@dxos/assistant 0.8.4-main.fd6878d → 0.8.4-staging.60fe92afc8

package/src/extraction/extraction-ner-function/extraction-ner-function.test.ts

@@ -1,116 +0,0 @@
-//
-// Copyright 2025 DXOS.org
-//
-
-import { beforeAll, describe, test } from 'vitest';
-
-import { type EchoDatabase } from '@dxos/echo-db';
-import { EchoTestBuilder } from '@dxos/echo-db/testing';
-import { FunctionExecutor, ServiceContainer } from '@dxos/functions';
-import { log } from '@dxos/log';
-import { DataType } from '@dxos/schema';
-import { Testing, createTestData } from '@dxos/schema/testing';
-import { range } from '@dxos/util';
-
-import { processTranscriptMessage } from '../extraction';
-
-import { extractionNerFn } from './extraction-ner-function';
-
-describe.skip('NER EntityExtraction', () => {
-  let builder: EchoTestBuilder;
-  let db: EchoDatabase;
-  let executor: FunctionExecutor;
-  let testData: {
-    transcriptJosiah: DataType.Message[];
-    transcriptWoflram: DataType.Message[];
-    transcriptMessages: DataType.Message[];
-    documents: Testing.DocumentType[];
-    contacts: Record<string, DataType.Person>;
-    organizations: Record<string, DataType.Organization>;
-  };
-  const TYPES = [DataType.Organization, DataType.Person, Testing.Contact, Testing.DocumentType];
-
-  beforeAll(async () => {
-    // TODO(dmaretskyi): Helper to scaffold this from a config.
-    builder = await new EchoTestBuilder().open();
-    const { db: db1 } = await builder.createDatabase({ indexing: { vector: true } });
-    db = db1;
-    db.graph.schemaRegistry.addSchema(TYPES);
-    const data = createTestData();
-    testData = {
-      transcriptJosiah: data.transcriptJosiah,
-      transcriptWoflram: data.transcriptWoflram,
-      transcriptMessages: data.transcriptMessages,
-      documents: data.documents.map((document) => db.add(document)),
-      contacts: Object.fromEntries(Object.entries(data.contacts).map(([key, value]) => [key, db.add(value)])),
-      organizations: Object.fromEntries(Object.entries(data.organizations).map(([key, value]) => [key, db.add(value)])),
-    };
-    await db.flush();
-
-    executor = new FunctionExecutor(
-      new ServiceContainer().setServices({
-        database: { db },
-      }),
-    );
-  }, 30_000);
-
-  test('should process a transcript block', async () => {
-    const { transcriptMessages, documents, contacts } = testData;
-    log.info('context', { documents, contacts });
-
-    for (const message of transcriptMessages) {
-      log.info('input', message);
-      const { message: enhancedMessage } = await processTranscriptMessage({
-        input: {
-          message,
-          objects: [...documents, ...Object.values(contacts)],
-        },
-        function: extractionNerFn,
-        executor,
-      });
-      log.info('output', enhancedMessage);
-    }
-  });
-
-  test('computational irreducibility', async () => {
-    const { transcriptWoflram, documents, contacts } = testData;
-
-    log.info('context', { documents, contacts });
-    const message = transcriptWoflram[0];
-    log.info('input', message);
-
-    await Promise.all(
-      range(10).map(async () => {
-        const { message: enhancedMessage, timeElapsed } = await processTranscriptMessage({
-          input: {
-            message,
-            objects: [...documents, ...Object.values(contacts)],
-          },
-          function: extractionNerFn,
-          executor,
-        });
-        log.info('output', { message: enhancedMessage.blocks[0], timeElapsed });
-      }),
-    );
-  });
-
-  test('org and document linking', async () => {
-    const { transcriptJosiah, documents, contacts, organizations } = testData;
-
-    log.info('context', { contacts, organizations, documents });
-
-    for (const message of transcriptJosiah) {
-      log.info('input', message);
-
-      const { message: enhancedMessage } = await processTranscriptMessage({
-        input: {
-          message,
-          objects: [...documents, ...Object.values(contacts), ...Object.values(organizations)],
-        },
-        function: extractionNerFn,
-        executor,
-      });
-      log.info('output', enhancedMessage);
-    }
-  });
-});

package/src/extraction/extraction-ner-function/extraction-ner-function.ts

@@ -1,57 +0,0 @@
-//
-// Copyright 2025 DXOS.org
-//
-
-// ISSUE(burdon): defineFunction
-// @ts-nocheck
-
-import { asyncTimeout } from '@dxos/async';
-import { create } from '@dxos/echo-schema';
-import { DatabaseService, defineFunction } from '@dxos/functions';
-import { invariant } from '@dxos/invariant';
-import { log } from '@dxos/log';
-import { DataType } from '@dxos/schema';
-
-import { ExtractionInput, ExtractionOutput } from '../extraction';
-import { findQuotes, insertReferences } from '../quotes';
-
-// TODO(mykola): Make it use vector index to query objects and not pass objects as input.
-export const extractionNerFn = defineFunction({
-  description: 'Extracts named entities from a text',
-  inputSchema: ExtractionInput,
-  outputSchema: ExtractionOutput,
-  handler: async ({ data: { message, options }, context }) => {
-    log.info('input', { message, options });
-    const startTime = performance.now();
-    const { db } = context.getService(DatabaseService);
-    const { extractFullEntities } = await import('./named-entity-recognition');
-
-    const entitiesPromise = Promise.all(
-      message.blocks.map(async (block) => {
-        invariant(block._tag === 'transcript' || block._tag === 'text', 'Block must have text');
-        return extractFullEntities(block.text);
-      }),
-    ).then((entities) => entities.flat());
-    const entities = options?.timeout ? await asyncTimeout(entitiesPromise, options.timeout) : await entitiesPromise;
-    log.info('entities', { entities });
-    const quoteReferences = await findQuotes(
-      entities.map((entity) => entity.word),
-      db,
-    );
-    const blocksWithReferences = message.blocks.map((block, i) => {
-      invariant(block._tag === 'transcript' || block._tag === 'text', 'Block must have text');
-      return { ...block, text: insertReferences(block._tag, quoteReferences) };
-    });
-    const messageWithReferences = create(DataType.Message, {
-      ...message,
-      blocks: blocksWithReferences,
-    });
-
-    log.info('output', { messageWithReferences });
-
-    return {
-      message: messageWithReferences,
-      timeElapsed: performance.now() - startTime,
-    };
-  },
-});

package/src/session/__snapshots__/session.test.ts.snap

@@ -1,32 +0,0 @@
-// Vitest Snapshot v1, https://vitest.dev/guide/snapshot.html
-
-exports[`AiSession > no tools 1`] = `
-[
-  {
-    "blocks": ReactiveArray [
-      {
-        "_tag": "text",
-        "text": "Hello world!",
-      },
-    ],
-    "created": "Sun, 20 Jul 2025 10:54:08 GMT",
-    "id": "01K0KQGFTFV3GBKGEWZW67K8HK",
-    "sender": {
-      "role": "user",
-    },
-  },
-  {
-    "blocks": ReactiveArray [
-      {
-        "_tag": "text",
-        "text": "Hi there! How can I help you today?",
-      },
-    ],
-    "created": "2025-07-20T10:54:10.801Z",
-    "id": "01K0KQGHVHJ37D4YG45WG090WV",
-    "sender": {
-      "role": "assistant",
-    },
-  },
-]
-`;

package/src/session/format.test.ts

@@ -1,60 +0,0 @@
-//
-// Copyright 2025 DXOS.org
-//
-
-import { beforeAll, describe, it } from '@effect/vitest';
-import { Effect } from 'effect';
-
-import { Blueprint, Template } from '@dxos/blueprints';
-import { Obj } from '@dxos/echo';
-import { type EchoDatabase } from '@dxos/echo-db';
-import { EchoTestBuilder } from '@dxos/echo-db/testing';
-import { DataType } from '@dxos/schema';
-import { trim } from '@dxos/util';
-
-import { createSystemPrompt } from '../templates/system';
-
-import { formatSystemPrompt } from './format';
-
-describe('format', () => {
-  let db: EchoDatabase;
-  beforeAll(async () => {
-    const builder = await new EchoTestBuilder().open();
-    ({ db } = await builder.createDatabase({
-      types: [DataType.Text, DataType.Organization, Blueprint.Blueprint],
-    }));
-  });
-
-  it.effect(
-    'should format',
-    Effect.fn(function* () {
-      const object = db.add(
-        Obj.make(DataType.Organization, {
-          name: 'Test',
-          website: 'https://www.test.com',
-        }),
-      );
-
-      const blueprint = db.add(
-        Blueprint.make({
-          key: 'example.com/blueprint/test',
-          name: 'Test',
-          instructions: Template.make({
-            source: trim`
-              Test
-              This is the test blueprint.
-            `,
-          }),
-        }),
-      );
-
-      const output = yield* formatSystemPrompt({
-        system: createSystemPrompt({}),
-        blueprints: [blueprint, blueprint],
-        objects: [object, object],
-      });
-
-      console.log(output);
-    }),
-  );
-});

package/src/session/session.test.ts

@@ -1,305 +0,0 @@
-//
-// Copyright 2025 DXOS.org
-//
-
-import { AiTool, AiToolkit } from '@effect/ai';
-import { describe, it } from '@effect/vitest';
-import { Effect, Layer, Schema } from 'effect';
-
-import { AiService, ToolExecutionService, ToolResolverService } from '@dxos/ai';
-import { AiServiceTestingPreset } from '@dxos/ai/testing';
-import { Obj, Type } from '@dxos/echo';
-import { log } from '@dxos/log';
-
-import { AiSession } from './session';
-
-// Define a calendar event artifact schema.
-const CalendarEventSchema = Schema.Struct({
-  title: Schema.String,
-  startTime: Schema.String,
-  endTime: Schema.String,
-  description: Schema.String,
-}).pipe(
-  Type.Obj({
-    typename: 'example.com/type/CalendarEvent',
-    version: '0.1.0',
-  }),
-);
-
-type CalendarEvent = Schema.Schema.Type<typeof CalendarEventSchema>;
-
-class TestToolkit extends AiToolkit.make(
-  AiTool.make('Calculator', {
-    description: 'Basic calculator tool',
-    parameters: {
-      input: Schema.String.annotations({
-        description: 'The calculation to perform.',
-      }),
-    },
-    success: Schema.Struct({
-      result: Schema.Number,
-    }),
-    failure: Schema.Never,
-  }),
-) {}
-
-// Tool handlers.
-const toolkitLayer = TestToolkit.toLayer({
-  Calculator: ({ input }) =>
-    Effect.gen(function* () {
-      const result = (() => {
-        // Restrict to basic arithmetic operations for safety.
-        const sanitizedInput = input.replace(/[^0-9+\-*/().\s]/g, '');
-        log.info('calculate', { sanitizedInput });
-
-        // eslint-disable-next-line @typescript-eslint/no-implied-eval
-        return Function(`"use strict"; return (${sanitizedInput})`)();
-      })();
-
-      return { result };
-    }),
-});
-
-describe.runIf(process.env.DX_RUN_SLOW_TESTS)('AiSession', () => {
-  it.effect(
-    'no tools',
-    Effect.fn(
-      function* ({ expect: _ }) {
-        const session = new AiSession({ operationModel: 'configured' });
-        const response = yield* session.run({
-          prompt: 'Hello world!',
-          history: [],
-        });
-        log.info('response', { response });
-      },
-      Effect.provide(
-        AiService.model('@anthropic/claude-3-5-sonnet-20241022').pipe(
-          Layer.provideMerge(ToolResolverService.layerEmpty),
-          Layer.provideMerge(ToolExecutionService.layerEmpty),
-          Layer.provideMerge(AiServiceTestingPreset('direct')),
-        ),
-      ),
-    ),
-  );
-
-  it.effect(
-    'calculator',
-    Effect.fn(
-      function* ({ expect: _ }) {
-        const session = new AiSession({ operationModel: 'configured' });
-        const response = yield* session.run({
-          prompt: 'What is 10 + 20?',
-          history: [],
-          toolkit: TestToolkit,
-        });
-        log.info('response', { response });
-      },
-      Effect.provide(
-        Layer.mergeAll(
-          toolkitLayer,
-          ToolResolverService.layerEmpty,
-          ToolExecutionService.layerEmpty,
-          AiService.model('@anthropic/claude-3-5-sonnet-20241022').pipe(
-            Layer.provideMerge(AiServiceTestingPreset('direct')),
-          ),
-        ),
-      ),
-    ),
-  );
-
-  // TODO(dmaretskyi): Revive test.
-  /*
-  it.skip('create calendar itinerary', { timeout: 60_000 }, async () => {
-    // overrides: { model: 'llama3.1:8b' },
-    const session = new AiSession({ operationModel: 'configured' });
-    const objects = new Set<string>();
-
-    // Define calendar artifact.
-    const calendarArtifact = defineArtifact({
-      id: 'artifact:dxos.org/plugin/calendar',
-      name: 'Calendar',
-      instructions: 'Use this to create and query calendar events.',
-      schema: CalendarEventSchema,
-      tools: [
-        createTool('calendar', {
-          name: 'query',
-          description: 'Query the calendar for events',
-          schema: Schema.Struct({}),
-          execute: async () => {
-            return ToolResult.Success(CALENDAR_EVENTS);
-          },
-        }),
-      ],
-    });
-
-    const tableArtifact = defineArtifact({
-      id: 'artifact:dxos.org/plugin/table',
-      name: 'Table',
-      instructions: 'Use this to create and manage tables. Each table has a unique id.',
-      schema: Schema.Struct({}),
-      tools: [
-        createTool('table', {
-          name: 'create',
-          description: 'Create a table',
-          schema: Schema.Struct({
-            data: Schema.Array(Schema.Any).annotations({ description: 'Array of data payloads to add as rows' }),
-          }),
-          execute: async ({ data }) => {
-            log('create table', { data });
-            const id = DXN.fromLocalObjectId(ObjectId.random()).toString();
-            objects.add(id);
-            // TODO(dmaretskyi): consider xml for refs instead of @dxn:echo:@:XXXXX
-            return ToolResult.Success(`table @${id}`);
-          },
-        }),
-      ],
-    });
-
-    const mapArtifact = defineArtifact({
-      id: 'artifact:dxos.org/plugin/map',
-      name: 'Map',
-      instructions:
-        'Use this to create and manage maps. Maps source data from tables. Table id is required to create a map.',
-      schema: Schema.Struct({}),
-      tools: [
-        createTool('map', {
-          name: 'create',
-          description: 'Create a map',
-          schema: Schema.Struct({
-            source: ArtifactId.annotations({
-              description: 'The table that will be used as the source of the map',
-            }),
-          }),
-          execute: async ({ source }) => {
-            // TODO(dmaretskyi): Use effect-schema decode instead of manual parsing.
-            const sourceId = ArtifactId.toDXN(source);
-            if (!objects.has(sourceId.toString())) {
-              return ToolResult.Error(`table id=${source} not found`);
-            }
-            log('create map', { sourceId });
-            const id = DXN.fromLocalObjectId(ObjectId.random()).toString();
-            objects.add(id);
-            // TODO(dmaretskyi): consider xml for refs instead of @dxn:echo:@:XXXXX
-            return ToolResult.Success(`map @${id}`);
-          },
-        }),
-      ],
-    });
-
-    const scriptArtifact = defineArtifact({
-      id: 'artifact:dxos.org/plugin/script',
-      name: 'Script',
-      instructions: 'Use this to create and manage scripts',
-      schema: Schema.Struct({}),
-      tools: [],
-    });
-
-    // session.streamEvent.on((event) => {
-    //   printStreamEvent(event);
-    // });
-
-    const printer = new ConsolePrinter();
-    session.message.on((message) => printer.printMessage(message));
-    session.userMessage.on((message) => printer.printMessage(message));
-    session.block.on((block) => printer.printContentBlock(block));
-
-    // session.update.on((update) => {
-    //   log('update', { update });
-    // });
-
-    // Test creating an itinerary
-    const response = await session.run({
-      client: aiClient,
-      tools: [],
-      artifacts: [calendarArtifact, tableArtifact, mapArtifact, scriptArtifact],
-      requiredArtifactIds: [calendarArtifact.id, tableArtifact.id, mapArtifact.id, scriptArtifact.id],
-      history: [],
-      generationOptions: {
-        model: '@anthropic/claude-3-5-haiku-20241022',
-      },
-      prompt: 'create a table and map for a travel itinerary based on events in my calendar',
-      toolResolver: new ToolRegistry([]),
-    });
-
-    log('result', {
-      objects,
-      finalMessage: response.at(-1),
-    });
-  });
-  */
-});
-
-// Travel to rome, florence, livorno, siena, madrid for conferences
-
-const _CALENDAR_EVENTS: CalendarEvent[] = [
-  Obj.make(CalendarEventSchema, {
-    title: 'Exploring Ancient Ruins in Rome',
-    startTime: '2024-01-01T10:00:00Z',
-    endTime: '2024-01-01T11:00:00Z',
-    description: 'Tech conference at the historic Colosseum with networking opportunities',
-  }),
-  Obj.make(CalendarEventSchema, {
-    title: 'Renaissance Tech Summit in Florence',
-    startTime: '2024-01-01T11:00:00Z',
-    endTime: '2024-01-01T12:00:00Z',
-    description: 'Discussing AI innovations surrounded by Renaissance art',
-  }),
-  Obj.make(CalendarEventSchema, {
-    title: 'Travel to Livorno',
-    startTime: '2024-01-01T12:00:00Z',
-    endTime: '2024-01-01T13:00:00Z',
-    description: 'Travel to Livorno',
-  }),
-  Obj.make(CalendarEventSchema, {
-    title: 'Travel to Siena',
-    startTime: '2024-01-01T13:00:00Z',
-    endTime: '2024-01-01T14:00:00Z',
-    description: 'Travel to Siena',
-  }),
-  Obj.make(CalendarEventSchema, {
-    title: 'Travel to Madrid',
-    startTime: '2024-01-01T14:00:00Z',
-    endTime: '2024-01-01T15:00:00Z',
-    description: 'Travel to Madrid',
-  }),
-];
-
-// const printStreamEvent = (event: GenerationStreamEvent) => {
-//   switch (event.type) {
-//     case 'message_start': {
-//       process.stdout.write(`${event.message.role.toUpperCase()}\n\n`);
-//       for (const content of event.message.content) {
-//         printContentBlock(content);
-//       }
-//       break;
-//     }
-//     case 'content_block_start': {
-//       printContentBlock(event.content);
-//       break;
-//     }
-//     case 'content_block_delta': {
-//       switch (event.delta.type) {
-//         case 'text_delta': {
-//           process.stdout.write(event.delta.text);
-//           break;
-//         }
-//         case 'input_json_delta': {
-//           process.stdout.write(event.delta.partial_json);
-//           break;
-//         }
-//       }
-//       break;
-//     }
-//     case 'content_block_stop': {
-//       process.stdout.write('\n');
-//       break;
-//     }
-//     case 'message_delta': {
-//       break;
-//     }
-//     case 'message_stop': {
-//       process.stdout.write('\n\n');
-//       break;
-//     }
-//   }
-// };