Merge pull request #755 from cipherstash/worktree-schema-build-v3-picker

feat(cli): per-column EQL v3 domain picker in stash schema build

Author: tobyhede

.changeset/schema-build-v3-domain-picker.md

@@ -0,0 +1,13 @@
+---
+'stash': patch
+---
+
+`stash schema build` now picks a concrete EQL v3 domain per column
+(`TextSearch`, `IntegerOrd`, `TextEq`, …) instead of the legacy v2
+"searchable capabilities" toggle. Boolean columns are assigned the
+storage-only `types.Boolean` domain automatically, while JSON columns are
+assigned the queryable `types.Json` domain, with encrypted containment and
+selector queries. Other columns default to the widest searchable domain,
+matching the previous behaviour. The internal `SearchOp` capability tuple
+and the `v3DomainFactory` translation shim are removed, unblocking EQL v2
+removal (#707, #751).

packages/cli/src/commands/init/__tests__/utils-codegen.test.ts

@@ -0,0 +1,96 @@
+import { describe, expect, it } from 'vitest'
+import type { SchemaDef } from '../types.js'
+import { generateClientFromSchemas } from '../utils.js'
+
+const schemas: SchemaDef[] = [
+  {
+    tableName: 'users',
+    columns: [
+      { name: 'email', domain: 'TextSearch' },
+      { name: 'age', domain: 'IntegerOrd' },
+      { name: 'verified', domain: 'Boolean' },
+    ],
+  },
+]
+
+describe('generateClientFromSchemas', () => {
+  it('emits the chosen v3 domain factory per column (generic/postgresql)', () => {
+    const out = generateClientFromSchemas('postgresql', schemas)
+    expect(out).toContain("email: types.TextSearch('email'),")
+    expect(out).toContain("age: types.IntegerOrd('age'),")
+    expect(out).toContain("verified: types.Boolean('verified'),")
+    expect(out).toContain("from '@cipherstash/stack/v3'")
+    expect(out).toContain('EncryptionV3(')
+  })
+
+  it('emits the chosen v3 domain factory per column (drizzle)', () => {
+    const out = generateClientFromSchemas('drizzle', schemas)
+    expect(out).toContain("email: types.TextSearch('email'),")
+    expect(out).toContain("age: types.IntegerOrd('age'),")
+    expect(out).toContain('extractEncryptionSchemaV3')
+    expect(out).toContain("from '@cipherstash/stack-drizzle/v3'")
+  })
+
+  it('carries no residual v2 capability vocabulary', () => {
+    const generic = generateClientFromSchemas('postgresql', schemas)
+    const drizzle = generateClientFromSchemas('drizzle', schemas)
+    for (const out of [generic, drizzle]) {
+      expect(out).not.toMatch(/searchOps/)
+      expect(out).not.toMatch(/freeTextSearch|orderAndRange|\.equality\(/)
+    }
+  })
+
+  it('routes supabase through the generic generator, not drizzle', () => {
+    // `supabase` and `postgresql` share generateGenericFromSchemas; a misroute
+    // (dropping the `case 'supabase':` fallthrough, or routing it to drizzle)
+    // is otherwise uncovered — the other cases only exercise postgresql/drizzle.
+    const out = generateClientFromSchemas('supabase', schemas)
+    expect(out).toContain("email: types.TextSearch('email'),")
+    expect(out).toContain("from '@cipherstash/stack/v3'")
+    expect(out).not.toContain('@cipherstash/stack-drizzle')
+  })
+
+  it('throws for prisma-next instead of returning an undefined client', () => {
+    // prisma-next is unreachable from schema/build.ts, but the switch must stay
+    // total: fail loudly rather than silently returning undefined (which would
+    // write a broken client file) if a caller ever routes it here.
+    expect(() => generateClientFromSchemas('prisma-next', schemas)).toThrow(
+      /does not generate a prisma-next client/,
+    )
+  })
+})
+
+// Exhaustive round-trip over the closed V3Domain union: the sample fixtures
+// above only exercise 3 of 13 domains, so every domain is proven to emit
+// verbatim through both generators. `V3Domain` and `DataType` are finite closed
+// unions, so enumeration is complete — a fast-check property would sample the
+// same finite set and add nothing (and `packages/cli` has no fast-check dep).
+const ALL_DOMAINS: import('../types.js').V3Domain[] = [
+  'Text',
+  'TextEq',
+  'TextOrd',
+  'TextMatch',
+  'TextSearch',
+  'Integer',
+  'IntegerEq',
+  'IntegerOrd',
+  'Date',
+  'DateEq',
+  'DateOrd',
+  'Boolean',
+  'Json',
+]
+
+describe.each([
+  'postgresql',
+  'drizzle',
+] as const)('generateClientFromSchemas domain round-trip (%s)', (integration) => {
+  it.each(ALL_DOMAINS)('emits types.%s verbatim', (domain) => {
+    const s: SchemaDef[] = [
+      { tableName: 'x', columns: [{ name: 'c', domain }] },
+    ]
+    expect(generateClientFromSchemas(integration, s)).toContain(
+      `c: types.${domain}('c'),`,
+    )
+  })
+})

packages/cli/src/commands/init/lib/__tests__/build-flow.integration.test.ts

@@ -0,0 +1,107 @@
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+
+const selectMock = vi.fn()
+const multiselectMock = vi.fn()
+const confirmMock = vi.fn()
+const CANCEL = Symbol('clack.cancel')
+vi.mock('@clack/prompts', () => ({
+  select: (...a: unknown[]) => selectMock(...a),
+  multiselect: (...a: unknown[]) => multiselectMock(...a),
+  confirm: (...a: unknown[]) => confirmMock(...a),
+  isCancel: (v: unknown) => v === CANCEL,
+  log: { info: vi.fn(), success: vi.fn(), error: vi.fn() },
+  spinner: () => ({ start: vi.fn(), stop: vi.fn() }),
+}))
+
+const queryMock = vi.fn()
+vi.mock('pg', () => ({
+  default: {
+    Client: vi.fn(() => ({
+      connect: vi.fn(async () => {}),
+      query: queryMock,
+      end: vi.fn(async () => {}),
+    })),
+  },
+}))
+
+const { buildSchemasFromDatabase } = await import('../introspect.js')
+const { generateClientFromSchemas } = await import('../../utils.js')
+
+describe('build flow (introspect → pick → codegen)', () => {
+  beforeEach(() => {
+    selectMock.mockReset()
+    multiselectMock.mockReset()
+    confirmMock.mockReset()
+    queryMock.mockResolvedValue({
+      rows: [
+        {
+          table_name: 'users',
+          column_name: 'email',
+          data_type: 'text',
+          udt_name: 'text',
+        },
+        {
+          table_name: 'orders',
+          column_name: 'total',
+          data_type: 'integer',
+          udt_name: 'int4',
+        },
+      ],
+    })
+  })
+
+  it('walks two tables and emits a v3 client with the chosen domains', async () => {
+    selectMock
+      .mockResolvedValueOnce('users')
+      .mockResolvedValueOnce('TextEq') // table 1 + email
+      .mockResolvedValueOnce('orders')
+      .mockResolvedValueOnce('IntegerOrd') // table 2 + total
+    multiselectMock
+      .mockResolvedValueOnce(['email'])
+      .mockResolvedValueOnce(['total'])
+    confirmMock.mockResolvedValueOnce(true) // "encrypt columns in another table?"
+
+    const schemas = await buildSchemasFromDatabase('postgresql://x')
+    expect(schemas).toHaveLength(2)
+    if (!schemas) throw new Error('expected schemas to be defined')
+
+    const out = generateClientFromSchemas('postgresql', schemas)
+    expect(out).toContain("email: types.TextEq('email'),")
+    expect(out).toContain("total: types.IntegerOrd('total'),")
+    expect(out).not.toMatch(/searchOps|v3DomainFactory/)
+  })
+
+  it('stops after the first table when the user declines another', async () => {
+    selectMock.mockResolvedValueOnce('users').mockResolvedValueOnce('TextEq')
+    multiselectMock.mockResolvedValueOnce(['email'])
+    confirmMock.mockResolvedValueOnce(false) // decline "another table?"
+
+    const schemas = await buildSchemasFromDatabase('postgresql://x')
+    expect(schemas).toEqual([
+      { tableName: 'users', columns: [{ name: 'email', domain: 'TextEq' }] },
+    ])
+  })
+
+  it('returns undefined for an empty public schema', async () => {
+    queryMock.mockResolvedValueOnce({ rows: [] })
+    expect(await buildSchemasFromDatabase('postgresql://x')).toBeUndefined()
+  })
+
+  it('returns undefined when the database connection/introspection fails', async () => {
+    // introspectDatabase throws → buildSchemasFromDatabase catches, stops the
+    // spinner, logs the error, and returns undefined rather than propagating.
+    queryMock.mockReset()
+    queryMock.mockRejectedValueOnce(new Error('ECONNREFUSED'))
+    expect(await buildSchemasFromDatabase('postgresql://x')).toBeUndefined()
+  })
+
+  it('returns undefined when the "another table?" prompt is cancelled', async () => {
+    // Distinct from declining (false): cancelling (Ctrl-C) mid-loop must abort
+    // the whole build, not return the tables gathered so far.
+    selectMock.mockResolvedValueOnce('users').mockResolvedValueOnce('TextEq')
+    multiselectMock.mockResolvedValueOnce(['email'])
+    confirmMock.mockResolvedValueOnce(CANCEL) // cancel the "another table?" prompt
+
+    expect(await buildSchemasFromDatabase('postgresql://x')).toBeUndefined()
+  })
+})