@dxos/assistant 0.8.4-main.67995b8 → 0.8.4-main.72ec0f3

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

@@ -1,7 +1,7 @@
 {
   "version": 3,
-  "sources": ["../../../../src/extraction/extraction.ts", "../../../../src/extraction/extraction-llm-function/extraction-llm-function.ts", "raw-loader:./instructions.tpl?raw", "../../../../src/extraction/quotes.ts", "../../../../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// TODO(burdon): Fix.\n// @ts-nocheck\n\nimport { ToolRegistry, AiService } from '@dxos/ai';\nimport { Obj } from '@dxos/echo';\nimport { create } from '@dxos/echo-schema';\nimport { defineFunction, type FunctionDefinition } from '@dxos/functions';\nimport { type ContentBlock, DataType } from '@dxos/schema';\n\nimport PROMPT from './instructions.tpl?raw';\nimport { AiSession } from '../../session';\nimport { ExtractionInput, ExtractionOutput } from '../extraction';\nimport { insertReferences, ReferencedQuotes } from '../quotes';\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);\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", "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\nimport { Schema } from 'effect';\n\nimport { Query, Filter } 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", "//\n// Copyright 2025 DXOS.org\n//\n\n// TODO(burdon): Fix.\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,cAAcC,iBAAiB;AACxC,SAASC,WAAW;AACpB,SAASC,cAAc;AACvB,SAASC,sBAA+C;AACxD,SAA4BC,YAAAA,iBAAgB;;;ACX5C;;;ACIA,SAASC,UAAAA,eAAc;AAEvB,SAASC,OAAOC,cAAc;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,MAAMvB,MAAMwB,OAAOvB,OAAOwB,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;;;AFvCO,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,SAAAA;AAC9B,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;;;AG9DA,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", "ToolRegistry", "AiService", "Obj", "create", "defineFunction", "DataType", "Schema", "Query", "Filter", "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"]
+  "sources": ["../../../../src/extraction/extraction.ts", "../../../../src/extraction/extraction-llm-function/extraction-llm-function.ts", "../../../../src/extraction/extraction-ner-function/extraction-ner-function.ts", "../../../../src/extraction/extraction-ner-function/named-entity-recognition.ts", "../../../../src/extraction/quotes.ts"],
+  "sourcesContent": ["//\n// Copyright 2025 DXOS.org\n//\n\nimport * as Schema from 'effect/Schema';\n\nimport { asyncTimeout } from '@dxos/async';\nimport { Expando } from '@dxos/echo/internal';\nimport { type FunctionDefinition } from '@dxos/functions';\nimport { type FunctionExecutor } from '@dxos/functions-runtime';\nimport { log } from '@dxos/log';\nimport { Message } from '@dxos/types';\n\nexport const ExtractionInput = Schema.Struct({\n  message: Message.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: Message.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\nimport { type FunctionDefinition, defineFunction } from '@dxos/functions';\n\nimport { ExtractionInput, ExtractionOutput } from '../extraction';\n\nexport const extractionAnthropicFunction: FunctionDefinition<ExtractionInput, ExtractionOutput> = defineFunction({\n  key: 'dxos.org/function/extraction/extract-entities',\n  name: 'Extract Entities',\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    throw new Error('not implemented');\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.Message, {\n    //       created: new Date().toISOString(),\n    //       sender: { role: 'user' },\n    //       blocks: [\n    //         {\n    //           _tag: 'text',\n    //           text: `<context>${JSON.stringify(objects)}</context>`,\n    //         } satisfies DataType.ContentBlock.Text,\n    //         {\n    //           _tag: 'text',\n    //           text: `<transcript>${JSON.stringify(message.blocks)}</transcript>`,\n    //         } satisfies DataType.ContentBlock.Text,\n    //       ],\n    //     }),\n    //   ],\n    //   artifacts: [],\n    //   prompt: '',\n    //   tools: [],\n    // } as any); // TODO(burdon): Rewrite test.\n\n    // return {\n    //   timeElapsed: performance.now() - startTime,\n    //   message: create(DataType.Message.Message, {\n    //     ...message,\n    //     blocks: message.blocks.map((block) =>\n    //       block._tag !== 'transcript'\n    //         ? block\n    //         : {\n    //             ...block,\n    //             text: insertReferences(block.text, result),\n    //           },\n    //     ),\n    //   }),\n    // };\n  },\n});\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport { defineFunction } from '@dxos/functions';\n\nimport { ExtractionInput, ExtractionOutput } from '../extraction';\n\n/**\n * Named Entity Recognition model\n */\n// TODO(mykola): Make it use vector index to query objects and not pass objects as input.\nexport const extractionNerFunction = defineFunction({\n  key: 'dxos.org/function/extraction/extract-entities',\n  name: 'Extract Entities',\n  description: 'Extracts named entities from a text',\n  inputSchema: ExtractionInput,\n  outputSchema: ExtractionOutput,\n  handler: async ({ data: { message, options }, context }) => {\n    throw new Error('not implemented');\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');\n    //     return extractFullEntities(block.text);\n    //   }),\n    // ).then((entities) => entities.flat());\n    // const entities = options?.timeout ? await asyncTimeout(entitiesPromise, options.timeout) : await entitiesPromise;\n\n    // log.info('entities', { entities });\n    // const quoteReferences = await findQuotes(\n    //   entities.map((entity) => entity.word),\n    //   db,\n    // );\n\n    // const blocksWithReferences = message.blocks.map((block) => {\n    //   invariant(block._tag === 'transcript' || block._tag === 'text');\n    //   return { ...block, text: insertReferences(block._tag, quoteReferences) };\n    // });\n\n    // const messageWithReferences = create(Message.Message, {\n    //   ...message,\n    //   blocks: blocksWithReferences,\n    // });\n\n    // log.info('output', { messageWithReferences });\n    // return {\n    //   message: messageWithReferences,\n    //   timeElapsed: performance.now() - startTime,\n    // };\n  },\n});\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport {\n  type TokenClassificationOutput,\n  type TokenClassificationPipelineType,\n  type TokenClassificationSingle,\n  pipeline,\n} from '@xenova/transformers';\n\nimport { log } from '@dxos/log';\n\nlet _ner: Promise<TokenClassificationPipelineType>;\n\n/**\n * Named Entity Recognition pipeline.\n * Initializes the pipeline on first call.\n * @returns The singleton promise that resolves to a token classification pipeline.\n */\nexport const getNer = () => {\n  if (!_ner) {\n    _ner = pipeline('ner', 'Xenova/bert-base-NER').then((ner) => {\n      log.info('NER model is ready');\n      return ner;\n    });\n  }\n\n  return _ner;\n};\n\n/**\n * Extracts entities from text using the named entity recognition pipeline and does simple grouping of tokens.\n */\nexport const extractFullEntities = async (text: string): Promise<TokenClassificationSingle[]> => {\n  const ner = await getNer();\n  const entities = (await ner(text)) as TokenClassificationOutput;\n  const tokenGroups = createTokenGroups(entities);\n  const combinedEntities = tokenGroups.map(combineNerTokens);\n  return combinedEntities;\n};\n\nexport const createTokenGroups = (tokens: TokenClassificationSingle[]): TokenClassificationSingle[][] => {\n  const tokenGroups: TokenClassificationSingle[][] = [];\n\n  const canBeCombined = (prevToken: TokenClassificationSingle, currentToken: TokenClassificationSingle) => {\n    const [currentState, currentEntity] = currentToken.entity.split('-'); // B, I, O\n    const [prevState, prevEntity] = prevToken.entity.split('-'); // B, I, O\n    if (prevEntity !== currentEntity) {\n      return false;\n    }\n    if (prevState === 'O') {\n      return false;\n    }\n    if (currentState === 'I') {\n      return true;\n    }\n    if ((prevState === 'B' || prevState === 'I') && currentState === 'B') {\n      return currentToken.word.startsWith('#');\n    }\n    return false;\n  };\n\n  for (const token of tokens) {\n    if (tokenGroups.length === 0) {\n      tokenGroups.push([token]);\n      continue;\n    }\n    const lastTokenGroup = tokenGroups.at(-1)!;\n    if (canBeCombined(lastTokenGroup.at(-1)!, token)) {\n      lastTokenGroup.push(token);\n      continue;\n    }\n    tokenGroups.push([token]);\n  }\n\n  return tokenGroups;\n};\n\nexport const combineNerTokens = (group: TokenClassificationSingle[]): TokenClassificationSingle => {\n  const combinedToken: TokenClassificationSingle = {\n    entity: group.at(0)!.entity.split('-')[1],\n    score: group.reduce((acc, token) => acc + token.score, 0) / group.length,\n    index: group.at(0)!.index,\n    word: group.at(0)!.word,\n    start: group.at(0)!.start,\n    end: group.at(-1)!.end,\n  };\n\n  for (const token of group.slice(1)) {\n    if (token.word.startsWith('#')) {\n      combinedToken.word += token.word.replace(/^#+/, '');\n    } else {\n      combinedToken.word += ' ' + token.word;\n    }\n  }\n\n  return combinedToken;\n};\n", "//\n// Copyright 2025 DXOS.org\n//\n\nimport * as Schema from 'effect/Schema';\n\nimport { Filter, Query } from '@dxos/echo';\nimport { ObjectId } from '@dxos/echo/internal';\nimport { type EchoDatabase } from '@dxos/echo-db';\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"],
+  "mappings": ";;;;AAIA,YAAYA,YAAY;AAExB,SAASC,oBAAoB;AAC7B,SAASC,eAAe;AAGxB,SAASC,WAAW;AACpB,SAASC,eAAe;;AAEjB,IAAMC,kBAAyBC,cAAO;EAC3CC,SAASH,QAAQA;EACjBI,SAAgBC,gBAAgBC,aAAMR,OAAAA,CAAAA;EACtCS,SAAgBF,gBACPH,cAAO;IACZM,SAAgBH,gBAAgBI,aAAM;IACtCC,eAAsBL,gBAAgBM,cAAO;EAC/C,CAAA,CAAA;AAEJ,CAAA;AAGO,IAAMC,mBAA0BV,cAAO;EAC5CC,SAASH,QAAQA;EACjBa,aAAoBJ;AACtB,CAAA;AA6BO,IAAMK,2BAA2B,OAAOC,WAAAA;AAC7C,MAAI;AACF,QAAIA,OAAOR,SAASC,WAAWO,OAAOR,QAAQC,UAAU,GAAG;AACzD,aAAO,MAAMX,aAAakB,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;AACjCX,UAAIsB,KAAK,uDAAuD;QAAED;MAAM,GAAA;;;;;;AACxE,aAAO;QACLjB,SAASY,OAAOI,MAAMhB;QACtBU,aAAa;MACf;IACF,OAAO;AACL,YAAMO;IACR;EACF;AACF;;;ACvEA,SAAkCE,sBAAsB;AAIjD,IAAMC,8BAAqFC,eAAe;EAC/GC,KAAK;EACLC,MAAM;EACNC,aAAa;EACbC,aAAaC;EACbC,cAAcC;EACdC,SAAS,OAAO,EAAEC,MAAM,EAAEC,SAASC,QAAO,GAAIC,QAAO,MAAE;AACrD,UAAM,IAAIC,MAAM,iBAAA;EA6ClB;AACF,CAAA;;;ACzDA,SAASC,kBAAAA,uBAAsB;AAQxB,IAAMC,wBAAwBC,gBAAe;EAClDC,KAAK;EACLC,MAAM;EACNC,aAAa;EACbC,aAAaC;EACbC,cAAcC;EACdC,SAAS,OAAO,EAAEC,MAAM,EAAEC,SAASC,QAAO,GAAIC,QAAO,MAAE;AACrD,UAAM,IAAIC,MAAM,iBAAA;EAmClB;AACF,CAAA;;;ACnDA,SAIEC,gBACK;AAEP,SAASC,OAAAA,YAAW;;AAEpB,IAAIC;AAOG,IAAMC,SAAS,MAAA;AACpB,MAAI,CAACD,MAAM;AACTA,WAAOF,SAAS,OAAO,sBAAA,EAAwBI,KAAK,CAACC,QAAAA;AACnDJ,MAAAA,KAAIK,KAAK,sBAAA,QAAA;;;;;;AACT,aAAOD;IACT,CAAA;EACF;AAEA,SAAOH;AACT;AAKO,IAAMK,sBAAsB,OAAOC,SAAAA;AACxC,QAAMH,MAAM,MAAMF,OAAAA;AAClB,QAAMM,WAAY,MAAMJ,IAAIG,IAAAA;AAC5B,QAAME,cAAcC,kBAAkBF,QAAAA;AACtC,QAAMG,mBAAmBF,YAAYG,IAAIC,gBAAAA;AACzC,SAAOF;AACT;AAEO,IAAMD,oBAAoB,CAACI,WAAAA;AAChC,QAAML,cAA6C,CAAA;AAEnD,QAAMM,gBAAgB,CAACC,WAAsCC,iBAAAA;AAC3D,UAAM,CAACC,cAAcC,aAAAA,IAAiBF,aAAaG,OAAOC,MAAM,GAAA;AAChE,UAAM,CAACC,WAAWC,UAAAA,IAAcP,UAAUI,OAAOC,MAAM,GAAA;AACvD,QAAIE,eAAeJ,eAAe;AAChC,aAAO;IACT;AACA,QAAIG,cAAc,KAAK;AACrB,aAAO;IACT;AACA,QAAIJ,iBAAiB,KAAK;AACxB,aAAO;IACT;AACA,SAAKI,cAAc,OAAOA,cAAc,QAAQJ,iBAAiB,KAAK;AACpE,aAAOD,aAAaO,KAAKC,WAAW,GAAA;IACtC;AACA,WAAO;EACT;AAEA,aAAWC,SAASZ,QAAQ;AAC1B,QAAIL,YAAYkB,WAAW,GAAG;AAC5BlB,kBAAYmB,KAAK;QAACF;OAAM;AACxB;IACF;AACA,UAAMG,iBAAiBpB,YAAYqB,GAAG,EAAC;AACvC,QAAIf,cAAcc,eAAeC,GAAG,EAAC,GAAKJ,KAAAA,GAAQ;AAChDG,qBAAeD,KAAKF,KAAAA;AACpB;IACF;AACAjB,gBAAYmB,KAAK;MAACF;KAAM;EAC1B;AAEA,SAAOjB;AACT;AAEO,IAAMI,mBAAmB,CAACkB,UAAAA;AAC/B,QAAMC,gBAA2C;IAC/CZ,QAAQW,MAAMD,GAAG,CAAA,EAAIV,OAAOC,MAAM,GAAA,EAAK,CAAA;IACvCY,OAAOF,MAAMG,OAAO,CAACC,KAAKT,UAAUS,MAAMT,MAAMO,OAAO,CAAA,IAAKF,MAAMJ;IAClES,OAAOL,MAAMD,GAAG,CAAA,EAAIM;IACpBZ,MAAMO,MAAMD,GAAG,CAAA,EAAIN;IACnBa,OAAON,MAAMD,GAAG,CAAA,EAAIO;IACpBC,KAAKP,MAAMD,GAAG,EAAC,EAAIQ;EACrB;AAEA,aAAWZ,SAASK,MAAMQ,MAAM,CAAA,GAAI;AAClC,QAAIb,MAAMF,KAAKC,WAAW,GAAA,GAAM;AAC9BO,oBAAcR,QAAQE,MAAMF,KAAKgB,QAAQ,OAAO,EAAA;IAClD,OAAO;AACLR,oBAAcR,QAAQ,MAAME,MAAMF;IACpC;EACF;AAEA,SAAOQ;AACT;;;AC9FA,YAAYS,aAAY;AAExB,SAASC,QAAQC,aAAa;AAC9B,SAASC,gBAAgB;AAEzB,SAASC,WAAW;AACpB,SAASC,OAAAA,YAAW;;AAEb,IAAMC,mBAA0BC,eAAO;EAC5CC,YAAmBC,cACVF,eAAO;IACZG,IAAWC;IACXC,OAAcD;EAChB,CAAA,CAAA,EAECE,KAAYC,eAAO,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;",
+  "names": ["Schema", "asyncTimeout", "Expando", "log", "Message", "ExtractionInput", "Struct", "message", "objects", "optional", "Array", "options", "timeout", "Number", "fallbackToRaw", "Boolean", "ExtractionOutput", "timeElapsed", "processTranscriptMessage", "params", "executor", "invoke", "function", "input", "error", "warn", "defineFunction", "extractionAnthropicFunction", "defineFunction", "key", "name", "description", "inputSchema", "ExtractionInput", "outputSchema", "ExtractionOutput", "handler", "data", "message", "objects", "context", "Error", "defineFunction", "extractionNerFunction", "defineFunction", "key", "name", "description", "inputSchema", "ExtractionInput", "outputSchema", "ExtractionOutput", "handler", "data", "message", "options", "context", "Error", "pipeline", "log", "_ner", "getNer", "then", "ner", "info", "extractFullEntities", "text", "entities", "tokenGroups", "createTokenGroups", "combinedEntities", "map", "combineNerTokens", "tokens", "canBeCombined", "prevToken", "currentToken", "currentState", "currentEntity", "entity", "split", "prevState", "prevEntity", "word", "startsWith", "token", "length", "push", "lastTokenGroup", "at", "group", "combinedToken", "score", "reduce", "acc", "index", "start", "end", "slice", "replace", "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"]
 }