCalledWith(
-            'id,name,age\n1,Alice,30\n2,Bob,25\n',
-            'users_export.csv',
-            'text/csv'
+        await expect(response.text()).resolves.toBe(
+            'id,name,age\n1,Alice,30\n2,Bob,25\n'
         )
-        expect(response.headers.get('Content-Type')).toBe('text/csv')
     })
 
     it('should return 404 if table does not exist', async () => {
-        vi.mocked(getTableData).mockResolvedValue(null)
+        vi.mocked(executeOperation).mockResolvedValueOnce([])
 
         const response = await exportTableToCsvRoute(
             'non_existent_table',
             mockDataSource,
             mockConfig
         )
 
-        expect(getTableData).toHaveBeenCalledWith(
-            'non_existent_table',
-            mockDataSource,
-            mockConfig
-        )
         expect(response.status).toBe(404)
 
         const jsonResponse: { error: string } = await response.json()
@@ -95,64 +91,118 @@ describe('CSV Export Module', () => {
         )
     })
 
-    it('should handle empty table (return only headers)', async () => {
-        vi.mocked(getTableData).mockResolvedValue([])
-
-        vi.mocked(createExportResponse).mockReturnValue(
-            new Response('mocked-csv-content', {
-                headers: { 'Content-Type': 'text/csv' },
-            })
-        )
+    it('should include headers for empty tables', async () => {
+        vi.mocked(executeOperation)
+            .mockResolvedValueOnce([{ name: 'empty_table' }])
+            .mockResolvedValueOnce([{ name: 'id' }, { name: 'name' }])
+            .mockResolvedValueOnce([
+                { sql: 'CREATE TABLE empty_table (id INTEGER, name TEXT);' },
+            ])
+            .mockResolvedValueOnce([])
 
         const response = await exportTableToCsvRoute(
             'empty_table',
             mockDataSource,
             mockConfig
         )
 
-        expect(getTableData).toHaveBeenCalledWith(
-            'empty_table',
+        await expect(response.text()).resolves.toBe('id,name\n')
+    })
+
+    it('should escape commas, quotes, and newlines in CSV values', async () => {
+        vi.mocked(executeOperation)
+            .mockResolvedValueOnce([{ name: 'special_chars' }])
+            .mockResolvedValueOnce([
+                { name: 'id' },
+                { name: 'name' },
+                { name: 'bio' },
+            ])
+            .mockResolvedValueOnce([
+                {
+                    sql: 'CREATE TABLE special_chars (id INTEGER, name TEXT, bio TEXT);',
+                },
+            ])
+            .mockResolvedValueOnce([
+                {
+                    id: 1,
+                    name: 'Sahithi, is',
+                    bio: 'my forever "penguin"\nline',
+                },
+            ])
+
+        const response = await exportTableToCsvRoute(
+            'special_chars',
             mockDataSource,
             mockConfig
         )
-        expect(createExportResponse).toHaveBeenCalledWith(
-            '',
-            'empty_table_export.csv',
-            'text/csv'
+
+        await expect(response.text()).resolves.toBe(
+            'id,name,bio\n1,"Sahithi, is","my forever ""penguin""\nline"\n'
         )
-        expect(response.headers.get('Content-Type')).toBe('text/csv')
     })
 
-    it('should escape commas and quotes in CSV values', async () => {
-        vi.mocked(getTableData).mockResolvedValue([
-            { id: 1, name: 'Sahithi, is', bio: 'my forever "penguin"' },
-        ])
-
-        vi.mocked(createExportResponse).mockReturnValue(
-            new Response('mocked-csv-content', {
-                headers: { 'Content-Type': 'text/csv' },
-            })
-        )
+    it('should page table data instead of loading the full table', async () => {
+        const firstPage = Array.from({ length: 1000 }, (_, index) => ({
+            __starbasedb_export_cursor_rowid: index + 1,
+            id: index + 1,
+            name: `User ${index + 1}`,
+        }))
+
+        vi.mocked(executeOperation)
+            .mockResolvedValueOnce([{ name: 'users' }])
+            .mockResolvedValueOnce([{ name: 'id' }, { name: 'name' }])
+            .mockResolvedValueOnce([
+                { sql: 'CREATE TABLE users (id INTEGER, name TEXT);' },
+            ])
+            .mockResolvedValueOnce(firstPage)
+            .mockResolvedValueOnce([
+                {
+                    __starbasedb_export_cursor_rowid: 1001,
+                    id: 1001,
+                    name: 'Last User',
+                },
+            ])
 
         const response = await exportTableToCsvRoute(
-            'special_chars',
+            'users',
             mockDataSource,
             mockConfig
         )
 
-        expect(createExportResponse).toHaveBeenCalledWith(
-            'id,name,bio\n1,"Sahithi, is","my forever ""penguin"""\n',
-            'special_chars_export.csv',
-            'text/csv'
+        const csv = await response.text()
+
+        expect(csv).toContain('1001,Last User\n')
+        expect(executeOperation).toHaveBeenNthCalledWith(
+            4,
+            [
+                {
+                    sql: 'SELECT rowid AS "__starbasedb_export_cursor_rowid", "id", "name" FROM "users" ORDER BY rowid LIMIT ?;',
+                    params: [1000],
+                },
+            ],
+            mockDataSource,
+            mockConfig
+        )
+        expect(executeOperation).toHaveBeenNthCalledWith(
+            5,
+            [
+                {
+                    sql: 'SELECT rowid AS "__starbasedb_export_cursor_rowid", "id", "name" FROM "users" WHERE rowid > ? ORDER BY rowid LIMIT ?;',
+                    params: [1000, 1000],
+                },
+            ],
+            mockDataSource,
+            mockConfig
         )
-        expect(response.headers.get('Content-Type')).toBe('text/csv')
     })
 
-    it('should return 500 on an unexpected error', async () => {
+    it('should return 500 on an unexpected error before streaming starts', async () => {
         const consoleErrorMock = vi
             .spyOn(console, 'error')
             .mockImplementation(() => {})
-        vi.mocked(getTableData).mockRejectedValue(new Error('Database Error'))
+        vi.mocked(executeOperation).mockRejectedValue(
+            new Error('Database Error')
+        )
 
         const response = await exportTableToCsvRoute(
             'users',
@@ -163,5 +213,6 @@ describe('CSV Export Module', () => {
         expect(response.status).toBe(500)
         const jsonResponse: { error: string } = await response.json()
         expect(jsonResponse.error).toBe('Failed to export table to CSV')
+        consoleErrorMock.mockRestore()
     })
 })

src/export/csv.ts

@@ -1,51 +1,60 @@
-import { getTableData, createExportResponse } from './index'
 import { createResponse } from '../utils'
 import { DataSource } from '../types'
 import { StarbaseDBConfiguration } from '../handler'
+import {
+    createStreamingExportResponse,
+    formatCsvValue,
+    getTableExportPlan,
+    iterateTableRows,
+    tableExists,
+    TableExportPlan,
+} from './streaming'
+
+async function* csvTableChunks(
+    tableName: string,
+    columns: string[],
+    dataSource: DataSource,
+    config: StarbaseDBConfiguration,
+    exportPlan: TableExportPlan
+): AsyncGenerator<string> {
+    if (columns.length) {
+        yield `${columns.map(formatCsvValue).join(',')}\n`
+    }
+
+    for await (const row of iterateTableRows(
+        tableName,
+        dataSource,
+        config,
+        undefined,
+        exportPlan
+    )) {
+        yield `${columns.map((column) => formatCsvValue(row[column])).join(',')}\n`
+    }
+}
 
 export async function exportTableToCsvRoute(
     tableName: string,
     dataSource: DataSource,
     config: StarbaseDBConfiguration
 ): Promise<Response> {
     try {
-        const data = await getTableData(tableName, dataSource, config)
-
-        if (data === null) {
+        if (!(await tableExists(tableName, dataSource, config))) {
             return createResponse(
                 undefined,
                 `Table '${tableName}' does not exist.`,
                 404
             )
         }
 
-        // Convert the result to CSV
-        let csvContent = ''
-        if (data.length > 0) {
-            // Add headers
-            csvContent += Object.keys(data[0]).join(',') + '\n'
-
-            // Add data rows
-            data.forEach((row: any) => {
-                csvContent +=
-                    Object.values(row)
-                        .map((value) => {
-                            if (
-                                typeof value === 'string' &&
-                                (value.includes(',') ||
-                                    value.includes('"') ||
-                                    value.includes('\n'))
-                            ) {
-                                return `"${value.replace(/"/g, '""')}"`
-                            }
-                            return value
-                        })
-                        .join(',') + '\n'
-            })
-        }
+        const exportPlan = await getTableExportPlan(
+            tableName,
+            dataSource,
+            config
+        )
+        const columns = exportPlan.columns
 
-        return createExportResponse(
-            csvContent,
+        return createStreamingExportResponse(
+            csvTableChunks(tableName, columns, dataSource, config, exportPlan),
             `${tableName}_export.csv`,
             'text/csv'
         )