@truefoundry/ai-gateway-sdk 0.0.0-stage → 0.0.2

package/dist/tfy-providers/custom/index.js

@@ -0,0 +1,85 @@
+import CustomAPIConfig from './api.js';
+import { CohereRerankV2Config as CohereRerankV2ConfigParam, CohereRerankV2ResponseTransform, } from './cohere-rerank-v2/rerank.js';
+import { InfinityEmbedConfig, InfinityEmbedResponseTransform } from './infinity/embed.js';
+import { InfinityRerankConfig } from './infinity/rerank.js';
+import { TEIEmbedConfig, TEIEmbedResponseTransform } from './tei/embed.js';
+import { VLLMOpenAIChatCompleteConfig, VLLMOpenAIChatCompleteResponseTransform } from './vllm-openai/chatComplete.js';
+import { VLLMOpenAICompleteConfig, VLLMOpenAICompleteResponseTransform } from './vllm-openai/complete.js';
+import { NemoRetrieverEmbedConfig, NemoRetrieverEmbedResponseTransform } from './nemo-retriever/embed.js';
+import { NemoRetrieverRerankConfig, NemoRetrieverRerankResponseTransform } from './nemo-retriever/rerank.js';
+import { OpenAICompatibleChatCompleteConfig, OpenAICompatibleChatCompleteResponseTransform, } from './openai-compatible/chatComplete.js';
+import { OpenAICompatibleCompleteConfig, OpenAICompatibleCompleteResponseTransform, } from './openai-compatible/complete.js';
+import { OpenAICompatibleEmbedConfig, OpenAICompatibleEmbedResponseTransform } from './openai-compatible/embed.js';
+const VLLMConfig = {
+    api: CustomAPIConfig,
+    getConfig: () => ({
+        complete: VLLMOpenAICompleteConfig,
+        chatComplete: VLLMOpenAIChatCompleteConfig,
+        responseTransforms: {
+            complete: VLLMOpenAICompleteResponseTransform,
+            chatComplete: VLLMOpenAIChatCompleteResponseTransform,
+        },
+    }),
+};
+const OpenAICompatibleConfig = {
+    api: CustomAPIConfig,
+    getConfig: () => ({
+        complete: OpenAICompatibleCompleteConfig,
+        chatComplete: OpenAICompatibleChatCompleteConfig,
+        embed: OpenAICompatibleEmbedConfig,
+        responseTransforms: {
+            complete: OpenAICompatibleCompleteResponseTransform,
+            chatComplete: OpenAICompatibleChatCompleteResponseTransform,
+            embed: OpenAICompatibleEmbedResponseTransform,
+        },
+    }),
+};
+const TEIConfig = {
+    api: CustomAPIConfig,
+    getConfig: () => ({
+        embed: TEIEmbedConfig,
+        responseTransforms: {
+            embed: TEIEmbedResponseTransform,
+        },
+    }),
+};
+const InfinityConfig = {
+    api: CustomAPIConfig,
+    getConfig: () => ({
+        embed: InfinityEmbedConfig,
+        rerank: InfinityRerankConfig,
+        responseTransforms: {
+            embed: InfinityEmbedResponseTransform,
+        },
+    }),
+};
+const CohereRerankV2Config = {
+    api: CustomAPIConfig,
+    getConfig: () => ({
+        rerank: CohereRerankV2ConfigParam,
+        responseTransforms: {
+            rerank: CohereRerankV2ResponseTransform,
+        },
+    }),
+};
+const NemoRetrieverConfig = {
+    api: CustomAPIConfig,
+    getConfig: () => ({
+        embed: NemoRetrieverEmbedConfig,
+        rerank: NemoRetrieverRerankConfig,
+        responseTransforms: {
+            embed: NemoRetrieverEmbedResponseTransform,
+            rerank: NemoRetrieverRerankResponseTransform,
+        },
+    }),
+};
+// ComfyUI video generation is blind-proxied via the `video-proxy` fn (see
+// `handlers/videoProxyHandler.ts`), which forwards the client's raw request/response verbatim
+// without ever consulting a provider's `getConfig()`/`requestHandlers` — only `api` (base URL +
+// auth headers, from `CustomAPIConfig`) is needed to route ComfyUI requests.
+const ComfyUIConfig = {
+    api: CustomAPIConfig,
+};
+// export all
+export { CohereRerankV2Config, ComfyUIConfig, InfinityConfig, NemoRetrieverConfig, OpenAICompatibleConfig, TEIConfig, VLLMConfig, };
+//# sourceMappingURL=index.js.map

package/dist/tfy-providers/custom/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/tfy-providers/custom/index.ts"],"names":[],"mappings":"AAMA,OAAO,eAAe,MAAM,OAAO,CAAC;AACpC,OAAO,EACL,oBAAoB,IAAI,yBAAyB,EACjD,+BAA+B,GAChC,MAAM,2BAA2B,CAAC;AACnC,OAAO,EAAE,mBAAmB,EAAE,8BAA8B,EAAE,MAAM,kBAAkB,CAAC;AACvF,OAAO,EAAE,oBAAoB,EAAE,MAAM,mBAAmB,CAAC;AACzD,OAAO,EAAE,cAAc,EAAE,yBAAyB,EAAE,MAAM,aAAa,CAAC;AACxE,OAAO,EAAE,4BAA4B,EAAE,uCAAuC,EAAE,MAAM,4BAA4B,CAAC;AACnH,OAAO,EAAE,wBAAwB,EAAE,mCAAmC,EAAE,MAAM,wBAAwB,CAAC;AAEvG,OAAO,EAAE,wBAAwB,EAAE,mCAAmC,EAAE,MAAM,wBAAwB,CAAC;AACvG,OAAO,EAAE,yBAAyB,EAAE,oCAAoC,EAAE,MAAM,yBAAyB,CAAC;AAC1G,OAAO,EACL,kCAAkC,EAClC,6CAA6C,GAC9C,MAAM,kCAAkC,CAAC;AAC1C,OAAO,EACL,8BAA8B,EAC9B,yCAAyC,GAC1C,MAAM,8BAA8B,CAAC;AACtC,OAAO,EAAE,2BAA2B,EAAE,sCAAsC,EAAE,MAAM,2BAA2B,CAAC;AAChH,MAAM,UAAU,GAAoB;IAClC,GAAG,EAAE,eAAe;IACpB,SAAS,EAAE,GAAG,EAAE,CAAC,CAAC;QAChB,QAAQ,EAAE,wBAAwB;QAClC,YAAY,EAAE,4BAA4B;QAC1C,kBAAkB,EAAE;YAClB,QAAQ,EAAE,mCAAmC;YAC7C,YAAY,EAAE,uCAAuC;SACtD;KACF,CAAC;CACH,CAAC;AAEF,MAAM,sBAAsB,GAAoB;IAC9C,GAAG,EAAE,eAAe;IACpB,SAAS,EAAE,GAAG,EAAE,CAAC,CAAC;QAChB,QAAQ,EAAE,8BAA8B;QACxC,YAAY,EAAE,kCAAkC;QAChD,KAAK,EAAE,2BAA2B;QAClC,kBAAkB,EAAE;YAClB,QAAQ,EAAE,yCAAyC;YACnD,YAAY,EAAE,6CAA6C;YAC3D,KAAK,EAAE,sCAAsC;SAC9C;KACF,CAAC;CACH,CAAC;AAEF,MAAM,SAAS,GAAoB;IACjC,GAAG,EAAE,eAAe;IACpB,SAAS,EAAE,GAAG,EAAE,CAAC,CAAC;QAChB,KAAK,EAAE,cAAc;QACrB,kBAAkB,EAAE;YAClB,KAAK,EAAE,yBAAyB;SACjC;KACF,CAAC;CACH,CAAC;AAEF,MAAM,cAAc,GAAoB;IACtC,GAAG,EAAE,eAAe;IACpB,SAAS,EAAE,GAAG,EAAE,CAAC,CAAC;QAChB,KAAK,EAAE,mBAAmB;QAC1B,MAAM,EAAE,oBAAoB;QAC5B,kBAAkB,EAAE;YAClB,KAAK,EAAE,8BAA8B;SACtC;KACF,CAAC;CACH,CAAC;AAEF,MAAM,oBAAoB,GAAoB;IAC5C,GAAG,EAAE,eAAe;IACpB,SAAS,EAAE,GAAG,EAAE,CAAC,CAAC;QAChB,MAAM,EAAE,yBAAyB;QACjC,kBAAkB,EAAE;YAClB,MAAM,EAAE,+BAA+B;SACxC;KACF,CAAC;CACH,CAAC;AAEF,MAAM,mBAAmB,GAAoB;IAC3C,GAAG,EAAE,eAAe;IACpB,SAAS,EAAE,GAAG,EAAE,CAAC,CAAC;QAChB,KAAK,EAAE,wBAAwB;QAC/B,MAAM,EAAE,yBAAyB;QACjC,kBAAkB,EAAE;YAClB,KAAK,EAAE,mCAAmC;YAC1C,MAAM,EAAE,oCAAoC;SAC7C;KACF,CAAC;CACH,CAAC;AAEF,0EAA0E;AAC1E,8FAA8F;AAC9F,gGAAgG;AAChG,6EAA6E;AAC7E,MAAM,aAAa,GAAoB;IACrC,GAAG,EAAE,eAAe;CACrB,CAAC;AAEF,aAAa;AACb,OAAO,EACL,oBAAoB,EACpB,aAAa,EACb,cAAc,EACd,mBAAmB,EACnB,sBAAsB,EACtB,SAAS,EACT,UAAU,GACX,CAAC"}

package/dist/tfy-providers/custom/infinity/embed.d.ts

@@ -0,0 +1,10 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+import { ErrorResponse, ProviderConfig } from '../../../providers/types.js';
+import { EmbedResponse } from '../../../types/embedRequestBody.js';
+import { Params } from '../../../types/requestBody.js';
+export declare const InfinityEmbedConfig: ProviderConfig;
+export declare const InfinityEmbedResponseTransform: (response: EmbedResponse | ErrorResponse, responseStatus: number, _responseHeaders: Headers, _strictOpenAiCompliance: boolean, gatewayRequestUrl: string, _gatewayRequest: Params) => EmbedResponse | ErrorResponse;

package/dist/tfy-providers/custom/infinity/embed.js

@@ -0,0 +1,24 @@
+import { CustomModelErrorResponseTransform } from '../vllm-openai/chatComplete.js';
+// TODOS: this configuration does not enforce the maximum token limit for the input parameter. If you want to enforce this, you might need to add a custom validation function or a max property to the ParameterConfig interface, and then use it in the input configuration. However, this might be complex because the token count is not a simple length check, but depends on the specific tokenization method used by the model.
+export const InfinityEmbedConfig = {
+    input: {
+        param: 'input',
+        required: true,
+    },
+    encoding_format: {
+        param: 'encoding_format',
+    },
+    dimensions: {
+        param: 'dimensions',
+    },
+    user: {
+        param: 'user',
+    },
+};
+export const InfinityEmbedResponseTransform = (response, responseStatus, _responseHeaders, _strictOpenAiCompliance, gatewayRequestUrl, _gatewayRequest) => {
+    if (responseStatus !== 200) {
+        return CustomModelErrorResponseTransform(response, responseStatus, gatewayRequestUrl);
+    }
+    return response;
+};
+//# sourceMappingURL=embed.js.map

package/dist/tfy-providers/custom/infinity/embed.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"embed.js","sourceRoot":"","sources":["../../../../src/tfy-providers/custom/infinity/embed.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,iCAAiC,EAAE,MAAM,6BAA6B,CAAC;AAEhF,saAAsa;AAEta,MAAM,CAAC,MAAM,mBAAmB,GAAmB;IACjD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;KACf;IACD,eAAe,EAAE;QACf,KAAK,EAAE,iBAAiB;KACzB;IACD,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;KACpB;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;CACF,CAAC;AAEF,MAAM,CAAC,MAAM,8BAA8B,GAON,CACnC,QAAQ,EACR,cAAc,EACd,gBAAgB,EAChB,uBAAuB,EACvB,iBAAiB,EACjB,eAAe,EACf,EAAE;IACF,IAAI,cAAc,KAAK,GAAG,EAAE,CAAC;QAC3B,OAAO,iCAAiC,CAAC,QAAQ,EAAE,cAAc,EAAE,iBAAiB,CAAC,CAAC;IACxF,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC,CAAC"}

package/dist/tfy-providers/custom/infinity/rerank.d.ts

@@ -0,0 +1,27 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+export declare const InfinityRerankConfig: {
+    query: {
+        param: string;
+        required: boolean;
+    };
+    top_n: {
+        param: string;
+        required: boolean;
+    };
+    documents: {
+        param: string;
+        required: boolean;
+    };
+    return_documents: {
+        param: string;
+        required: boolean;
+    };
+    max_tokens_per_doc: {
+        param: string;
+        required: boolean;
+    };
+};

package/dist/tfy-providers/custom/infinity/rerank.js

@@ -0,0 +1,28 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+export const InfinityRerankConfig = {
+    query: {
+        param: 'query',
+        required: true,
+    },
+    top_n: {
+        param: 'top_n',
+        required: false,
+    },
+    documents: {
+        param: 'documents',
+        required: true,
+    },
+    return_documents: {
+        param: 'return_documents',
+        required: false,
+    },
+    max_tokens_per_doc: {
+        param: 'max_tokens_per_doc',
+        required: false,
+    },
+};
+//# sourceMappingURL=rerank.js.map

package/dist/tfy-providers/custom/infinity/rerank.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"rerank.js","sourceRoot":"","sources":["../../../../src/tfy-providers/custom/infinity/rerank.ts"],"names":[],"mappings":"AAAA;;;;GAIG;AACH,MAAM,CAAC,MAAM,oBAAoB,GAAG;IAClC,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;KACf;IACD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,KAAK;KAChB;IACD,SAAS,EAAE;QACT,KAAK,EAAE,WAAW;QAClB,QAAQ,EAAE,IAAI;KACf;IACD,gBAAgB,EAAE;QAChB,KAAK,EAAE,kBAAkB;QACzB,QAAQ,EAAE,KAAK;KAChB;IACD,kBAAkB,EAAE;QAClB,KAAK,EAAE,oBAAoB;QAC3B,QAAQ,EAAE,KAAK;KAChB;CACF,CAAC"}

package/dist/tfy-providers/custom/nemo-retriever/embed.d.ts

@@ -0,0 +1,10 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+import { ErrorResponse, ProviderConfig } from '../../../providers/types.js';
+import { EmbedResponse } from '../../../types/embedRequestBody.js';
+import { Params } from '../../../types/requestBody.js';
+export declare const NemoRetrieverEmbedConfig: ProviderConfig;
+export declare const NemoRetrieverEmbedResponseTransform: (response: EmbedResponse | ErrorResponse, responseStatus: number, _responseHeaders: Headers, _strictOpenAiCompliance: boolean, gatewayRequestUrl: string, _gatewayRequest: Params) => EmbedResponse | ErrorResponse;

package/dist/tfy-providers/custom/nemo-retriever/embed.js

@@ -0,0 +1,28 @@
+import { CustomModelErrorResponseTransform } from '../vllm-openai/chatComplete.js';
+// TODOS: this configuration does not enforce the maximum token limit for the input parameter. If you want to enforce this, you might need to add a custom validation function or a max property to the ParameterConfig interface, and then use it in the input configuration. However, this might be complex because the token count is not a simple length check, but depends on the specific tokenization method used by the model.
+export const NemoRetrieverEmbedConfig = {
+    model: {
+        param: 'model',
+        required: true,
+    },
+    input: {
+        param: 'input',
+        required: true,
+    },
+    encoding_format: {
+        param: 'encoding_format',
+    },
+    dimensions: {
+        param: 'dimensions',
+    },
+    user: {
+        param: 'user',
+    },
+};
+export const NemoRetrieverEmbedResponseTransform = (response, responseStatus, _responseHeaders, _strictOpenAiCompliance, gatewayRequestUrl, _gatewayRequest) => {
+    if (responseStatus !== 200) {
+        return CustomModelErrorResponseTransform(response, responseStatus, gatewayRequestUrl);
+    }
+    return response;
+};
+//# sourceMappingURL=embed.js.map

package/dist/tfy-providers/custom/nemo-retriever/embed.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"embed.js","sourceRoot":"","sources":["../../../../src/tfy-providers/custom/nemo-retriever/embed.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,iCAAiC,EAAE,MAAM,6BAA6B,CAAC;AAEhF,saAAsa;AAEta,MAAM,CAAC,MAAM,wBAAwB,GAAmB;IACtD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;KACf;IACD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;KACf;IACD,eAAe,EAAE;QACf,KAAK,EAAE,iBAAiB;KACzB;IACD,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;KACpB;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;CACF,CAAC;AAEF,MAAM,CAAC,MAAM,mCAAmC,GAOX,CACnC,QAAQ,EACR,cAAc,EACd,gBAAgB,EAChB,uBAAuB,EACvB,iBAAiB,EACjB,eAAe,EACf,EAAE;IACF,IAAI,cAAc,KAAK,GAAG,EAAE,CAAC;QAC3B,OAAO,iCAAiC,CAAC,QAAQ,EAAE,cAAc,EAAE,iBAAiB,CAAC,CAAC;IACxF,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC,CAAC"}

package/dist/tfy-providers/custom/nemo-retriever/rerank.d.ts

@@ -0,0 +1,45 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+import { ErrorResponse, RerankResponse } from '../../../providers/types.js';
+import { Params } from '../../../types/requestBody.js';
+export declare const NemoRetrieverRerankConfig: {
+    model: {
+        param: string;
+        required: boolean;
+    };
+    query: {
+        param: string;
+        required: boolean;
+        transform: (params: Params) => {
+            text: string | undefined;
+        };
+    };
+    documents: {
+        param: string;
+        required: boolean;
+        transform: (params: Params) => {
+            text: string;
+        }[] | undefined;
+    };
+    max_tokens_per_doc: {
+        param: string;
+        required: boolean;
+        transform: (params: Params) => "END" | "NONE";
+    };
+};
+export interface NemoRetrieverRerankResponse {
+    rankings: [
+        {
+            index: number;
+            logit: number;
+        }
+    ];
+    usage?: {
+        prompt_tokens: number;
+        total_tokens: number;
+    };
+}
+export declare const NemoRetrieverRerankResponseTransform: (response: NemoRetrieverRerankResponse, responseStatus: number, _responseHeaders: Headers, _strictOpenAiCompliance: boolean, gatewayRequestUrl: string, request: Params) => RerankResponse | ErrorResponse;

package/dist/tfy-providers/custom/nemo-retriever/rerank.js

@@ -0,0 +1,53 @@
+import { CustomModelErrorResponseTransform } from '../vllm-openai/chatComplete.js';
+export const NemoRetrieverRerankConfig = {
+    model: {
+        param: 'model',
+        required: true,
+    },
+    query: {
+        param: 'query',
+        required: true,
+        transform: (params) => {
+            return { text: params.query };
+        },
+    },
+    documents: {
+        param: 'passages',
+        required: true,
+        transform: (params) => params.documents?.map((doc) => {
+            return { text: doc };
+        }),
+    },
+    max_tokens_per_doc: {
+        param: 'truncate',
+        required: false,
+        transform: (params) => {
+            if (params.max_tokens_per_doc && params.max_tokens_per_doc > 0) {
+                return 'END';
+            }
+            return 'NONE';
+        },
+    },
+};
+export const NemoRetrieverRerankResponseTransform = (response, responseStatus, _responseHeaders, _strictOpenAiCompliance, gatewayRequestUrl, request) => {
+    if (responseStatus !== 200) {
+        return CustomModelErrorResponseTransform(response, responseStatus, gatewayRequestUrl);
+    }
+    const rerankResponse = {
+        id: crypto.randomUUID(),
+        results: response.rankings.map(ranking => ({
+            index: ranking.index,
+            relevance_score: ranking.logit,
+        })),
+        usage: {
+            prompt_tokens: response.usage?.prompt_tokens || 0,
+            total_tokens: response.usage?.total_tokens || 0,
+            completion_tokens: 0,
+        },
+    };
+    if (request.top_n && request.top_n > 0) {
+        rerankResponse.results = rerankResponse.results.slice(0, request.top_n);
+    }
+    return rerankResponse;
+};
+//# sourceMappingURL=rerank.js.map

package/dist/tfy-providers/custom/nemo-retriever/rerank.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"rerank.js","sourceRoot":"","sources":["../../../../src/tfy-providers/custom/nemo-retriever/rerank.ts"],"names":[],"mappings":"AAOA,OAAO,EAAE,iCAAiC,EAAE,MAAM,6BAA6B,CAAC;AAEhF,MAAM,CAAC,MAAM,yBAAyB,GAAG;IACvC,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;KACf;IACD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;QACd,SAAS,EAAE,CAAC,MAAc,EAAE,EAAE;YAC5B,OAAO,EAAE,IAAI,EAAE,MAAM,CAAC,KAAK,EAAE,CAAC;QAChC,CAAC;KACF;IACD,SAAS,EAAE;QACT,KAAK,EAAE,UAAU;QACjB,QAAQ,EAAE,IAAI;QACd,SAAS,EAAE,CAAC,MAAc,EAAE,EAAE,CAC5B,MAAM,CAAC,SAAS,EAAE,GAAG,CAAC,CAAC,GAAW,EAAE,EAAE;YACpC,OAAO,EAAE,IAAI,EAAE,GAAG,EAAE,CAAC;QACvB,CAAC,CAAC;KACL;IACD,kBAAkB,EAAE;QAClB,KAAK,EAAE,UAAU;QACjB,QAAQ,EAAE,KAAK;QACf,SAAS,EAAE,CAAC,MAAc,EAAE,EAAE;YAC5B,IAAI,MAAM,CAAC,kBAAkB,IAAI,MAAM,CAAC,kBAAkB,GAAG,CAAC,EAAE,CAAC;gBAC/D,OAAO,KAAK,CAAC;YACf,CAAC;YACD,OAAO,MAAM,CAAC;QAChB,CAAC;KACF;CACF,CAAC;AAeF,MAAM,CAAC,MAAM,oCAAoC,GAOX,CACpC,QAAQ,EACR,cAAc,EACd,gBAAgB,EAChB,uBAAuB,EACvB,iBAAiB,EACjB,OAAO,EACP,EAAE;IACF,IAAI,cAAc,KAAK,GAAG,EAAE,CAAC;QAC3B,OAAO,iCAAiC,CAAC,QAAQ,EAAE,cAAc,EAAE,iBAAiB,CAAC,CAAC;IACxF,CAAC;IACD,MAAM,cAAc,GAAmB;QACrC,EAAE,EAAE,MAAM,CAAC,UAAU,EAAE;QACvB,OAAO,EAAE,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC;YACzC,KAAK,EAAE,OAAO,CAAC,KAAK;YACpB,eAAe,EAAE,OAAO,CAAC,KAAK;SAC/B,CAAC,CAAC;QACH,KAAK,EAAE;YACL,aAAa,EAAE,QAAQ,CAAC,KAAK,EAAE,aAAa,IAAI,CAAC;YACjD,YAAY,EAAE,QAAQ,CAAC,KAAK,EAAE,YAAY,IAAI,CAAC;YAC/C,iBAAiB,EAAE,CAAC;SACrB;KACF,CAAC;IACF,IAAI,OAAO,CAAC,KAAK,IAAI,OAAO,CAAC,KAAK,GAAG,CAAC,EAAE,CAAC;QACvC,cAAc,CAAC,OAAO,GAAG,cAAc,CAAC,OAAO,CAAC,KAAK,CAAC,CAAC,EAAE,OAAO,CAAC,KAAK,CAAC,CAAC;IAC1E,CAAC;IACD,OAAO,cAAc,CAAC;AACxB,CAAC,CAAC"}

package/dist/tfy-providers/custom/openai-compatible/chatComplete.d.ts

@@ -0,0 +1,12 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+import { ChatCompletionResponse, ErrorResponse, ProviderConfig } from '../../../providers/types.js';
+import { Params } from '../../../types/requestBody.js';
+export declare const OpenAICompatibleChatCompleteConfig: ProviderConfig;
+export interface OpenAICompatibleChatCompleteResponse extends ChatCompletionResponse {
+    system_fingerprint: string;
+}
+export declare const OpenAICompatibleChatCompleteResponseTransform: (response: OpenAICompatibleChatCompleteResponse | ErrorResponse, responseStatus: number, _responseHeaders: Headers, _strictOpenAiCompliance: boolean, gatewayRequestUrl: string, _gatewayRequest: Params) => ChatCompletionResponse | ErrorResponse;

package/dist/tfy-providers/custom/openai-compatible/chatComplete.js

@@ -0,0 +1,109 @@
+import { CustomModelErrorResponseTransform } from '../vllm-openai/chatComplete.js';
+// TODOS: this configuration does not enforce the maximum token limit for the input parameter. If you want to enforce this, you might need to add a custom validation function or a max property to the ParameterConfig interface, and then use it in the input configuration. However, this might be complex because the token count is not a simple length check, but depends on the specific tokenization method used by the model.
+export const OpenAICompatibleChatCompleteConfig = {
+    model: {
+        param: 'model',
+        transform: (params) => {
+            return params.model;
+        },
+    },
+    messages: {
+        param: 'messages',
+        default: '',
+    },
+    functions: {
+        param: 'functions',
+    },
+    function_call: {
+        param: 'function_call',
+    },
+    max_tokens: {
+        param: 'max_tokens',
+        default: 100,
+        min: 0,
+    },
+    temperature: {
+        param: 'temperature',
+        default: 1,
+        min: 0,
+        max: 2,
+    },
+    top_p: {
+        param: 'top_p',
+        default: 1,
+        min: 0,
+        max: 1,
+    },
+    n: {
+        param: 'n',
+        default: 1,
+    },
+    stream: {
+        param: 'stream',
+        default: false,
+    },
+    stop: {
+        param: 'stop',
+    },
+    presence_penalty: {
+        param: 'presence_penalty',
+        min: -2,
+        max: 2,
+    },
+    frequency_penalty: {
+        param: 'frequency_penalty',
+        min: -2,
+        max: 2,
+    },
+    logit_bias: {
+        param: 'logit_bias',
+    },
+    user: {
+        param: 'user',
+    },
+    seed: {
+        param: 'seed',
+    },
+    tools: {
+        param: 'tools',
+        required: false,
+        transform: (params) => {
+            if (params.tools && params.tools.length > 0) {
+                return params.tools;
+            }
+            return [];
+        },
+    },
+    tool_choice: {
+        param: 'tool_choice',
+    },
+    response_format: {
+        param: 'response_format',
+    },
+    logprobs: {
+        param: 'logprobs',
+        default: false,
+    },
+    top_logprobs: {
+        param: 'top_logprobs',
+    },
+    stream_options: {
+        param: 'stream_options',
+        required: (params) => params.stream === true,
+        default: { include_usage: true },
+        transform: (params) => {
+            const existingStreamOptions = params.stream_options || {};
+            return {
+                ...existingStreamOptions,
+                include_usage: true,
+            };
+        },
+    },
+};
+export const OpenAICompatibleChatCompleteResponseTransform = (response, responseStatus, _responseHeaders, _strictOpenAiCompliance, gatewayRequestUrl, _gatewayRequest) => {
+    if (responseStatus !== 200) {
+        return CustomModelErrorResponseTransform(response, responseStatus, gatewayRequestUrl);
+    }
+    return response;
+};
+//# sourceMappingURL=chatComplete.js.map

package/dist/tfy-providers/custom/openai-compatible/chatComplete.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"chatComplete.js","sourceRoot":"","sources":["../../../../src/tfy-providers/custom/openai-compatible/chatComplete.ts"],"names":[],"mappings":"AAOA,OAAO,EAAE,iCAAiC,EAAE,MAAM,6BAA6B,CAAC;AAEhF,saAAsa;AAEta,MAAM,CAAC,MAAM,kCAAkC,GAAmB;IAChE,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,SAAS,EAAE,CAAC,MAAc,EAAE,EAAE;YAC5B,OAAO,MAAM,CAAC,KAAK,CAAC;QACtB,CAAC;KACF;IACD,QAAQ,EAAE;QACR,KAAK,EAAE,UAAU;QACjB,OAAO,EAAE,EAAE;KACZ;IACD,SAAS,EAAE;QACT,KAAK,EAAE,WAAW;KACnB;IACD,aAAa,EAAE;QACb,KAAK,EAAE,eAAe;KACvB;IACD,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;QACnB,OAAO,EAAE,GAAG;QACZ,GAAG,EAAE,CAAC;KACP;IACD,WAAW,EAAE;QACX,KAAK,EAAE,aAAa;QACpB,OAAO,EAAE,CAAC;QACV,GAAG,EAAE,CAAC;QACN,GAAG,EAAE,CAAC;KACP;IACD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,OAAO,EAAE,CAAC;QACV,GAAG,EAAE,CAAC;QACN,GAAG,EAAE,CAAC;KACP;IACD,CAAC,EAAE;QACD,KAAK,EAAE,GAAG;QACV,OAAO,EAAE,CAAC;KACX;IACD,MAAM,EAAE;QACN,KAAK,EAAE,QAAQ;QACf,OAAO,EAAE,KAAK;KACf;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;IACD,gBAAgB,EAAE;QAChB,KAAK,EAAE,kBAAkB;QACzB,GAAG,EAAE,CAAC,CAAC;QACP,GAAG,EAAE,CAAC;KACP;IACD,iBAAiB,EAAE;QACjB,KAAK,EAAE,mBAAmB;QAC1B,GAAG,EAAE,CAAC,CAAC;QACP,GAAG,EAAE,CAAC;KACP;IACD,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;KACpB;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;IACD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,KAAK;QACf,SAAS,EAAE,CAAC,MAAc,EAAE,EAAE;YAC5B,IAAI,MAAM,CAAC,KAAK,IAAI,MAAM,CAAC,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBAC5C,OAAO,MAAM,CAAC,KAAK,CAAC;YACtB,CAAC;YACD,OAAO,EAAE,CAAC;QACZ,CAAC;KACF;IACD,WAAW,EAAE;QACX,KAAK,EAAE,aAAa;KACrB;IACD,eAAe,EAAE;QACf,KAAK,EAAE,iBAAiB;KACzB;IACD,QAAQ,EAAE;QACR,KAAK,EAAE,UAAU;QACjB,OAAO,EAAE,KAAK;KACf;IACD,YAAY,EAAE;QACZ,KAAK,EAAE,cAAc;KACtB;IACD,cAAc,EAAE;QACd,KAAK,EAAE,gBAAgB;QACvB,QAAQ,EAAE,CAAC,MAAc,EAAE,EAAE,CAAC,MAAM,CAAC,MAAM,KAAK,IAAI;QACpD,OAAO,EAAE,EAAE,aAAa,EAAE,IAAI,EAAE;QAChC,SAAS,EAAE,CAAC,MAAW,EAAE,EAAE;YACzB,MAAM,qBAAqB,GAAG,MAAM,CAAC,cAAc,IAAI,EAAE,CAAC;YAC1D,OAAO;gBACL,GAAG,qBAAqB;gBACxB,aAAa,EAAE,IAAI;aACpB,CAAC;QACJ,CAAC;KACF;CACF,CAAC;AAMF,MAAM,CAAC,MAAM,6CAA6C,GAOZ,CAC5C,QAAQ,EACR,cAAc,EACd,gBAAgB,EAChB,uBAAuB,EACvB,iBAAiB,EACjB,eAAe,EACf,EAAE;IACF,IAAI,cAAc,KAAK,GAAG,EAAE,CAAC;QAC3B,OAAO,iCAAiC,CAAC,QAAQ,EAAE,cAAc,EAAE,iBAAiB,CAAC,CAAC;IACxF,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC,CAAC"}

package/dist/tfy-providers/custom/openai-compatible/complete.d.ts

@@ -0,0 +1,12 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+import { CompletionResponse, ErrorResponse, ProviderConfig } from '../../../providers/types.js';
+import { Params } from '../../../types/requestBody.js';
+export declare const OpenAICompatibleCompleteConfig: ProviderConfig;
+export interface OpenAICompatibleCompleteResponse extends CompletionResponse {
+    system_fingerprint: string;
+}
+export declare const OpenAICompatibleCompleteResponseTransform: (response: OpenAICompatibleCompleteResponse | ErrorResponse, responseStatus: number, _responseHeaders: Headers, _strictOpenAiCompliance: boolean, gatewayRequestUrl: string, _gatewayRequest: Params) => CompletionResponse | ErrorResponse;

package/dist/tfy-providers/custom/openai-compatible/complete.js

@@ -0,0 +1,82 @@
+import { CustomModelErrorResponseTransform } from '../vllm-openai/chatComplete.js';
+// TODOS: this configuration does not enforce the maximum token limit for the input parameter. If you want to enforce this, you might need to add a custom validation function or a max property to the ParameterConfig interface, and then use it in the input configuration. However, this might be complex because the token count is not a simple length check, but depends on the specific tokenization method used by the model.
+export const OpenAICompatibleCompleteConfig = {
+    model: {
+        param: 'model',
+        transform: (params) => {
+            return params.model;
+        },
+    },
+    prompt: {
+        param: 'prompt',
+        default: '',
+    },
+    max_tokens: {
+        param: 'max_tokens',
+        default: 100,
+        min: 0,
+    },
+    temperature: {
+        param: 'temperature',
+        default: 1,
+        min: 0,
+        max: 2,
+    },
+    top_p: {
+        param: 'top_p',
+        default: 1,
+        min: 0,
+        max: 1,
+    },
+    n: {
+        param: 'n',
+        default: 1,
+    },
+    stream: {
+        param: 'stream',
+        default: false,
+    },
+    logprobs: {
+        param: 'logprobs',
+        max: 5,
+    },
+    echo: {
+        param: 'echo',
+        default: false,
+    },
+    stop: {
+        param: 'stop',
+    },
+    presence_penalty: {
+        param: 'presence_penalty',
+        min: -2,
+        max: 2,
+    },
+    frequency_penalty: {
+        param: 'frequency_penalty',
+        min: -2,
+        max: 2,
+    },
+    best_of: {
+        param: 'best_of',
+    },
+    logit_bias: {
+        param: 'logit_bias',
+    },
+    user: {
+        param: 'user',
+    },
+    seed: {
+        param: 'seed',
+    },
+    suffix: {
+        param: 'suffix',
+    },
+};
+export const OpenAICompatibleCompleteResponseTransform = (response, responseStatus, _responseHeaders, _strictOpenAiCompliance, gatewayRequestUrl, _gatewayRequest) => {
+    if (responseStatus !== 200) {
+        return CustomModelErrorResponseTransform(response, responseStatus, gatewayRequestUrl);
+    }
+    return response;
+};
+//# sourceMappingURL=complete.js.map

package/dist/tfy-providers/custom/openai-compatible/complete.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"complete.js","sourceRoot":"","sources":["../../../../src/tfy-providers/custom/openai-compatible/complete.ts"],"names":[],"mappings":"AAOA,OAAO,EAAE,iCAAiC,EAAE,MAAM,6BAA6B,CAAC;AAEhF,saAAsa;AAEta,MAAM,CAAC,MAAM,8BAA8B,GAAmB;IAC5D,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,SAAS,EAAE,CAAC,MAAc,EAAE,EAAE;YAC5B,OAAO,MAAM,CAAC,KAAK,CAAC;QACtB,CAAC;KACF;IACD,MAAM,EAAE;QACN,KAAK,EAAE,QAAQ;QACf,OAAO,EAAE,EAAE;KACZ;IACD,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;QACnB,OAAO,EAAE,GAAG;QACZ,GAAG,EAAE,CAAC;KACP;IACD,WAAW,EAAE;QACX,KAAK,EAAE,aAAa;QACpB,OAAO,EAAE,CAAC;QACV,GAAG,EAAE,CAAC;QACN,GAAG,EAAE,CAAC;KACP;IACD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,OAAO,EAAE,CAAC;QACV,GAAG,EAAE,CAAC;QACN,GAAG,EAAE,CAAC;KACP;IACD,CAAC,EAAE;QACD,KAAK,EAAE,GAAG;QACV,OAAO,EAAE,CAAC;KACX;IACD,MAAM,EAAE;QACN,KAAK,EAAE,QAAQ;QACf,OAAO,EAAE,KAAK;KACf;IACD,QAAQ,EAAE;QACR,KAAK,EAAE,UAAU;QACjB,GAAG,EAAE,CAAC;KACP;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;QACb,OAAO,EAAE,KAAK;KACf;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;IACD,gBAAgB,EAAE;QAChB,KAAK,EAAE,kBAAkB;QACzB,GAAG,EAAE,CAAC,CAAC;QACP,GAAG,EAAE,CAAC;KACP;IACD,iBAAiB,EAAE;QACjB,KAAK,EAAE,mBAAmB;QAC1B,GAAG,EAAE,CAAC,CAAC;QACP,GAAG,EAAE,CAAC;KACP;IACD,OAAO,EAAE;QACP,KAAK,EAAE,SAAS;KACjB;IACD,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;KACpB;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;IACD,MAAM,EAAE;QACN,KAAK,EAAE,QAAQ;KAChB;CACF,CAAC;AAMF,MAAM,CAAC,MAAM,yCAAyC,GAOZ,CACxC,QAAQ,EACR,cAAc,EACd,gBAAgB,EAChB,uBAAuB,EACvB,iBAAiB,EACjB,eAAe,EACf,EAAE;IACF,IAAI,cAAc,KAAK,GAAG,EAAE,CAAC;QAC3B,OAAO,iCAAiC,CAAC,QAAQ,EAAE,cAAc,EAAE,iBAAiB,CAAC,CAAC;IACxF,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC,CAAC"}

package/dist/tfy-providers/custom/openai-compatible/embed.d.ts

@@ -0,0 +1,10 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+import { ErrorResponse, ProviderConfig } from '../../../providers/types.js';
+import { EmbedResponse } from '../../../types/embedRequestBody.js';
+import { Params } from '../../../types/requestBody.js';
+export declare const OpenAICompatibleEmbedConfig: ProviderConfig;
+export declare const OpenAICompatibleEmbedResponseTransform: (response: EmbedResponse | ErrorResponse, responseStatus: number, _responseHeaders: Headers, _strictOpenAiCompliance: boolean, gatewayRequestUrl: string, _gatewayRequest: Params) => EmbedResponse | ErrorResponse;

package/dist/tfy-providers/custom/openai-compatible/embed.js

@@ -0,0 +1,28 @@
+import { CustomModelErrorResponseTransform } from '../vllm-openai/chatComplete.js';
+// TODOS: this configuration does not enforce the maximum token limit for the input parameter. If you want to enforce this, you might need to add a custom validation function or a max property to the ParameterConfig interface, and then use it in the input configuration. However, this might be complex because the token count is not a simple length check, but depends on the specific tokenization method used by the model.
+export const OpenAICompatibleEmbedConfig = {
+    model: {
+        param: 'model',
+        required: true,
+    },
+    input: {
+        param: 'input',
+        required: true,
+    },
+    encoding_format: {
+        param: 'encoding_format',
+    },
+    dimensions: {
+        param: 'dimensions',
+    },
+    user: {
+        param: 'user',
+    },
+};
+export const OpenAICompatibleEmbedResponseTransform = (response, responseStatus, _responseHeaders, _strictOpenAiCompliance, gatewayRequestUrl, _gatewayRequest) => {
+    if (responseStatus !== 200) {
+        return CustomModelErrorResponseTransform(response, responseStatus, gatewayRequestUrl);
+    }
+    return response;
+};
+//# sourceMappingURL=embed.js.map

package/dist/tfy-providers/custom/openai-compatible/embed.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"embed.js","sourceRoot":"","sources":["../../../../src/tfy-providers/custom/openai-compatible/embed.ts"],"names":[],"mappings":"AAQA,OAAO,EAAE,iCAAiC,EAAE,MAAM,6BAA6B,CAAC;AAEhF,saAAsa;AAEta,MAAM,CAAC,MAAM,2BAA2B,GAAmB;IACzD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;KACf;IACD,KAAK,EAAE;QACL,KAAK,EAAE,OAAO;QACd,QAAQ,EAAE,IAAI;KACf;IACD,eAAe,EAAE;QACf,KAAK,EAAE,iBAAiB;KACzB;IACD,UAAU,EAAE;QACV,KAAK,EAAE,YAAY;KACpB;IACD,IAAI,EAAE;QACJ,KAAK,EAAE,MAAM;KACd;CACF,CAAC;AAEF,MAAM,CAAC,MAAM,sCAAsC,GAOd,CACnC,QAAQ,EACR,cAAc,EACd,gBAAgB,EAChB,uBAAuB,EACvB,iBAAiB,EACjB,eAAe,EACf,EAAE;IACF,IAAI,cAAc,KAAK,GAAG,EAAE,CAAC;QAC3B,OAAO,iCAAiC,CAAC,QAAQ,EAAE,cAAc,EAAE,iBAAiB,CAAC,CAAC;IACxF,CAAC;IACD,OAAO,QAAQ,CAAC;AAClB,CAAC,CAAC"}

package/dist/tfy-providers/custom/tei/embed.d.ts

@@ -0,0 +1,10 @@
+/**
+ * ================================
+ * TFY Managed File
+ * ================================
+ */
+import { ErrorResponse, ProviderConfig } from '../../../providers/types.js';
+import { EmbedResponse } from '../../../types/embedRequestBody.js';
+import { Params } from '../../../types/requestBody.js';
+export declare const TEIEmbedConfig: ProviderConfig;
+export declare const TEIEmbedResponseTransform: (response: EmbedResponse | ErrorResponse, responseStatus: number, _responseHeaders: Headers, _strictOpenAiCompliance: boolean, gatewayRequestUrl: string, _gatewayRequest: Params) => EmbedResponse | ErrorResponse;

package/dist/tfy-providers/custom/tei/embed.js

@@ -0,0 +1,24 @@
+import { CustomModelErrorResponseTransform } from '../vllm-openai/chatComplete.js';
+// TODOS: this configuration does not enforce the maximum token limit for the input parameter. If you want to enforce this, you might need to add a custom validation function or a max property to the ParameterConfig interface, and then use it in the input configuration. However, this might be complex because the token count is not a simple length check, but depends on the specific tokenization method used by the model.
+export const TEIEmbedConfig = {
+    input: {
+        param: 'input',
+        required: true,
+    },
+    encoding_format: {
+        param: 'encoding_format',
+    },
+    dimensions: {
+        param: 'dimensions',
+    },
+    user: {
+        param: 'user',
+    },
+};
+export const TEIEmbedResponseTransform = (response, responseStatus, _responseHeaders, _strictOpenAiCompliance, gatewayRequestUrl, _gatewayRequest) => {
+    if (responseStatus !== 200) {
+        return CustomModelErrorResponseTransform(response, responseStatus, gatewayRequestUrl);
+    }
+    return response;
+};
+//# sourceMappingURL=embed.js.map