@sentry/core/build/esm/tracing/langchain/embeddings.js

Base implementation for all Sentry JavaScript SDKs

import { captureException } from '../../exports.js';
import { SEMANTIC_ATTRIBUTE_SENTRY_OP, SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN } from '../../semanticAttributes.js';
import { startSpan } from '../trace.js';
import { GEN_AI_EMBEDDINGS_INPUT_ATTRIBUTE, GEN_AI_EMBEDDINGS_OPERATION_ATTRIBUTE, GEN_AI_REQUEST_MODEL_ATTRIBUTE, GEN_AI_SYSTEM_ATTRIBUTE, GEN_AI_REQUEST_DIMENSIONS_ATTRIBUTE, GEN_AI_REQUEST_ENCODING_FORMAT_ATTRIBUTE, GEN_AI_OPERATION_NAME_ATTRIBUTE } from '../ai/gen-ai-attributes.js';
import { resolveAIRecordingOptions } from '../ai/utils.js';
import { LANGCHAIN_ORIGIN } from './constants.js';

/**
 * Infers the AI provider system name from the embedding class instance.
 */
function inferSystemFromInstance(instance) {
  const name = (instance.constructor )?.name ?? '';
  if (name.includes('OpenAI')) return 'openai';
  if (name.includes('Google')) return 'google_genai';
  if (name.includes('Mistral')) return 'mistralai';
  if (name.includes('Vertex')) return 'google_vertexai';
  if (name.includes('Bedrock')) return 'aws_bedrock';
  if (name.includes('Ollama')) return 'ollama';
  if (name.includes('Cloudflare')) return 'cloudflare';
  if (name.includes('Cohere')) return 'cohere';
  return 'langchain';
}

/**
 * Extracts span attributes from a LangChain embedding class instance.
 */
function extractEmbeddingAttributes(instance) {
  const embeddingsInstance = (instance ?? {}) ;

  const attributes = {
    [SEMANTIC_ATTRIBUTE_SENTRY_ORIGIN]: LANGCHAIN_ORIGIN,
    [SEMANTIC_ATTRIBUTE_SENTRY_OP]: GEN_AI_EMBEDDINGS_OPERATION_ATTRIBUTE,
    [GEN_AI_OPERATION_NAME_ATTRIBUTE]: 'embeddings',
    [GEN_AI_REQUEST_MODEL_ATTRIBUTE]: embeddingsInstance.model ?? 'unknown',
  };

  attributes[GEN_AI_SYSTEM_ATTRIBUTE] = inferSystemFromInstance(embeddingsInstance);
  if ('dimensions' in embeddingsInstance) {
    attributes[GEN_AI_REQUEST_DIMENSIONS_ATTRIBUTE] = embeddingsInstance.dimensions;
  }
  if ('encodingFormat' in embeddingsInstance) {
    attributes[GEN_AI_REQUEST_ENCODING_FORMAT_ATTRIBUTE] = embeddingsInstance.encodingFormat;
  }

  return attributes;
}

/**
 * Wraps a LangChain embedding method (embedQuery or embedDocuments) to create Sentry spans.
 *
 * Used internally by the Node.js auto-instrumentation to patch embedding class prototypes.
 */
function instrumentEmbeddingMethod(
  originalMethod,
  options = {},
) {
  const { recordInputs } = resolveAIRecordingOptions(options);

  return new Proxy(originalMethod, {
    apply(target, thisArg, args) {
      const attributes = extractEmbeddingAttributes(thisArg);
      const modelName = attributes[GEN_AI_REQUEST_MODEL_ATTRIBUTE] || 'unknown';

      if (recordInputs) {
        const input = args[0];
        if (input != null) {
          attributes[GEN_AI_EMBEDDINGS_INPUT_ATTRIBUTE] = typeof input === 'string' ? input : JSON.stringify(input);
        }
      }

      return startSpan(
        {
          name: `embeddings ${modelName}`,
          op: GEN_AI_EMBEDDINGS_OPERATION_ATTRIBUTE,
          attributes: attributes ,
        },
        () => {
          return Reflect.apply(target, thisArg, args).then(undefined, error => {
            captureException(error, {
              mechanism: { handled: false, type: 'auto.ai.langchain' },
            });
            throw error;
          });
        },
      );
    },
  }) ;
}

/**
 * Wraps a LangChain embeddings instance to create Sentry spans for `embedQuery` and `embedDocuments` calls.
 *
 * Use this in non-Node runtimes (Cloudflare, browser, etc.) where auto-instrumentation is not available.
 *
 * @example
 * ```javascript
 * import * as Sentry from '@sentry/cloudflare';
 * import { OpenAIEmbeddings } from '@langchain/openai';
 *
 * const embeddings = Sentry.instrumentLangChainEmbeddings(
 *   new OpenAIEmbeddings({ model: 'text-embedding-3-small' })
 * );
 *
 * await embeddings.embedQuery('Hello world');
 * await embeddings.embedDocuments(['doc1', 'doc2']);
 * ```
 */
function instrumentLangChainEmbeddings(instance, options) {
  const embeddingsInstance = instance ;

  if (typeof embeddingsInstance.embedQuery === 'function') {
    embeddingsInstance.embedQuery = instrumentEmbeddingMethod(
      embeddingsInstance.embedQuery ,
      options,
    );
  }

  if (typeof embeddingsInstance.embedDocuments === 'function') {
    embeddingsInstance.embedDocuments = instrumentEmbeddingMethod(
      embeddingsInstance.embedDocuments ,
      options,
    );
  }

  return instance;
}

export { instrumentEmbeddingMethod, instrumentLangChainEmbeddings };
//# sourceMappingURL=embeddings.js.map