@tanstack/ai-cohere 0.4.6 → 0.4.9

package/dist/esm/adapters/embedding.js

@@ -2,7 +2,7 @@ import { getCohereApiKeyFromEnv, resolveCohereTransport } from "../utils/client.
 import { BaseEmbeddingAdapter } from "@tanstack/ai/adapters";
 import { toRunErrorPayload } from "@tanstack/ai/adapter-internals";
 import { arrayBufferToBase64, generateId } from "@tanstack/ai-utils";
-import { resolveEmbeddingInput } from "@tanstack/ai";
+import { isFileSource, resolveEmbeddingInput, unsupportedFileSourceError } from "@tanstack/ai";
 //#region src/adapters/embedding.ts
 var DEFAULT_TIMEOUT_MS = 3e4;
 /**
@@ -138,6 +138,7 @@ var CohereEmbeddingAdapter = class extends BaseEmbeddingAdapter {
 	*/
 	async resolveImageUrl(image) {
 		const source = image.source;
+		if (isFileSource(source)) throw unsupportedFileSourceError(this.name);
 		if (source.type === "data") return `data:${source.mimeType};base64,${source.value}`;
 		if (source.value.startsWith("data:")) return source.value;
 		if (!this.clientConfig.allowUrlFetch) throw new Error("Cohere does not fetch remote image URLs; pass base64 data or a data: URI (or enable config.allowUrlFetch to have the adapter download it)");

package/dist/esm/adapters/embedding.js.map

@@ -1 +1 @@
-{"version":3,"file":"embedding.js","names":[],"sources":["../../../src/adapters/embedding.ts"],"sourcesContent":["import { BaseEmbeddingAdapter } from '@tanstack/ai/adapters'\nimport { toRunErrorPayload } from '@tanstack/ai/adapter-internals'\nimport { arrayBufferToBase64, generateId } from '@tanstack/ai-utils'\nimport { resolveEmbeddingInput } from '@tanstack/ai'\nimport { getCohereApiKeyFromEnv, resolveCohereTransport } from '../utils/client'\nimport type {\n  EmbeddingOptions,\n  EmbeddingResult,\n  ImagePart,\n  TokenUsage,\n} from '@tanstack/ai'\nimport type {\n  CohereEmbeddingModel,\n  CohereEmbeddingModelInputModalitiesByName,\n  CohereEmbeddingModelProviderOptionsByName,\n} from '../model-meta'\nimport type { CohereEmbeddingProviderOptions } from '../embedding/embedding-provider-options'\nimport type { CohereClientConfig } from '../utils/client'\n\n/**\n * Configuration for Cohere embedding adapter.\n */\nexport interface CohereEmbeddingConfig extends CohereClientConfig {}\n\nconst DEFAULT_TIMEOUT_MS = 30_000\n\n/**\n * Returns true when `url` is malformed, non-http(s), or targets a private /\n * loopback / link-local host. Used to block SSRF via `allowUrlFetch`.\n */\nfunction isPrivateOrInternalUrl(url: string): boolean {\n  let parsed: URL\n  try {\n    parsed = new URL(url)\n  } catch {\n    return true\n  }\n  if (parsed.protocol !== 'http:' && parsed.protocol !== 'https:') {\n    return true\n  }\n  const host = parsed.hostname.toLowerCase()\n  if (\n    host === 'localhost' ||\n    host.endsWith('.localhost') ||\n    host === '::1' ||\n    host === '[::1]' ||\n    host.startsWith('127.') ||\n    host.startsWith('10.') ||\n    host.startsWith('192.168.') ||\n    host.startsWith('169.254.') ||\n    /^172\\.(1[6-9]|2\\d|3[01])\\./.test(host)\n  ) {\n    return true\n  }\n  return false\n}\n\nasync function fetchWithTimeout(\n  url: string,\n  init: RequestInit | undefined,\n  timeoutMs: number,\n): Promise<Response> {\n  const controller = new AbortController()\n  const timeoutId = setTimeout(() => controller.abort(), timeoutMs)\n  try {\n    return await fetch(url, { ...init, signal: controller.signal })\n  } finally {\n    clearTimeout(timeoutId)\n  }\n}\n\n/** One content part of a Cohere v2/embed fused input. */\ntype CohereEmbedContentPart =\n  | { type: 'text'; text: string }\n  | { type: 'image_url'; image_url: { url: string } }\n\n/** Wire shape of the Cohere v2/embed request body. */\ninterface CohereEmbedRequestBody {\n  model: string\n  inputs: Array<{ content: Array<CohereEmbedContentPart> }>\n  input_type: CohereEmbeddingProviderOptions['inputType']\n  embedding_types: ['float']\n  truncate?: 'NONE' | 'START' | 'END'\n  output_dimension?: number\n}\n\n/** Wire shape of the Cohere v2/embed response (fields the adapter reads). */\ninterface CohereEmbedResponse {\n  id?: string\n  embeddings?: {\n    float?: Array<Array<number>>\n  }\n  meta?: {\n    billed_units?: {\n      input_tokens?: number\n      images?: number\n    }\n  }\n}\n\n/**\n * Cohere Embedding Adapter\n *\n * Tree-shakeable adapter for Cohere multimodal embeddings (embed-v4.0),\n * implemented with plain `fetch` against the v2/embed endpoint — no Cohere\n * SDK dependency.\n *\n * Features:\n * - Batch embedding (one request for the whole input array)\n * - Multimodal inputs: text, images, and fused text+image items (one vector\n *   per input item)\n * - Matryoshka dimension reduction via the top-level `dimensions` option\n *   (mapped to Cohere's `output_dimension`)\n */\nexport class CohereEmbeddingAdapter<\n  TModel extends CohereEmbeddingModel,\n> extends BaseEmbeddingAdapter<\n  TModel,\n  CohereEmbeddingProviderOptions,\n  CohereEmbeddingModelProviderOptionsByName,\n  CohereEmbeddingModelInputModalitiesByName\n> {\n  readonly name = 'cohere' as const\n\n  protected clientConfig: CohereEmbeddingConfig\n\n  constructor(config: CohereEmbeddingConfig, model: TModel) {\n    super(model, {})\n    this.clientConfig = config\n  }\n\n  async createEmbeddings(\n    options: EmbeddingOptions<CohereEmbeddingProviderOptions>,\n  ): Promise<EmbeddingResult> {\n    const { model, logger, modelOptions } = options\n\n    try {\n      // The provider options type makes `modelOptions` required at the\n      // embed() call site; this guard covers untyped/dynamic callers.\n      const inputType: CohereEmbeddingProviderOptions['inputType'] | undefined =\n        modelOptions?.inputType\n      if (!inputType) {\n        throw new Error(\n          `Cohere embeddings require modelOptions.inputType ('search_document' | 'search_query' | 'classification' | 'clustering').`,\n        )\n      }\n\n      const resolved = resolveEmbeddingInput(options.input)\n      const inputs = await Promise.all(\n        resolved.map(async (item) => {\n          const content: Array<CohereEmbedContentPart> = item.texts.map(\n            (text) => ({ type: 'text', text }),\n          )\n          for (const image of item.images) {\n            content.push({\n              type: 'image_url',\n              image_url: { url: await this.resolveImageUrl(image) },\n            })\n          }\n          return { content }\n        }),\n      )\n\n      // embedding_types is pinned to ['float'] (overriding any disagreeing\n      // modelOptions.embeddingTypes) so vectors are always number[].\n      const body: CohereEmbedRequestBody = {\n        model,\n        inputs,\n        input_type: inputType,\n        embedding_types: ['float'],\n      }\n      const truncate = modelOptions?.truncate\n      if (truncate !== undefined) {\n        body.truncate = truncate\n      }\n      if (options.dimensions !== undefined) {\n        body.output_dimension = options.dimensions\n      }\n\n      logger.request(\n        `activity=embed provider=${this.name} model=${model} inputs=${inputs.length}`,\n        { provider: this.name, model },\n      )\n\n      const timeoutMs = this.clientConfig.timeout ?? DEFAULT_TIMEOUT_MS\n      const transport = resolveCohereTransport(this.clientConfig)\n      const response = await fetchWithTimeout(\n        `${transport.baseUrl}/v2/embed`,\n        {\n          method: 'POST',\n          headers: {\n            Authorization: `Bearer ${this.clientConfig.apiKey}`,\n            'Content-Type': 'application/json',\n            ...transport.headers,\n          },\n          body: JSON.stringify(body),\n        },\n        timeoutMs,\n      )\n\n      if (!response.ok) {\n        const bodyText = await response.text()\n        let message = bodyText\n        try {\n          const parsed: unknown = JSON.parse(bodyText)\n          if (\n            typeof parsed === 'object' &&\n            parsed !== null &&\n            'message' in parsed &&\n            typeof parsed.message === 'string'\n          ) {\n            message = parsed.message\n          }\n        } catch {\n          // Not JSON — fall back to the raw body text.\n        }\n        throw new Error(`Cohere embed failed (${response.status}): ${message}`)\n      }\n\n      const data = (await response.json()) as CohereEmbedResponse\n\n      const vectors = data.embeddings?.float\n      if (!vectors) {\n        throw new Error(\n          'Cohere embed response did not include float embeddings',\n        )\n      }\n      if (vectors.length !== inputs.length) {\n        throw new Error(\n          `Cohere embed returned ${vectors.length} embeddings for ${inputs.length} inputs`,\n        )\n      }\n\n      const result: EmbeddingResult = {\n        id: generateId(this.name),\n        model,\n        embeddings: vectors.map((vector, index) => ({ vector, index })),\n      }\n\n      const inputTokens = data.meta?.billed_units?.input_tokens\n      if (inputTokens !== undefined) {\n        const usage: TokenUsage = {\n          promptTokens: inputTokens,\n          completionTokens: 0,\n          totalTokens: inputTokens,\n        }\n        result.usage = usage\n      }\n\n      return result\n    } catch (error: unknown) {\n      logger.errors(`${this.name}.createEmbeddings fatal`, {\n        error: toRunErrorPayload(error, `${this.name}.createEmbeddings failed`),\n        source: `${this.name}.createEmbeddings`,\n      })\n      throw error\n    }\n  }\n\n  /**\n   * Resolves an image part to a URL Cohere accepts. Cohere does not fetch\n   * remote image URLs, so everything is normalized to a `data:` URI unless\n   * the caller already provided one.\n   */\n  protected async resolveImageUrl(image: ImagePart): Promise<string> {\n    const source = image.source\n\n    if (source.type === 'data') {\n      return `data:${source.mimeType};base64,${source.value}`\n    }\n\n    if (source.value.startsWith('data:')) {\n      return source.value\n    }\n\n    if (!this.clientConfig.allowUrlFetch) {\n      throw new Error(\n        'Cohere does not fetch remote image URLs; pass base64 data or a data: URI (or enable config.allowUrlFetch to have the adapter download it)',\n      )\n    }\n\n    if (isPrivateOrInternalUrl(source.value)) {\n      throw new Error(\n        `Refusing to fetch internal or private URL for Cohere embedding: ${source.value}`,\n      )\n    }\n\n    const response = await fetchWithTimeout(\n      source.value,\n      undefined,\n      this.clientConfig.timeout ?? DEFAULT_TIMEOUT_MS,\n    )\n    if (!response.ok) {\n      throw new Error(\n        `Failed to fetch image URL for Cohere embedding (${response.status}): ${source.value}`,\n      )\n    }\n    const mimeType =\n      response.headers.get('content-type') ??\n      source.mimeType ??\n      'application/octet-stream'\n    const base64 = arrayBufferToBase64(await response.arrayBuffer())\n    return `data:${mimeType};base64,${base64}`\n  }\n}\n\n/**\n * Creates a Cohere embedding adapter with explicit API key.\n * Type resolution happens here at the call site.\n *\n * @param model - The model name (e.g., 'embed-v4.0')\n * @param apiKey - Your Cohere API key\n * @param config - Optional additional configuration\n * @returns Configured Cohere embedding adapter instance with resolved types\n *\n * @example\n * ```typescript\n * const adapter = createCohereEmbedding('embed-v4.0', 'api_key');\n *\n * const result = await embed({\n *   adapter,\n *   input: 'a red guitar',\n *   modelOptions: { inputType: 'search_document' }\n * });\n * ```\n */\nexport function createCohereEmbedding<TModel extends CohereEmbeddingModel>(\n  model: TModel,\n  apiKey: string,\n  config?: Omit<CohereEmbeddingConfig, 'apiKey'>,\n): CohereEmbeddingAdapter<TModel> {\n  return new CohereEmbeddingAdapter({ apiKey, ...config }, model)\n}\n\n/**\n * Creates a Cohere embedding adapter using the `COHERE_API_KEY` environment variable.\n * Type resolution happens here at the call site.\n *\n * Looks for `COHERE_API_KEY` in:\n * - `process.env` (Node.js)\n * - `window.env` (Browser with injected env)\n *\n * @param model - The model name (e.g., 'embed-v4.0')\n * @param config - Optional configuration (excluding apiKey which is auto-detected)\n * @returns Configured Cohere embedding adapter instance with resolved types\n * @throws Error if COHERE_API_KEY is not found in environment\n *\n * @example\n * ```typescript\n * // Automatically uses COHERE_API_KEY from environment\n * const adapter = cohereEmbedding('embed-v4.0');\n *\n * const result = await embed({\n *   adapter,\n *   input: ['a red guitar', 'a blue drum kit'],\n *   modelOptions: { inputType: 'search_query' },\n *   dimensions: 1024\n * });\n *\n * console.log(result.embeddings[0].vector)\n * ```\n */\nexport function cohereEmbedding<TModel extends CohereEmbeddingModel>(\n  model: TModel,\n  config?: Omit<CohereEmbeddingConfig, 'apiKey'>,\n): CohereEmbeddingAdapter<TModel> {\n  const apiKey = getCohereApiKeyFromEnv()\n  return createCohereEmbedding(model, apiKey, config)\n}\n"],"mappings":";;;;;;AAwBA,IAAM,qBAAqB;;;;;AAM3B,SAAS,uBAAuB,KAAsB;CACpD,IAAI;CACJ,IAAI;EACF,SAAS,IAAI,IAAI,GAAG;CACtB,QAAQ;EACN,OAAO;CACT;CACA,IAAI,OAAO,aAAa,WAAW,OAAO,aAAa,UACrD,OAAO;CAET,MAAM,OAAO,OAAO,SAAS,YAAY;CACzC,IACE,SAAS,eACT,KAAK,SAAS,YAAY,KAC1B,SAAS,SACT,SAAS,WACT,KAAK,WAAW,MAAM,KACtB,KAAK,WAAW,KAAK,KACrB,KAAK,WAAW,UAAU,KAC1B,KAAK,WAAW,UAAU,KAC1B,6BAA6B,KAAK,IAAI,GAEtC,OAAO;CAET,OAAO;AACT;AAEA,eAAe,iBACb,KACA,MACA,WACmB;CACnB,MAAM,aAAa,IAAI,gBAAgB;CACvC,MAAM,YAAY,iBAAiB,WAAW,MAAM,GAAG,SAAS;CAChE,IAAI;EACF,OAAO,MAAM,MAAM,KAAK;GAAE,GAAG;GAAM,QAAQ,WAAW;EAAO,CAAC;CAChE,UAAU;EACR,aAAa,SAAS;CACxB;AACF;;;;;;;;;;;;;;;AA6CA,IAAa,yBAAb,cAEU,qBAKR;CACA,OAAgB;CAEhB;CAEA,YAAY,QAA+B,OAAe;EACxD,MAAM,OAAO,CAAC,CAAC;EACf,KAAK,eAAe;CACtB;CAEA,MAAM,iBACJ,SAC0B;EAC1B,MAAM,EAAE,OAAO,QAAQ,iBAAiB;EAExC,IAAI;GAGF,MAAM,YACJ,cAAc;GAChB,IAAI,CAAC,WACH,MAAM,IAAI,MACR,0HACF;GAGF,MAAM,WAAW,sBAAsB,QAAQ,KAAK;GACpD,MAAM,SAAS,MAAM,QAAQ,IAC3B,SAAS,IAAI,OAAO,SAAS;IAC3B,MAAM,UAAyC,KAAK,MAAM,KACvD,UAAU;KAAE,MAAM;KAAQ;IAAK,EAClC;IACA,KAAK,MAAM,SAAS,KAAK,QACvB,QAAQ,KAAK;KACX,MAAM;KACN,WAAW,EAAE,KAAK,MAAM,KAAK,gBAAgB,KAAK,EAAE;IACtD,CAAC;IAEH,OAAO,EAAE,QAAQ;GACnB,CAAC,CACH;GAIA,MAAM,OAA+B;IACnC;IACA;IACA,YAAY;IACZ,iBAAiB,CAAC,OAAO;GAC3B;GACA,MAAM,WAAW,cAAc;GAC/B,IAAI,aAAa,KAAA,GACf,KAAK,WAAW;GAElB,IAAI,QAAQ,eAAe,KAAA,GACzB,KAAK,mBAAmB,QAAQ;GAGlC,OAAO,QACL,2BAA2B,KAAK,KAAK,SAAS,MAAM,UAAU,OAAO,UACrE;IAAE,UAAU,KAAK;IAAM;GAAM,CAC/B;GAEA,MAAM,YAAY,KAAK,aAAa,WAAW;GAC/C,MAAM,YAAY,uBAAuB,KAAK,YAAY;GAC1D,MAAM,WAAW,MAAM,iBACrB,GAAG,UAAU,QAAQ,YACrB;IACE,QAAQ;IACR,SAAS;KACP,eAAe,UAAU,KAAK,aAAa;KAC3C,gBAAgB;KAChB,GAAG,UAAU;IACf;IACA,MAAM,KAAK,UAAU,IAAI;GAC3B,GACA,SACF;GAEA,IAAI,CAAC,SAAS,IAAI;IAChB,MAAM,WAAW,MAAM,SAAS,KAAK;IACrC,IAAI,UAAU;IACd,IAAI;KACF,MAAM,SAAkB,KAAK,MAAM,QAAQ;KAC3C,IACE,OAAO,WAAW,YAClB,WAAW,QACX,aAAa,UACb,OAAO,OAAO,YAAY,UAE1B,UAAU,OAAO;IAErB,QAAQ,CAER;IACA,MAAM,IAAI,MAAM,wBAAwB,SAAS,OAAO,KAAK,SAAS;GACxE;GAEA,MAAM,OAAQ,MAAM,SAAS,KAAK;GAElC,MAAM,UAAU,KAAK,YAAY;GACjC,IAAI,CAAC,SACH,MAAM,IAAI,MACR,wDACF;GAEF,IAAI,QAAQ,WAAW,OAAO,QAC5B,MAAM,IAAI,MACR,yBAAyB,QAAQ,OAAO,kBAAkB,OAAO,OAAO,QAC1E;GAGF,MAAM,SAA0B;IAC9B,IAAI,WAAW,KAAK,IAAI;IACxB;IACA,YAAY,QAAQ,KAAK,QAAQ,WAAW;KAAE;KAAQ;IAAM,EAAE;GAChE;GAEA,MAAM,cAAc,KAAK,MAAM,cAAc;GAC7C,IAAI,gBAAgB,KAAA,GAMlB,OAAO,QAAQ;IAJb,cAAc;IACd,kBAAkB;IAClB,aAAa;GAEA;GAGjB,OAAO;EACT,SAAS,OAAgB;GACvB,OAAO,OAAO,GAAG,KAAK,KAAK,0BAA0B;IACnD,OAAO,kBAAkB,OAAO,GAAG,KAAK,KAAK,yBAAyB;IACtE,QAAQ,GAAG,KAAK,KAAK;GACvB,CAAC;GACD,MAAM;EACR;CACF;;;;;;CAOA,MAAgB,gBAAgB,OAAmC;EACjE,MAAM,SAAS,MAAM;EAErB,IAAI,OAAO,SAAS,QAClB,OAAO,QAAQ,OAAO,SAAS,UAAU,OAAO;EAGlD,IAAI,OAAO,MAAM,WAAW,OAAO,GACjC,OAAO,OAAO;EAGhB,IAAI,CAAC,KAAK,aAAa,eACrB,MAAM,IAAI,MACR,2IACF;EAGF,IAAI,uBAAuB,OAAO,KAAK,GACrC,MAAM,IAAI,MACR,mEAAmE,OAAO,OAC5E;EAGF,MAAM,WAAW,MAAM,iBACrB,OAAO,OACP,KAAA,GACA,KAAK,aAAa,WAAW,kBAC/B;EACA,IAAI,CAAC,SAAS,IACZ,MAAM,IAAI,MACR,mDAAmD,SAAS,OAAO,KAAK,OAAO,OACjF;EAOF,OAAO,QAJL,SAAS,QAAQ,IAAI,cAAc,KACnC,OAAO,YACP,2BAEsB,UADT,oBAAoB,MAAM,SAAS,YAAY,CAC5B;CACpC;AACF;;;;;;;;;;;;;;;;;;;;;AAsBA,SAAgB,sBACd,OACA,QACA,QACgC;CAChC,OAAO,IAAI,uBAAuB;EAAE;EAAQ,GAAG;CAAO,GAAG,KAAK;AAChE;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA8BA,SAAgB,gBACd,OACA,QACgC;CAEhC,OAAO,sBAAsB,OADd,uBACqB,GAAQ,MAAM;AACpD"}
+{"version":3,"file":"embedding.js","names":[],"sources":["../../../src/adapters/embedding.ts"],"sourcesContent":["import { BaseEmbeddingAdapter } from '@tanstack/ai/adapters'\nimport { toRunErrorPayload } from '@tanstack/ai/adapter-internals'\nimport { arrayBufferToBase64, generateId } from '@tanstack/ai-utils'\nimport {\n  isFileSource,\n  resolveEmbeddingInput,\n  unsupportedFileSourceError,\n} from '@tanstack/ai'\nimport { getCohereApiKeyFromEnv, resolveCohereTransport } from '../utils/client'\nimport type {\n  EmbeddingOptions,\n  EmbeddingResult,\n  ImagePart,\n  TokenUsage,\n} from '@tanstack/ai'\nimport type {\n  CohereEmbeddingModel,\n  CohereEmbeddingModelInputModalitiesByName,\n  CohereEmbeddingModelProviderOptionsByName,\n} from '../model-meta'\nimport type { CohereEmbeddingProviderOptions } from '../embedding/embedding-provider-options'\nimport type { CohereClientConfig } from '../utils/client'\n\n/**\n * Configuration for Cohere embedding adapter.\n */\nexport interface CohereEmbeddingConfig extends CohereClientConfig {}\n\nconst DEFAULT_TIMEOUT_MS = 30_000\n\n/**\n * Returns true when `url` is malformed, non-http(s), or targets a private /\n * loopback / link-local host. Used to block SSRF via `allowUrlFetch`.\n */\nfunction isPrivateOrInternalUrl(url: string): boolean {\n  let parsed: URL\n  try {\n    parsed = new URL(url)\n  } catch {\n    return true\n  }\n  if (parsed.protocol !== 'http:' && parsed.protocol !== 'https:') {\n    return true\n  }\n  const host = parsed.hostname.toLowerCase()\n  if (\n    host === 'localhost' ||\n    host.endsWith('.localhost') ||\n    host === '::1' ||\n    host === '[::1]' ||\n    host.startsWith('127.') ||\n    host.startsWith('10.') ||\n    host.startsWith('192.168.') ||\n    host.startsWith('169.254.') ||\n    /^172\\.(1[6-9]|2\\d|3[01])\\./.test(host)\n  ) {\n    return true\n  }\n  return false\n}\n\nasync function fetchWithTimeout(\n  url: string,\n  init: RequestInit | undefined,\n  timeoutMs: number,\n): Promise<Response> {\n  const controller = new AbortController()\n  const timeoutId = setTimeout(() => controller.abort(), timeoutMs)\n  try {\n    return await fetch(url, { ...init, signal: controller.signal })\n  } finally {\n    clearTimeout(timeoutId)\n  }\n}\n\n/** One content part of a Cohere v2/embed fused input. */\ntype CohereEmbedContentPart =\n  | { type: 'text'; text: string }\n  | { type: 'image_url'; image_url: { url: string } }\n\n/** Wire shape of the Cohere v2/embed request body. */\ninterface CohereEmbedRequestBody {\n  model: string\n  inputs: Array<{ content: Array<CohereEmbedContentPart> }>\n  input_type: CohereEmbeddingProviderOptions['inputType']\n  embedding_types: ['float']\n  truncate?: 'NONE' | 'START' | 'END'\n  output_dimension?: number\n}\n\n/** Wire shape of the Cohere v2/embed response (fields the adapter reads). */\ninterface CohereEmbedResponse {\n  id?: string\n  embeddings?: {\n    float?: Array<Array<number>>\n  }\n  meta?: {\n    billed_units?: {\n      input_tokens?: number\n      images?: number\n    }\n  }\n}\n\n/**\n * Cohere Embedding Adapter\n *\n * Tree-shakeable adapter for Cohere multimodal embeddings (embed-v4.0),\n * implemented with plain `fetch` against the v2/embed endpoint — no Cohere\n * SDK dependency.\n *\n * Features:\n * - Batch embedding (one request for the whole input array)\n * - Multimodal inputs: text, images, and fused text+image items (one vector\n *   per input item)\n * - Matryoshka dimension reduction via the top-level `dimensions` option\n *   (mapped to Cohere's `output_dimension`)\n */\nexport class CohereEmbeddingAdapter<\n  TModel extends CohereEmbeddingModel,\n> extends BaseEmbeddingAdapter<\n  TModel,\n  CohereEmbeddingProviderOptions,\n  CohereEmbeddingModelProviderOptionsByName,\n  CohereEmbeddingModelInputModalitiesByName\n> {\n  readonly name = 'cohere' as const\n\n  protected clientConfig: CohereEmbeddingConfig\n\n  constructor(config: CohereEmbeddingConfig, model: TModel) {\n    super(model, {})\n    this.clientConfig = config\n  }\n\n  async createEmbeddings(\n    options: EmbeddingOptions<CohereEmbeddingProviderOptions>,\n  ): Promise<EmbeddingResult> {\n    const { model, logger, modelOptions } = options\n\n    try {\n      // The provider options type makes `modelOptions` required at the\n      // embed() call site; this guard covers untyped/dynamic callers.\n      const inputType: CohereEmbeddingProviderOptions['inputType'] | undefined =\n        modelOptions?.inputType\n      if (!inputType) {\n        throw new Error(\n          `Cohere embeddings require modelOptions.inputType ('search_document' | 'search_query' | 'classification' | 'clustering').`,\n        )\n      }\n\n      const resolved = resolveEmbeddingInput(options.input)\n      const inputs = await Promise.all(\n        resolved.map(async (item) => {\n          const content: Array<CohereEmbedContentPart> = item.texts.map(\n            (text) => ({ type: 'text', text }),\n          )\n          for (const image of item.images) {\n            content.push({\n              type: 'image_url',\n              image_url: { url: await this.resolveImageUrl(image) },\n            })\n          }\n          return { content }\n        }),\n      )\n\n      // embedding_types is pinned to ['float'] (overriding any disagreeing\n      // modelOptions.embeddingTypes) so vectors are always number[].\n      const body: CohereEmbedRequestBody = {\n        model,\n        inputs,\n        input_type: inputType,\n        embedding_types: ['float'],\n      }\n      const truncate = modelOptions?.truncate\n      if (truncate !== undefined) {\n        body.truncate = truncate\n      }\n      if (options.dimensions !== undefined) {\n        body.output_dimension = options.dimensions\n      }\n\n      logger.request(\n        `activity=embed provider=${this.name} model=${model} inputs=${inputs.length}`,\n        { provider: this.name, model },\n      )\n\n      const timeoutMs = this.clientConfig.timeout ?? DEFAULT_TIMEOUT_MS\n      const transport = resolveCohereTransport(this.clientConfig)\n      const response = await fetchWithTimeout(\n        `${transport.baseUrl}/v2/embed`,\n        {\n          method: 'POST',\n          headers: {\n            Authorization: `Bearer ${this.clientConfig.apiKey}`,\n            'Content-Type': 'application/json',\n            ...transport.headers,\n          },\n          body: JSON.stringify(body),\n        },\n        timeoutMs,\n      )\n\n      if (!response.ok) {\n        const bodyText = await response.text()\n        let message = bodyText\n        try {\n          const parsed: unknown = JSON.parse(bodyText)\n          if (\n            typeof parsed === 'object' &&\n            parsed !== null &&\n            'message' in parsed &&\n            typeof parsed.message === 'string'\n          ) {\n            message = parsed.message\n          }\n        } catch {\n          // Not JSON — fall back to the raw body text.\n        }\n        throw new Error(`Cohere embed failed (${response.status}): ${message}`)\n      }\n\n      const data = (await response.json()) as CohereEmbedResponse\n\n      const vectors = data.embeddings?.float\n      if (!vectors) {\n        throw new Error(\n          'Cohere embed response did not include float embeddings',\n        )\n      }\n      if (vectors.length !== inputs.length) {\n        throw new Error(\n          `Cohere embed returned ${vectors.length} embeddings for ${inputs.length} inputs`,\n        )\n      }\n\n      const result: EmbeddingResult = {\n        id: generateId(this.name),\n        model,\n        embeddings: vectors.map((vector, index) => ({ vector, index })),\n      }\n\n      const inputTokens = data.meta?.billed_units?.input_tokens\n      if (inputTokens !== undefined) {\n        const usage: TokenUsage = {\n          promptTokens: inputTokens,\n          completionTokens: 0,\n          totalTokens: inputTokens,\n        }\n        result.usage = usage\n      }\n\n      return result\n    } catch (error: unknown) {\n      logger.errors(`${this.name}.createEmbeddings fatal`, {\n        error: toRunErrorPayload(error, `${this.name}.createEmbeddings failed`),\n        source: `${this.name}.createEmbeddings`,\n      })\n      throw error\n    }\n  }\n\n  /**\n   * Resolves an image part to a URL Cohere accepts. Cohere does not fetch\n   * remote image URLs, so everything is normalized to a `data:` URI unless\n   * the caller already provided one.\n   */\n  protected async resolveImageUrl(image: ImagePart): Promise<string> {\n    const source = image.source\n    if (isFileSource(source)) throw unsupportedFileSourceError(this.name)\n\n    if (source.type === 'data') {\n      return `data:${source.mimeType};base64,${source.value}`\n    }\n\n    if (source.value.startsWith('data:')) {\n      return source.value\n    }\n\n    if (!this.clientConfig.allowUrlFetch) {\n      throw new Error(\n        'Cohere does not fetch remote image URLs; pass base64 data or a data: URI (or enable config.allowUrlFetch to have the adapter download it)',\n      )\n    }\n\n    if (isPrivateOrInternalUrl(source.value)) {\n      throw new Error(\n        `Refusing to fetch internal or private URL for Cohere embedding: ${source.value}`,\n      )\n    }\n\n    const response = await fetchWithTimeout(\n      source.value,\n      undefined,\n      this.clientConfig.timeout ?? DEFAULT_TIMEOUT_MS,\n    )\n    if (!response.ok) {\n      throw new Error(\n        `Failed to fetch image URL for Cohere embedding (${response.status}): ${source.value}`,\n      )\n    }\n    const mimeType =\n      response.headers.get('content-type') ??\n      source.mimeType ??\n      'application/octet-stream'\n    const base64 = arrayBufferToBase64(await response.arrayBuffer())\n    return `data:${mimeType};base64,${base64}`\n  }\n}\n\n/**\n * Creates a Cohere embedding adapter with explicit API key.\n * Type resolution happens here at the call site.\n *\n * @param model - The model name (e.g., 'embed-v4.0')\n * @param apiKey - Your Cohere API key\n * @param config - Optional additional configuration\n * @returns Configured Cohere embedding adapter instance with resolved types\n *\n * @example\n * ```typescript\n * const adapter = createCohereEmbedding('embed-v4.0', 'api_key');\n *\n * const result = await embed({\n *   adapter,\n *   input: 'a red guitar',\n *   modelOptions: { inputType: 'search_document' }\n * });\n * ```\n */\nexport function createCohereEmbedding<TModel extends CohereEmbeddingModel>(\n  model: TModel,\n  apiKey: string,\n  config?: Omit<CohereEmbeddingConfig, 'apiKey'>,\n): CohereEmbeddingAdapter<TModel> {\n  return new CohereEmbeddingAdapter({ apiKey, ...config }, model)\n}\n\n/**\n * Creates a Cohere embedding adapter using the `COHERE_API_KEY` environment variable.\n * Type resolution happens here at the call site.\n *\n * Looks for `COHERE_API_KEY` in:\n * - `process.env` (Node.js)\n * - `window.env` (Browser with injected env)\n *\n * @param model - The model name (e.g., 'embed-v4.0')\n * @param config - Optional configuration (excluding apiKey which is auto-detected)\n * @returns Configured Cohere embedding adapter instance with resolved types\n * @throws Error if COHERE_API_KEY is not found in environment\n *\n * @example\n * ```typescript\n * // Automatically uses COHERE_API_KEY from environment\n * const adapter = cohereEmbedding('embed-v4.0');\n *\n * const result = await embed({\n *   adapter,\n *   input: ['a red guitar', 'a blue drum kit'],\n *   modelOptions: { inputType: 'search_query' },\n *   dimensions: 1024\n * });\n *\n * console.log(result.embeddings[0].vector)\n * ```\n */\nexport function cohereEmbedding<TModel extends CohereEmbeddingModel>(\n  model: TModel,\n  config?: Omit<CohereEmbeddingConfig, 'apiKey'>,\n): CohereEmbeddingAdapter<TModel> {\n  const apiKey = getCohereApiKeyFromEnv()\n  return createCohereEmbedding(model, apiKey, config)\n}\n"],"mappings":";;;;;;AA4BA,IAAM,qBAAqB;;;;;AAM3B,SAAS,uBAAuB,KAAsB;CACpD,IAAI;CACJ,IAAI;EACF,SAAS,IAAI,IAAI,GAAG;CACtB,QAAQ;EACN,OAAO;CACT;CACA,IAAI,OAAO,aAAa,WAAW,OAAO,aAAa,UACrD,OAAO;CAET,MAAM,OAAO,OAAO,SAAS,YAAY;CACzC,IACE,SAAS,eACT,KAAK,SAAS,YAAY,KAC1B,SAAS,SACT,SAAS,WACT,KAAK,WAAW,MAAM,KACtB,KAAK,WAAW,KAAK,KACrB,KAAK,WAAW,UAAU,KAC1B,KAAK,WAAW,UAAU,KAC1B,6BAA6B,KAAK,IAAI,GAEtC,OAAO;CAET,OAAO;AACT;AAEA,eAAe,iBACb,KACA,MACA,WACmB;CACnB,MAAM,aAAa,IAAI,gBAAgB;CACvC,MAAM,YAAY,iBAAiB,WAAW,MAAM,GAAG,SAAS;CAChE,IAAI;EACF,OAAO,MAAM,MAAM,KAAK;GAAE,GAAG;GAAM,QAAQ,WAAW;EAAO,CAAC;CAChE,UAAU;EACR,aAAa,SAAS;CACxB;AACF;;;;;;;;;;;;;;;AA6CA,IAAa,yBAAb,cAEU,qBAKR;CACA,OAAgB;CAEhB;CAEA,YAAY,QAA+B,OAAe;EACxD,MAAM,OAAO,CAAC,CAAC;EACf,KAAK,eAAe;CACtB;CAEA,MAAM,iBACJ,SAC0B;EAC1B,MAAM,EAAE,OAAO,QAAQ,iBAAiB;EAExC,IAAI;GAGF,MAAM,YACJ,cAAc;GAChB,IAAI,CAAC,WACH,MAAM,IAAI,MACR,0HACF;GAGF,MAAM,WAAW,sBAAsB,QAAQ,KAAK;GACpD,MAAM,SAAS,MAAM,QAAQ,IAC3B,SAAS,IAAI,OAAO,SAAS;IAC3B,MAAM,UAAyC,KAAK,MAAM,KACvD,UAAU;KAAE,MAAM;KAAQ;IAAK,EAClC;IACA,KAAK,MAAM,SAAS,KAAK,QACvB,QAAQ,KAAK;KACX,MAAM;KACN,WAAW,EAAE,KAAK,MAAM,KAAK,gBAAgB,KAAK,EAAE;IACtD,CAAC;IAEH,OAAO,EAAE,QAAQ;GACnB,CAAC,CACH;GAIA,MAAM,OAA+B;IACnC;IACA;IACA,YAAY;IACZ,iBAAiB,CAAC,OAAO;GAC3B;GACA,MAAM,WAAW,cAAc;GAC/B,IAAI,aAAa,KAAA,GACf,KAAK,WAAW;GAElB,IAAI,QAAQ,eAAe,KAAA,GACzB,KAAK,mBAAmB,QAAQ;GAGlC,OAAO,QACL,2BAA2B,KAAK,KAAK,SAAS,MAAM,UAAU,OAAO,UACrE;IAAE,UAAU,KAAK;IAAM;GAAM,CAC/B;GAEA,MAAM,YAAY,KAAK,aAAa,WAAW;GAC/C,MAAM,YAAY,uBAAuB,KAAK,YAAY;GAC1D,MAAM,WAAW,MAAM,iBACrB,GAAG,UAAU,QAAQ,YACrB;IACE,QAAQ;IACR,SAAS;KACP,eAAe,UAAU,KAAK,aAAa;KAC3C,gBAAgB;KAChB,GAAG,UAAU;IACf;IACA,MAAM,KAAK,UAAU,IAAI;GAC3B,GACA,SACF;GAEA,IAAI,CAAC,SAAS,IAAI;IAChB,MAAM,WAAW,MAAM,SAAS,KAAK;IACrC,IAAI,UAAU;IACd,IAAI;KACF,MAAM,SAAkB,KAAK,MAAM,QAAQ;KAC3C,IACE,OAAO,WAAW,YAClB,WAAW,QACX,aAAa,UACb,OAAO,OAAO,YAAY,UAE1B,UAAU,OAAO;IAErB,QAAQ,CAER;IACA,MAAM,IAAI,MAAM,wBAAwB,SAAS,OAAO,KAAK,SAAS;GACxE;GAEA,MAAM,OAAQ,MAAM,SAAS,KAAK;GAElC,MAAM,UAAU,KAAK,YAAY;GACjC,IAAI,CAAC,SACH,MAAM,IAAI,MACR,wDACF;GAEF,IAAI,QAAQ,WAAW,OAAO,QAC5B,MAAM,IAAI,MACR,yBAAyB,QAAQ,OAAO,kBAAkB,OAAO,OAAO,QAC1E;GAGF,MAAM,SAA0B;IAC9B,IAAI,WAAW,KAAK,IAAI;IACxB;IACA,YAAY,QAAQ,KAAK,QAAQ,WAAW;KAAE;KAAQ;IAAM,EAAE;GAChE;GAEA,MAAM,cAAc,KAAK,MAAM,cAAc;GAC7C,IAAI,gBAAgB,KAAA,GAMlB,OAAO,QAAQ;IAJb,cAAc;IACd,kBAAkB;IAClB,aAAa;GAEA;GAGjB,OAAO;EACT,SAAS,OAAgB;GACvB,OAAO,OAAO,GAAG,KAAK,KAAK,0BAA0B;IACnD,OAAO,kBAAkB,OAAO,GAAG,KAAK,KAAK,yBAAyB;IACtE,QAAQ,GAAG,KAAK,KAAK;GACvB,CAAC;GACD,MAAM;EACR;CACF;;;;;;CAOA,MAAgB,gBAAgB,OAAmC;EACjE,MAAM,SAAS,MAAM;EACrB,IAAI,aAAa,MAAM,GAAG,MAAM,2BAA2B,KAAK,IAAI;EAEpE,IAAI,OAAO,SAAS,QAClB,OAAO,QAAQ,OAAO,SAAS,UAAU,OAAO;EAGlD,IAAI,OAAO,MAAM,WAAW,OAAO,GACjC,OAAO,OAAO;EAGhB,IAAI,CAAC,KAAK,aAAa,eACrB,MAAM,IAAI,MACR,2IACF;EAGF,IAAI,uBAAuB,OAAO,KAAK,GACrC,MAAM,IAAI,MACR,mEAAmE,OAAO,OAC5E;EAGF,MAAM,WAAW,MAAM,iBACrB,OAAO,OACP,KAAA,GACA,KAAK,aAAa,WAAW,kBAC/B;EACA,IAAI,CAAC,SAAS,IACZ,MAAM,IAAI,MACR,mDAAmD,SAAS,OAAO,KAAK,OAAO,OACjF;EAOF,OAAO,QAJL,SAAS,QAAQ,IAAI,cAAc,KACnC,OAAO,YACP,2BAEsB,UADT,oBAAoB,MAAM,SAAS,YAAY,CAC5B;CACpC;AACF;;;;;;;;;;;;;;;;;;;;;AAsBA,SAAgB,sBACd,OACA,QACA,QACgC;CAChC,OAAO,IAAI,uBAAuB;EAAE;EAAQ,GAAG;CAAO,GAAG,KAAK;AAChE;;;;;;;;;;;;;;;;;;;;;;;;;;;;;AA8BA,SAAgB,gBACd,OACA,QACgC;CAEhC,OAAO,sBAAsB,OADd,uBACqB,GAAQ,MAAM;AACpD"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tanstack/ai-cohere",
-  "version": "0.4.6",
+  "version": "0.4.9",
   "description": "Cohere adapter for TanStack AI — multimodal embeddings and document reranking.",
   "author": "Tanner Linsley",
   "license": "MIT",
@@ -53,12 +53,12 @@
     "retrieval"
   ],
   "peerDependencies": {
-    "@tanstack/ai": "^0.58.0"
+    "@tanstack/ai": "^0.61.0"
   },
   "devDependencies": {
     "@vitest/coverage-v8": "4.1.10",
     "vite": "^8.2.1",
-    "@tanstack/ai": "0.58.0"
+    "@tanstack/ai": "0.61.0"
   },
   "dependencies": {
     "@tanstack/ai-utils": "^0.4.1"

package/src/adapters/embedding.ts

@@ -1,7 +1,11 @@
 import { BaseEmbeddingAdapter } from '@tanstack/ai/adapters'
 import { toRunErrorPayload } from '@tanstack/ai/adapter-internals'
 import { arrayBufferToBase64, generateId } from '@tanstack/ai-utils'
-import { resolveEmbeddingInput } from '@tanstack/ai'
+import {
+  isFileSource,
+  resolveEmbeddingInput,
+  unsupportedFileSourceError,
+} from '@tanstack/ai'
 import { getCohereApiKeyFromEnv, resolveCohereTransport } from '../utils/client'
 import type {
   EmbeddingOptions,
@@ -264,6 +268,7 @@ export class CohereEmbeddingAdapter<
    */
   protected async resolveImageUrl(image: ImagePart): Promise<string> {
     const source = image.source
+    if (isFileSource(source)) throw unsupportedFileSourceError(this.name)
 
     if (source.type === 'data') {
       return `data:${source.mimeType};base64,${source.value}`