feat(DAL): support for uuid (#547)

This PR adds support for UUIDs to the DAL.

Summary of changes:
- The generation script (`yarn client:generate`) was slightly modified
to add a `/// @zod.string.uuid()` comment to the introspected Prisma
schema. This comment instructs the Zod generator to extend the Zod
schema with a uuid validator.
- The generator is changed to expose the UUID type in the generated
Electric client
- An e2e test was added to test that UUIDs can sync to and from Electric
and that writes with invalid UUIDs are rejected by Zod.

.changeset/rude-guests-sit.md

@@ -0,0 +1,6 @@
+---
+"electric-sql": patch
+"@electric-sql/prisma-generator": patch
+---
+
+Adds client-side support for int2 and int4 types.

.changeset/twelve-cheetahs-taste.md

@@ -0,0 +1,6 @@
+---
+"electric-sql": patch
+"@electric-sql/prisma-generator": patch
+---
+
+Add client-side validations for UUIDs.

clients/typescript/src/cli/migrations/migrate.ts

@@ -172,6 +172,9 @@ async function _generate(opts: Omit<GeneratorOptions, 'watch'>) {
     // Introspect the created DB to update the Prisma schema
     await introspectDB(prismaSchema)
 
+    // Add custom validators (such as uuid) to the Prisma schema
+    await addValidators(prismaSchema)
+
     // Modify snake_case table names to PascalCase
     await pascalCaseTableNames(prismaSchema)
 
@@ -326,6 +329,60 @@ async function introspectDB(prismaSchema: string): Promise<void> {
   )
 }
 
+/**
+ * Adds validators to the Prisma schema.
+ * @param prismaSchema Path to the Prisma schema
+ */
+async function addValidators(prismaSchema: string): Promise<void> {
+  const lines = await getFileLines(removeComments(prismaSchema))
+  const newLines = lines.map(addValidator)
+  // Write the modified Prisma schema to the file
+  await fs.writeFile(prismaSchema, newLines.join('\n'))
+}
+
+/**
+ * Adds a validator to the Prisma schema line if needed.
+ * @param ln A line from the Prisma schema
+ */
+function addValidator(ln: string): string {
+  const field = parseFields(ln)[0] // try to parse a field (the line could be something else than a field)
+
+  if (field) {
+    const intValidator = '@zod.number.int().gte(-2147483648).lte(2147483647)'
+
+    // Map attributes to validators
+    const attributeValidatorMapping = new Map([
+      ['@db.Uuid', '@zod.string.uuid()'],
+      ['@db.SmallInt', '@zod.number.int().gte(-32768).lte(32767)'],
+      ['@db.Int', intValidator],
+    ])
+    const attribute = field.attributes
+      .map((a) => a.type)
+      .find((a) => attributeValidatorMapping.has(a))
+
+    if (attribute) {
+      return ln + ' /// ' + attributeValidatorMapping.get(attribute)!
+    } else {
+      // No attribute validators,
+      // check if the field's type requires a validator
+      const typeValidatorMapping = new Map([
+        ['Int', intValidator],
+        ['Int?', intValidator],
+        ['Int[]', intValidator],
+      ])
+      const typeValidator = typeValidatorMapping.get(field.type)
+
+      if (typeValidator) {
+        return ln + ' /// ' + typeValidator
+      } else {
+        return ln
+      }
+    }
+  } else {
+    return ln
+  }
+}
+
 async function generateElectricClient(prismaSchema: string): Promise<void> {
   await executeShellCommand(
     `npx prisma generate --schema="${prismaSchema}"`,
@@ -426,3 +483,76 @@ function isSnakeCased(name: string): boolean {
 function snake2PascalCase(name: string): string {
   return name.split('_').map(capitaliseFirstLetter).join('')
 }
+
+// The below is duplicated code from the generator
+// TODO: move it to a separate helper package
+//       also move the model parsing to the package
+//       also move the removing comments function
+
+export type Attribute = {
+  type: `@${string}`
+  args: Array<string>
+}
+export type Field = {
+  field: string
+  type: string
+  attributes: Array<Attribute>
+}
+
+/**
+ * Removes all line comments from a string.
+ * A line comment is a comment that starts with *exactly* `//`.
+ * It does not remove comments starting with `///`.
+ */
+function removeComments(str: string): string {
+  const commentRegex = /(?<=[^/])\/\/(?=[^/]).*$/g // matches // until end of the line (does not match more than 2 slashes)
+  return str.replaceAll(commentRegex, '')
+}
+
+/**
+ * Takes the body of a model and returns
+ * an array of fields defined by the model.
+ * @param body Body of a model
+ * @returns Fields defined by the model
+ */
+function parseFields(body: string): Array<Field> {
+  // The regex below matches the fields of a model (it assumes there are no comments at the end of the line)
+  // It uses named captured groups to capture the field name, its type, and optional attributes
+  // the type can be `type` or `type?` or `type[]`
+  const fieldRegex =
+    /^\s*(?<field>\w+)\s+(?<type>[\w]+(\?|(\[]))?)\s*(?<attributes>((@[\w.]+\s*)|(@[\w.]+\(.*\)+\s*))+)?\s*$/gm
+  const fieldMatches = [...body.matchAll(fieldRegex)]
+  const fs = fieldMatches.map(
+    (match) =>
+      match.groups as { field: string; type: string; attributes?: string }
+  )
+
+  return fs.map((f) => ({
+    ...f,
+    attributes: parseAttributes(f.attributes ?? ''),
+  }))
+}
+
+/**
+ * Takes a string of attributes, e.g. `@id @db.Timestamp(2)`,
+ * and returns an array of attributes, e.g. `['@id', '@db.Timestamp(2)]`.
+ * @param attributes String of attributes
+ * @returns Array of attributes.
+ */
+function parseAttributes(attributes: string): Array<Attribute> {
+  // Matches each attribute in a string of attributes
+  // e.g. @id @db.Timestamp(2)
+  // The optional args capture group matches anything
+  // but not @or newline because that would be the start of a new attribute
+  const attributeRegex = /(?<type>@[\w.]+)(?<args>\([^@\n\r]+\))?/g
+  const matches = [...attributes.matchAll(attributeRegex)]
+  return matches.map((m) => {
+    const { type, args } = m.groups! as { type: string; args?: string }
+    const noParens = args?.substring(1, args.length - 1) // arguments without starting '(' and closing ')'
+    const parsedArgs = noParens?.split(',')?.map((arg) => arg.trim()) ?? []
+    return {
+      type: type as `@${string}`,
+      args: parsedArgs,
+    }
+  })
+}

clients/typescript/test/client/conversions/input.test.ts

@@ -31,7 +31,7 @@ await tbl.sync()
 function setupDB() {
   db.exec('DROP TABLE IF EXISTS DataTypes')
   db.exec(
-    "CREATE TABLE DataTypes('id' int PRIMARY KEY, 'date' varchar, 'time' varchar, 'timetz' varchar, 'timestamp' varchar, 'timestamptz' varchar, 'bool' int, 'relatedId' int);"
+    "CREATE TABLE DataTypes('id' int PRIMARY KEY, 'date' varchar, 'time' varchar, 'timetz' varchar, 'timestamp' varchar, 'timestamptz' varchar, 'bool' int, 'uuid' varchar, 'int2' int2, 'int4' int4, 'relatedId' int);"
   )
 
   db.exec('DROP TABLE IF EXISTS Dummy')
@@ -215,6 +215,9 @@ const dateNulls = {
   timestamp: null,
   timestamptz: null,
   bool: null,
+  int2: null,
+  int4: null,
+  uuid: null,
 }
 
 const nulls = {
@@ -535,13 +538,12 @@ test.serial('deleteMany transforms JS objects to SQLite', async (t) => {
 
   t.is(deleteRes.count, 2)
 
-  const fetchRes = await tbl.findMany()
-
-  t.deepEqual(fetchRes, [
-    {
-      ...dateNulls,
-      ...o3,
-      relatedId: null,
+  const fetchRes = await tbl.findMany({
+    select: {
+      id: true,
+      timestamp: true,
     },
-  ])
+  })
+
+  t.deepEqual(fetchRes, [o3])
 })

clients/typescript/test/client/conversions/sqlite.test.ts

@@ -30,7 +30,7 @@ await tbl.sync()
 function setupDB() {
   db.exec('DROP TABLE IF EXISTS DataTypes')
   db.exec(
-    "CREATE TABLE DataTypes('id' int PRIMARY KEY, 'date' varchar, 'time' varchar, 'timetz' varchar, 'timestamp' varchar, 'timestamptz' varchar, 'bool' int, 'relatedId' int);"
+    "CREATE TABLE DataTypes('id' int PRIMARY KEY, 'date' varchar, 'time' varchar, 'timetz' varchar, 'timestamp' varchar, 'timestamptz' varchar, 'bool' int, 'uuid' varchar, 'int2' int2, 'int4' int4, 'relatedId' int);"
   )
 }
 

clients/typescript/test/client/generated/client/index-browser.js

@@ -94,6 +94,9 @@ exports.Prisma.DataTypesScalarFieldEnum = {
   timestamp: 'timestamp',
   timestamptz: 'timestamptz',
   bool: 'bool',
+  uuid: 'uuid',
+  int2: 'int2',
+  int4: 'int4',
   relatedId: 'relatedId'
 };