runbooks 0.7.5__py3-none-any.whl → 0.7.6__py3-none-any.whl

jupyter-agent/app.py

@@ -1,256 +0,0 @@
-"""
-app.py
-
-Main entry point for the Jupyter Agent. This file sets up the Gradio user interface,
-handles session-specific sandbox initialization, file management, and orchestrates
-the interactive notebook generation and code execution.
-"""
-
-import json
-import os
-from pathlib import Path
-from typing import Any, Dict, List, Tuple
-
-import gradio as gr
-from e2b_code_interpreter import Sandbox
-from gradio.utils import get_space
-from huggingface_hub import InferenceClient
-from transformers import AutoTokenizer
-
-## Import helper functions from utils with type hints
-from utils import (
-    create_base_notebook,
-    run_interactive_notebook,
-    update_notebook_display,
-)
-
-from runbooks.utils.logger import configure_logger
-
-## ✅ Configure Logger
-logger = configure_logger(__name__)
-
-## Load environment variables if not running in a Hugging Face Space
-if not get_space():
-    try:
-        from dotenv import load_dotenv
-
-        load_dotenv()
-    except (ImportError, ModuleNotFoundError):
-        logger.warning("python-dotenv not installed; proceeding without .env support.")
-        pass
-
-## Global configurations
-## For sandbox execution and Hugging Face API authentication
-E2B_API_KEY = os.environ["E2B_API_KEY"]
-HF_TOKEN = os.environ["HF_TOKEN"]
-## Set a limit on the number of new tokens generated per request.
-DEFAULT_MAX_TOKENS = 512
-## A dictionary used to keep track of separate execution environments (sandboxes) for each session.
-SANDBOXES = {}
-## For storing temporary files (e.g., generated notebooks)
-TMP_DIR = "./tmp/"
-## Ensure the temporary directory exists
-if not os.path.exists(TMP_DIR):
-    os.makedirs(TMP_DIR)
-
-## Initialize a base notebook on startup
-notebook_data = create_base_notebook([])[0]
-with open(TMP_DIR + "jupyter-agent.ipynb", "w", encoding="utf-8") as f:
-    json.dump(notebook_data, f, indent=2)
-
-## Read default system prompt
-try:
-    with open("ds-system-prompt.txt", "r", encoding="utf-8") as f:
-        DEFAULT_SYSTEM_PROMPT: str = f.read()
-except FileNotFoundError:
-    logger.error("ds-system-prompt.txt not found. Please ensure it is available.")
-    DEFAULT_SYSTEM_PROMPT = ""
-
-
-## --- Main Execution Function ---
-def execute_jupyter_agent(
-    system_prompt: str,
-    user_input: str,
-    max_new_tokens: int,
-    model: str,
-    files: List[str],
-    message_history: List[Dict[str, Any]],
-    request: gr.Request,
-) -> Tuple[str, List[Dict[str, Any]], str]:
-    """
-    Core callback function that orchestrates the interactive notebook generation.
-
-    :param system_prompt: The system prompt template.
-    :param user_input: User's input command.
-    :param max_new_tokens: Maximum number of tokens to generate.
-    :param model: Identifier for the language model.
-    :param files: List of file paths uploaded by the user.
-    :param message_history: History of conversation messages.
-    :param request: Gradio request object with session details.
-    :return: A tuple containing the updated notebook HTML, updated message history,
-             and the path to the generated notebook file.
-    """
-    ## Retrieve or create a sandbox instance for the given session.
-    if request.session_hash not in SANDBOXES:
-        ## Create a new Sandbox with the E2B API key and stored
-        SANDBOXES[request.session_hash] = Sandbox(api_key=E2B_API_KEY)
-    sbx = SANDBOXES[request.session_hash]
-
-    ## Create session-specific directory for saving notebook
-    save_dir = os.path.join(TMP_DIR, request.session_hash)
-    os.makedirs(save_dir, exist_ok=True)
-    save_dir = os.path.join(save_dir, "jupyter-agent.ipynb")
-
-    ## Initializes an inference client for text generation using the provided Hugging Face token.
-    client = InferenceClient(api_key=HF_TOKEN)
-
-    ## Loads the tokenizer corresponding to the chosen model.
-    tokenizer = AutoTokenizer.from_pretrained(model)
-    # model = "meta-llama/Llama-3.1-8B-Instruct"
-
-    ## Process uploaded files
-    filenames = []
-    if files is not None:
-        for filepath in files:
-            filpath = Path(filepath)
-            with open(filepath, "rb") as file:
-                print(f"uploading {filepath}...")
-                ## Write the file into the sandbox’s file system (allowing the agent to access it during execution).
-                sbx.files.write(filpath.name, file)
-                filenames.append(filpath.name)
-
-    ## Initialize message_history if it doesn't exist
-    if len(message_history) == 0:
-        message_history.append(
-            {
-                "role": "system",
-                "content": system_prompt.format("- " + "\n- ".join(filenames)),
-            }
-        )
-    message_history.append({"role": "user", "content": user_input})
-
-    ## Outputs the current conversation history for debugging purposes.
-    logger.debug(f"Message history: {message_history}")
-
-    ## Generate notebook updates by streaming responses
-    for notebook_html, notebook_data, messages in run_interactive_notebook(
-        client, model, tokenizer, message_history, sbx, max_new_tokens=max_new_tokens
-    ):
-        message_history = messages
-
-        ## Yield intermediate UI updates with a fixed download path (initial version)
-        yield notebook_html, message_history, TMP_DIR + "jupyter-agent.ipynb"
-
-    ## Save the final notebook JSON data to a specified path.
-    with open(save_dir, "w", encoding="utf-8") as f:
-        json.dump(notebook_data, f, indent=2)
-    yield notebook_html, message_history, save_dir
-    logger.info(f"Notebook saved to {save_dir}")
-
-
-def clear(msg_state):
-    msg_state = []
-    return update_notebook_display(create_base_notebook([])[0]), msg_state
-
-
-## Gradio components fill the full height of the viewport, allow scrolling, and have appropriate padding
-custom_css = """
-#component-0 {
-    height: 100vh;
-    overflow-y: auto;
-    padding: 20px;
-}
-
-.gradio-container {
-    height: 100vh !important;
-}
-
-.contain {
-    height: 100vh !important;
-}
-"""
-## TODO
-# footer {
-#         visibility: hidden;
-# }
-
-
-## Build and return the Gradio Blocks interface for the Jupyter Agent.
-# with gr.Blocks(css=custom_css) as poc:
-with gr.Blocks() as poc:
-    msg_state = gr.State(value=[])
-
-    html_output = gr.HTML(value=update_notebook_display(create_base_notebook([])[0]))
-
-    user_input = gr.Textbox(
-        value="Solve the Bayes' theorem equation and plot the results.",
-        lines=3,
-        label="User input",
-    )
-
-    with gr.Row():
-        generate_btn = gr.Button("▶️ Let's go!")
-        clear_btn = gr.Button("🧹 Clear")
-
-    file = gr.File(
-        TMP_DIR + "jupyter-agent.ipynb", label="💾 Download Jupyter Notebook"
-    )
-
-    with gr.Accordion("Upload files", open=False):
-        files = gr.File(label="Upload files to use", file_count="multiple")
-
-    with gr.Accordion("Advanced Settings", open=False):
-        system_input = gr.Textbox(
-            label="System Prompt",
-            value=DEFAULT_SYSTEM_PROMPT,
-            elem_classes="input-box",
-            lines=8,
-        )
-        with gr.Row():
-            max_tokens = gr.Number(
-                label="Max New Tokens",
-                value=DEFAULT_MAX_TOKENS,
-                minimum=128,
-                maximum=2048,
-                step=8,
-                interactive=True,
-            )
-
-            model = gr.Dropdown(
-                # value="meta-llama/Llama-3.1-8B-Instruct",
-                value="meta-llama/Llama-3.3-70B-Instruct",
-                choices=[
-                    ## Text only instruct-tuned model in 70B size (text in/text out).
-                    "meta-llama/Llama-3.3-70B-Instruct",
-                    ## Pretrained and fine-tuned text models with sizes
-                    "meta-llama/Llama-3.1-8B-Instruct",
-                    ## pretrained and instruction-tuned image reasoning generative models in 11B and 90B sizes (text + images in / text out)
-                    "meta-llama/Llama-3.2-3B-Instruct",
-                    "meta-llama/Llama-3.2-11B-Vision-Instruct",
-                ],
-                label="Models",
-            )
-
-    generate_btn.click(
-        fn=execute_jupyter_agent,
-        inputs=[system_input, user_input, max_tokens, model, files, msg_state],
-        outputs=[html_output, msg_state, file],
-    )
-
-    clear_btn.click(fn=clear, inputs=[msg_state], outputs=[html_output, msg_state])
-
-    poc.load(
-        fn=None,
-        inputs=None,
-        outputs=None,
-        js=""" () => {
-    if (document.querySelectorAll('.dark').length) {
-        document.querySelectorAll('.dark').forEach(el => el.classList.remove('dark'));
-    }
-}
-""",
-    )
-
-## Main entry point: launch the Gradio interface.
-## Disable server-side rendering (i.e., client-side rendering is used).
-poc.launch(ssr_mode=False, pwa=True, share=True)

jupyter-agent/ds-system-prompt.txt

@@ -1,154 +0,0 @@
-# Data Science Agent Protocol
-
-You are an intelligent data science assistant with access to an IPython interpreter. Your primary goal is to solve analytical tasks through careful, iterative exploration and execution of code. You must avoid making assumptions and instead verify everything through code execution.
-
-## Core Principles
-1. Always execute code to verify assumptions
-2. Break down complex problems into smaller steps
-3. Learn from execution results
-4. Maintain clear communication about your process
-
-## Available Packages
-You have access to these pre-installed packages:
-
-### Core Data Science
-- numpy (1.26.4)
-- pandas (1.5.3)
-- scipy (1.12.0)
-- scikit-learn (1.4.1.post1)
-
-### Visualization
-- matplotlib (3.9.2)
-- seaborn (0.13.2)
-- plotly (5.19.0)
-- bokeh (3.3.4)
-- e2b_charts (latest)
-
-### Image & Signal Processing
-- opencv-python (4.9.0.80)
-- pillow (9.5.0)
-- scikit-image (0.22.0)
-- imageio (2.34.0)
-
-### Text & NLP
-- nltk (3.8.1)
-- spacy (3.7.4)
-- gensim (4.3.2)
-- textblob (0.18.0)
-
-### Audio Processing
-- librosa (0.10.1)
-- soundfile (0.12.1)
-
-### File Handling
-- python-docx (1.1.0)
-- openpyxl (3.1.2)
-- xlrd (2.0.1)
-
-### Other Utilities
-- requests (2.26.0)
-- beautifulsoup4 (4.12.3)
-- sympy (1.12)
-- xarray (2024.2.0)
-- joblib (1.3.2)
-
-## Environment Constraints
-- You cannot install new packages or libraries
-- Work only with pre-installed packages in the environment
-- If a solution requires a package that's not available:
-  1. Check if the task can be solved with base libraries
-  2. Propose alternative approaches using available packages
-  3. Inform the user if the task cannot be completed with current limitations
-
-## Analysis Protocol
-
-### 1. Initial Assessment
-- Acknowledge the user's task and explain your high-level approach
-- List any clarifying questions needed before proceeding
-- Identify which available files might be relevant from: {}
-- Verify which required packages are available in the environment
-
-### 2. Data Exploration
-Execute code to:
-- Read and validate each relevant file
-- Determine file formats (CSV, JSON, etc.)
-- Check basic properties:
-  - Number of rows/records
-  - Column names and data types
-  - Missing values
-  - Basic statistical summaries
-- Share key insights about the data structure
-
-### 3. Execution Planning
-- Based on the exploration results, outline specific steps to solve the task
-- Break down complex operations into smaller, verifiable steps
-- Identify potential challenges or edge cases
-
-### 4. Iterative Solution Development
-For each step in your plan:
-- Write and execute code for that specific step
-- Verify the results meet expectations
-- Debug and adjust if needed
-- Document any unexpected findings
-- Only proceed to the next step after current step is working
-
-### 5. Result Validation
-- Verify the solution meets all requirements
-- Check for edge cases
-- Ensure results are reproducible
-- Document any assumptions or limitations
-
-## Error Handling Protocol
-When encountering errors:
-1. Show the error message
-2. Analyze potential causes
-3. Propose specific fixes
-4. Execute modified code
-5. Verify the fix worked
-6. Document the solution for future reference
-
-## Communication Guidelines
-- Explain your reasoning at each step
-- Share relevant execution results
-- Highlight important findings or concerns
-- Ask for clarification when needed
-- Provide context for your decisions
-
-## Code Execution Rules
-- Execute code through the IPython interpreter directly
-- Understand that the environment is stateful (like a Jupyter notebook):
-  - Variables and objects from previous executions persist
-  - Reference existing variables instead of recreating them
-  - Only rerun code if variables are no longer in memory or need updating
-- Don't rewrite or re-execute code unnecessarily:
-  - Use previously computed results when available
-  - Only rewrite code that needs modification
-  - Indicate when you're using existing variables from previous steps
-- Run code after each significant change
-- Don't show code blocks without executing them
-- Verify results before proceeding
-- Keep code segments focused and manageable
-
-## Memory Management Guidelines
-- Track important variables and objects across steps
-- Clear large objects when they're no longer needed
-- Inform user about significant objects kept in memory
-- Consider memory impact when working with large datasets:
-  - Avoid creating unnecessary copies of large data
-  - Use inplace operations when appropriate
-  - Clean up intermediate results that won't be needed later
-
-## Best Practices
-- Use descriptive variable names
-- Include comments for complex operations
-- Handle errors gracefully
-- Clean up resources when done
-- Document any dependencies
-- Prefer base Python libraries when possible
-- Verify package availability before using
-- Leverage existing computations:
-  - Check if required data is already in memory
-  - Reference previous results instead of recomputing
-  - Document which existing variables you're using
-
-Remember: Verification through execution is always better than assumption!

jupyter-agent/llama3_template.jinja

@@ -1,123 +0,0 @@
-{{- bos_token }}
-{%- if custom_tools is defined %}
-    {%- set tools = custom_tools %}
-{%- endif %}
-{%- if not tools_in_user_message is defined %}
-    {%- set tools_in_user_message = true %}
-{%- endif %}
-{%- if not date_string is defined %}
-    {%- set date_string = "26 Jul 2024" %}
-{%- endif %}
-{%- if not tools is defined %}
-    {%- set tools = none %}
-{%- endif %}
-
-{#- This block extracts the system message, so we can slot it into the right place. #}
-{%- if messages[0]['role'] == 'system' %}
-    {%- set system_message = messages[0]['content']|trim %}
-    {%- set messages = messages[1:] %}
-{%- else %}
-    {%- set system_message = "" %}
-{%- endif %}
-
-{#- System message + builtin tools #}
-{{- "<|start_header_id|>system<|end_header_id|>\n\n" }}
-{%- if builtin_tools is defined or tools is not none %}
-    {{- "Environment: ipython\n" }}
-{%- endif %}
-{%- if builtin_tools is defined %}
-    {%- set filtered_tools = builtin_tools | reject('equalto', 'code_interpreter') | list %}
-    {%- if filtered_tools | length > 0 %}
-        {{- "Tools: " + filtered_tools | join(", ") + "\n\n"}}
-    {%- else %}
-        {{- "\n" }}
-    {%- endif %}
-{%- endif %}
-{{- "Cutting Knowledge Date: December 2023\n" }}
-{{- "Today Date: " + date_string + "\n\n" }}
-{%- if tools is not none and not tools_in_user_message %}
-    {{- "You have access to the following functions. To call a function, please respond with JSON for a function call." }}
-    {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
-    {{- "Do not use variables.\n\n" }}
-    {%- for t in tools %}
-        {{- t | tojson(indent=4) }}
-        {{- "\n\n" }}
-    {%- endfor %}
-{%- endif %}
-{{- system_message }}
-{{- "<|eot_id|>" }}
-
-{#- Custom tools are passed in a user message with some extra guidance #}
-{%- if tools_in_user_message and not tools is none %}
-    {#- Extract the first user message so we can plug it in here #}
-    {%- if messages | length != 0 %}
-        {%- set first_user_message = messages[0]['content']|trim %}
-        {%- set messages = messages[1:] %}
-    {%- else %}
-        {{- raise_exception("Cannot put tools in the first user message when there's no first user message!") }}
-{%- endif %}
-    {{- '<|start_header_id|>user<|end_header_id|>\n\n' -}}
-    {{- "Given the following functions, please respond with a JSON for a function call " }}
-    {{- "with its proper arguments that best answers the given prompt.\n\n" }}
-    {{- 'Respond in the format {"name": function name, "parameters": dictionary of argument name and its value}.' }}
-    {{- "Do not use variables.\n\n" }}
-    {%- for t in tools %}
-        {{- t | tojson(indent=4) }}
-        {{- "\n\n" }}
-    {%- endfor %}
-    {{- first_user_message + "<|eot_id|>"}}
-{%- endif %}
-
-{%- for message in messages %}
-    {%- if not (message.role == 'ipython' or message.role == 'tool' or 'tool_calls' in message) %}
-        {{- '<|start_header_id|>' + message['role'] + '<|end_header_id|>\n\n'+ message['content'] | trim}}
-        {%- if loop.nextitem and loop.nextitem.role == 'assistant' and message.role == 'assistant' %}
-            {{- '<|eom_id|>' }}
-        {%- else %}
-            {{- '<|eot_id|>' }}
-        {%- endif %}
-    {%- elif 'tool_calls' in message %}
-        {%- if not message.tool_calls|length == 1 %}
-            {{- raise_exception("This model only supports single tool-calls at once!") }}
-        {%- endif %}
-        {%- set tool_call = message.tool_calls[0].function %}
-        {%- if builtin_tools is defined and tool_call.name in builtin_tools %}
-            {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
-            {%- if tool_call.name == "code_interpreter" %}
-                {{- "<|python_tag|>" + tool_call.arguments.code }}
-            {%- else %}
-                {{- "<|python_tag|>" + tool_call.name + ".call(" }}
-                {%- for arg_name, arg_val in tool_call.arguments | items %}
-                    {{- arg_name + '="' + arg_val + '"' }}
-                    {%- if not loop.last %}
-                        {{- ", " }}
-                    {%- endif %}
-                {%- endfor %}
-                {{- ")" }}
-            {%- endif %}
-        {%- else  %}
-            {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' -}}
-            {{- '{"name": "' + tool_call.name + '", ' }}
-            {{- '"parameters": ' }}
-            {{- tool_call.arguments | tojson }}
-            {{- "}" }}
-        {%- endif %}
-        {%- if builtin_tools is defined %}
-            {#- This means we're in ipython mode #}
-            {{- "<|eom_id|>"  }}
-        {%- else %}
-            {{- "<|eot_id|>" }}
-        {%- endif %}
-    {%- elif message.role == "tool" or message.role == "ipython" %}
-        {{- "<|start_header_id|>ipython<|end_header_id|>\n\n" }}
-        {%- if not message.content is string and (message.content is mapping or message.content is iterable) %}
-            {{- message.content | tojson }}
-        {%- else %}
-            {{- message.content }}
-        {%- endif %}
-        {{- "<|eot_id|>" }}
-    {%- endif %}
-{%- endfor %}
-{%- if add_generation_prompt %}
-    {{- '<|start_header_id|>assistant<|end_header_id|>\n\n' }}
-{%- endif %}

jupyter-agent/requirements.txt

@@ -1,9 +0,0 @@
-nbformat
-nbconvert
-huggingface_hub
-e2b-code-interpreter
-transformers
-traitlets
-gradio
-## For local development, to load environment variables from a .env file
-python-dotenv