@promptbook/openai 0.70.0-1 → 0.72.0-0

package/README.md

@@ -202,15 +202,21 @@ See the other models available in the Promptbook package:
 
 Rest of the documentation is common for **entire promptbook ecosystem**:
 
+# ✨ New Features
+
+-   ✨ **Support [OpenAI o1 model](https://openai.com/o1/)**
+
 ## 🤍 The Promptbook Whitepaper
 
+
+
 If you have a simple, single prompt for ChatGPT, GPT-4, Anthropic Claude, Google Gemini, Llama 2, or whatever, it doesn't matter how you integrate it. Whether it's calling a REST API directly, using the SDK, hardcoding the prompt into the source code, or importing a text file, the process remains the same.
 
 But often you will struggle with the limitations of LLMs, such as hallucinations, off-topic responses, poor quality output, language drift, word repetition repetition repetition repetition or misuse, lack of context, or just plain w𝒆𝐢rd responses. When this happens, you generally have three options:
 
 1. **Fine-tune** the model to your specifications or even train your own.
 2. **Prompt-engineer** the prompt to the best shape you can achieve.
-3. Use **multiple prompts** in a [pipeline](https://github.com/webgptorg/promptbook/discussions/64) to get the best result.
+3. Orchestrate **multiple prompts** in a [pipeline](https://github.com/webgptorg/promptbook/discussions/64) to get the best result.
 
 In all of these situations, but especially in 3., the Promptbook library can make your life easier.
 
@@ -222,7 +228,9 @@ In all of these situations, but especially in 3., the Promptbook library can mak
 -   Promptbook has built in versioning. You can test multiple **A/B versions** of pipelines and see which one works best.
 -   Promptbook is designed to do [**RAG** (Retrieval-Augmented Generation)](https://github.com/webgptorg/promptbook/discussions/41) and other advanced techniques. You can use **knowledge** to improve the quality of the output.
 
-## 🧔 Promptbook _(for prompt-engeneers)_
+
+
+## 🧔 Pipeline _(for prompt-engeneers)_
 
 **P**romp**t** **b**oo**k** markdown file (or `.ptbk.md` file) is document that describes a **pipeline** - a series of prompts that are chained together to form somewhat reciepe for transforming natural language input.
 
@@ -539,6 +547,8 @@ The following glossary is used to clarify certain concepts:
 -   When you want to **version** your prompts and **test multiple versions**
 -   When you want to **log** the execution of prompts and backtrace the issues
 
+[See more](https://github.com/webgptorg/promptbook/discussions/111)
+
 ### ➖ When not to use
 
 -   When you have already implemented single simple prompt and it works fine for your job
@@ -548,6 +558,8 @@ The following glossary is used to clarify certain concepts:
 -   When your main focus is on something other than text - like images, audio, video, spreadsheets _(other media types may be added in the future, [see discussion](https://github.com/webgptorg/promptbook/discussions/103))_
 -   When you need to use recursion _([see the discussion](https://github.com/webgptorg/promptbook/discussions/38))_
 
+[See more](https://github.com/webgptorg/promptbook/discussions/112)
+
 ## 🐜 Known issues
 
 -   [🤸‍♂️ Iterations not working yet](https://github.com/webgptorg/promptbook/discussions/55)
@@ -560,63 +572,15 @@ The following glossary is used to clarify certain concepts:
 
 ## ❔ FAQ
 
-
-
 If you have a question [start a discussion](https://github.com/webgptorg/promptbook/discussions/), [open an issue](https://github.com/webgptorg/promptbook/issues) or [write me an email](https://www.pavolhejny.com/contact).
 
-### Why not just use the OpenAI SDK / Anthropic Claude SDK / ...?
-
-Different levels of abstraction. OpenAI library is for direct use of OpenAI API. This library is for a higher level of abstraction. It define pipelines that are independent of the underlying library, LLM model, or even LLM provider.
-
-### How is it different from the Langchain library?
-
-Langchain is primarily aimed at ML developers working in Python. This library is for developers working in javascript/typescript and creating applications for end users.
-
-We are considering creating a bridge/converter between these two libraries.
-
-
-
-### Promptbooks vs. OpenAI`s GPTs
-
-GPTs are chat assistants that can be assigned to specific tasks and materials. But they are still chat assistants. Promptbooks are a way to orchestrate many more predefined tasks to have much tighter control over the process. Promptbooks are not a good technology for creating human-like chatbots, GPTs are not a good technology for creating outputs with specific requirements.
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-### Where should I store my promptbooks?
-
-If you use raw SDKs, you just put prompts in the sourcecode, mixed in with typescript, javascript, python or whatever programming language you use.
-
-If you use promptbooks, you can store them in several places, each with its own advantages and disadvantages:
-
-1. As **source code**, typically git-committed. In this case you can use the versioning system and the promptbooks will be tightly coupled with the version of the application. You still get the power of promptbooks, as you separate the concerns of the prompt-engineer and the programmer.
-
-2. As data in a **database** In this case, promptbooks are like posts / articles on the blog. They can be modified independently of the application. You don't need to redeploy the application to change the promptbooks. You can have multiple versions of promptbooks for each user. You can have a web interface for non-programmers to create and modify promptbooks. But you lose the versioning system and you still have to consider the interface between the promptbooks and the application _(= input and output parameters)_.
-
-3. In a **configuration** in environment variables. This is a good way to store promptbooks if you have an application with multiple deployments and you want to have different but simple promptbooks for each deployment and you don't need to change them often.
-
-### What should I do when I need same promptbook in multiple human languages?
-
-A single promptbook can be written for several _(human)_ languages at once. However, we recommend that you have separate promptbooks for each language.
-
-In large language models, you will get better results if you have prompts in the same language as the user input.
-
-The best way to manage this is to have suffixed promptbooks like `write-website-content.en.ptbk.md` and `write-website-content.cs.ptbk.md` for each supported language.
-
-
-
-
+-   [❔ Why not just use the OpenAI SDK / Anthropic Claude SDK / ...?](https://github.com/webgptorg/promptbook/discussions/114)
+-   [❔ How is it different from the OpenAI`s GPTs?](https://github.com/webgptorg/promptbook/discussions/118)
+-   [❔ How is it different from the Langchain?](https://github.com/webgptorg/promptbook/discussions/115)
+-   [❔ How is it different from the DSPy?](https://github.com/webgptorg/promptbook/discussions/117)
+-   [❔ How is it different from _anything_?](https://github.com/webgptorg/promptbook/discussions?discussions_q=is%3Aopen+label%3A%22Promptbook+vs%22)
+-   [❔ Is Promptbook using RAG _(Retrieval-Augmented Generation)_?](https://github.com/webgptorg/promptbook/discussions/123)
+-   [❔ Is Promptbook using function calling?](https://github.com/webgptorg/promptbook/discussions/124)
 
 ## ⌚ Changelog
 

package/esm/index.es.js

@@ -6,8 +6,8 @@ import spaceTrim$1, { spaceTrim } from 'spacetrim';
 /**
  * The version of the Promptbook library
  */
-var PROMPTBOOK_VERSION = '0.70.0-0';
-// TODO: !!!! List here all the versions and annotate + put into script
+var PROMPTBOOK_VERSION = '0.68.5';
+// TODO:[main] !!!! List here all the versions and annotate + put into script
 
 /*! *****************************************************************************
 Copyright (c) Microsoft Corporation.
@@ -340,7 +340,7 @@ function checkSerializableAsJson(name, value) {
 }
 /**
  * TODO: [🧠][🛣] More elegant way to tracking than passing `name`
- * TODO: [🧠] !!! In-memory cache of same values to prevent multiple checks
+ * TODO: [🧠][main] !!! In-memory cache of same values to prevent multiple checks
  * Note: [🐠] This is how `checkSerializableAsJson` + `isSerializableAsJson` together can just retun true/false or rich error message
  */
 
@@ -1196,6 +1196,7 @@ var OPENAI_MODELS = $asDeeplyFrozenSerializableJson('OPENAI_MODELS', [
             prompt: computeUsage("$5.00 / 1M tokens"),
             output: computeUsage("$15.00 / 1M tokens"),
         },
+        //TODO:[main] !!!!!! Add gpt-4o-mini-2024-07-18 and all others to be up to date
     },
     /**/
     /**/
@@ -1210,6 +1211,51 @@ var OPENAI_MODELS = $asDeeplyFrozenSerializableJson('OPENAI_MODELS', [
     },
     /**/
     /**/
+    {
+        modelVariant: 'CHAT',
+        modelTitle: 'o1-preview',
+        modelName: 'o1-preview',
+        pricing: {
+            prompt: computeUsage("$15.00 / 1M tokens"),
+            output: computeUsage("$60.00 / 1M tokens"),
+        },
+    },
+    /**/
+    /**/
+    {
+        modelVariant: 'CHAT',
+        modelTitle: 'o1-preview-2024-09-12',
+        modelName: 'o1-preview-2024-09-12',
+        //             <- TODO:[main] !!!!!! Some better system to organize theese date suffixes and versions
+        pricing: {
+            prompt: computeUsage("$15.00 / 1M tokens"),
+            output: computeUsage("$60.00 / 1M tokens"),
+        },
+    },
+    /**/
+    /**/
+    {
+        modelVariant: 'CHAT',
+        modelTitle: 'o1-mini',
+        modelName: 'o1-mini',
+        pricing: {
+            prompt: computeUsage("$3.00 / 1M tokens"),
+            output: computeUsage("$12.00 / 1M tokens"),
+        },
+    },
+    /**/
+    /**/
+    {
+        modelVariant: 'CHAT',
+        modelTitle: 'o1-mini-2024-09-12',
+        modelName: 'o1-mini-2024-09-12',
+        pricing: {
+            prompt: computeUsage("$3.00 / 1M tokens"),
+            output: computeUsage("$12.00 / 1M tokens"),
+        },
+    },
+    /**/
+    /**/
     {
         modelVariant: 'CHAT',
         modelTitle: 'gpt-3.5-turbo-16k-0613',
@@ -1619,7 +1665,7 @@ var OpenAiExecutionTools = /** @class */ (function () {
  * @public exported from `@promptbook/openai`
  */
 var createOpenAiExecutionTools = Object.assign(function (options) {
-    // TODO: [🧠] !!!! If browser, auto add `dangerouslyAllowBrowser`
+    // TODO: [🧠][main] !!!! If browser, auto add `dangerouslyAllowBrowser`
     if (($isRunningInBrowser() || $isRunningInWebWorker()) && !options.dangerouslyAllowBrowser) {
         options = __assign(__assign({}, options), { dangerouslyAllowBrowser: true });
     }
@@ -1633,6 +1679,245 @@ var createOpenAiExecutionTools = Object.assign(function (options) {
  * TODO: [🎶] Naming "constructor" vs "creator" vs "factory"
  */
 
+/**
+ * This error type indicates that some part of the code is not implemented yet
+ *
+ * @public exported from `@promptbook/core`
+ */
+var NotYetImplementedError = /** @class */ (function (_super) {
+    __extends(NotYetImplementedError, _super);
+    function NotYetImplementedError(message) {
+        var _this = _super.call(this, spaceTrim(function (block) { return "\n                    ".concat(block(message), "\n\n                    Note: This feature is not implemented yet but it will be soon.\n\n                    If you want speed up the implementation or just read more, look here:\n                    https://github.com/webgptorg/promptbook\n\n                    Or contact us on me@pavolhejny.com\n\n                "); })) || this;
+        _this.name = 'NotYetImplementedError';
+        Object.setPrototypeOf(_this, NotYetImplementedError.prototype);
+        return _this;
+    }
+    return NotYetImplementedError;
+}(Error));
+
+/**
+ * Represents the usage with no resources consumed
+ *
+ * @public exported from `@promptbook/core`
+ */
+$deepFreeze({
+    price: { value: 0 },
+    input: {
+        tokensCount: { value: 0 },
+        charactersCount: { value: 0 },
+        wordsCount: { value: 0 },
+        sentencesCount: { value: 0 },
+        linesCount: { value: 0 },
+        paragraphsCount: { value: 0 },
+        pagesCount: { value: 0 },
+    },
+    output: {
+        tokensCount: { value: 0 },
+        charactersCount: { value: 0 },
+        wordsCount: { value: 0 },
+        sentencesCount: { value: 0 },
+        linesCount: { value: 0 },
+        paragraphsCount: { value: 0 },
+        pagesCount: { value: 0 },
+    },
+});
+/**
+ * Represents the usage with unknown resources consumed
+ *
+ * @public exported from `@promptbook/core`
+ */
+var UNCERTAIN_USAGE = $deepFreeze({
+    price: { value: 0, isUncertain: true },
+    input: {
+        tokensCount: { value: 0, isUncertain: true },
+        charactersCount: { value: 0, isUncertain: true },
+        wordsCount: { value: 0, isUncertain: true },
+        sentencesCount: { value: 0, isUncertain: true },
+        linesCount: { value: 0, isUncertain: true },
+        paragraphsCount: { value: 0, isUncertain: true },
+        pagesCount: { value: 0, isUncertain: true },
+    },
+    output: {
+        tokensCount: { value: 0, isUncertain: true },
+        charactersCount: { value: 0, isUncertain: true },
+        wordsCount: { value: 0, isUncertain: true },
+        sentencesCount: { value: 0, isUncertain: true },
+        linesCount: { value: 0, isUncertain: true },
+        paragraphsCount: { value: 0, isUncertain: true },
+        pagesCount: { value: 0, isUncertain: true },
+    },
+});
+
+/**
+ * Execution Tools for calling OpenAI API Assistants
+ *
+ * This is usefull for calling OpenAI API with a single assistant, for more wide usage use `OpenAiExecutionTools`.
+ *
+ * @public exported from `@promptbook/openai`
+ */
+var OpenAiAssistantExecutionTools = /** @class */ (function (_super) {
+    __extends(OpenAiAssistantExecutionTools, _super);
+    /**
+     * Creates OpenAI Execution Tools.
+     *
+     * @param options which are relevant are directly passed to the OpenAI client
+     */
+    function OpenAiAssistantExecutionTools(options) {
+        if (options === void 0) { options = {}; }
+        return _super.call(this, options) || this;
+    }
+    Object.defineProperty(OpenAiAssistantExecutionTools.prototype, "title", {
+        get: function () {
+            return 'OpenAI Assistant';
+        },
+        enumerable: false,
+        configurable: true
+    });
+    Object.defineProperty(OpenAiAssistantExecutionTools.prototype, "description", {
+        get: function () {
+            return 'Use single assistant provided by OpenAI';
+        },
+        enumerable: false,
+        configurable: true
+    });
+    /**
+     * Calls OpenAI API to use a chat model.
+     */
+    OpenAiAssistantExecutionTools.prototype.callChatModel = function (prompt) {
+        var _a, _b, _c;
+        return __awaiter(this, void 0, void 0, function () {
+            var content, parameters, modelRequirements /*, format*/, client, _d, _e, key, rawPromptContent, rawRequest, start, complete, stream, rawResponse, resultContent, usage;
+            var e_1, _f;
+            var _this = this;
+            return __generator(this, function (_g) {
+                switch (_g.label) {
+                    case 0:
+                        if (this.options.isVerbose) {
+                            console.info('💬 OpenAI callChatModel call', { prompt: prompt });
+                        }
+                        content = prompt.content, parameters = prompt.parameters, modelRequirements = prompt.modelRequirements;
+                        return [4 /*yield*/, this.getClient()];
+                    case 1:
+                        client = _g.sent();
+                        // TODO: [☂] Use here more modelRequirements
+                        if (modelRequirements.modelVariant !== 'CHAT') {
+                            throw new PipelineExecutionError('Use callChatModel only for CHAT variant');
+                        }
+                        try {
+                            // TODO: [👨‍👨‍👧‍👧] Remove:
+                            for (_d = __values(['maxTokens', 'modelName', 'seed', 'temperature']), _e = _d.next(); !_e.done; _e = _d.next()) {
+                                key = _e.value;
+                                if (modelRequirements[key] !== undefined) {
+                                    throw new NotYetImplementedError("In `OpenAiAssistantExecutionTools` you cannot specify `".concat(key, "`"));
+                                }
+                            }
+                        }
+                        catch (e_1_1) { e_1 = { error: e_1_1 }; }
+                        finally {
+                            try {
+                                if (_e && !_e.done && (_f = _d.return)) _f.call(_d);
+                            }
+                            finally { if (e_1) throw e_1.error; }
+                        }
+                        rawPromptContent = replaceParameters(content, __assign(__assign({}, parameters), { modelName: 'assistant' }));
+                        rawRequest = {
+                            // [👨‍👨‍👧‍👧] ...modelSettings,
+                            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: [
+                                    // TODO: !!!!!! Unhardcode
+                                    // TODO: !!!!!! Allow threads to be passed
+                                    { role: 'user', content: 'What is the meaning of life? I want breathtaking speech.' },
+                                ],
+                            },
+                            // !!!!!! user: this.options.user,
+                        };
+                        start = getCurrentIsoDate();
+                        if (this.options.isVerbose) {
+                            console.info(colors.bgWhite('rawRequest'), JSON.stringify(rawRequest, null, 4));
+                        }
+                        return [4 /*yield*/, client.beta.threads.createAndRunStream(rawRequest)];
+                    case 2:
+                        stream = _g.sent();
+                        stream.on('connect', function () {
+                            if (_this.options.isVerbose) {
+                                console.info('connect', stream.currentEvent);
+                            }
+                        });
+                        stream.on('messageDelta', function (messageDelta) {
+                            var _a;
+                            if (_this.options.isVerbose &&
+                                messageDelta &&
+                                messageDelta.content &&
+                                messageDelta.content[0] &&
+                                messageDelta.content[0].type === 'text') {
+                                console.info('messageDelta', (_a = messageDelta.content[0].text) === null || _a === void 0 ? void 0 : _a.value);
+                            }
+                            // TODO: !!!!!! report progress
+                        });
+                        stream.on('messageCreated', function (message) {
+                            if (_this.options.isVerbose) {
+                                console.info('messageCreated', message);
+                            }
+                        });
+                        stream.on('messageDone', function (message) {
+                            if (_this.options.isVerbose) {
+                                console.info('messageDone', message);
+                            }
+                        });
+                        return [4 /*yield*/, stream.finalMessages()];
+                    case 3:
+                        rawResponse = _g.sent();
+                        if (this.options.isVerbose) {
+                            console.info(colors.bgWhite('rawResponse'), JSON.stringify(rawResponse, null, 4));
+                        }
+                        if (rawResponse.length !== 1) {
+                            throw new PipelineExecutionError("There is NOT 1 BUT ".concat(rawResponse.length, " finalMessages from OpenAI"));
+                        }
+                        if (rawResponse[0].content.length !== 1) {
+                            throw new PipelineExecutionError("There is NOT 1 BUT ".concat(rawResponse[0].content.length, " finalMessages content from OpenAI"));
+                        }
+                        if (((_a = rawResponse[0].content[0]) === null || _a === void 0 ? void 0 : _a.type) !== 'text') {
+                            throw new PipelineExecutionError("There is NOT 'text' BUT ".concat((_b = rawResponse[0].content[0]) === null || _b === void 0 ? void 0 : _b.type, " finalMessages content type from OpenAI"));
+                        }
+                        resultContent = (_c = rawResponse[0].content[0]) === null || _c === void 0 ? void 0 : _c.text.value;
+                        //                                                     <- TODO: !!!!!! There are also annotations, maybe use them
+                        // eslint-disable-next-line prefer-const
+                        complete = getCurrentIsoDate();
+                        usage = UNCERTAIN_USAGE;
+                        // TODO: !!!!!!> = computeOpenAiUsage(content, resultContent || '', rawResponse);
+                        if (resultContent === null) {
+                            throw new PipelineExecutionError('No response message from OpenAI');
+                        }
+                        return [2 /*return*/, $asDeeplyFrozenSerializableJson('OpenAiAssistantExecutionTools ChatPromptResult', {
+                                content: resultContent,
+                                modelName: 'assistant',
+                                //          <- TODO: !!!!!! Can we detect really used model: rawResponse.model || modelName,
+                                timing: {
+                                    start: start,
+                                    complete: complete,
+                                },
+                                usage: usage,
+                                rawPromptContent: rawPromptContent,
+                                rawRequest: rawRequest,
+                                rawResponse: rawResponse,
+                                // <- [🗯]
+                            })];
+                }
+            });
+        });
+    };
+    return OpenAiAssistantExecutionTools;
+}(OpenAiExecutionTools));
+/**
+ * TODO: !!!!!! DO not use colors - can be used in browser
+ * TODO: [🧠][🧙‍♂️] Maybe there can be some wizzard for thoose who want to use just OpenAI
+ * TODO: Maybe make custom OpenAiError
+ * TODO: [🧠][🈁] Maybe use `isDeterministic` from options
+ * TODO: [🧠][🌰] Allow to pass `title` for tracking purposes
+ */
+
 /**
  * @@@
  *
@@ -1705,5 +1990,5 @@ var _OpenAiRegistration = $llmToolsRegister.register(createOpenAiExecutionTools)
  * TODO: [🎶] Naming "constructor" vs "creator" vs "factory"
  */
 
-export { OPENAI_MODELS, OpenAiExecutionTools, PROMPTBOOK_VERSION, _OpenAiRegistration, createOpenAiExecutionTools };
+export { OPENAI_MODELS, OpenAiAssistantExecutionTools, OpenAiExecutionTools, PROMPTBOOK_VERSION, _OpenAiRegistration, createOpenAiExecutionTools };
 //# sourceMappingURL=index.es.js.map