fix: generate data native types (#1412)

Signed-off-by: Alexis Rico <[email protected]>
Co-authored-by: Alexis Rico <[email protected]>

packages/importer/src/random-data.ts

@@ -1,5 +1,12 @@
 import { fakerEN as faker } from '@faker-js/faker';
 import { Schemas } from '@xata.io/client';
+import { z } from 'zod';
+
+// TODO: Remove this once we migrate pgroll branches
+type PgRollColumn = Schemas.Column & {
+  comment?: string;
+  pgType?: string;
+};
 
 export function generateRandomData(table: Schemas.Table, size: number) {
   const records: Record<string, unknown>[] = [];
@@ -11,61 +18,61 @@ export function generateRandomData(table: Schemas.Table, size: number) {
   return records;
 }
 
-function randomRecord(columns: Schemas.Column[]) {
+function randomRecord(columns: PgRollColumn[]) {
   const record: Record<string, unknown> = {};
   for (const column of columns) {
     record[column.name] = randomData(column);
   }
   return record;
 }
 
-function randomData(column: Schemas.Column) {
-  switch (column.type) {
-    case 'text':
-      return faker.lorem.paragraphs(rand(2, 3));
-    case 'email':
-      return faker.internet.email({ provider: 'acme.pets' });
+function randomData(column: PgRollColumn) {
+  const columnCommentType = narrowStringType(column.comment);
+  // Note that this is a best effort and seeding may fail for invalid Xata columns
+  // that are foreign keys, or have constraints such as length attached to them.
+  switch (column.pgType) {
+    case 'boolean':
+      return rand(0, 1) === 1;
+    case 'bigint':
+    case 'int8':
+    case 'integer':
     case 'int':
+    case 'int4':
+    case 'smallint':
       return rand(1, 100);
-    case 'float':
+    case 'double precision':
+    case 'float8':
+    case 'real':
       return rand(1, 10000) / rand(1, 100);
-    case 'bool':
-      return rand(0, 1) === 1;
-    case 'multiple':
-      return faker.word.words(rand(1, 3)).split(' ');
-    case 'string':
-      return randomString(column.name);
-    case 'datetime':
+    case 'text':
+    case 'varchar':
+    case 'character varying':
+      if (columnCommentType === 'email') return faker.internet.email({ provider: 'acme.pets' });
+      return faker.word.words(3);
+    case 'timestamptz':
       return faker.date.recent({ days: rand(1, 10) });
-    default:
-      return undefined;
+    case 'text[]':
+      return faker.word.words(rand(1, 3)).split(' ');
   }
+
+  if (column.pgType?.startsWith('character(') || column.pgType?.startsWith('varchar(')) return faker.word.words(3);
+  if (column.pgType?.startsWith('numeric(')) return rand(1, 10000) / rand(1, 100);
+
+  return undefined;
 }
 
 function rand(min: number, max: number) {
   return Math.floor(Math.random() * (max - min) + min);
 }
 
-const generators: Record<string, () => string> = {
-  city: () => faker.location.city(),
-  country: () => faker.location.country(),
-  county: () => faker.location.county(),
-  state: () => faker.location.state(),
-  street: () => faker.location.street(),
-  timezone: () => faker.location.timeZone(),
-  tz: () => faker.location.timeZone(),
-  zipcode: () => faker.location.zipCode(),
-  zip: () => faker.location.zipCode(),
-  department: () => faker.commerce.department(),
-  product: () => faker.commerce.product(),
-  company: () => faker.company.name(),
-  firstName: () => faker.person.firstName(),
-  lastName: () => faker.person.lastName(),
-  phone: () => faker.phone.number('501-###-###')
-};
+export const xataStringColumns = ['email', 'text', 'string'] as const;
 
-function randomString(columnName: string) {
-  const gen = generators[columnName.toLowerCase()];
-  if (gen) return gen();
-  return faker.word.words(2);
-}
+const XataStringColumn = z.object({
+  ['xata.type']: z.enum(xataStringColumns)
+});
+
+const narrowStringType = (comment?: string): Schemas.Column['type'] => {
+  if (!comment) return 'text';
+  const result = XataStringColumn.safeParse(JSON.parse(comment));
+  return result.success ? result.data['xata.type'] : 'text';
+};