Add langserve endpoint (#1009)

* Add support for langserve endpoints

* Add support for langserve endpoints

* Fix linting

* Fix linting issues

* Fix issue import

---------

Co-authored-by: antoniora <antonio.ramos@adyen.com>

README.md

@@ -618,6 +618,24 @@ MODELS=`[
 
 ```
 
+##### LangServe
+
+LangChain applications that are deployed using LangServe can be called with the following config:
+
+```
+MODELS=`[
+//...
+    {
+       "name": "summarization-chain", //model-name
+       "endpoints" : [{
+         "type": "langserve",
+         "url" : "http://127.0.0.1:8100",
+       }]
+     },
+]`
+
+```
+
 ### Custom endpoint authorization
 
 #### Basic and Bearer

src/lib/server/endpoints/endpoints.ts

@@ -17,6 +17,9 @@ import endpointCloudflare, {
 	endpointCloudflareParametersSchema,
 } from "./cloudflare/endpointCloudflare";
 import { endpointCohere, endpointCohereParametersSchema } from "./cohere/endpointCohere";
+import endpointLangserve, {
+	endpointLangserveParametersSchema,
+} from "./langserve/endpointLangserve";
 
 // parameters passed when generating text
 export interface EndpointParameters {
@@ -48,6 +51,7 @@ export const endpoints = {
 	vertex: endpointVertex,
 	cloudflare: endpointCloudflare,
 	cohere: endpointCohere,
+	langserve: endpointLangserve,
 };
 
 export const endpointSchema = z.discriminatedUnion("type", [
@@ -60,5 +64,6 @@ export const endpointSchema = z.discriminatedUnion("type", [
 	endpointVertexParametersSchema,
 	endpointCloudflareParametersSchema,
 	endpointCohereParametersSchema,
+	endpointLangserveParametersSchema,
 ]);
 export default endpoints;

src/lib/server/endpoints/langserve/endpointLangserve.ts

@@ -0,0 +1,128 @@
+import { buildPrompt } from "$lib/buildPrompt";
+import { z } from "zod";
+import type { Endpoint } from "../endpoints";
+import type { TextGenerationStreamOutput } from "@huggingface/inference";
+
+export const endpointLangserveParametersSchema = z.object({
+	weight: z.number().int().positive().default(1),
+	model: z.any(),
+	type: z.literal("langserve"),
+	url: z.string().url(),
+});
+
+export function endpointLangserve(
+	input: z.input<typeof endpointLangserveParametersSchema>
+): Endpoint {
+	const { url, model } = endpointLangserveParametersSchema.parse(input);
+
+	return async ({ messages, preprompt, continueMessage }) => {
+		const prompt = await buildPrompt({
+			messages,
+			continueMessage,
+			preprompt,
+			model,
+		});
+
+		const r = await fetch(`${url}/stream`, {
+			method: "POST",
+			headers: {
+				"Content-Type": "application/json",
+			},
+			body: JSON.stringify({
+				input: { text: prompt },
+			}),
+		});
+
+		if (!r.ok) {
+			throw new Error(`Failed to generate text: ${await r.text()}`);
+		}
+
+		const encoder = new TextDecoderStream();
+		const reader = r.body?.pipeThrough(encoder).getReader();
+
+		return (async function* () {
+			let stop = false;
+			let generatedText = "";
+			let tokenId = 0;
+			let accumulatedData = ""; // Buffer to accumulate data chunks
+
+			while (!stop) {
+				// Read the stream and log the outputs to console
+				const out = (await reader?.read()) ?? { done: false, value: undefined };
+
+				// If it's done, we cancel
+				if (out.done) {
+					reader?.cancel();
+					return;
+				}
+
+				if (!out.value) {
+					return;
+				}
+
+				// Accumulate the data chunk
+				accumulatedData += out.value;
+				// Keep read data to check event type
+				const eventData = out.value;
+
+				// Process each complete JSON object in the accumulated data
+				while (accumulatedData.includes("\n")) {
+					// Assuming each JSON object ends with a newline
+					const endIndex = accumulatedData.indexOf("\n");
+					let jsonString = accumulatedData.substring(0, endIndex).trim();
+					// Remove the processed part from the buffer
+
+					accumulatedData = accumulatedData.substring(endIndex + 1);
+
+					// Stopping with end event
+					if (eventData.startsWith("event: end")) {
+						stop = true;
+						yield {
+							token: {
+								id: tokenId++,
+								text: "",
+								logprob: 0,
+								special: true,
+							},
+							generated_text: generatedText,
+							details: null,
+						} satisfies TextGenerationStreamOutput;
+						reader?.cancel();
+						continue;
+					}
+
+					if (eventData.startsWith("event: data") && jsonString.startsWith("data: ")) {
+						jsonString = jsonString.slice(6);
+						let data = null;
+
+						// Handle the parsed data
+						try {
+							data = JSON.parse(jsonString);
+						} catch (e) {
+							console.error("Failed to parse JSON", e);
+							console.error("Problematic JSON string:", jsonString);
+							continue; // Skip this iteration and try the next chunk
+						}
+						// Assuming content within data is a plain string
+						if (data) {
+							generatedText += data;
+							const output: TextGenerationStreamOutput = {
+								token: {
+									id: tokenId++,
+									text: data,
+									logprob: 0,
+									special: false,
+								},
+								generated_text: null,
+								details: null,
+							};
+							yield output;
+						}
+					}
+				}
+			}
+		})();
+	};
+}
+
+export default endpointLangserve;

src/lib/server/models.ts

@@ -177,6 +177,8 @@ const addEndpoint = (m: Awaited<ReturnType<typeof processModel>>) => ({
 						return await endpoints.cloudflare(args);
 					case "cohere":
 						return await endpoints.cohere(args);
+					case "langserve":
+						return await endpoints.langserve(args);
 					default:
 						// for legacy reason
 						return endpoints.tgi(args);