Create a new GraphGenerator instance from subgraph data.

* Handle API errors with proper error types

Extract subgraph data around specified nodes.

Main LLM generation method Args: messages: List of message dicts with 'role' and 'content' model: Mo

Helper to get text from a tree-sitter node.

Count tokens in content using LangChain's approximate token counting Args: content: Content to count (can be str

* Extract error message from API response

Build LLM context from one or more subgraphs. Args: subgraphs: Single GraphGenerator or list of GraphGenerators

Perform fuzzy search and return a subgraph. Args: query: Search query string categories: Optional li

Get configuration for a specific model

Check if the access token is valid and not a placeholder.

Returns the 1-based start line number of a node. Returns -1 if line number cannot be determined.

Clean up temporary directories.

Automatically detect provider from model name

* Generic function to handle all API operations (text, graph, structure) * for both GitHub repos and ZIP files

Print formatted subheader

Create a dummy user for testing

Find repository by ID using ObjectId or repository identifier Args: repo_id: Repository ObjectId, identifier (format: owner/repo

Parse GitHub repository URL into owner, repo, and optional branch/path.

Print formatted header

Returns the 1-based end line number of a node. Returns -1 if line number cannot be determined.

Check if file exists.

Get a LangChain chat model instance - this is the main method you'll use Args: model: Model name (e.g., "gpt-4o-

Get or create a GitVizz graph for the repository

Print formatted header

Print formatted header

Print formatted subheader

Extracts the full code snippet for a given AST node.

Find common anti-patterns in the code. Args: pattern_type: Type of anti-pattern ('god_class', 'long_method', 'de

Get API key for request (user's or system)

Get available models per provider

Analyze architectural layers and dependencies. Returns: GraphGenerator subgraph showing layered architecture

Get the storage path for a specific user.

Resolve the branch to use for a repository. Priority: 1. If requested_branch is provided and exists, use it 2. If requested_branch i

Filter files to include only source code and exclude images, binaries, etc. For .ipynb files, content is extracted from cells.

Test a single source with given configuration.

Internal function to decode JWT token and return user

Generate content using llm_service with robust error handling.

Helper to run async functions in sync context

Add a message to the conversation

Check if repository already exists for user and if it's up to date.

Combine multiple subgraphs into one. Args: *subgraphs: Variable number of GraphGenerator subgraphs

Decrypt API key for usage

Encrypt API key for secure storage

Extract files from a ZIP archive and return file list with paths.

Format repository contents into LLM-friendly text with directory structure.

Create a GraphGenerator directly from a ZIP file or folder. Args: source_path: Path to ZIP file or directory

Fuzzy search using sequence matching

Generate a unique identifier for the repository using owner/repo/branch format.

Get the latest commit SHA for a GitHub repository branch.

Get maximum context length for a model

Get repository context for LLM processing by processing the locally stored ZIP file Args: repository: Repository

* Helper function to extract owner and repo from GitHub URL

Parses the given file content using tree-sitter. Args: relative_file_path: The path to the file, relative to the project

Search nodes by code content

Search nodes by name (exact and partial matches) - optimized with index

* Stop streaming and close the connection

Check if model supports function calling

Test advanced features like NetworkX, I/O, visualization.

Convert the graph to a NetworkX DiGraph.

Calculate similarity between query and text using SequenceMatcher.

Safely get end_lineno for an AST node.

Create a realistic codebase for testing.

Download a test source ZIP file.

Filter nodes by category and return subgraph. Args: categories: List of categories to include depth:

Find circular dependencies in the code. Returns: GraphGenerator subgraph containing nodes involved in cycles

Find potential interface/abstraction violations. Returns: GraphGenerator subgraph containing violation instances

Find code that appears to lack test coverage. Returns: GraphGenerator subgraph containing untested code

Find potentially unused code (nodes with no incoming connections). Args: min_depth: Minimum depth from entry poi