Add Jan provider support (#932)

* jan provier info

* refactor: update logging and variable naming 🛠️

docs/src/content/docs/getting-started/configuration.mdx

@@ -1184,6 +1184,10 @@ script({
 
 Follow [this guide](https://huggingface.co/blog/yagilb/lms-hf) to load Hugging Face models into LMStudio.
 
+## Jan
+
+The `jan` provider connects to the [Jan](https://jan.ai/) local server.
+
 ## LocalAI
 
 [LocalAI](https://localai.io/) act as a drop-in replacement REST API that’s compatible
@@ -1233,9 +1237,8 @@ that allows you to run an LLM locally.
 
 The provider is `llamafile` and the model name is ignored.
 
-## Jan, LLaMA.cpp
+## LLaMA.cpp
 
-[Jan](https://jan.ai/),
 [LLaMA.cpp](https://github.com/ggerganov/llama.cpp/tree/master/examples/server)
 also allow running models locally or interfacing with other LLM vendors.
 

packages/cli/src/nodehost.ts

@@ -96,13 +96,16 @@ class ModelManager implements ModelService {
     ): Promise<ResponseStatus> {
         const { trace } = options || {}
         const { provider, model } = parseModelIdentifier(modelid)
-        if (provider === MODEL_PROVIDER_OLLAMA) {
-            if (this.pulled.includes(modelid)) return { ok: true }
+        if (this.pulled.includes(modelid)) return { ok: true }
 
-            if (!isQuiet) logVerbose(`ollama pull ${model}`)
+        if (provider === MODEL_PROVIDER_OLLAMA) {
+            logVerbose(`${provider}: pull ${model}`)
             try {
                 const conn = await this.getModelToken(modelid)
-                const res = await fetch(`${conn.base}/api/pull`, {
+
+                let res: Response
+                // OLLAMA
+                res = await fetch(`${conn.base}/api/pull`, {
                     method: "POST",
                     headers: {
                         "User-Agent": TOOL_ID,
@@ -115,13 +118,14 @@ class ModelManager implements ModelService {
                     ),
                 })
                 if (res.ok) {
-                    const resp = await res.json()
+                    const resj = await res.json()
+                    //logVerbose(JSON.stringify(resj, null, 2))
                 }
                 if (res.ok) this.pulled.push(modelid)
                 return { ok: res.ok, status: res.status }
             } catch (e) {
-                logError(`failed to pull model ${model}`)
-                trace?.error("pull model failed", e)
+                logError(`${provider}: failed to pull model ${model}`)
+                trace?.error(`${provider}: pull model failed`, e)
                 return { ok: false, status: 500, error: serializeError(e) }
             }
         }

packages/core/src/connection.ts

@@ -32,8 +32,9 @@ import {
     MISTRAL_API_BASE,
     MODEL_PROVIDER_LMSTUDIO,
     LMSTUDIO_API_BASE,
+    MODEL_PROVIDER_JAN,
+    JAN_API_BASE,
 } from "./constants"
-import { fileExists, readText, writeText } from "./fs"
 import {
     OpenAIAPIType,
     host,
@@ -481,6 +482,19 @@ export async function parseTokenFromEnv(
         }
     }
 
+    if (provider === MODEL_PROVIDER_JAN) {
+        const base = findEnvVar(env, "JAN", BASE_SUFFIX)?.value || JAN_API_BASE
+        if (!URL.canParse(base)) throw new Error(`${base} must be a valid URL`)
+        return {
+            provider,
+            model,
+            base,
+            token: "lmstudio",
+            type: "openai",
+            source: "env: JAN_API_...",
+        }
+    }
+
     if (provider === MODEL_PROVIDER_TRANSFORMERS) {
         return {
             provider,

packages/core/src/constants.ts

@@ -154,6 +154,7 @@ export const ALIBABA_BASE =
     "https://dashscope-intl.aliyuncs.com/compatible-mode/v1"
 export const MISTRAL_API_BASE = "https://api.mistral.ai/v1"
 export const LMSTUDIO_API_BASE = "http://localhost:1234/v1"
+export const JAN_API_BASE = "http://localhost:1337/v1"
 
 export const PROMPTFOO_CACHE_PATH = ".genaiscript/cache/tests"
 export const PROMPTFOO_CONFIG_DIR = ".genaiscript/config/tests"
@@ -184,6 +185,7 @@ export const MODEL_PROVIDER_TRANSFORMERS = "transformers"
 export const MODEL_PROVIDER_ALIBABA = "alibaba"
 export const MODEL_PROVIDER_MISTRAL = "mistral"
 export const MODEL_PROVIDER_LMSTUDIO = "lmstudio"
+export const MODEL_PROVIDER_JAN = "jan"
 
 export const TRACE_FILE_PREVIEW_MAX_LENGTH = 240
 
@@ -210,6 +212,8 @@ export const DOCS_CONFIGURATION_OLLAMA_URL =
     "https://microsoft.github.io/genaiscript/getting-started/configuration/#ollama"
 export const DOCS_CONFIGURATION_LMSTUDIO_URL =
     "https://microsoft.github.io/genaiscript/getting-started/configuration/#lmstudio"
+export const DOCS_CONFIGURATION_JAN_URL =
+    "https://microsoft.github.io/genaiscript/getting-started/configuration/#jan"
 export const DOCS_CONFIGURATION_LLAMAFILE_URL =
     "https://microsoft.github.io/genaiscript/getting-started/configuration/#llamafile"
 export const DOCS_CONFIGURATION_LITELLM_URL =
@@ -311,6 +315,11 @@ export const MODEL_PROVIDERS: readonly {
         detail: "LM Studio local server",
         url: DOCS_CONFIGURATION_LMSTUDIO_URL,
     },
+    {
+        id: MODEL_PROVIDER_LMSTUDIO,
+        detail: "Jan local server",
+        url: DOCS_CONFIGURATION_JAN_URL,
+    },
     {
         id: MODEL_PROVIDER_ALIBABA,
         detail: "Alibaba models",

packages/core/src/promptdom.ts

@@ -17,22 +17,12 @@ import { YAMLStringify } from "./yaml"
 import {
     DEFAULT_FENCE_FORMAT,
     MARKDOWN_PROMPT_FENCE,
-    MODEL_PROVIDER_ALIBABA,
-    MODEL_PROVIDER_ANTHROPIC,
-    MODEL_PROVIDER_AZURE_OPENAI,
-    MODEL_PROVIDER_AZURE_SERVERLESS_MODELS,
-    MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI,
-    MODEL_PROVIDER_LLAMAFILE,
-    MODEL_PROVIDER_LMSTUDIO,
-    MODEL_PROVIDER_OLLAMA,
-    MODEL_PROVIDER_OPENAI,
     PROMPT_FENCE,
     PROMPTY_REGEX,
     SANITIZED_PROMPT_INJECTION,
     TEMPLATE_ARG_DATA_SLICE_SAMPLE,
     TEMPLATE_ARG_FILE_MAX_TOKENS,
 } from "./constants"
-import { parseModelIdentifier } from "./models"
 import {
     appendAssistantMessage,
     appendSystemMessage,

packages/core/src/types/prompt_template.d.ts

@@ -150,6 +150,7 @@ type ModelType = OptionsOrString<
     | "anthropic:claude-2.1"
     | "anthropic:claude-instant-1.2"
     | "huggingface:microsoft/Phi-3-mini-4k-instruct"
+    | "jan:llama3.2-3b-instruct"
     | "google:gemini-1.5-flash"
     | "google:gemini-1.5-flash-latest"
     | "google:gemini-1.5-flash-8b"

packages/vscode/src/lmaccess.ts

@@ -1,141 +1,13 @@
 /* eslint-disable @typescript-eslint/naming-convention */
 import * as vscode from "vscode"
 import { ExtensionState } from "./state"
-import {
-    MODEL_PROVIDER_OLLAMA,
-    MODEL_PROVIDER_LLAMAFILE,
-    MODEL_PROVIDER_AICI,
-    MODEL_PROVIDER_AZURE_OPENAI,
-    MODEL_PROVIDER_LITELLM,
-    MODEL_PROVIDER_OPENAI,
-    MODEL_PROVIDER_CLIENT,
-    MODEL_PROVIDER_GITHUB,
-    TOOL_NAME,
-    MODEL_PROVIDER_AZURE_SERVERLESS_MODELS,
-    MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI,
-    DOCS_CONFIGURATION_URL,
-    MODEL_PROVIDER_GOOGLE,
-    MODEL_PROVIDER_ALIBABA,
-    MODEL_PROVIDER_LMSTUDIO,
-} from "../../core/src/constants"
-import { OpenAIAPIType } from "../../core/src/host"
-import { parseModelIdentifier } from "../../core/src/models"
 import { ChatCompletionMessageParam } from "../../core/src/chattypes"
 import { LanguageModelChatRequest } from "../../core/src/server/client"
 import { ChatStart } from "../../core/src/server/messages"
 import { serializeError } from "../../core/src/error"
 import { logVerbose } from "../../core/src/util"
 import { renderMessageContent } from "../../core/src/chatrender"
 
-async function generateLanguageModelConfiguration(
-    state: ExtensionState,
-    modelId: string
-) {
-    const { provider } = parseModelIdentifier(modelId)
-    const supportedProviders = [
-        MODEL_PROVIDER_OLLAMA,
-        MODEL_PROVIDER_LLAMAFILE,
-        MODEL_PROVIDER_AICI,
-        MODEL_PROVIDER_AZURE_OPENAI,
-        MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI,
-        MODEL_PROVIDER_AZURE_SERVERLESS_MODELS,
-        MODEL_PROVIDER_LITELLM,
-        MODEL_PROVIDER_LMSTUDIO,
-        MODEL_PROVIDER_GOOGLE,
-        MODEL_PROVIDER_ALIBABA,
-    ]
-    if (supportedProviders.includes(provider)) {
-        return { provider }
-    }
-
-    const languageChatModels = await state.languageChatModels()
-    if (Object.keys(languageChatModels).length)
-        return { provider: MODEL_PROVIDER_CLIENT, model: "*" }
-
-    const items: (vscode.QuickPickItem & {
-        model?: string
-        provider?: string
-        apiType?: OpenAIAPIType
-    })[] = []
-    if (isLanguageModelsAvailable()) {
-        const models = await vscode.lm.selectChatModels()
-        if (models.length)
-            items.push({
-                label: "Visual Studio Language Chat Models",
-                detail: `Use a registered LLM such as GitHub Copilot.`,
-                model: "*",
-                provider: MODEL_PROVIDER_CLIENT,
-            })
-    }
-    items.push(
-        {
-            label: "OpenAI",
-            detail: `Use a personal OpenAI subscription.`,
-            provider: MODEL_PROVIDER_OPENAI,
-        },
-        {
-            label: "Azure OpenAI",
-            detail: `Use a Azure-hosted OpenAI subscription.`,
-            provider: MODEL_PROVIDER_AZURE_OPENAI,
-            apiType: "azure",
-        },
-        {
-            label: "Azure AI OpenAI (serverless deployment)",
-            detail: `Use a Azure OpenAI serverless model deployment through Azure AI Studio.`,
-            provider: MODEL_PROVIDER_AZURE_SERVERLESS_OPENAI,
-            apiType: "azure_serverless",
-        },
-        {
-            label: "Azure AI Models (serverless deployment)",
-            detail: `Use a Azure serverless model deployment through Azure AI Studio.`,
-            provider: MODEL_PROVIDER_AZURE_SERVERLESS_MODELS,
-            apiType: "azure_serverless_models",
-        },
-        {
-            label: "GitHub Models",
-            detail: `Use a GitHub Models with a GitHub subscription.`,
-            provider: MODEL_PROVIDER_GITHUB,
-        },
-        {
-            label: "Alibaba Cloud",
-            detail: "Use Alibaba Cloud models.",
-            provider: MODEL_PROVIDER_ALIBABA,
-        },
-        {
-            label: "LocalAI",
-            description: "https://localai.io/",
-            detail: "Use local LLMs instead OpenAI. Requires LocalAI and Docker.",
-            provider: MODEL_PROVIDER_OPENAI,
-            apiType: "localai",
-        },
-        {
-            label: "Ollama",
-            description: "https://ollama.com/",
-            detail: "Run a open source LLMs locally. Requires Ollama",
-            provider: MODEL_PROVIDER_OLLAMA,
-        },
-        {
-            label: "AICI",
-            description: "http://github.com/microsoft/aici",
-            detail: "Generate AICI javascript prompts.",
-            provider: MODEL_PROVIDER_AICI,
-        }
-    )
-
-    const res: { model?: string; provider?: string; apiType?: OpenAIAPIType } =
-        await vscode.window.showQuickPick<
-            vscode.QuickPickItem & {
-                model?: string
-                provider?: string
-                apiType?: OpenAIAPIType
-            }
-        >(items, {
-            title: `Configure a Language Model for ${modelId}`,
-        })
-
-    return res
-}
-
 async function pickChatModel(
     state: ExtensionState,
     model: string
@@ -165,26 +37,6 @@ async function pickChatModel(
     return chatModel
 }
 
-export async function pickLanguageModel(
-    state: ExtensionState,
-    modelId: string
-) {
-    const res = await generateLanguageModelConfiguration(state, modelId)
-    if (res === undefined) return undefined
-
-    if (res.model) return res.model
-    else {
-        const configure = "Configure..."
-        vscode.window.showWarningMessage(
-            `${TOOL_NAME} - model connection not configured.`,
-            configure
-        )
-        if (configure)
-            vscode.env.openExternal(vscode.Uri.parse(DOCS_CONFIGURATION_URL))
-        return undefined
-    }
-}
-
 export function isLanguageModelsAvailable() {
     return (
         typeof vscode.lm !== "undefined" &&