gnn-parsing

name: gnn-parsing description: GNN file discovery, parsing, and multi-format serialization. Use when reading GNN model files, parsing StateSpaceBlock definitions, extracting connections, validating GNN syntax, or converting between GNN formats.

GNN Parsing (Step 3)

Purpose

Discovers, parses, and validates GNN model files. Extracts structured data from Markdown-based GNN specifications including state spaces, connections, parameterizations, and ontology annotations.

Key Commands

# Parse all GNN files in a directory
python src/3_gnn.py --target-dir input/gnn_files --output-dir output --verbose

# As part of pipeline
python src/main.py --only-steps 3 --verbose

GNN File Sections

API

import logging
from pathlib import Path
from gnn import (
    discover_gnn_files,
    parse_gnn_file,
    process_gnn_directory,
    process_gnn_multi_format,
    validate_gnn,
)
from gnn import GNNFormalParser

logger = logging.getLogger(__name__)

# Discover GNN files in a directory
files = discover_gnn_files("input/gnn_files/")

# Parse a single file (lightweight dict-style summary)
model = parse_gnn_file("input/gnn_files/my_model.md")

# Process entire directory (used by pipeline)
results = process_gnn_directory("input/gnn_files/", "output/")

# Multi-format processing (requires a Logger — same contract as 3_gnn.py)
process_gnn_multi_format(
    Path("input/gnn_files"),
    Path("output"),
    logger,
    verbose=True,
)

# Formal parsing and validation
is_valid, errors = validate_gnn(content_string)

Key Exports

• discover_gnn_files — find .md GNN files in a directory tree
• parse_gnn_file — parse a single GNN file into structured data
• process_gnn_directory — process all files in a directory
• process_gnn_multi_format — full multi-format serialization (needs logging.Logger)
• validate_gnn_structure — structural validation of a parsed model
• validate_gnn — content-level syntax validation
• GNNParsingSystem, GNNFormat — registry-backed multi-format I/O
• GNNFormalParser, ParsedGNN, ParsedGNNFormal — formal parser types

Output

Parsed models are consumed by downstream steps:

• Step 5 (Type Check): validates matrix dimensions
• Steps 7–8 (Export/Viz): generates outputs
• Step 11 (Render): generates simulation code

MCP Tools

This module registers tools with the GNN MCP server (see mcp.py):

• get_gnn_documentation
• get_gnn_module_info
• get_gnn_schema_info
• parse_gnn_content
• process_gnn_directory
• run_round_trip_tests
• validate_cross_format_consistency_content
• validate_directory_cross_format_consistency
• validate_gnn_content
• validate_schema_definitions_consistency

References

• AGENTS.md — Module documentation
• README.md — Usage guide
• SPEC.md — Module specification
• ../../doc/gnn/reference/gnn_syntax.md — Complete syntax reference
• ../../doc/gnn/tutorials/gnn_examples_doc.md — Example models

Documentation

• README: Module Overview
• AGENTS: Agentic Workflows
• SPEC: Architectural Specification
• SKILL: Capability API