sarvam-ai-sdk 0.0.1 → 0.0.2

package/README.md

@@ -5,18 +5,18 @@ contains language model support for the Sarvam chat completion, Text-to-Speech a
 
 ## Setup
 
-The **[Sarvam](http://sarvam.ai)** provider is available in the `@ai-sdk/sarvam` module. You can install it with
+The **[Sarvam](http://sarvam.ai)** provider is available in the `sarvam-ai-sdk` module. You can install it with
 
 ```bash
-npm i @ai-sdk/sarvam
+npm i sarvam-ai-sdk
 ```
 
 ## Provider Instance
 
-You can import the default provider instance `sarvam` from `@ai-sdk/sarvam`:
+You can import the default provider instance `sarvam` from `sarvam-ai-sdk`:
 
 ```ts
-import { sarvam } from '@ai-sdk/sarvam';
+import { sarvam } from 'sarvam-ai-sdk';
 ```
 
 Create `.env` file with API key from **[Sarvam Dashboard](https://dashboard.sarvam.ai/)**
@@ -27,7 +27,7 @@ SARVAM_API_KEY="your_api_key"
 ## Example
 
 ```ts
-import { sarvam } from '@ai-sdk/sarvam';
+import { sarvam } from 'sarvam-ai-sdk';
 import { generateText } from 'ai';
 
 const { text } = await generateText({
@@ -41,7 +41,7 @@ console.log(text); // പാചകം തുടരൂ, സുഹൃത്തു
 ## Text-to-Speech
 
 ```ts
-import { sarvam } from "@ai-sdk/sarvam";
+import { sarvam } from "sarvam-ai-sdk";
 import { experimental_generateSpeech as generateSpeech } from "ai";
 import { writeFile } from "fs/promises";
 
@@ -57,7 +57,7 @@ await writeFile("./src/transcript-test.wav", audioBuffer);
 ## Speech-to-Text
 
 ```ts
-import { sarvam } from "@ai-sdk/sarvam";
+import { sarvam } from "sarvam-ai-sdk";
 import { experimental_transcribe as transcribe } from "ai";
 import { readFile } from "fs/promises";
 
@@ -69,6 +69,40 @@ const { text } = await transcribe({
 console.log(text); // പാചകം തുടരും സുഹൃത്തുക്കളെ
 ```
 
+## Tool Calling
+
+> [!WARNING]
+> Latest `sarvam-m` model isn't trained on native tool calling feature (aka JSON mode). So we simulate this with prompt engineering technique.
+
+```ts
+import { z } from "zod";
+import { generateText, tool } from "ai";
+import { sarvam } from "sarvam-ai-sdk";
+
+
+const result = await generateText({
+  model: sarvam("sarvam-m", {
+    simulateToolCalling: true, // ⚠️ important
+  }),
+  tools: {
+    weather: tool({
+      description: "Get the weather in a location",
+      parameters: z.object({
+        location: z.string().describe("The location to get the weather for"),
+      }),
+      execute: async ({ location }) => ({
+        location,
+        temperature: 72 + Math.floor(Math.random() * 21) - 10,
+      }),
+    }),
+  },
+  system: "Your are a helpful AI",
+  prompt: "കൊച്ചിയിലെ കാലാവസ്ഥ എന്താണ്?",
+});
+
+console.log(result.toolResults);
+```
+
 ## Documentation
 
 Please check out the **[Sarvam provider documentation](https://ai-sdk.dev/providers/ai-sdk-providers/sarvam)** and **[Sarvam API documentation](https://docs.sarvam.ai)** for more information.

package/dist/index.cjs

@@ -55,22 +55,25 @@ __export(index_exports, {
 module.exports = __toCommonJS(index_exports);
 
 // src/sarvam-provider.ts
-var import_provider_utils6 = require("@ai-sdk/provider-utils");
+var import_provider_utils7 = require("@ai-sdk/provider-utils");
 
 // src/sarvam-chat-language-model.ts
 var import_provider3 = require("@ai-sdk/provider");
-var import_provider_utils3 = require("@ai-sdk/provider-utils");
+var import_provider_utils4 = require("@ai-sdk/provider-utils");
 var import_zod2 = require("zod");
 
 // src/convert-to-sarvam-chat-messages.ts
 var import_provider = require("@ai-sdk/provider");
 var import_provider_utils = require("@ai-sdk/provider-utils");
-function convertToSarvamChatMessages(prompt) {
+function convertToSarvamChatMessages(prompt, fakeToolSystemPrompt) {
   const messages = [];
   for (const { role, content } of prompt) {
     switch (role) {
       case "system": {
-        messages.push({ role: "system", content });
+        const contentData = fakeToolSystemPrompt ? `${content}
+
+${fakeToolSystemPrompt}` : content;
+        messages.push({ role: "system", content: contentData });
         break;
       }
       case "user": {
@@ -165,6 +168,23 @@ function getResponseMetadata({
   };
 }
 
+// src/map-sarvam-finish-reason.ts
+function mapSarvamFinishReason(finishReason) {
+  switch (finishReason) {
+    case "stop":
+      return "stop";
+    case "length":
+      return "length";
+    case "content_filter":
+      return "content-filter";
+    case "function_call":
+    case "tool_calls":
+      return "tool-calls";
+    default:
+      return "unknown";
+  }
+}
+
 // src/sarvam-error.ts
 var import_zod = require("zod");
 var import_provider_utils2 = require("@ai-sdk/provider-utils");
@@ -181,6 +201,8 @@ var sarvamFailedResponseHandler = (0, import_provider_utils2.createJsonErrorResp
 
 // src/sarvam-prepare-tools.ts
 var import_provider2 = require("@ai-sdk/provider");
+var import_provider_utils3 = require("@ai-sdk/provider-utils");
+var import_json_schema_to_typescript = require("json-schema-to-typescript");
 function prepareTools({
   mode
 }) {
@@ -234,24 +256,69 @@ function prepareTools({
     }
   }
 }
-
-// src/map-sarvam-finish-reason.ts
-function mapSarvamFinishReason(finishReason) {
-  switch (finishReason) {
-    case "stop":
-      return "stop";
-    case "length":
-      return "length";
-    case "content_filter":
-      return "content-filter";
-    case "function_call":
-    case "tool_calls":
-      return "tool-calls";
-    default:
-      return "unknown";
+var simulateToolCalling = async (tools) => {
+  const context = [];
+  const names = [];
+  for (const tool of tools) {
+    names.push(tool.function.name);
+    const tsType = await (0, import_json_schema_to_typescript.compile)(
+      tool.function.parameters,
+      tool.function.name,
+      {
+        bannerComment: "",
+        format: false,
+        declareExternallyReferenced: true,
+        enableConstEnums: true,
+        unreachableDefinitions: false,
+        strictIndexSignatures: false
+      }
+    );
+    const toolContext = tsType.replace(
+      /export interface (\w+) \{/,
+      (_, name) => `type ${tool.function.name} = {`
+    ).replace(/\/\*\*\s*\n\s*\*\s*(.*?)\s*\n\s*\*\//, "// $1");
+    context.push(`// ${tool.function.description}
+${toolContext}`);
   }
+  const text = `These are the available tool you can execute.
+
+${context.join("\n")}
+
+type YourToolChoices = {
+ 'toolName': '${names.join("' | '")}',
+ 'toolData': ${names.join(" | ")}
 }
 
+Respond normally.
+If user request to execute any tool, respond with pure JSON format
+Make sure to translate toolData to English.
+
+eg:
+const myChoice: YourToolChoices = {
+    "toolName": <name>,
+    "toolData": <data>
+}`;
+  return text;
+};
+var extractToolCallData = (text) => {
+  const jsonRegex = /\{(?:[^{}]*|\{[^{}]*\})*\}/g;
+  const jsonMatches = text.match(jsonRegex);
+  if (jsonMatches && jsonMatches[0]) {
+    try {
+      const toolFunction = JSON.parse(jsonMatches[0]);
+      if (!("toolName" in toolFunction)) return;
+      if (!("toolData" in toolFunction)) return;
+      return {
+        args: JSON.stringify(toolFunction.toolData),
+        toolCallId: (0, import_provider_utils3.generateId)(),
+        toolCallType: "function",
+        toolName: toolFunction.toolName
+      };
+    } catch (error) {
+    }
+  }
+};
+
 // src/sarvam-chat-language-model.ts
 var SarvamChatLanguageModel = class {
   constructor(modelId, settings, config) {
@@ -268,7 +335,7 @@ var SarvamChatLanguageModel = class {
   get supportsImageUrls() {
     return !this.settings.downloadImages;
   }
-  getArgs({
+  async getArgs({
     mode,
     prompt,
     maxTokens,
@@ -298,14 +365,14 @@ var SarvamChatLanguageModel = class {
         details: "JSON response format schema is not supported"
       });
     }
-    const sarvamOptions = (0, import_provider_utils3.parseProviderOptions)({
+    const sarvamOptions = (0, import_provider_utils4.parseProviderOptions)({
       provider: "sarvam",
       providerOptions: providerMetadata,
       schema: import_zod2.z.object({
         reasoningFormat: import_zod2.z.enum(["parsed", "raw", "hidden"]).nullish()
       })
     });
-    const baseArgs = {
+    const baseArgs = (prompt2, fakeToolSystemPrompt) => ({
       // model id:
       model: this.modelId,
       // model specific settings:
@@ -327,15 +394,16 @@ var SarvamChatLanguageModel = class {
       // provider options:
       reasoning_format: sarvamOptions == null ? void 0 : sarvamOptions.reasoningFormat,
       // messages:
-      messages: convertToSarvamChatMessages(prompt)
-    };
+      messages: convertToSarvamChatMessages(prompt2, fakeToolSystemPrompt)
+    });
     switch (type) {
       case "regular": {
         const { tools, tool_choice, toolWarnings } = prepareTools({
           mode
         });
+        const fakeSystemPrompt = tools && this.settings.simulateToolCalling ? await simulateToolCalling(tools) : void 0;
         return {
-          args: __spreadProps(__spreadValues({}, baseArgs), {
+          args: __spreadProps(__spreadValues({}, baseArgs(prompt, fakeSystemPrompt)), {
             tools,
             tool_choice
           }),
@@ -344,7 +412,7 @@ var SarvamChatLanguageModel = class {
       }
       case "object-json": {
         return {
-          args: __spreadProps(__spreadValues({}, baseArgs), {
+          args: __spreadProps(__spreadValues({}, baseArgs(prompt)), {
             response_format: (
               // json object response format is not supported for streaming:
               stream === false ? { type: "json_object" } : void 0
@@ -355,7 +423,7 @@ var SarvamChatLanguageModel = class {
       }
       case "object-tool": {
         return {
-          args: __spreadProps(__spreadValues({}, baseArgs), {
+          args: __spreadProps(__spreadValues({}, baseArgs(prompt)), {
             tool_choice: {
               type: "function",
               function: { name: mode.tool.name }
@@ -382,21 +450,23 @@ var SarvamChatLanguageModel = class {
   }
   async doGenerate(options) {
     var _b, _c, _d, _e, _f, _g, _h;
-    const { args, warnings } = this.getArgs(__spreadProps(__spreadValues({}, options), { stream: false }));
+    const { args, warnings } = await this.getArgs(__spreadProps(__spreadValues({}, options), {
+      stream: false
+    }));
     const body = JSON.stringify(args);
     const {
       responseHeaders,
       value: response,
       rawValue: rawResponse
-    } = await (0, import_provider_utils3.postJsonToApi)({
+    } = await (0, import_provider_utils4.postJsonToApi)({
       url: this.config.url({
         path: "/chat/completions",
         modelId: this.modelId
       }),
-      headers: (0, import_provider_utils3.combineHeaders)(this.config.headers(), options.headers),
+      headers: (0, import_provider_utils4.combineHeaders)(this.config.headers(), options.headers),
       body: args,
       failedResponseHandler: sarvamFailedResponseHandler,
-      successfulResponseHandler: (0, import_provider_utils3.createJsonResponseHandler)(
+      successfulResponseHandler: (0, import_provider_utils4.createJsonResponseHandler)(
         sarvamChatResponseSchema
       ),
       abortSignal: options.abortSignal,
@@ -404,18 +474,29 @@ var SarvamChatLanguageModel = class {
     });
     const _a = args, { messages: rawPrompt } = _a, rawSettings = __objRest(_a, ["messages"]);
     const choice = response.choices[0];
+    let text = (_b = choice.message.content) != null ? _b : void 0;
+    let toolCalls = (_c = choice.message.tool_calls) == null ? void 0 : _c.map((toolCall) => {
+      var _a2;
+      return {
+        toolCallType: "function",
+        toolCallId: (_a2 = toolCall.id) != null ? _a2 : (0, import_provider_utils4.generateId)(),
+        toolName: toolCall.function.name,
+        args: toolCall.function.arguments
+      };
+    });
+    if (this.settings.simulateToolCalling) {
+      if (text && text.length !== 0 && (!toolCalls || (toolCalls == null ? void 0 : toolCalls.length) === 0)) {
+        const newTools = extractToolCallData(text);
+        if (newTools) {
+          toolCalls = [newTools];
+          text = void 0;
+        }
+      }
+    }
     return {
-      text: (_b = choice.message.content) != null ? _b : void 0,
-      reasoning: (_c = choice.message.reasoning) != null ? _c : void 0,
-      toolCalls: (_d = choice.message.tool_calls) == null ? void 0 : _d.map((toolCall) => {
-        var _a2;
-        return {
-          toolCallType: "function",
-          toolCallId: (_a2 = toolCall.id) != null ? _a2 : (0, import_provider_utils3.generateId)(),
-          toolName: toolCall.function.name,
-          args: toolCall.function.arguments
-        };
-      }),
+      text,
+      toolCalls,
+      reasoning: (_d = choice.message.reasoning) != null ? _d : void 0,
       finishReason: mapSarvamFinishReason(choice.finish_reason),
       usage: {
         promptTokens: (_f = (_e = response.usage) == null ? void 0 : _e.prompt_tokens) != null ? _f : NaN,
@@ -429,19 +510,21 @@ var SarvamChatLanguageModel = class {
     };
   }
   async doStream(options) {
-    const { args, warnings } = this.getArgs(__spreadProps(__spreadValues({}, options), { stream: true }));
+    const { args, warnings } = await this.getArgs(__spreadProps(__spreadValues({}, options), { stream: true }));
     const body = JSON.stringify(__spreadProps(__spreadValues({}, args), { stream: true }));
-    const { responseHeaders, value: response } = await (0, import_provider_utils3.postJsonToApi)({
+    const { responseHeaders, value: response } = await (0, import_provider_utils4.postJsonToApi)({
       url: this.config.url({
         path: "/chat/completions",
         modelId: this.modelId
       }),
-      headers: (0, import_provider_utils3.combineHeaders)(this.config.headers(), options.headers),
+      headers: (0, import_provider_utils4.combineHeaders)(this.config.headers(), options.headers),
       body: __spreadProps(__spreadValues({}, args), {
         stream: true
       }),
       failedResponseHandler: sarvamFailedResponseHandler,
-      successfulResponseHandler: (0, import_provider_utils3.createEventSourceResponseHandler)(sarvamChatChunkSchema),
+      successfulResponseHandler: (0, import_provider_utils4.createEventSourceResponseHandler)(
+        sarvamChatChunkSchema
+      ),
       abortSignal: options.abortSignal,
       fetch: this.config.fetch
     });
@@ -490,9 +573,7 @@ var SarvamChatLanguageModel = class {
             }
             const choice = value.choices[0];
             if ((choice == null ? void 0 : choice.finish_reason) != null) {
-              finishReason = mapSarvamFinishReason(
-                choice.finish_reason
-              );
+              finishReason = mapSarvamFinishReason(choice.finish_reason);
             }
             if ((choice == null ? void 0 : choice.delta) == null) {
               return;
@@ -552,13 +633,11 @@ var SarvamChatLanguageModel = class {
                         argsTextDelta: toolCall2.function.arguments
                       });
                     }
-                    if ((0, import_provider_utils3.isParsableJson)(
-                      toolCall2.function.arguments
-                    )) {
+                    if ((0, import_provider_utils4.isParsableJson)(toolCall2.function.arguments)) {
                       controller.enqueue({
                         type: "tool-call",
                         toolCallType: "function",
-                        toolCallId: (_h = toolCall2.id) != null ? _h : (0, import_provider_utils3.generateId)(),
+                        toolCallId: (_h = toolCall2.id) != null ? _h : (0, import_provider_utils4.generateId)(),
                         toolName: toolCall2.function.name,
                         args: toolCall2.function.arguments
                       });
@@ -581,11 +660,11 @@ var SarvamChatLanguageModel = class {
                   toolName: toolCall.function.name,
                   argsTextDelta: (_l = toolCallDelta.function.arguments) != null ? _l : ""
                 });
-                if (((_m = toolCall.function) == null ? void 0 : _m.name) != null && ((_n = toolCall.function) == null ? void 0 : _n.arguments) != null && (0, import_provider_utils3.isParsableJson)(toolCall.function.arguments)) {
+                if (((_m = toolCall.function) == null ? void 0 : _m.name) != null && ((_n = toolCall.function) == null ? void 0 : _n.arguments) != null && (0, import_provider_utils4.isParsableJson)(toolCall.function.arguments)) {
                   controller.enqueue({
                     type: "tool-call",
                     toolCallType: "function",
-                    toolCallId: (_o = toolCall.id) != null ? _o : (0, import_provider_utils3.generateId)(),
+                    toolCallId: (_o = toolCall.id) != null ? _o : (0, import_provider_utils4.generateId)(),
                     toolName: toolCall.function.name,
                     args: toolCall.function.arguments
                   });
@@ -680,7 +759,7 @@ var sarvamChatChunkSchema = import_zod2.z.union([
 ]);
 
 // src/sarvam-transcription-model.ts
-var import_provider_utils4 = require("@ai-sdk/provider-utils");
+var import_provider_utils5 = require("@ai-sdk/provider-utils");
 var import_zod4 = require("zod");
 
 // src/sarvam-transcription-settings.ts
@@ -722,7 +801,7 @@ var SarvamTranscriptionModel = class {
       throw new Error(
         "Language code unknown is not supported for model saarika:v1"
       );
-    const sarvamOptions = (0, import_provider_utils4.parseProviderOptions)({
+    const sarvamOptions = (0, import_provider_utils5.parseProviderOptions)({
       provider: "sarvam",
       providerOptions: {
         sarvam: __spreadValues(__spreadValues({}, providerOptions == null ? void 0 : providerOptions.sarvam), this.config.transcription)
@@ -763,15 +842,15 @@ var SarvamTranscriptionModel = class {
       value: response,
       responseHeaders,
       rawValue: rawResponse
-    } = await (0, import_provider_utils4.postFormDataToApi)({
+    } = await (0, import_provider_utils5.postFormDataToApi)({
       url: this.config.url({
         path: "/speech-to-text",
         modelId: this.modelId
       }),
-      headers: (0, import_provider_utils4.combineHeaders)(this.config.headers(), options.headers),
+      headers: (0, import_provider_utils5.combineHeaders)(this.config.headers(), options.headers),
       formData,
       failedResponseHandler: sarvamFailedResponseHandler,
-      successfulResponseHandler: (0, import_provider_utils4.createJsonResponseHandler)(
+      successfulResponseHandler: (0, import_provider_utils5.createJsonResponseHandler)(
         sarvamTranscriptionResponseSchema
       ),
       abortSignal: options.abortSignal,
@@ -818,7 +897,7 @@ var sarvamTranscriptionResponseSchema = import_zod4.z.object({
 });
 
 // src/sarvam-speech-model.ts
-var import_provider_utils5 = require("@ai-sdk/provider-utils");
+var import_provider_utils6 = require("@ai-sdk/provider-utils");
 
 // src/sarvam-speech-settings.ts
 var import_zod5 = require("zod");
@@ -878,7 +957,7 @@ var SarvamSpeechModel = class {
     providerOptions
   }) {
     const warnings = [];
-    const sarvamOptions = (0, import_provider_utils5.parseProviderOptions)({
+    const sarvamOptions = (0, import_provider_utils6.parseProviderOptions)({
       provider: "sarvam",
       providerOptions: {
         sarvam: __spreadValues(__spreadValues({}, providerOptions == null ? void 0 : providerOptions.sarvam), this.config.speech)
@@ -942,15 +1021,15 @@ var SarvamSpeechModel = class {
       value,
       responseHeaders,
       rawValue: rawResponse
-    } = await (0, import_provider_utils5.postJsonToApi)({
+    } = await (0, import_provider_utils6.postJsonToApi)({
       url: this.config.url({
         path: "/text-to-speech",
         modelId: this.modelId
       }),
-      headers: (0, import_provider_utils5.combineHeaders)(this.config.headers(), options.headers),
+      headers: (0, import_provider_utils6.combineHeaders)(this.config.headers(), options.headers),
       body: requestBody,
       failedResponseHandler: sarvamFailedResponseHandler,
-      successfulResponseHandler: (0, import_provider_utils5.createJsonResponseHandler)(
+      successfulResponseHandler: (0, import_provider_utils6.createJsonResponseHandler)(
         import_zod6.z.object({
           request_id: import_zod6.z.string(),
           audios: import_zod6.z.array(import_zod6.z.string())
@@ -979,8 +1058,8 @@ var SarvamSpeechModel = class {
 // src/sarvam-provider.ts
 function createSarvam(options = {}) {
   var _a;
-  const baseURL = (_a = (0, import_provider_utils6.withoutTrailingSlash)(options.baseURL)) != null ? _a : "https://api.sarvam.ai";
-  const ApiKey = (0, import_provider_utils6.loadApiKey)({
+  const baseURL = (_a = (0, import_provider_utils7.withoutTrailingSlash)(options.baseURL)) != null ? _a : "https://api.sarvam.ai";
+  const ApiKey = (0, import_provider_utils7.loadApiKey)({
     apiKey: options.apiKey,
     environmentVariableName: "SARVAM_API_KEY",
     description: "Sarvam"