change document to text and binary content

packages/runtime/src/services/database/migrations/migrations.001-init.ts

@@ -22,12 +22,12 @@ const init: Migration = {
       table.datetime('createdAt').notNullable();
       table.datetime('deletedAt').nullable();
       table.string('contentType').nullable();
-      table.text('content').nullable();
+      table.binary('content').nullable();
+      table.text('text').nullable();
       table.string('source').nullable();
       table.string('sourceId').nullable();
       table.string('type').notNullable();
       table.integer('typeVersion').nullable();
-      table.text('searchText').nullable();
       table.jsonb('metadata').nullable();
 
       table.index(['source', 'sourceId']);
@@ -74,12 +74,12 @@ type DocumentRow = {
   createdAt: Date;
   deletedAt: Date | null;
   contentType: string | null;
-  content: string | null;
+  content: Buffer | null;
+  text: string | null;
   source: string | null;
   sourceId: string | null;
   type: string;
   typeVersion: number | null;
-  searchText: string | null;
   metadata: unknown;
 };
 

packages/runtime/src/services/documents/documents.schemas.ts

@@ -11,12 +11,11 @@ const documentSchema = z
     updatedAt: z.iso.datetime(),
     deletedAt: z.iso.datetime().nullable(),
     contentType: z.string().nullable(),
-    content: z.string().nullable(),
+    text: z.string().nullable(),
     source: z.string().nullable(),
     sourceId: z.string().nullable(),
     type: z.string(),
     typeVersion: z.int().nullable(),
-    searchText: z.string().nullable(),
     metadata: z.unknown(),
   })
   .meta({ id: 'Document' });
@@ -29,17 +28,17 @@ const documentUpsertSchema = z
     owner: z.string().nullish(),
     contentType: z.string().nullish(),
     content: z.string().nullish(),
+    text: z.string().nullish(),
     source: z.string().nullish(),
     sourceId: z.string().nullish(),
     type: z.string().optional(),
     typeVersion: z.int().nullish(),
-    searchText: z.string().nullish(),
     metadata: z.unknown().nullish(),
   })
   .meta({
     id: 'DocumentUpsert',
     example: {
-      content: 'the cat is yellow',
+      text: 'the cat is yellow',
       contentType: 'text/plain',
       source: 'test',
       sourceId: 'test',

packages/runtime/src/services/documents/documents.ts

@@ -6,6 +6,7 @@ import { EventEmitter } from '../../utils/utils.event-emitter.js';
 import type { Services } from '../../utils/utils.services.js';
 import { compareObjectKeys } from '../../utils/utils.compare.js';
 import { applyQueryFilter } from '../../utils/utils.query.js';
+import { base64ToMaybeBuffer } from '../../utils/utils.binary.js';
 
 import { mapFromDocumentRow } from './documents.mapping.js';
 import type {
@@ -85,7 +86,6 @@ class DocumentsService extends EventEmitter<DocumentsServiceEvents> {
             'source',
             'content',
             'contentType',
-            'searchText',
             'type',
             'typeVersion',
             'metadata',
@@ -100,6 +100,7 @@ class DocumentsService extends EventEmitter<DocumentsServiceEvents> {
         await trx<TableRows['documents']>(tableNames.documents)
           .update({
             ...document,
+            content: base64ToMaybeBuffer(document.content),
             id,
             updatedAt: now,
           })
@@ -107,6 +108,7 @@ class DocumentsService extends EventEmitter<DocumentsServiceEvents> {
         const resultDocument: Document = mapFromDocumentRow({
           ...current,
           ...document,
+          content: base64ToMaybeBuffer(document.content ?? current.content) || null,
           id,
         });
         this.emit('updated', resultDocument);
@@ -121,21 +123,22 @@ class DocumentsService extends EventEmitter<DocumentsServiceEvents> {
           metadata: {},
           type: 'raw',
           ...document,
+          content: base64ToMaybeBuffer(document.content),
           id,
           createdAt: now,
           updatedAt: now,
         });
         const resultDocument: Document = mapFromDocumentRow({
           type: 'raw',
+          text: null,
           owner: null,
           contentType: null,
-          content: null,
           source: null,
           sourceId: null,
           typeVersion: null,
-          searchText: null,
           metadata: {},
           ...document,
+          content: base64ToMaybeBuffer(document.content) || null,
           deletedAt: null,
           id,
           createdAt: now,

packages/runtime/src/services/splitter/splitters/splitters.text.ts

@@ -3,13 +3,13 @@ import { RecursiveCharacterTextSplitter } from '@langchain/textsplitters';
 import type { Splitter } from '../splitter.types.js';
 
 const textSplitter: Splitter = {
-  match: (document) => !!document.content,
+  match: (document) => !!document.text,
   chunk: async (document) => {
-    if (!document.content) {
+    if (!document.text) {
       return [];
     }
     const splitter = new RecursiveCharacterTextSplitter({ chunkSize: 100, chunkOverlap: 0 });
-    const texts = await splitter.splitText(document.content);
+    const texts = await splitter.splitText(document.text);
     return texts;
   },
 };

packages/runtime/src/utils/utils.binary.ts

@@ -0,0 +1,14 @@
+const base64ToMaybeBuffer = (input?: string | null | Buffer) => {
+  if (input === null) {
+    return input;
+  }
+  if (!input) {
+    return undefined;
+  }
+  if (typeof input === 'object') {
+    return input;
+  }
+  return Buffer.from(input, 'base64');
+};
+
+export { base64ToMaybeBuffer };

packages/runtime/src/utils/utils.compare.ts

@@ -1,9 +1,13 @@
 import deepEqual from 'deep-equal';
 
-const compareObjectKeys = <T extends Record<string, unknown>>(a: T, b: T, keys: (keyof T)[]) => {
+const compareObjectKeys = <A extends Record<string, unknown>, B extends Record<string, unknown>>(
+  a: A,
+  b: B,
+  keys: (keyof (A & B))[],
+) => {
   for (const key of keys) {
-    const avalue = a[key];
-    const bvalue = b[key];
+    const avalue = a[key as keyof A];
+    const bvalue = b[key as keyof B];
     if (!deepEqual(avalue, bvalue)) {
       return false;
     }