@dxos/assistant 0.8.4-main.f9ba587 → 0.8.4-main.fd6878d

package/dist/lib/node-esm/chunk-XOOPGSGW.mjs.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../../src/session/artifact-diff.ts", "../../../src/errors.ts", "../../../src/session/session.ts", "../../../src/session/format.ts", "../../../src/session/version-pin.ts"],
+  "sourcesContent": ["//\n// Copyright 2025 DXOS.org\n//\n\nimport { Context } from 'effect';\n\nimport { type ObjectVersion } from '@dxos/echo-db';\nimport { type ObjectId } from '@dxos/echo-schema';\n\n/**\n * Resolves artifact ids to their versions.\n * Used to give the model a sense of the changes to the artifacts made by users during the conversation.\n * The artifacts versions are pinned in the history, and whenever the artifact changes in-between assistant's steps,\n * a diff is inserted into the conversation.\n *\n * Can be optionally provided to the session run call.\n */\n// TODO(burdon): Move.\n// TODO(dmaretskyi): Convert to Context.Reference\nexport class ArtifactDiffResolver extends Context.Tag('@dxos/assistant/ArtifactDiffResolver')<\n  ArtifactDiffResolver,\n  ArtifactDiffResolver.Service\n>() {}\n\nexport namespace ArtifactDiffResolver {\n  export type Service = {\n    resolve: (artifacts: { id: ObjectId; lastVersion: ObjectVersion }[]) => Promise<\n      Map<\n        ObjectId,\n        {\n          version: ObjectVersion;\n          diff?: string;\n        }\n      >\n    >;\n  };\n}\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport { BaseError } from '@dxos/errors';\n\n/**\n * Generic error for AI agent execution.\n */\nexport class AiAssistantError extends BaseError.extend('AI_ASSISTANT_ERROR') {}\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport { type AiError, AiLanguageModel, type AiResponse, type AiTool, AiToolkit } from '@effect/ai';\nimport { Chunk, type Context, Effect, Function, Option, Queue, type Schema, Stream } from 'effect';\n\nimport {\n  type AiInputPreprocessingError,\n  AiParser,\n  AiPreprocessor,\n  type AiToolNotFoundError,\n  type ConsolePrinter,\n  ToolExecutionService,\n  ToolResolverService,\n  callTools,\n  getToolCalls,\n} from '@dxos/ai';\nimport { type Blueprint } from '@dxos/blueprints';\nimport { todo } from '@dxos/debug';\nimport { Obj } from '@dxos/echo';\nimport { log } from '@dxos/log';\nimport { type ContentBlock, DataType } from '@dxos/schema';\nimport { isNotFalsy } from '@dxos/util';\n\nimport { type AiAssistantError } from '../errors';\n\nimport { formatSystemPrompt, formatUserPrompt } from './format';\n\nexport type AiSessionOptions = {};\n\n/**\n * Live observer of the generation process.\n */\nexport interface GenerationObserver {\n  /**\n   * Unparsed content block parts from the model.\n   */\n  onPart: (part: AiResponse.Part) => Effect.Effect<void>;\n\n  /**\n   * Parsed content blocks from the model.\n   * NOTE: Use block.pending to determine if the block was completed.\n   * For each block this will be called 0..n times with a pending block and then once with the final state of the block.\n   *\n   * Example:\n   *  1. { pending: true, text: \"Hello\"}\n   *  2. { pending: true, text: \"Hello, I am a\"}\n   *  3. { pending: false, text: \"Hello, I am a helpful assistant!\"}\n   */\n  onBlock: (block: ContentBlock.Any) => Effect.Effect<void>;\n\n  /**\n   * Complete messages fired during the session, both from the model and from the user.\n   * This message will contain all message blocks emitted through the `onBlock` callback.\n   */\n  onMessage: (message: DataType.Message) => Effect.Effect<void>;\n}\n\nexport const GenerationObserver = Object.freeze({\n  make: ({\n    onPart = Function.constant(Effect.void),\n    onBlock = Function.constant(Effect.void),\n    onMessage = Function.constant(Effect.void),\n  }: Partial<GenerationObserver> = {}): GenerationObserver => ({\n    onPart,\n    onBlock,\n    onMessage,\n  }),\n\n  noop: () => GenerationObserver.make(),\n\n  /**\n   * Debug printer to be used in unit-tests and browser devtools.\n   */\n  fromPrinter: (printer: ConsolePrinter) =>\n    GenerationObserver.make({\n      onBlock: (block) =>\n        Effect.sync(() => {\n          if (block.pending) {\n            return; // Only prints full blocks (better for unit-tests and browser devtools).\n          }\n          printer.printContentBlock(block);\n        }),\n      onMessage: (message) =>\n        Effect.sync(() => {\n          if (message.sender.role === 'assistant') {\n            return; // Skip assistant messages since they are printed in the `onBlock` callback.\n          }\n          printer.printMessage(message);\n        }),\n    }),\n});\n\nexport type SessionRunParams<Tools extends AiTool.Any> = {\n  prompt: string;\n  system?: string;\n  history?: DataType.Message[];\n  objects?: Obj.Any[]; // TODO(burdon): Meta only (typename and id -- write to binder).\n  blueprints?: Blueprint.Blueprint[];\n  toolkit?: AiToolkit.AiToolkit<Tools>;\n  observer?: GenerationObserver;\n};\n\n/**\n * Contains message history, tools, current context.\n * Current context means the state of the app, time of day, and other contextual information.\n * It makes requests to the model, its a state machine.\n * It keeps track of the current goal.\n * It manages the context window.\n * Tracks the success criteria of reaching the goal, exposing metrics (stretch).\n * Could be run locally in the app or remotely.\n * Could be personal or shared.\n */\nexport class AiSession {\n  /** Complete messages fired during the session, both from the model and from the user. */\n  public readonly messageQueue = Effect.runSync(Queue.unbounded<DataType.Message>());\n\n  // TODO(dmaretskyi): Remove the queues and convert everything to observer.\n\n  /**\n   * Blocks streaming from the model during the session.\n   * @deprecated Use `observer.onBlock` instead.\n   */\n  public readonly blockQueue = Effect.runSync(Queue.unbounded<Option.Option<ContentBlock.Any>>());\n\n  /** Unparsed events from the underlying generation stream. */\n  public readonly eventQueue = Effect.runSync(Queue.unbounded<AiResponse.Part>());\n\n  /** Prevents concurrent execution of session. */\n  private readonly _semaphore = Effect.runSync(Effect.makeSemaphore(1));\n\n  /** Pending messages (incl. the current user request). */\n  private _pending: DataType.Message[] = [];\n\n  /** Prior history from queue. */\n  private _history: DataType.Message[] = [];\n\n  constructor(private readonly _options: AiSessionOptions = {}) {}\n\n  /**\n   * Runs the AI model loop interacting with tools and artifacts.\n   * @returns The messages generated by the session, including the user's prompt.\n   */\n  // TODO(dmaretskyi): Toolkit context doesn't get added to the effect type.\n  run = <Tools extends AiTool.Any>(\n    params: SessionRunParams<Tools>,\n  ): Effect.Effect<\n    DataType.Message[],\n    AiAssistantError | AiInputPreprocessingError | AiToolNotFoundError | AiError.AiError,\n    AiLanguageModel.AiLanguageModel | ToolResolverService | ToolExecutionService | AiTool.ToHandler<Tools>\n  > =>\n    Effect.gen(this, function* () {\n      const observer = params.observer ?? GenerationObserver.noop();\n\n      // Create toolkit.\n      const toolkit: AiToolkit.ToHandler<Tools> = yield* createToolkit(params);\n\n      // Generate system prompt.\n      // TODO(budon): Dynamically resolve template variables.\n      const system = yield* formatSystemPrompt(params);\n      console.log(system);\n\n      // Generate user prompt.\n      const promptMessages = yield* formatUserPrompt(params);\n      yield* this.messageQueue.offer(promptMessages);\n      yield* observer.onMessage(promptMessages);\n\n      this._history = [...(params.history ?? [])];\n      this._pending = [promptMessages];\n\n      // Potential tool-use loop.\n      do {\n        log.info('request', {\n          prompt: promptMessages,\n          system: { snippet: [system.slice(0, 32), '...', system.slice(-32)].join(''), length: system.length },\n          pending: this._pending.length,\n          history: this._history.length,\n          objects: params.objects?.length ?? 0,\n          blueprints: params.blueprints?.length ?? 0,\n          toolkit: Object.values(toolkit.tools).map((tool: AiTool.Any) => tool.name),\n        });\n\n        // Generate the prompt and make request.\n        const prompt = yield* AiPreprocessor.preprocessAiInput([...this._history, ...this._pending]);\n        const blocks = yield* AiLanguageModel.streamText({\n          prompt,\n          system,\n          toolkit,\n          // TODO(burdon): Despite this flag, the model still calls tools.\n          //  Flag is only used in generateText (not streamText); patch and submit bug.\n          //  https://github.com/Effect-TS/effect/blob/main/packages/ai/ai/src/AiLanguageModel.ts#L401\n          disableToolCallResolution: true,\n        }).pipe(\n          AiParser.parseResponse({\n            onBlock: (block) =>\n              Effect.all([this.blockQueue.offer(Option.some(block)), observer.onBlock(block)], { discard: true }),\n            onPart: (part) => Effect.all([this.eventQueue.offer(part), observer.onPart(part)], { discard: true }),\n          }),\n          Stream.runCollect,\n          Effect.map(Chunk.toArray),\n        );\n\n        // Signal to stream consumers that message blocks are complete.\n        // Allows for coordination between the block and message queues\n        //   to prevent the streaming blocks from being rendered twice when the message is produced.\n        // TODO(wittjosiah): The block queue should probably be drained at this point in the case that there is no consumer.\n        yield* this.blockQueue.offer(Option.none());\n\n        // Create response message.\n        const response = Obj.make(DataType.Message, {\n          created: new Date().toISOString(),\n          sender: { role: 'assistant' },\n          blocks,\n        });\n        this._pending.push(response);\n        yield* this.messageQueue.offer(response);\n        yield* observer.onMessage(response);\n\n        // Parse response for tool calls.\n        const toolCalls = getToolCalls(response);\n        if (toolCalls.length === 0) {\n          break;\n        }\n\n        // TODO(burdon): Report errors to user; with proposed actions.\n        const toolResults = yield* callTools(toolkit, toolCalls);\n        const toolResultsMessage = Obj.make(DataType.Message, {\n          created: new Date().toISOString(),\n          sender: { role: 'user' },\n          blocks: toolResults,\n        });\n\n        this._pending.push(toolResultsMessage);\n        yield* this.messageQueue.offer(toolResultsMessage);\n        yield* observer.onMessage(toolResultsMessage);\n      } while (true);\n\n      // Signals to stream consumers that the session has completed and no more messages are coming.\n      yield* Queue.shutdown(this.messageQueue);\n      yield* Queue.shutdown(this.blockQueue);\n      yield* Queue.shutdown(this.eventQueue);\n\n      log('done', { pending: this._pending.length });\n      return this._pending;\n    }).pipe(this._semaphore.withPermits(1), Effect.withSpan('AiSession.run'));\n\n  // TODO(burdon): Implement.\n  async runStructured<S extends Schema.Schema.AnyNoContext>(\n    _schema: S,\n    _options: SessionRunParams<AiTool.Any>,\n  ): Promise<Schema.Schema.Type<S>> {\n    return todo();\n    // const parser = structuredOutputParser(schema);\n    // const result = await this.run({\n    //   ...options,\n    //   executableTools: [...(options.executableTools ?? []), parser.tool],\n    // });\n    // return parser.getResult(result);\n  }\n}\n\n/**\n * Build a combined toolkit from the blueprint tools and the provided toolkit.\n */\nconst createToolkit = <Tools extends AiTool.Any>({\n  toolkit,\n  blueprints = [],\n}: Pick<SessionRunParams<Tools>, 'toolkit' | 'blueprints'>) =>\n  Effect.gen(function* () {\n    const blueprintToolkit = yield* ToolResolverService.resolveToolkit(blueprints.flatMap(({ tools }) => tools));\n    const blueprintToolkitHandler: Context.Context<AiTool.ToHandler<AiTool.Any>> = yield* blueprintToolkit.toContext(\n      ToolExecutionService.handlersFor(blueprintToolkit),\n    );\n\n    return yield* AiToolkit.merge(...[toolkit, blueprintToolkit].filter(isNotFalsy)).pipe(\n      Effect.provide(blueprintToolkitHandler),\n    ) as Effect.Effect<AiToolkit.ToHandler<any>, never, AiTool.ToHandler<Tools>>;\n  });\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport { Array, Effect, Option, pipe } from 'effect';\n\nimport { Template } from '@dxos/blueprints';\nimport { Obj } from '@dxos/echo';\nimport { ObjectVersion } from '@dxos/echo-db';\nimport { type ObjectId } from '@dxos/echo-schema';\nimport { DatabaseService } from '@dxos/functions';\nimport { log } from '@dxos/log';\nimport { type ContentBlock, DataType } from '@dxos/schema';\nimport { trim } from '@dxos/util';\n\nimport { AiAssistantError } from '../errors';\n\nimport { ArtifactDiffResolver } from './artifact-diff';\nimport { type SessionRunParams } from './session';\n\n/**\n * Formats the system prompt.\n */\n// TODO(burdon): Move to AiPreprocessor.\nexport const formatSystemPrompt = ({\n  system,\n  blueprints = [],\n  objects = [],\n}: Pick<SessionRunParams<any>, 'system' | 'blueprints' | 'objects'>) =>\n  Effect.gen(function* () {\n    // TOOD(burdon): Should process templates.\n    const blueprintDefs = yield* pipe(\n      blueprints,\n      Effect.forEach((blueprint) => Effect.succeed(blueprint.instructions)),\n      Effect.flatMap(Effect.forEach((template) => DatabaseService.load(template.source))),\n      Effect.map(\n        Array.map(\n          (template) => trim`\n            <blueprint>\n              ${Template.process(template.content)}\n            </blueprint>\n          `,\n        ),\n      ),\n      Effect.map((blueprints) =>\n        blueprints.length > 0 ? ['## Blueprints Definitions', ...blueprints].join('\\n\\n') : undefined,\n      ),\n    );\n\n    const objectDefs = yield* pipe(\n      objects,\n      Effect.forEach((object) =>\n        Effect.succeed(trim`\n          <object>\n            <dxn>${Obj.getDXN(object)}</dxn>\n            <typename>${Obj.getTypename(object)}</typename>\n          </object>\n        `),\n      ),\n      Effect.map((objects) => (objects.length > 0 ? ['## Context Objects', ...objects].join('\\n\\n') : undefined)),\n    );\n\n    return yield* pipe(\n      Effect.succeed([system, blueprintDefs, objectDefs].filter((def): def is string => def !== undefined)),\n      Effect.map((parts) => parts.join('\\n\\n')),\n    );\n  });\n\n/**\n * Formats the user prompt.\n */\n// TODO(burdon): Move to AiPreprocessor.\n// TODO(burdon): Convert util below to `Effect.fn` (to preserve stack info)\nexport const formatUserPrompt = ({ prompt, history = [] }: Pick<SessionRunParams<any>, 'prompt' | 'history'>) =>\n  Effect.gen(function* () {\n    const prelude: ContentBlock.Any[] = [];\n\n    // TODO(dmaretskyi): Evaluate other approaches as `serviceOption` isn't represented in the type system.\n    const artifactDiffResolver = yield* Effect.serviceOption(ArtifactDiffResolver);\n    if (Option.isSome(artifactDiffResolver)) {\n      const versions = gatherObjectVersions(history);\n      const artifactDiff = yield* Effect.tryPromise({\n        try: () =>\n          artifactDiffResolver.value.resolve(\n            [...versions.entries()].map(([id, version]) => ({ id, lastVersion: version })),\n          ),\n        catch: AiAssistantError.wrap('Artifact diff resolution error'),\n      });\n\n      log.info('version', { artifactDiff, versions });\n      for (const [id, { version }] of [...artifactDiff.entries()]) {\n        if (ObjectVersion.equals(version, versions.get(id)!)) {\n          artifactDiff.delete(id);\n          continue;\n        }\n\n        prelude.push({ _tag: 'anchor', objectId: id, version });\n      }\n\n      if (artifactDiff.size > 0) {\n        prelude.push(createArtifactUpdateBlock(artifactDiff));\n      }\n    }\n\n    return Obj.make(DataType.Message, {\n      created: new Date().toISOString(),\n      sender: { role: 'user' },\n      blocks: [...prelude, { _tag: 'text', text: prompt }],\n    });\n  });\n\nconst gatherObjectVersions = (messages: DataType.Message[]): Map<ObjectId, ObjectVersion> => {\n  const artifactIds = new Map<ObjectId, ObjectVersion>();\n  for (const message of messages) {\n    for (const block of message.blocks) {\n      if (block._tag === 'anchor') {\n        artifactIds.set(block.objectId, block.version as ObjectVersion);\n      }\n    }\n  }\n\n  return artifactIds;\n};\n\nconst createArtifactUpdateBlock = (\n  artifactDiff: Map<ObjectId, { version: ObjectVersion; diff?: string }>,\n): ContentBlock.Any => {\n  return {\n    _tag: 'text',\n    // TODO(dmaretskyi): Does this need to be a special content-block?\n    disposition: 'artifact-update',\n    text: trim`\n      The following artifacts have been updated since the last message:\n      ${[...artifactDiff.entries()]\n        .map(([id, { diff }]) => `<changed-artifact id=\"${id}\">${diff ? `\\n${diff}` : ''}</changed-artifact>`)\n        .join('\\n')}\n    `,\n  };\n};\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport { Schema } from 'effect';\n\nimport { type ObjectVersion } from '@dxos/echo-db';\nimport { getVersion } from '@dxos/echo-db';\nimport { type AnyEchoObject, ObjectId } from '@dxos/echo-schema';\nimport { type ContentBlock } from '@dxos/schema';\n\n// TODO(dmaretskyi): Extract.\nconst ObjectVersionSchema = Schema.Unknown as Schema.Schema<ObjectVersion>;\n\nconst VersionPinSchema = Schema.Struct({\n  // TODO(dmaretskyi): Use Ref when those support encoding.\n  objectId: ObjectId,\n  // TODO(dmaretskyi): Could be opaque.\n  version: ObjectVersionSchema,\n});\n\n/**\n * Used to pin a specific version of an object during the course of a conversation.\n * This allows us to detect when the object changes during the course of the conversation.\n */\nexport interface VersionPin extends Schema.Schema.Type<typeof VersionPinSchema> {}\n\nexport const VersionPin: typeof VersionPinSchema & {\n  DISPOSITION: 'version-pin';\n  fromObject: (object: AnyEchoObject) => VersionPin;\n  createBlock: (pin: VersionPin) => ContentBlock.Any;\n} = class extends VersionPinSchema {\n  static readonly DISPOSITION = 'version-pin';\n  static fromObject(object: AnyEchoObject): VersionPin {\n    return VersionPin.make({\n      objectId: object.id,\n      version: getVersion(object),\n    });\n  }\n\n  static createBlock(pin: VersionPin): ContentBlock.Any {\n    return {\n      _tag: 'json',\n      disposition: VersionPin.DISPOSITION,\n      data: JSON.stringify(pin),\n    };\n  }\n};\n"],
+  "mappings": ";;;AAIA,SAASA,eAAe;AAejB,IAAMC,uBAAN,cAAmCC,QAAQC,IAAI,sCAAA,EAAA,EAAA;AAGjD;;;AClBL,SAASC,iBAAiB;AAKnB,IAAMC,mBAAN,cAA+BC,UAAUC,OAAO,oBAAA,EAAA;AAAuB;;;ACL9E,SAAuBC,iBAA+CC,iBAAiB;AACvF,SAASC,OAAqBC,UAAAA,SAAQC,UAAUC,UAAAA,SAAQC,OAAoBC,cAAc;AAE1F,SAEEC,UACAC,gBAGAC,sBACAC,qBACAC,WACAC,oBACK;AAEP,SAASC,YAAY;AACrB,SAASC,OAAAA,YAAW;AACpB,SAASC,OAAAA,YAAW;AACpB,SAA4BC,YAAAA,iBAAgB;AAC5C,SAASC,kBAAkB;;;ACnB3B,SAASC,OAAOC,QAAQC,QAAQC,YAAY;AAE5C,SAASC,gBAAgB;AACzB,SAASC,WAAW;AACpB,SAASC,qBAAqB;AAE9B,SAASC,uBAAuB;AAChC,SAASC,WAAW;AACpB,SAA4BC,gBAAgB;AAC5C,SAASC,YAAY;;AAWd,IAAMC,qBAAqB,CAAC,EACjCC,QACAC,aAAa,CAAA,GACbC,UAAU,CAAA,EAAE,MAEZC,OAAOC,IAAI,aAAA;AAET,QAAMC,gBAAgB,OAAOC,KAC3BL,YACAE,OAAOI,QAAQ,CAACC,cAAcL,OAAOM,QAAQD,UAAUE,YAAY,CAAA,GACnEP,OAAOQ,QAAQR,OAAOI,QAAQ,CAACK,aAAaC,gBAAgBC,KAAKF,SAASG,MAAM,CAAA,CAAA,GAChFZ,OAAOa,IACLC,MAAMD,IACJ,CAACJ,aAAaM;;gBAERC,SAASC,QAAQR,SAASS,OAAO,CAAA;;WAEtC,CAAA,GAGLlB,OAAOa,IAAI,CAACf,gBACVA,YAAWqB,SAAS,IAAI;IAAC;OAAgCrB;IAAYsB,KAAK,MAAA,IAAUC,MAAAA,CAAAA;AAIxF,QAAMC,aAAa,OAAOnB,KACxBJ,SACAC,OAAOI,QAAQ,CAACmB,WACdvB,OAAOM,QAAQS;;mBAEJS,IAAIC,OAAOF,MAAAA,CAAAA;wBACNC,IAAIE,YAAYH,MAAAA,CAAAA;;SAE/B,CAAA,GAEHvB,OAAOa,IAAI,CAACd,aAAaA,SAAQoB,SAAS,IAAI;IAAC;OAAyBpB;IAASqB,KAAK,MAAA,IAAUC,MAAAA,CAAAA;AAGlG,SAAO,OAAOlB,KACZH,OAAOM,QAAQ;IAACT;IAAQK;IAAeoB;IAAYK,OAAO,CAACC,QAAuBA,QAAQP,MAAAA,CAAAA,GAC1FrB,OAAOa,IAAI,CAACgB,UAAUA,MAAMT,KAAK,MAAA,CAAA,CAAA;AAErC,CAAA;AAOK,IAAMU,mBAAmB,CAAC,EAAEC,QAAQC,UAAU,CAAA,EAAE,MACrDhC,OAAOC,IAAI,aAAA;AACT,QAAMgC,UAA8B,CAAA;AAGpC,QAAMC,uBAAuB,OAAOlC,OAAOmC,cAAcC,oBAAAA;AACzD,MAAIC,OAAOC,OAAOJ,oBAAAA,GAAuB;AACvC,UAAMK,WAAWC,qBAAqBR,OAAAA;AACtC,UAAMS,eAAe,OAAOzC,OAAO0C,WAAW;MAC5CC,KAAK,MACHT,qBAAqBU,MAAMC,QACzB;WAAIN,SAASO,QAAO;QAAIjC,IAAI,CAAC,CAACkC,IAAIC,OAAAA,OAAc;QAAED;QAAIE,aAAaD;MAAQ,EAAA,CAAA;MAE/EE,OAAOC,iBAAiBC,KAAK,gCAAA;IAC/B,CAAA;AAEAC,QAAIC,KAAK,WAAW;MAAEb;MAAcF;IAAS,GAAA;;;;;;AAC7C,eAAW,CAACQ,IAAI,EAAEC,QAAO,CAAE,KAAK;SAAIP,aAAaK,QAAO;OAAK;AAC3D,UAAIS,cAAcC,OAAOR,SAAST,SAASkB,IAAIV,EAAAA,CAAAA,GAAO;AACpDN,qBAAaiB,OAAOX,EAAAA;AACpB;MACF;AAEAd,cAAQ0B,KAAK;QAAEC,MAAM;QAAUC,UAAUd;QAAIC;MAAQ,CAAA;IACvD;AAEA,QAAIP,aAAaqB,OAAO,GAAG;AACzB7B,cAAQ0B,KAAKI,0BAA0BtB,YAAAA,CAAAA;IACzC;EACF;AAEA,SAAOjB,IAAIwC,KAAKC,SAASC,SAAS;IAChCC,UAAS,oBAAIC,KAAAA,GAAOC,YAAW;IAC/BC,QAAQ;MAAEC,MAAM;IAAO;IACvBC,QAAQ;SAAIvC;MAAS;QAAE2B,MAAM;QAAQa,MAAM1C;MAAO;;EACpD,CAAA;AACF,CAAA;AAEF,IAAMS,uBAAuB,CAACkC,aAAAA;AAC5B,QAAMC,cAAc,oBAAIC,IAAAA;AACxB,aAAWC,WAAWH,UAAU;AAC9B,eAAWI,SAASD,QAAQL,QAAQ;AAClC,UAAIM,MAAMlB,SAAS,UAAU;AAC3Be,oBAAYI,IAAID,MAAMjB,UAAUiB,MAAM9B,OAAO;MAC/C;IACF;EACF;AAEA,SAAO2B;AACT;AAEA,IAAMZ,4BAA4B,CAChCtB,iBAAAA;AAEA,SAAO;IACLmB,MAAM;;IAENoB,aAAa;IACbP,MAAM1D;;QAEF;SAAI0B,aAAaK,QAAO;MACvBjC,IAAI,CAAC,CAACkC,IAAI,EAAEkC,KAAI,CAAE,MAAM,yBAAyBlC,EAAAA,KAAOkC,OAAO;EAAKA,IAAAA,KAAS,EAAA,qBAAuB,EACpG7D,KAAK,IAAA,CAAA;;EAEZ;AACF;;;;AD/EO,IAAM8D,qBAAqBC,OAAOC,OAAO;EAC9CC,MAAM,CAAC,EACLC,SAASC,SAASC,SAASC,QAAOC,IAAI,GACtCC,UAAUJ,SAASC,SAASC,QAAOC,IAAI,GACvCE,YAAYL,SAASC,SAASC,QAAOC,IAAI,EAAC,IACX,CAAC,OAA2B;IAC3DJ;IACAK;IACAC;EACF;EAEAC,MAAM,MAAMX,mBAAmBG,KAAI;;;;EAKnCS,aAAa,CAACC,YACZb,mBAAmBG,KAAK;IACtBM,SAAS,CAACK,UACRP,QAAOQ,KAAK,MAAA;AACV,UAAID,MAAME,SAAS;AACjB;MACF;AACAH,cAAQI,kBAAkBH,KAAAA;IAC5B,CAAA;IACFJ,WAAW,CAACQ,YACVX,QAAOQ,KAAK,MAAA;AACV,UAAIG,QAAQC,OAAOC,SAAS,aAAa;AACvC;MACF;AACAP,cAAQQ,aAAaH,OAAAA;IACvB,CAAA;EACJ,CAAA;AACJ,CAAA;AAsBO,IAAMI,YAAN,MAAMA;;;EAEKC;;;;;;EAQAC;;EAGAC;;EAGCC;;EAGTC;;EAGAC;EAER,YAA6BC,WAA6B,CAAC,GAAG;SAAjCA,WAAAA;SAtBbN,eAAehB,QAAOuB,QAAQC,MAAMC,UAAS,CAAA;SAQ7CR,aAAajB,QAAOuB,QAAQC,MAAMC,UAAS,CAAA;SAG3CP,aAAalB,QAAOuB,QAAQC,MAAMC,UAAS,CAAA;SAG1CN,aAAanB,QAAOuB,QAAQvB,QAAO0B,cAAc,CAAA,CAAA;SAG1DN,WAA+B,CAAA;SAG/BC,WAA+B,CAAA;SASvCM,MAAM,CACJC,WAMA5B,QAAO6B,IAAI,MAAM,aAAA;AACf,YAAMC,WAAWF,OAAOE,YAAYrC,mBAAmBW,KAAI;AAG3D,YAAM2B,UAAsC,OAAOC,cAAcJ,MAAAA;AAIjE,YAAMK,SAAS,OAAOC,mBAAmBN,MAAAA;AACzCO,cAAQC,IAAIH,MAAAA;AAGZ,YAAMI,iBAAiB,OAAOC,iBAAiBV,MAAAA;AAC/C,aAAO,KAAKZ,aAAauB,MAAMF,cAAAA;AAC/B,aAAOP,SAAS3B,UAAUkC,cAAAA;AAE1B,WAAKhB,WAAW;WAAKO,OAAOY,WAAW,CAAA;;AACvC,WAAKpB,WAAW;QAACiB;;AAGjB,SAAG;AACDD,QAAAA,KAAIK,KAAK,WAAW;UAClBC,QAAQL;UACRJ,QAAQ;YAAEU,SAAS;cAACV,OAAOW,MAAM,GAAG,EAAA;cAAK;cAAOX,OAAOW,MAAM,GAAC;cAAKC,KAAK,EAAA;YAAKC,QAAQb,OAAOa;UAAO;UACnGrC,SAAS,KAAKW,SAAS0B;UACvBN,SAAS,KAAKnB,SAASyB;UACvBC,SAASnB,OAAOmB,SAASD,UAAU;UACnCE,YAAYpB,OAAOoB,YAAYF,UAAU;UACzCf,SAASrC,OAAOuD,OAAOlB,QAAQmB,KAAK,EAAEC,IAAI,CAACC,SAAqBA,KAAKC,IAAI;QAC3E,GAAA;;;;;;AAGA,cAAMX,SAAS,OAAOY,eAAeC,kBAAkB;aAAI,KAAKlC;aAAa,KAAKD;SAAS;AAC3F,cAAMoC,SAAS,OAAOC,gBAAgBC,WAAW;UAC/ChB;UACAT;UACAF;;;;UAIA4B,2BAA2B;QAC7B,CAAA,EAAGC,KACDC,SAASC,cAAc;UACrB5D,SAAS,CAACK,UACRP,QAAO+D,IAAI;YAAC,KAAK9C,WAAWsB,MAAMyB,QAAOC,KAAK1D,KAAAA,CAAAA;YAASuB,SAAS5B,QAAQK,KAAAA;aAAS;YAAE2D,SAAS;UAAK,CAAA;UACnGrE,QAAQ,CAACsE,SAASnE,QAAO+D,IAAI;YAAC,KAAK7C,WAAWqB,MAAM4B,IAAAA;YAAOrC,SAASjC,OAAOsE,IAAAA;aAAQ;YAAED,SAAS;UAAK,CAAA;QACrG,CAAA,GACAE,OAAOC,YACPrE,QAAOmD,IAAImB,MAAMC,OAAO,CAAA;AAO1B,eAAO,KAAKtD,WAAWsB,MAAMyB,QAAOQ,KAAI,CAAA;AAGxC,cAAMC,WAAWC,KAAI9E,KAAK+E,UAASC,SAAS;UAC1CC,UAAS,oBAAIC,KAAAA,GAAOC,YAAW;UAC/BnE,QAAQ;YAAEC,MAAM;UAAY;UAC5B2C;QACF,CAAA;AACA,aAAKpC,SAAS4D,KAAKP,QAAAA;AACnB,eAAO,KAAKzD,aAAauB,MAAMkC,QAAAA;AAC/B,eAAO3C,SAAS3B,UAAUsE,QAAAA;AAG1B,cAAMQ,YAAYC,aAAaT,QAAAA;AAC/B,YAAIQ,UAAUnC,WAAW,GAAG;AAC1B;QACF;AAGA,cAAMqC,cAAc,OAAOC,UAAUrD,SAASkD,SAAAA;AAC9C,cAAMI,qBAAqBX,KAAI9E,KAAK+E,UAASC,SAAS;UACpDC,UAAS,oBAAIC,KAAAA,GAAOC,YAAW;UAC/BnE,QAAQ;YAAEC,MAAM;UAAO;UACvB2C,QAAQ2B;QACV,CAAA;AAEA,aAAK/D,SAAS4D,KAAKK,kBAAAA;AACnB,eAAO,KAAKrE,aAAauB,MAAM8C,kBAAAA;AAC/B,eAAOvD,SAAS3B,UAAUkF,kBAAAA;MAC5B,SAAS;AAGT,aAAO7D,MAAM8D,SAAS,KAAKtE,YAAY;AACvC,aAAOQ,MAAM8D,SAAS,KAAKrE,UAAU;AACrC,aAAOO,MAAM8D,SAAS,KAAKpE,UAAU;AAErCkB,MAAAA,KAAI,QAAQ;QAAE3B,SAAS,KAAKW,SAAS0B;MAAO,GAAA;;;;;;AAC5C,aAAO,KAAK1B;IACd,CAAA,EAAGwC,KAAK,KAAKzC,WAAWoE,YAAY,CAAA,GAAIvF,QAAOwF,SAAS,eAAA,CAAA;EA3GK;EAO/D7D;;EAuGA,MAAM8D,cACJC,SACApE,UACgC;AAChC,WAAOqE,KAAAA;EAOT;AACF;AAKA,IAAM3D,gBAAgB,CAA2B,EAC/CD,SACAiB,aAAa,CAAA,EAAE,MAEfhD,QAAO6B,IAAI,aAAA;AACT,QAAM+D,mBAAmB,OAAOC,oBAAoBC,eAAe9C,WAAW+C,QAAQ,CAAC,EAAE7C,MAAK,MAAOA,KAAAA,CAAAA;AACrG,QAAM8C,0BAAyE,OAAOJ,iBAAiBK,UACrGC,qBAAqBC,YAAYP,gBAAAA,CAAAA;AAGnC,SAAO,OAAOQ,UAAUC,MAAK,GAAI;IAACtE;IAAS6D;IAAkBU,OAAOC,UAAAA,CAAAA,EAAa3C,KAC/E5D,QAAOwG,QAAQR,uBAAAA,CAAAA;AAEnB,CAAA;;;AElRF,SAASS,cAAc;AAGvB,SAASC,kBAAkB;AAC3B,SAA6BC,gBAAgB;AAI7C,IAAMC,sBAAsBC,OAAOC;AAEnC,IAAMC,mBAAmBF,OAAOG,OAAO;;EAErCC,UAAUC;;EAEVC,SAASP;AACX,CAAA;AAQO,IAAMQ,aAIT,cAAcL,iBAAAA;EAChB,OAAgBM,cAAc;EAC9B,OAAOC,WAAWC,QAAmC;AACnD,WAAOH,WAAWI,KAAK;MACrBP,UAAUM,OAAOE;MACjBN,SAASO,WAAWH,MAAAA;IACtB,CAAA;EACF;EAEA,OAAOI,YAAYC,KAAmC;AACpD,WAAO;MACLC,MAAM;MACNC,aAAaV,WAAWC;MACxBU,MAAMC,KAAKC,UAAUL,GAAAA;IACvB;EACF;AACF;",
+  "names": ["Context", "ArtifactDiffResolver", "Context", "Tag", "BaseError", "AiAssistantError", "BaseError", "extend", "AiLanguageModel", "AiToolkit", "Chunk", "Effect", "Function", "Option", "Queue", "Stream", "AiParser", "AiPreprocessor", "ToolExecutionService", "ToolResolverService", "callTools", "getToolCalls", "todo", "Obj", "log", "DataType", "isNotFalsy", "Array", "Effect", "Option", "pipe", "Template", "Obj", "ObjectVersion", "DatabaseService", "log", "DataType", "trim", "formatSystemPrompt", "system", "blueprints", "objects", "Effect", "gen", "blueprintDefs", "pipe", "forEach", "blueprint", "succeed", "instructions", "flatMap", "template", "DatabaseService", "load", "source", "map", "Array", "trim", "Template", "process", "content", "length", "join", "undefined", "objectDefs", "object", "Obj", "getDXN", "getTypename", "filter", "def", "parts", "formatUserPrompt", "prompt", "history", "prelude", "artifactDiffResolver", "serviceOption", "ArtifactDiffResolver", "Option", "isSome", "versions", "gatherObjectVersions", "artifactDiff", "tryPromise", "try", "value", "resolve", "entries", "id", "version", "lastVersion", "catch", "AiAssistantError", "wrap", "log", "info", "ObjectVersion", "equals", "get", "delete", "push", "_tag", "objectId", "size", "createArtifactUpdateBlock", "make", "DataType", "Message", "created", "Date", "toISOString", "sender", "role", "blocks", "text", "messages", "artifactIds", "Map", "message", "block", "set", "disposition", "diff", "GenerationObserver", "Object", "freeze", "make", "onPart", "Function", "constant", "Effect", "void", "onBlock", "onMessage", "noop", "fromPrinter", "printer", "block", "sync", "pending", "printContentBlock", "message", "sender", "role", "printMessage", "AiSession", "messageQueue", "blockQueue", "eventQueue", "_semaphore", "_pending", "_history", "_options", "runSync", "Queue", "unbounded", "makeSemaphore", "run", "params", "gen", "observer", "toolkit", "createToolkit", "system", "formatSystemPrompt", "console", "log", "promptMessages", "formatUserPrompt", "offer", "history", "info", "prompt", "snippet", "slice", "join", "length", "objects", "blueprints", "values", "tools", "map", "tool", "name", "AiPreprocessor", "preprocessAiInput", "blocks", "AiLanguageModel", "streamText", "disableToolCallResolution", "pipe", "AiParser", "parseResponse", "all", "Option", "some", "discard", "part", "Stream", "runCollect", "Chunk", "toArray", "none", "response", "Obj", "DataType", "Message", "created", "Date", "toISOString", "push", "toolCalls", "getToolCalls", "toolResults", "callTools", "toolResultsMessage", "shutdown", "withPermits", "withSpan", "runStructured", "_schema", "todo", "blueprintToolkit", "ToolResolverService", "resolveToolkit", "flatMap", "blueprintToolkitHandler", "toContext", "ToolExecutionService", "handlersFor", "AiToolkit", "merge", "filter", "isNotFalsy", "provide", "Schema", "getVersion", "ObjectId", "ObjectVersionSchema", "Schema", "Unknown", "VersionPinSchema", "Struct", "objectId", "ObjectId", "version", "VersionPin", "DISPOSITION", "fromObject", "object", "make", "id", "getVersion", "createBlock", "pin", "_tag", "disposition", "data", "JSON", "stringify"]
+}

package/dist/lib/node-esm/extraction/index.mjs

@@ -0,0 +1,270 @@
+import { createRequire } from 'node:module';const require = createRequire(import.meta.url);
+import {
+  AiSession
+} from "../chunk-XOOPGSGW.mjs";
+import {
+  combineNerTokens,
+  createTokenGroups,
+  extractFullEntities,
+  getNer
+} from "../chunk-3UU5V4FS.mjs";
+
+// src/extraction/extraction.ts
+import { Schema } from "effect";
+import { asyncTimeout } from "@dxos/async";
+import { Expando } from "@dxos/echo-schema";
+import { log } from "@dxos/log";
+import { DataType } from "@dxos/schema";
+var __dxlog_file = "/__w/dxos/dxos/packages/core/assistant/src/extraction/extraction.ts";
+var ExtractionInput = Schema.Struct({
+  message: DataType.Message,
+  objects: Schema.optional(Schema.Array(Expando)),
+  options: Schema.optional(Schema.Struct({
+    timeout: Schema.optional(Schema.Number),
+    fallbackToRaw: Schema.optional(Schema.Boolean)
+  }))
+});
+var ExtractionOutput = Schema.Struct({
+  message: DataType.Message,
+  timeElapsed: Schema.Number
+});
+var processTranscriptMessage = async (params) => {
+  try {
+    if (params.options?.timeout && params.options.timeout > 0) {
+      return await asyncTimeout(params.executor.invoke(params.function, params.input), params.options.timeout);
+    } else {
+      return await params.executor.invoke(params.function, params.input);
+    }
+  } catch (error) {
+    if (params.options?.fallbackToRaw) {
+      log.warn("failed to process message, falling back to raw text", {
+        error
+      }, {
+        F: __dxlog_file,
+        L: 66,
+        S: void 0,
+        C: (f, a) => f(...a)
+      });
+      return {
+        message: params.input.message,
+        timeElapsed: 0
+      };
+    } else {
+      throw error;
+    }
+  }
+};
+
+// src/extraction/extraction-llm-function/extraction-llm-function.ts
+import { AiService, ToolRegistry } from "@dxos/ai";
+import { Obj } from "@dxos/echo";
+import { create } from "@dxos/echo-schema";
+import { defineFunction } from "@dxos/functions";
+import { DataType as DataType2 } from "@dxos/schema";
+
+// src/extraction/quotes.ts
+import { Schema as Schema2 } from "effect";
+import { Filter, Query } from "@dxos/echo";
+import { ObjectId } from "@dxos/echo-schema";
+import { DXN } from "@dxos/keys";
+import { log as log2 } from "@dxos/log";
+var __dxlog_file2 = "/__w/dxos/dxos/packages/core/assistant/src/extraction/quotes.ts";
+var ReferencedQuotes = Schema2.Struct({
+  references: Schema2.Array(Schema2.Struct({
+    id: Schema2.String,
+    quote: Schema2.String
+  })).pipe(Schema2.mutable).annotations({
+    // TODO(burdon): Does this description make sense?
+    description: `
+      The references to the context objects that are mentioned in the transcript. 
+      Quote should match the original transcript text exactly, while id is the id of the context object.
+    `
+  })
+});
+var findQuotes = async (quotes, db) => {
+  const references = await Promise.all(quotes.map(async (quote) => {
+    const { objects } = await db.query(Query.select(Filter.text(quote, {
+      type: "vector"
+    }))).run();
+    log2.info("found quotes in vector index", {
+      objects,
+      quote
+    }, {
+      F: __dxlog_file2,
+      L: 35,
+      S: void 0,
+      C: (f, a) => f(...a)
+    });
+    return objects.length > 0 ? [
+      {
+        id: objects[0].id.toString(),
+        quote
+      }
+    ] : [];
+  }));
+  return {
+    references: references.flat()
+  };
+};
+var insertReferences = (text, quotes) => {
+  for (const quote of quotes.references) {
+    if (!ObjectId.isValid(quote.id)) {
+      continue;
+    }
+    const regex = new RegExp(quote.quote.replace(/[.*+?^${}()|[\]\\]/g, "\\$&"), "gi");
+    text = text.replace(regex, `[${quote.quote}][${DXN.fromLocalObjectId(quote.id).toString()}]`);
+  }
+  return text;
+};
+
+// raw-loader:./instructions.tpl?raw
+var instructions_default = 'You are an AI assistant specialized in processing and enhancing transcripts with contextual information.\nYour task is to analyze a given transcript and extract references to the context objects.\n\nThe context and transcript are provided in the user message in `context` and `transcript` XML tags.\n\nHere are your instructions:\n\n1. Read the transcript and context information provided above.\n\n2. Process each segment of the transcript separately.\n\n3. For each segment:\n   a. Analyze the content and identify relevant information from the context.\n   b. Extract a number of references to the context objects that are mentioned in the segment.\n   c. Output the references in the format `{ "quote": "earnings report", "id": "01JT0JP9AX0XKGZX4MV4B69VT6" }`.\n   d. If there are no references, just output an empty array.\n\n4. Maintain the exact order and structure of the original transcript segments.\n\nOutput Format:\nYou have been provided with a tool that specifies output format\n\nRemember:\n- Only quote the text that is in the transcript.\n- Analyze each segment carefully before processing it.\n- Only reference objects that appear in the context and exactly by their id, If the object is not in the context, do not insert a reference. \n\nNow, process the transcript and context provided below according to these instructions.';
+
+// src/extraction/extraction-llm-function/extraction-llm-function.ts
+var extractionAnthropicFn = defineFunction({
+  description: "Extract entities from the transcript message and add them to the message.",
+  inputSchema: ExtractionInput,
+  outputSchema: ExtractionOutput,
+  handler: async ({ data: { message, objects }, context }) => {
+    const startTime = performance.now();
+    const ai = context.getService(AiService.AiService);
+    const session = new AiSession({
+      operationModel: "configured"
+    });
+    const result = await session.runStructured(ReferencedQuotes, {
+      generationOptions: {
+        model: "@anthropic/claude-3-5-haiku-20241022"
+      },
+      client: ai.client,
+      systemPrompt: instructions_default,
+      history: [
+        Obj.make(DataType2.Message, {
+          created: (/* @__PURE__ */ new Date()).toISOString(),
+          sender: {
+            role: "user"
+          },
+          blocks: [
+            {
+              _tag: "text",
+              text: `<context>${JSON.stringify(objects)}</context>`
+            },
+            {
+              _tag: "text",
+              text: `<transcript>${JSON.stringify(message.blocks)}</transcript>`
+            }
+          ]
+        })
+      ],
+      artifacts: [],
+      prompt: "",
+      tools: [],
+      toolResolver: new ToolRegistry([])
+    });
+    return {
+      message: create(DataType2.Message, {
+        ...message,
+        blocks: message.blocks.map((block, i) => block._tag !== "transcript" ? block : {
+          ...block,
+          text: insertReferences(block.text, result)
+        })
+      }),
+      timeElapsed: performance.now() - startTime
+    };
+  }
+});
+
+// src/extraction/extraction-ner-function/extraction-ner-function.ts
+import { asyncTimeout as asyncTimeout2 } from "@dxos/async";
+import { create as create2 } from "@dxos/echo-schema";
+import { DatabaseService, defineFunction as defineFunction2 } from "@dxos/functions";
+import { invariant } from "@dxos/invariant";
+import { log as log3 } from "@dxos/log";
+import { DataType as DataType3 } from "@dxos/schema";
+var __dxlog_file3 = "/__w/dxos/dxos/packages/core/assistant/src/extraction/extraction-ner-function/extraction-ner-function.ts";
+var extractionNerFn = defineFunction2({
+  description: "Extracts named entities from a text",
+  inputSchema: ExtractionInput,
+  outputSchema: ExtractionOutput,
+  handler: async ({ data: { message, options }, context }) => {
+    log3.info("input", {
+      message,
+      options
+    }, {
+      F: __dxlog_file3,
+      L: 24,
+      S: void 0,
+      C: (f, a) => f(...a)
+    });
+    const startTime = performance.now();
+    const { db } = context.getService(DatabaseService);
+    const { extractFullEntities: extractFullEntities2 } = await import("../named-entity-recognition-W4VMRCH6.mjs");
+    const entitiesPromise = Promise.all(message.blocks.map(async (block) => {
+      invariant(block._tag === "transcript" || block._tag === "text", "Block must have text", {
+        F: __dxlog_file3,
+        L: 31,
+        S: void 0,
+        A: [
+          "block._tag === 'transcript' || block._tag === 'text'",
+          "'Block must have text'"
+        ]
+      });
+      return extractFullEntities2(block.text);
+    })).then((entities2) => entities2.flat());
+    const entities = options?.timeout ? await asyncTimeout2(entitiesPromise, options.timeout) : await entitiesPromise;
+    log3.info("entities", {
+      entities
+    }, {
+      F: __dxlog_file3,
+      L: 36,
+      S: void 0,
+      C: (f, a) => f(...a)
+    });
+    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", {
+        F: __dxlog_file3,
+        L: 42,
+        S: void 0,
+        A: [
+          "block._tag === 'transcript' || block._tag === 'text'",
+          "'Block must have text'"
+        ]
+      });
+      return {
+        ...block,
+        text: insertReferences(block._tag, quoteReferences)
+      };
+    });
+    const messageWithReferences = create2(DataType3.Message, {
+      ...message,
+      blocks: blocksWithReferences
+    });
+    log3.info("output", {
+      messageWithReferences
+    }, {
+      F: __dxlog_file3,
+      L: 50,
+      S: void 0,
+      C: (f, a) => f(...a)
+    });
+    return {
+      message: messageWithReferences,
+      timeElapsed: performance.now() - startTime
+    };
+  }
+});
+export {
+  ExtractionInput,
+  ExtractionOutput,
+  ReferencedQuotes,
+  combineNerTokens,
+  createTokenGroups,
+  extractFullEntities,
+  extractionAnthropicFn,
+  extractionNerFn,
+  findQuotes,
+  getNer,
+  insertReferences,
+  processTranscriptMessage
+};
+//# sourceMappingURL=index.mjs.map

package/dist/lib/node-esm/extraction/index.mjs.map

@@ -0,0 +1,7 @@
+{
+  "version": 3,
+  "sources": ["../../../../src/extraction/extraction.ts", "../../../../src/extraction/extraction-llm-function/extraction-llm-function.ts", "../../../../src/extraction/quotes.ts", "raw-loader:./instructions.tpl?raw", "../../../../src/extraction/extraction-ner-function/extraction-ner-function.ts"],
+  "sourcesContent": ["//\n// Copyright 2025 DXOS.org\n//\n\nimport { Schema } from 'effect';\n\nimport { asyncTimeout } from '@dxos/async';\nimport { Expando } from '@dxos/echo-schema';\nimport { type FunctionDefinition, type FunctionExecutor } from '@dxos/functions';\nimport { log } from '@dxos/log';\nimport { DataType } from '@dxos/schema';\n\nexport const ExtractionInput = Schema.Struct({\n  message: DataType.Message,\n  objects: Schema.optional(Schema.Array(Expando)),\n  options: Schema.optional(\n    Schema.Struct({\n      timeout: Schema.optional(Schema.Number),\n      fallbackToRaw: Schema.optional(Schema.Boolean),\n    }),\n  ),\n});\nexport interface ExtractionInput extends Schema.Schema.Type<typeof ExtractionInput> {}\n\nexport const ExtractionOutput = Schema.Struct({\n  message: DataType.Message,\n  timeElapsed: Schema.Number,\n});\nexport interface ExtractionOutput extends Schema.Schema.Type<typeof ExtractionOutput> {}\n\nexport type ProcessTranscriptMessageParams = {\n  input: ExtractionInput;\n  function: FunctionDefinition<ExtractionInput, ExtractionOutput>;\n  executor: FunctionExecutor;\n\n  options?: {\n    /**\n     * Timeout for the entity extraction.\n     */\n    timeout?: number;\n\n    /**\n     * Fallback to raw text if the entity extraction fails.\n     * Otherwise the function will throw an error.\n     * @default false\n     */\n    fallbackToRaw?: boolean;\n  };\n};\n\nexport type ExtractionFunction = FunctionDefinition<ExtractionInput, ExtractionOutput>;\n\n/**\n * Extract entities from the transcript message and add them to the message.\n */\n// TODO(dmaretskyi): Move context to a vector search index.\nexport const processTranscriptMessage = async (params: ProcessTranscriptMessageParams): Promise<ExtractionOutput> => {\n  try {\n    if (params.options?.timeout && params.options.timeout > 0) {\n      return await asyncTimeout(params.executor.invoke(params.function, params.input), params.options.timeout);\n    } else {\n      return await params.executor.invoke(params.function, params.input);\n    }\n  } catch (error) {\n    if (params.options?.fallbackToRaw) {\n      log.warn('failed to process message, falling back to raw text', { error });\n      return {\n        message: params.input.message,\n        timeElapsed: 0,\n      };\n    } else {\n      throw error;\n    }\n  }\n};\n", "//\n// Copyright 2025 DXOS.org\n//\n\n// ISSUE(burdon): defineFunction\n// @ts-nocheck\n\nimport { AiService, ToolRegistry } from '@dxos/ai';\nimport { Obj } from '@dxos/echo';\nimport { create } from '@dxos/echo-schema';\nimport { type FunctionDefinition, defineFunction } from '@dxos/functions';\nimport { type ContentBlock, DataType } from '@dxos/schema';\n\nimport { AiSession } from '../../session';\nimport { ExtractionInput, ExtractionOutput } from '../extraction';\nimport { ReferencedQuotes, insertReferences } from '../quotes';\n\nimport PROMPT from './instructions.tpl?raw';\n\nexport const extractionAnthropicFn: FunctionDefinition<ExtractionInput, ExtractionOutput> = defineFunction({\n  description: 'Extract entities from the transcript message and add them to the message.',\n  inputSchema: ExtractionInput,\n  outputSchema: ExtractionOutput,\n  handler: async ({ data: { message, objects }, context }) => {\n    const startTime = performance.now();\n    const ai = context.getService(AiService.AiService);\n    const session = new AiSession({ operationModel: 'configured' });\n    const result = await session.runStructured(ReferencedQuotes, {\n      generationOptions: {\n        model: '@anthropic/claude-3-5-haiku-20241022',\n      },\n      client: ai.client,\n      systemPrompt: PROMPT,\n      history: [\n        Obj.make(DataType.Message, {\n          created: new Date().toISOString(),\n          sender: { role: 'user' },\n          blocks: [\n            {\n              _tag: 'text',\n              text: `<context>${JSON.stringify(objects)}</context>`,\n            } satisfies ContentBlock.Text,\n            {\n              _tag: 'text',\n              text: `<transcript>${JSON.stringify(message.blocks)}</transcript>`,\n            } satisfies ContentBlock.Text,\n          ],\n        }),\n      ],\n      artifacts: [],\n      prompt: '',\n      tools: [],\n      toolResolver: new ToolRegistry([]),\n    } as any); // TODO(burdon): Rewrite test.\n\n    return {\n      message: create(DataType.Message, {\n        ...message,\n        blocks: message.blocks.map((block, i) =>\n          block._tag !== 'transcript'\n            ? block\n            : {\n                ...block,\n                text: insertReferences(block.text, result),\n              },\n        ),\n      }),\n      timeElapsed: performance.now() - startTime,\n    };\n  },\n});\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport { Schema } from 'effect';\n\nimport { Filter, Query } from '@dxos/echo';\nimport { type EchoDatabase } from '@dxos/echo-db';\nimport { ObjectId } from '@dxos/echo-schema';\nimport { DXN } from '@dxos/keys';\nimport { log } from '@dxos/log';\n\nexport const ReferencedQuotes = Schema.Struct({\n  references: Schema.Array(\n    Schema.Struct({\n      id: Schema.String,\n      quote: Schema.String, // TODO(burdon): Quote?\n    }),\n  )\n    .pipe(Schema.mutable)\n    .annotations({\n      // TODO(burdon): Does this description make sense?\n      description: `\n      The references to the context objects that are mentioned in the transcript. \n      Quote should match the original transcript text exactly, while id is the id of the context object.\n    `,\n    }),\n});\nexport type ReferencedQuotes = Schema.Schema.Type<typeof ReferencedQuotes>;\n\nexport const findQuotes = async (quotes: string[], db: EchoDatabase): Promise<ReferencedQuotes> => {\n  const references = await Promise.all(\n    quotes.map(async (quote) => {\n      const { objects } = await db.query(Query.select(Filter.text(quote, { type: 'vector' }))).run();\n      log.info('found quotes in vector index', { objects, quote });\n      return objects.length > 0 ? [{ id: objects[0].id.toString(), quote }] : [];\n    }),\n  );\n  return { references: references.flat() };\n};\n\n/**\n * Finds and replaces all quotes with DXNs references.\n */\n// TODO(dmaretskyi): Lookup and verifiy ids from provided context.\nexport const insertReferences = (text: string, quotes: ReferencedQuotes) => {\n  for (const quote of quotes.references) {\n    if (!ObjectId.isValid(quote.id)) {\n      continue;\n    }\n\n    // Use a case-insensitive regular expression to replace the quote.\n    const regex = new RegExp(quote.quote.replace(/[.*+?^${}()|[\\]\\\\]/g, '\\\\$&'), 'gi');\n    text = text.replace(regex, `[${quote.quote}][${DXN.fromLocalObjectId(quote.id).toString()}]`);\n  }\n\n  return text;\n};\n", "You are an AI assistant specialized in processing and enhancing transcripts with contextual information.\nYour task is to analyze a given transcript and extract references to the context objects.\n\nThe context and transcript are provided in the user message in `context` and `transcript` XML tags.\n\nHere are your instructions:\n\n1. Read the transcript and context information provided above.\n\n2. Process each segment of the transcript separately.\n\n3. For each segment:\n   a. Analyze the content and identify relevant information from the context.\n   b. Extract a number of references to the context objects that are mentioned in the segment.\n   c. Output the references in the format `{ \"quote\": \"earnings report\", \"id\": \"01JT0JP9AX0XKGZX4MV4B69VT6\" }`.\n   d. If there are no references, just output an empty array.\n\n4. Maintain the exact order and structure of the original transcript segments.\n\nOutput Format:\nYou have been provided with a tool that specifies output format\n\nRemember:\n- Only quote the text that is in the transcript.\n- Analyze each segment carefully before processing it.\n- Only reference objects that appear in the context and exactly by their id, If the object is not in the context, do not insert a reference. \n\nNow, process the transcript and context provided below according to these instructions.", "//\n// Copyright 2025 DXOS.org\n//\n\n// ISSUE(burdon): defineFunction\n// @ts-nocheck\n\nimport { asyncTimeout } from '@dxos/async';\nimport { create } from '@dxos/echo-schema';\nimport { DatabaseService, defineFunction } from '@dxos/functions';\nimport { invariant } from '@dxos/invariant';\nimport { log } from '@dxos/log';\nimport { DataType } from '@dxos/schema';\n\nimport { ExtractionInput, ExtractionOutput } from '../extraction';\nimport { findQuotes, insertReferences } from '../quotes';\n\n// TODO(mykola): Make it use vector index to query objects and not pass objects as input.\nexport const extractionNerFn = defineFunction({\n  description: 'Extracts named entities from a text',\n  inputSchema: ExtractionInput,\n  outputSchema: ExtractionOutput,\n  handler: async ({ data: { message, options }, context }) => {\n    log.info('input', { message, options });\n    const startTime = performance.now();\n    const { db } = context.getService(DatabaseService);\n    const { extractFullEntities } = await import('./named-entity-recognition');\n\n    const entitiesPromise = Promise.all(\n      message.blocks.map(async (block) => {\n        invariant(block._tag === 'transcript' || block._tag === 'text', 'Block must have text');\n        return extractFullEntities(block.text);\n      }),\n    ).then((entities) => entities.flat());\n    const entities = options?.timeout ? await asyncTimeout(entitiesPromise, options.timeout) : await entitiesPromise;\n    log.info('entities', { entities });\n    const quoteReferences = await findQuotes(\n      entities.map((entity) => entity.word),\n      db,\n    );\n    const blocksWithReferences = message.blocks.map((block, i) => {\n      invariant(block._tag === 'transcript' || block._tag === 'text', 'Block must have text');\n      return { ...block, text: insertReferences(block._tag, quoteReferences) };\n    });\n    const messageWithReferences = create(DataType.Message, {\n      ...message,\n      blocks: blocksWithReferences,\n    });\n\n    log.info('output', { messageWithReferences });\n\n    return {\n      message: messageWithReferences,\n      timeElapsed: performance.now() - startTime,\n    };\n  },\n});\n"],
+  "mappings": ";;;;;;;;;;;;AAIA,SAASA,cAAc;AAEvB,SAASC,oBAAoB;AAC7B,SAASC,eAAe;AAExB,SAASC,WAAW;AACpB,SAASC,gBAAgB;;AAElB,IAAMC,kBAAkBL,OAAOM,OAAO;EAC3CC,SAASH,SAASI;EAClBC,SAAST,OAAOU,SAASV,OAAOW,MAAMT,OAAAA,CAAAA;EACtCU,SAASZ,OAAOU,SACdV,OAAOM,OAAO;IACZO,SAASb,OAAOU,SAASV,OAAOc,MAAM;IACtCC,eAAef,OAAOU,SAASV,OAAOgB,OAAO;EAC/C,CAAA,CAAA;AAEJ,CAAA;AAGO,IAAMC,mBAAmBjB,OAAOM,OAAO;EAC5CC,SAASH,SAASI;EAClBU,aAAalB,OAAOc;AACtB,CAAA;AA6BO,IAAMK,2BAA2B,OAAOC,WAAAA;AAC7C,MAAI;AACF,QAAIA,OAAOR,SAASC,WAAWO,OAAOR,QAAQC,UAAU,GAAG;AACzD,aAAO,MAAMZ,aAAamB,OAAOC,SAASC,OAAOF,OAAOG,UAAUH,OAAOI,KAAK,GAAGJ,OAAOR,QAAQC,OAAO;IACzG,OAAO;AACL,aAAO,MAAMO,OAAOC,SAASC,OAAOF,OAAOG,UAAUH,OAAOI,KAAK;IACnE;EACF,SAASC,OAAO;AACd,QAAIL,OAAOR,SAASG,eAAe;AACjCZ,UAAIuB,KAAK,uDAAuD;QAAED;MAAM,GAAA;;;;;;AACxE,aAAO;QACLlB,SAASa,OAAOI,MAAMjB;QACtBW,aAAa;MACf;IACF,OAAO;AACL,YAAMO;IACR;EACF;AACF;;;ACnEA,SAASE,WAAWC,oBAAoB;AACxC,SAASC,WAAW;AACpB,SAASC,cAAc;AACvB,SAAkCC,sBAAsB;AACxD,SAA4BC,YAAAA,iBAAgB;;;ACP5C,SAASC,UAAAA,eAAc;AAEvB,SAASC,QAAQC,aAAa;AAE9B,SAASC,gBAAgB;AACzB,SAASC,WAAW;AACpB,SAASC,OAAAA,YAAW;;AAEb,IAAMC,mBAAmBN,QAAOO,OAAO;EAC5CC,YAAYR,QAAOS,MACjBT,QAAOO,OAAO;IACZG,IAAIV,QAAOW;IACXC,OAAOZ,QAAOW;EAChB,CAAA,CAAA,EAECE,KAAKb,QAAOc,OAAO,EACnBC,YAAY;;IAEXC,aAAa;;;;EAIf,CAAA;AACJ,CAAA;AAGO,IAAMC,aAAa,OAAOC,QAAkBC,OAAAA;AACjD,QAAMX,aAAa,MAAMY,QAAQC,IAC/BH,OAAOI,IAAI,OAAOV,UAAAA;AAChB,UAAM,EAAEW,QAAO,IAAK,MAAMJ,GAAGK,MAAMtB,MAAMuB,OAAOxB,OAAOyB,KAAKd,OAAO;MAAEe,MAAM;IAAS,CAAA,CAAA,CAAA,EAAKC,IAAG;AAC5FvB,IAAAA,KAAIwB,KAAK,gCAAgC;MAAEN;MAASX;IAAM,GAAA;;;;;;AAC1D,WAAOW,QAAQO,SAAS,IAAI;MAAC;QAAEpB,IAAIa,QAAQ,CAAA,EAAGb,GAAGqB,SAAQ;QAAInB;MAAM;QAAK,CAAA;EAC1E,CAAA,CAAA;AAEF,SAAO;IAAEJ,YAAYA,WAAWwB,KAAI;EAAG;AACzC;AAMO,IAAMC,mBAAmB,CAACP,MAAcR,WAAAA;AAC7C,aAAWN,SAASM,OAAOV,YAAY;AACrC,QAAI,CAACL,SAAS+B,QAAQtB,MAAMF,EAAE,GAAG;AAC/B;IACF;AAGA,UAAMyB,QAAQ,IAAIC,OAAOxB,MAAMA,MAAMyB,QAAQ,uBAAuB,MAAA,GAAS,IAAA;AAC7EX,WAAOA,KAAKW,QAAQF,OAAO,IAAIvB,MAAMA,KAAK,KAAKR,IAAIkC,kBAAkB1B,MAAMF,EAAE,EAAEqB,SAAQ,CAAA,GAAK;EAC9F;AAEA,SAAOL;AACT;;;ACzDA;;;AFmBO,IAAMa,wBAA+EC,eAAe;EACzGC,aAAa;EACbC,aAAaC;EACbC,cAAcC;EACdC,SAAS,OAAO,EAAEC,MAAM,EAAEC,SAASC,QAAO,GAAIC,QAAO,MAAE;AACrD,UAAMC,YAAYC,YAAYC,IAAG;AACjC,UAAMC,KAAKJ,QAAQK,WAAWC,UAAUA,SAAS;AACjD,UAAMC,UAAU,IAAIC,UAAU;MAAEC,gBAAgB;IAAa,CAAA;AAC7D,UAAMC,SAAS,MAAMH,QAAQI,cAAcC,kBAAkB;MAC3DC,mBAAmB;QACjBC,OAAO;MACT;MACAC,QAAQX,GAAGW;MACXC,cAAcC;MACdC,SAAS;QACPC,IAAIC,KAAKC,UAASC,SAAS;UACzBC,UAAS,oBAAIC,KAAAA,GAAOC,YAAW;UAC/BC,QAAQ;YAAEC,MAAM;UAAO;UACvBC,QAAQ;YACN;cACEC,MAAM;cACNC,MAAM,YAAYC,KAAKC,UAAUjC,OAAAA,CAAAA;YACnC;YACA;cACE8B,MAAM;cACNC,MAAM,eAAeC,KAAKC,UAAUlC,QAAQ8B,MAAM,CAAA;YACpD;;QAEJ,CAAA;;MAEFK,WAAW,CAAA;MACXC,QAAQ;MACRC,OAAO,CAAA;MACPC,cAAc,IAAIC,aAAa,CAAA,CAAE;IACnC,CAAA;AAEA,WAAO;MACLvC,SAASwC,OAAOjB,UAASC,SAAS;QAChC,GAAGxB;QACH8B,QAAQ9B,QAAQ8B,OAAOW,IAAI,CAACC,OAAOC,MACjCD,MAAMX,SAAS,eACXW,QACA;UACE,GAAGA;UACHV,MAAMY,iBAAiBF,MAAMV,MAAMpB,MAAAA;QACrC,CAAA;MAER,CAAA;MACAiC,aAAazC,YAAYC,IAAG,IAAKF;IACnC;EACF;AACF,CAAA;;;AG/DA,SAAS2C,gBAAAA,qBAAoB;AAC7B,SAASC,UAAAA,eAAc;AACvB,SAASC,iBAAiBC,kBAAAA,uBAAsB;AAChD,SAASC,iBAAiB;AAC1B,SAASC,OAAAA,YAAW;AACpB,SAASC,YAAAA,iBAAgB;;AAMlB,IAAMC,kBAAkBC,gBAAe;EAC5CC,aAAa;EACbC,aAAaC;EACbC,cAAcC;EACdC,SAAS,OAAO,EAAEC,MAAM,EAAEC,SAASC,QAAO,GAAIC,QAAO,MAAE;AACrDC,IAAAA,KAAIC,KAAK,SAAS;MAAEJ;MAASC;IAAQ,GAAA;;;;;;AACrC,UAAMI,YAAYC,YAAYC,IAAG;AACjC,UAAM,EAAEC,GAAE,IAAKN,QAAQO,WAAWC,eAAAA;AAClC,UAAM,EAAEC,qBAAAA,qBAAmB,IAAK,MAAM,OAAO,0CAAA;AAE7C,UAAMC,kBAAkBC,QAAQC,IAC9Bd,QAAQe,OAAOC,IAAI,OAAOC,UAAAA;AACxBC,gBAAUD,MAAME,SAAS,gBAAgBF,MAAME,SAAS,QAAQ,wBAAA;;;;;;;;;AAChE,aAAOR,qBAAoBM,MAAMG,IAAI;IACvC,CAAA,CAAA,EACAC,KAAK,CAACC,cAAaA,UAASC,KAAI,CAAA;AAClC,UAAMD,WAAWrB,SAASuB,UAAU,MAAMC,cAAab,iBAAiBX,QAAQuB,OAAO,IAAI,MAAMZ;AACjGT,IAAAA,KAAIC,KAAK,YAAY;MAAEkB;IAAS,GAAA;;;;;;AAChC,UAAMI,kBAAkB,MAAMC,WAC5BL,SAASN,IAAI,CAACY,WAAWA,OAAOC,IAAI,GACpCrB,EAAAA;AAEF,UAAMsB,uBAAuB9B,QAAQe,OAAOC,IAAI,CAACC,OAAOc,MAAAA;AACtDb,gBAAUD,MAAME,SAAS,gBAAgBF,MAAME,SAAS,QAAQ,wBAAA;;;;;;;;;AAChE,aAAO;QAAE,GAAGF;QAAOG,MAAMY,iBAAiBf,MAAME,MAAMO,eAAAA;MAAiB;IACzE,CAAA;AACA,UAAMO,wBAAwBC,QAAOC,UAASC,SAAS;MACrD,GAAGpC;MACHe,QAAQe;IACV,CAAA;AAEA3B,IAAAA,KAAIC,KAAK,UAAU;MAAE6B;IAAsB,GAAA;;;;;;AAE3C,WAAO;MACLjC,SAASiC;MACTI,aAAa/B,YAAYC,IAAG,IAAKF;IACnC;EACF;AACF,CAAA;",
+  "names": ["Schema", "asyncTimeout", "Expando", "log", "DataType", "ExtractionInput", "Struct", "message", "Message", "objects", "optional", "Array", "options", "timeout", "Number", "fallbackToRaw", "Boolean", "ExtractionOutput", "timeElapsed", "processTranscriptMessage", "params", "executor", "invoke", "function", "input", "error", "warn", "AiService", "ToolRegistry", "Obj", "create", "defineFunction", "DataType", "Schema", "Filter", "Query", "ObjectId", "DXN", "log", "ReferencedQuotes", "Struct", "references", "Array", "id", "String", "quote", "pipe", "mutable", "annotations", "description", "findQuotes", "quotes", "db", "Promise", "all", "map", "objects", "query", "select", "text", "type", "run", "info", "length", "toString", "flat", "insertReferences", "isValid", "regex", "RegExp", "replace", "fromLocalObjectId", "extractionAnthropicFn", "defineFunction", "description", "inputSchema", "ExtractionInput", "outputSchema", "ExtractionOutput", "handler", "data", "message", "objects", "context", "startTime", "performance", "now", "ai", "getService", "AiService", "session", "AiSession", "operationModel", "result", "runStructured", "ReferencedQuotes", "generationOptions", "model", "client", "systemPrompt", "PROMPT", "history", "Obj", "make", "DataType", "Message", "created", "Date", "toISOString", "sender", "role", "blocks", "_tag", "text", "JSON", "stringify", "artifacts", "prompt", "tools", "toolResolver", "ToolRegistry", "create", "map", "block", "i", "insertReferences", "timeElapsed", "asyncTimeout", "create", "DatabaseService", "defineFunction", "invariant", "log", "DataType", "extractionNerFn", "defineFunction", "description", "inputSchema", "ExtractionInput", "outputSchema", "ExtractionOutput", "handler", "data", "message", "options", "context", "log", "info", "startTime", "performance", "now", "db", "getService", "DatabaseService", "extractFullEntities", "entitiesPromise", "Promise", "all", "blocks", "map", "block", "invariant", "_tag", "text", "then", "entities", "flat", "timeout", "asyncTimeout", "quoteReferences", "findQuotes", "entity", "word", "blocksWithReferences", "i", "insertReferences", "messageWithReferences", "create", "DataType", "Message", "timeElapsed"]
+}