@promptbook/cli 0.112.0-99 → 0.113.0-0

package/src/llm-providers/ollama/playground/playground.ts

@@ -1,120 +0,0 @@
-#!/usr/bin/env ts-node
-
-import * as dotenv from 'dotenv';
-
-dotenv.config({ path: '.env' });
-
-import colors from 'colors'; // <- TODO: [🔶] Make system to put color and style to both node and browser
-import { embeddingVectorToString } from '../../../execution/embeddingVectorToString';
-import { usageToHuman } from '../../../execution/utils/usageToHuman';
-import type { Prompt } from '../../../types/Prompt';
-import { keepUnused } from '../../../utils/organization/keepUnused';
-// import { OllamaAssistantExecutionTools } from '../OllamaAssistantExecutionTools';
-import type { Usage } from '../../../execution/Usage';
-import { countUsage } from '../../_common/utils/count-total-usage/countUsage';
-import { createOllamaExecutionTools } from '../createOllamaExecutionTools';
-
-playground()
-    .catch((error) => {
-        console.error(colors.bgRed(error.name || 'NamelessError'));
-        console.error(error);
-        process.exit(1);
-    })
-    .then(() => {
-        process.exit(0);
-    });
-
-/**
- * Handles playground.
- */
-async function playground() {
-    console.info(`🧸  Ollama Playground`);
-
-    // Do here stuff you want to test
-    //========================================>
-
-    const ollamaExecutionTools = createOllamaExecutionTools({
-        isVerbose: true,
-    });
-
-    const ollamaExecutionToolsWithUsage = countUsage(ollamaExecutionTools);
-
-    ollamaExecutionToolsWithUsage.spending().subscribe((usage: Usage) => {
-        const wordCount = (usage?.input?.wordsCount?.value || 0) + (usage?.output?.wordsCount?.value || 0);
-        console.log(`[💸] Spending ${wordCount} words`);
-    });
-
-    keepUnused(ollamaExecutionTools);
-    keepUnused(embeddingVectorToString);
-    keepUnused(usageToHuman);
-    keepUnused<Prompt>();
-
-    /*/
-    const models = await ollamaExecutionTools.listModels();
-    console.info({ models });
-    /**/
-
-    /*/
-    const completionPrompt = {
-        title: 'Hello',
-        parameters: {},
-        content: `Hello, my name is Alice.`,
-        modelRequirements: {
-            modelVariant: 'COMPLETION',
-        },
-    } as const satisfies Prompt;
-    const completionPromptResult = await ollamaExecutionTools.callCompletionModel(completionPrompt);
-    console.info({ completionPromptResult });
-    console.info(colors.cyan(usageToHuman(chatPromptResult.usage)));
-    console.info(colors.green(completionPrompt.content + completionPromptResult.content));
-    /**/
-
-    /**/
-    const chatPrompt = {
-        title: 'Promptbook speech',
-        parameters: {},
-        content: `Hello`,
-        modelRequirements: {
-            modelVariant: 'CHAT',
-            // modelName: 'mistral',
-            systemMessage: 'You are an assistant who only speaks in rhymes.',
-            temperature: 1.5,
-        },
-    } as const satisfies Prompt;
-    const chatPromptResult = await ollamaExecutionToolsWithUsage.callChatModel!(chatPrompt);
-    console.info({ chatPromptResult });
-    console.info(colors.cyan(usageToHuman(chatPromptResult.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(chatPrompt.content));
-    console.info(colors.bgGreen(' Chat: ') + colors.green(chatPromptResult.content));
-    /**/
-
-    /*/
-    // TODO: Test Translations in playground
-    /**/
-
-    /*/
-    const prompt = {
-        title: 'Hello',
-        parameters: {},
-        content: `Hello, my name is Alice.`,
-        modelRequirements: {
-            modelVariant: 'EMBEDDING',
-            // modelName: 'text-embedding-ada-002',
-        },
-    } as const satisfies Prompt;
-    const promptResult = await ollamaExecutionTools.callEmbeddingModel(prompt);
-    console.info({ promptResult });
-    console.info(colors.cyan(usageToHuman(chatPromptResult.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(prompt.content));
-    console.info(colors.bgGreen(' Embedding: ') + colors.green(embeddingVectorToString(promptResult.content)));
-    /**/
-
-    /*/
-    // <- Note: [🤖] Test here new model variant if needed
-    /**/
-
-    //========================================/
-}
-
-// Note: [⚫] Code for playground [ollama playground](src/llm-providers/ollama/playground/playground.ts) should never be published in any package
-// TODO: [main] !!3 Test here that `systemMessage`, `temperature` and `seed` are working correctly

package/src/llm-providers/ollama/playground/tsconfig.json

@@ -1,19 +0,0 @@
-{
-    "compilerOptions": {
-        "target": "esnext",
-        "module": "commonjs",
-        "downlevelIteration": true,
-        "allowJs": true,
-        "moduleResolution": "node",
-        "forceConsistentCasingInFileNames": true,
-        "noImplicitReturns": true,
-        "noImplicitThis": true,
-        "noImplicitAny": true,
-        "strictNullChecks": true,
-        "experimentalDecorators": true,
-        "noUnusedLocals": false,
-        "resolveJsonModule": true,
-        "esModuleInterop": true
-    },
-    "exclude": ["node_modules"]
-}

package/src/llm-providers/openai/playground/playground.ts

@@ -1,406 +0,0 @@
-#!/usr/bin/env ts-node
-
-import * as dotenv from 'dotenv';
-
-dotenv.config({ path: '.env' });
-
-import colors from 'colors'; // <- TODO: [🔶] Make system to put color and style to both node and browser
-import { embeddingVectorToString } from '../../../execution/embeddingVectorToString';
-import { usageToHuman } from '../../../execution/utils/usageToHuman';
-import { JavascriptEvalExecutionTools } from '../../../scripting/javascript/JavascriptEvalExecutionTools';
-import type { ChatPrompt, Prompt } from '../../../types/Prompt';
-import { keepUnused } from '../../../utils/organization/keepUnused';
-// import { OpenAiAssistantExecutionTools } from '../OpenAiAssistantExecutionTools';
-import { join } from 'path';
-import { DEFAULT_EXECUTION_CACHE_DIRNAME } from '../../../config';
-import type { Usage } from '../../../execution/Usage';
-import { $provideFilesystemForNode } from '../../../scrapers/_common/register/$provideFilesystemForNode';
-import { FileCacheStorage } from '../../../storage/file-cache-storage/FileCacheStorage';
-import { cacheLlmTools } from '../../_common/utils/cache/cacheLlmTools';
-import { countUsage } from '../../_common/utils/count-total-usage/countUsage';
-import { OpenAiAgentKitExecutionTools } from '../OpenAiAgentKitExecutionTools';
-import { OpenAiExecutionTools } from '../OpenAiExecutionTools';
-
-playground()
-    .catch((error) => {
-        console.error(colors.bgRed(error.name || 'NamelessError'));
-        console.error(error);
-        process.exit(1);
-    })
-    .then(() => {
-        process.exit(0);
-    });
-
-/**
- * Handles playground.
- */
-async function playground() {
-    console.info(`🧸  OpenAI Playground`);
-
-    // Do here stuff you want to test
-    //========================================>
-
-    const script = new JavascriptEvalExecutionTools({
-        functions: {
-            // TODO: !!!! Normalize function names:
-            get_coupon_code: async (params: { productCategory: string; discountPercentage: number }) => {
-                // Simulate generating a coupon code based on product category and discount percentage
-                const { productCategory, discountPercentage } = params;
-                const code = `SAVE${discountPercentage}${productCategory.toUpperCase().slice(0, 3)}2025`;
-
-                console.log(
-                    `!!!! [🛠️] Generated coupon code: ${code} for category: ${productCategory} with discount: ${discountPercentage}%`,
-                );
-
-                return code;
-            },
-        },
-    });
-
-    let openAiExecutionTools = new OpenAiExecutionTools(
-        //            <- TODO: [🧱] Implement in a functional (not new Class) way
-        {
-            isVerbose: true,
-            userId: 'playground',
-            apiKey: process.env.OPENAI_API_KEY!,
-            executionTools: { script },
-        },
-    );
-
-    openAiExecutionTools = cacheLlmTools(openAiExecutionTools, {
-        storage: new FileCacheStorage(
-            { fs: $provideFilesystemForNode() },
-            {
-                rootFolderPath: join(
-                    process.cwd(),
-                    DEFAULT_EXECUTION_CACHE_DIRNAME,
-                    // <- TODO: [🦒] Allow to override (pass different value into the function)
-                ),
-            },
-        ),
-        // isCacheReloaded: isCacheReloaded,
-    });
-
-    const openAiExecutionToolsWithUsage = countUsage(openAiExecutionTools);
-
-    openAiExecutionToolsWithUsage.spending().subscribe((usage: Usage) => {
-        const wordCount = (usage?.input?.wordsCount?.value || 0) + (usage?.output?.wordsCount?.value || 0);
-        console.log(`[💸] Spending ${wordCount} words`);
-    });
-
-    /*/
-    const openAiAssistantExecutionTools = new OpenAiAssistantExecutionTools(
-        //            <- TODO: [🧱] Implement in a functional (not new Class) way
-        {
-            isVerbose: true,
-            userId: 'playground',
-            apiKey: process.env.OPENAI_API_KEY!,
-            assistantId: 'asst_CJCZzFCbBL0f2D4OWMXVTdBB',
-            //            <- Note: This is not a private information, just ID of the assistant which is accessible only with correct API key
-        },
-    );
-    /**/
-
-    /*/
-    const openAiAgentKitExecutionTools = new OpenAiAgentKitExecutionTools({
-        isVerbose: true,
-        userId: 'playground',
-        apiKey: process.env.OPENAI_API_KEY!,
-        executionTools: { script },
-    });
-
-    const agentKitPrompt = {
-        title: 'AgentKit Coupon Demo',
-        parameters: {},
-        content: `Generate a coupon code for electronics with 15% off using the tool.`,
-        tools: [
-            {
-                name: 'get_coupon_code',
-                description: 'Generate a coupon code based on product category and discount percentage',
-                parameters: {
-                    type: 'object',
-                    properties: {
-                        productCategory: {
-                            type: 'string',
-                            description: 'The category of the product (e.g., electronics, clothing)',
-                        },
-                        discountPercentage: {
-                            type: 'number',
-                            description: 'The discount percentage to apply',
-                        },
-                    },
-                    required: ['productCategory', 'discountPercentage'],
-                },
-            },
-        ],
-        modelRequirements: {
-            modelVariant: 'CHAT',
-            systemMessage:
-                'Use the get_coupon_code tool to generate the coupon. Do not invent coupon codes.',
-        },
-    } satisfies ChatPrompt;
-    const agentKitResult = await openAiAgentKitExecutionTools.callChatModel!(agentKitPrompt);
-    console.info({ agentKitResult });
-    console.info(colors.cyan(usageToHuman(agentKitResult.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(agentKitPrompt.content));
-    console.info(colors.bgGreen(' AgentKit: ') + colors.green(agentKitResult.content));
-    /**/
-
-    keepUnused(openAiExecutionTools);
-    keepUnused(OpenAiAgentKitExecutionTools);
-    // keepUnused(openAiAssistantExecutionTools);
-    keepUnused(embeddingVectorToString);
-    keepUnused(usageToHuman);
-    keepUnused<Prompt>();
-
-    /*/
-    const models = await openAiExecutionTools.listModels();
-    console.info({ models });
-    /**/
-
-    /*/
-    const imagePrompt = {
-        title: 'Hello',
-        parameters: {},
-        content: `Coffee cup floating in space, surrealistic, vibrant colors, high detail`,
-        modelRequirements: {
-            modelVariant: 'IMAGE_GENERATION',
-        },
-    } as const satisfies Prompt;
-    const imagePromptResult = await openAiExecutionTools.callImageGenerationModel(imagePrompt);
-    console.info({ imagePromptResult });
-    console.info(colors.cyan(usageToHuman(imagePromptResult.usage)));
-    console.info(colors.green(imagePromptResult.content));
-    /**/
-
-    /*/
-    const completionPrompt = {
-        title: 'Hello',
-        parameters: {},
-        content: `Hello, my name is Alice.`,
-        modelRequirements: {
-            modelVariant: 'COMPLETION',
-        },
-    } as const satisfies Prompt;
-    const completionPromptResult = await openAiExecutionTools.callCompletionModel(completionPrompt);
-    console.info({ completionPromptResult });
-    console.info(colors.cyan(usageToHuman(chatPromptResult.usage)));
-    console.info(colors.green(completionPrompt.content + completionPromptResult.content));
-    /**/
-
-    /**/
-    const chatPrompt = {
-        title: 'Prague',
-        parameters: {},
-        // thread: [
-        //     // <- TODO: !!! Maybe rename to `previousMessages`
-        //     {
-        //         sender: 'user', // <- [👥] TODO: Standardize to `role: 'USER' | 'ASSISTANT'
-        //         content: 'Where is Prague ',
-        //     },
-        //     {
-        //         sender: 'assistant',
-        //         content: 'Prague is a beautiful city located in the Czech Republic.',
-        //     },
-        // ],
-        content: `Give me a coupons for grocery shopping in electronics store with a 20% discount.`,
-
-        tools: [
-            {
-                name: 'get_coupon_code',
-
-                description: 'Generate a coupon code based on product category and discount percentage',
-
-                parameters: {
-                    type: 'object',
-                    properties: {
-                        productCategory: {
-                            type: 'string',
-                            description: 'The category of the product (e.g., electronics, clothing)',
-                        },
-                        discountPercentage: {
-                            type: 'number',
-                            description: 'The discount percentage to apply',
-                        },
-                    },
-                    required: ['productCategory', 'discountPercentage'],
-                },
-            },
-        ],
-
-        modelRequirements: {
-            modelVariant: 'CHAT',
-            systemMessage:
-                'You are an helpful assistant who can provide coupon codes, via the get_coupon_code tool. Do not make up coupon codes on your own. Do not provide more that 25% discount and more than 5 coupons per message.',
-            // modelName: 'gpt-3.5-turbo',
-            modelName: 'gpt-5',
-            temperature: 1.5,
-        },
-    } /* as const */ satisfies ChatPrompt; /* <- Note: [🤛] */
-    const chatPromptResult = await openAiExecutionToolsWithUsage.callChatModel!(chatPrompt);
-    console.info({ chatPromptResult });
-    console.info(colors.cyan(usageToHuman(chatPromptResult.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(chatPrompt.content));
-    console.info(colors.bgGreen(' Chat: ') + colors.green(chatPromptResult.content));
-    /**/
-
-    /**/
-    // TODO: !!!!! Test
-    const chatPromptWithFiles = {
-        title: 'Chat with files',
-        parameters: {},
-        content: `What is in these images?`,
-
-        files: [
-            // Note: In node.js we need to mock the File object or use a real one if available
-            // For playground purposes, we can use a small transparent pixel
-            new File(
-                // cspell:disable
-                [Buffer.from('R0lGODlhAQABAIAAAAAAAP///yH5BAEAAAAALAAAAAABAAEAAAIBRAA7', 'base64')],
-                // cspell:enable
-                'pixel.gif',
-                { type: 'image/gif' },
-            ),
-        ],
-        modelRequirements: {
-            modelVariant: 'CHAT',
-            modelName: 'gpt-4o',
-        },
-    } satisfies ChatPrompt;
-    const chatPromptWithFilesResult = await openAiExecutionToolsWithUsage.callChatModel!(chatPromptWithFiles);
-    console.info({ chatPromptWithFilesResult });
-    console.info(colors.bgBlue(' User: ') + colors.blue(chatPromptWithFiles.content));
-    console.info(colors.bgGreen(' Chat: ') + colors.green(chatPromptWithFilesResult.content));
-    /**/
-
-    /*/
-    const chatPromptResult2 = await openAiExecutionToolsWithUsage.callChatModel!(chatPrompt);
-    console.info({ chatPromptResult2 });
-    console.info(colors.cyan(usageToHuman(chatPromptResult2.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(chatPrompt.content));
-    console.info(colors.bgGreen(' Chat: ') + colors.green(chatPromptResult2.content));
-    /**/
-
-    /*/
-    // TODO: Test Translations in playground
-    /**/
-
-    /*/
-    const prompt = {
-        title: 'Hello',
-        parameters: {},
-        content: `Hello, my name is Alice.`,
-        modelRequirements: {
-            modelVariant: 'EMBEDDING',
-            // modelName: 'text-embedding-ada-002',
-        },
-    } as const satisfies Prompt;
-    const promptResult = await openAiExecutionTools.callEmbeddingModel(prompt);
-    console.info({ promptResult });
-    console.info(colors.cyan(usageToHuman(chatPromptResult.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(prompt.content));
-    console.info(colors.bgGreen(' Embedding: ') + colors.green(embeddingVectorToString(promptResult.content)));
-    /**/
-
-    /*/
-    const chatPrompt = {
-        title: 'Promptbook speech',
-        parameters: {},
-        content: `Write me speech about Promptbook and how it can help me to build the most beautiful chatbot and change the world`,
-        modelRequirements: {
-            modelVariant: 'CHAT',
-            // TODO: [👨‍👨‍👧‍👧] systemMessage: 'You are an assistant who only speaks in rhymes.',
-            // TODO: [👨‍👨‍👧‍👧] temperature: 1.5,
-        },
-
-        /*
-        [🗯]
-        replyingTo: {
-
-        }
-        * /
-    } as const satisfies Prompt;
-    const chatPromptResult = await openAiAssistantExecutionTools.callChatModel(chatPrompt);
-    console.info({ chatPromptResult });
-    console.info(colors.cyan(usageToHuman(chatPromptResult.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(chatPrompt.content));
-    console.info(colors.bgGreen(' Assistant: ') + colors.green(chatPromptResult.content));
-    /**/
-
-    /*/
-    const openai = await openAiExecutionTools.getClient();
-    const stream = openai.beta.threads.createAndRunStream({
-        stream: true,
-        assistant_id: 'asst_CJCZzFCbBL0f2D4OWMXVTdBB',
-        //             <- Note: This is not a private information, just ID of the assistant which is accessible only with correct API key
-        thread: {
-            messages: [{ role: 'user', content: 'What is the meaning of life? I want breathtaking speech.' }],
-        },
-    });
-
-    console.log(stream);
-
-    stream.on('connect', () => {
-        console.log('connect', stream.currentEvent);
-    });
-
-    stream.on('messageDelta', (messageDelta) => {
-        console.log('messageDelta', (messageDelta as any).content[0].text);
-    });
-
-    stream.on('messageCreated', (message) => {
-        console.log('messageCreated', message);
-    });
-
-    stream.on('messageDone', (message) => {
-        console.log('messageDone', message);
-    });
-
-    const finalMessages = await stream.finalMessages();
-    console.log('finalMessages', finalMessages, finalMessages[0]!.content[0]!);
-
-    /**/
-
-    /*/
-    const chatWithToolsPrompt = {
-        title: 'Chat with tools',
-        parameters: {},
-        content: `What is the current weather in Prague?`,
-        modelRequirements: {
-            modelVariant: 'CHAT',
-            modelName: 'gpt-4o',
-            tools: [
-                {
-                    name: 'get_current_weather',
-                    description: 'Get the current weather in a given location',
-                    parameters: {
-                        type: 'object',
-                        properties: {
-                            location: {
-                                type: 'string',
-                                description: 'The city and state, e.g. San Francisco, CA',
-                            },
-                            unit: { type: 'string', enum: ['celsius', 'fahrenheit'] },
-                        },
-                        required: ['location'],
-                    },
-                },
-            ],
-        },
-    } as const satisfies Prompt;
-    const chatWithToolsResult = await openAiExecutionTools.callChatModel(chatWithToolsPrompt);
-    console.info({ chatWithToolsResult });
-    console.info(colors.cyan(usageToHuman(chatWithToolsResult.usage)));
-    console.info(colors.bgBlue(' User: ') + colors.blue(chatWithToolsPrompt.content));
-    console.info(colors.bgGreen(' Chat: ') + colors.green(chatWithToolsResult.content));
-    /**/
-
-    /*/
-    // <- Note: [🤖] Test here new model variant if needed
-    /**/
-
-    //========================================/
-}
-
-// Note: [⚫] Code for playground [openai playground](src/llm-providers/openai/playground/playground.ts) should never be published in any package
-// TODO: [main] !!3 Test here that `systemMessage`, `temperature` and `seed` are working correctly

package/src/llm-providers/openai/playground/tsconfig.json

@@ -1,19 +0,0 @@
-{
-    "compilerOptions": {
-        "target": "esnext",
-        "module": "commonjs",
-        "downlevelIteration": true,
-        "allowJs": true,
-        "moduleResolution": "node",
-        "forceConsistentCasingInFileNames": true,
-        "noImplicitReturns": true,
-        "noImplicitThis": true,
-        "noImplicitAny": true,
-        "strictNullChecks": true,
-        "experimentalDecorators": true,
-        "noUnusedLocals": false,
-        "resolveJsonModule": true,
-        "esModuleInterop": true
-    },
-    "exclude": ["node_modules"]
-}