fastgx 2.0.5

High-performance in-memory graph database

FastGraph

FastGraph is a high-performance, in-memory graph database engineered for Python applications that demand speed, low latency, and zero nonsense. Built for developers who want graph operations that actually keep up.

Created by BRAHMAI with ❤️ & ☕.

---

Key Features

• High-performance adjacency-list engine with O(1) edge lookups
• Memory-efficient internals and lightweight subgraph views
• Automatic and manual indexes for ultra-fast queries
• JSON/YAML configuration system with sensible defaults
• Powerful CLI for common graph management tasks
• Persistence support across msgpack, pickle, and JSON
• Full thread safety for concurrent read/write workloads
• Integrated performance monitoring and instrumentation

---

Quick Start

Installation

pip install fastgx

Basic Usage

from fastgraph import FastGraph

graph = FastGraph("my_graph")

graph.add_node("alice", name="Alice", age=30, type="Person")
graph.add_node("bob", name="Bob", age=25, type="Person")
graph.add_node("company", name="TechCorp", type="Company")

graph.add_edge("alice", "bob", "friends", since=2021)
graph.add_edge("alice", "company", "works_at", role="Engineer")
graph.add_edge("bob", "company", "works_at", role="Manager")

people = graph.find_nodes(type="Person")
print(len(people))

friends = graph.neighbors_out("alice", rel="friends")
print([n for n, edge in friends])

graph.save("my_graph.msgpack")

---

Configuration

FastGraph supports JSON/YAML configs for repeatable setups.

Example default configuration:

engine:
  name: "FastGraph"
  version: "2.0.0"

storage:
  data_dir: "~/.cache/fastgraph/data"
  default_format: "msgpack"

memory:
  query_cache_size: 128
  cache_ttl: 3600

indexing:
  auto_index: true
  default_indexes: ["id", "type", "name"]

performance:
  thread_pool_size: 4
  batch_threshold: 100

cli:
  default_output_format: "table"
  verbose: false

Usage:

from fastgraph import FastGraph
from fastgraph.config import ConfigManager

config = ConfigManager("my_config.yaml")
graph = FastGraph("my_graph", config=config)

---

CLI Tools

fastgraph create --name "social_network"
fastgraph import data.json --format json --save social.msgpack
fastgraph export social.msgpack --format json --output export.json
fastgraph stats social.msgpack --detailed --components
fastgraph config --show
fastgraph config --set storage.default_format=json

---

Advanced Capabilities

Indexing

graph.build_node_index("type")
graph.build_node_index("age")

people = graph.find_nodes(type="Person")
engineers = graph.find_nodes(role="Engineer")

stats = graph.get_index_stats()
print(stats["global"]["index_hits"])

Batch Operations

graph.add_nodes_batch([
    ("user1", {"name": "John", "age": 28}),
    ("user2", {"name": "Jane", "age": 32}),
    ("user3", {"name": "Bob",  "age": 25})
])

graph.add_edges_batch([
    ("user1", "user2", "friends"),
    ("user2", "user3", "colleagues"),
    ("user1", "user3", "acquaintances")
])

Subgraph Views

view = graph.create_subgraph_view(
    "people",
    lambda nid, attrs: attrs.get("type") == "Person"
)

print(view.node_count)
components = graph.traversal_ops.connected_components()

Traversal Algorithms

bfs = graph.traversal_ops.bfs("alice", max_depth=2)
path = graph.traversal_ops.shortest_path("alice", "bob")

for p in graph.traversal_ops.find_paths("alice", "bob", max_length=3):
    print(p)

---

Performance Monitoring

from fastgraph.utils.performance import performance_monitor

@performance_monitor("query_op")
def q():
    return graph.find_nodes(type="Person", age__gte=25)

monitor = graph.get_performance_monitor()
print(monitor.get_stats()["query_op"].avg_duration)

---

Testing

pip install fastgx[dev]
pytest
pytest --cov=fastgraph --cov-report=html

---

Development

Setup

git clone https://github.com/cognition-brahmai/fastgraph.git
cd fastgraph
pip install -e ".[dev]"
pre-commit install

Quality

black fastgraph tests examples
flake8 fastgraph tests examples
mypy fastgraph
pre-commit run --all-files

---

Benchmarks

FastGraph is tuned for high throughput and low latency.

Operation	Complexity
Node lookup	O(1)
Edge lookup	O(1)
Neighbor query	O(degree)
Indexed search	O(log n)
Batch inserts	O(n)

Detailed benchmarks live in benchmarks/.

---

Contributing

1. Fork the repo
2. Create a branch
3. Commit your changes
4. Push and open a PR

We keep contributions simple and frictionless.

---

License

MIT. See the LICENSE file.

---

Acknowledgments

FastGraph builds on proven graph database principles and a relentless focus on performance and developer experience. Thanks to all contributors and the Python ecosystem that made it possible.