aisquare-pipe 0.1.0

Universal anything-to-anything connector framework

aisquare.pipe

Universal anything-to-anything connector framework. Instead of building N² bespoke integrations between services, each service implements a single Source (pulls data) and/or Sink (pushes data) connector against a universal DataEnvelope spec. The framework handles type matching, pipeline orchestration, and plugin discovery.

Install

pip install aisquare-pipe                       # framework only
pip install "aisquare-pipe[popular]"            # framework + the canonical core connectors
pip install "aisquare-pipe[full]"               # framework + all connectors maintained in this repo
pip install aisquare-pipe-<service>             # framework + a single connector (e.g. aisquare-pipe-dropbox)

For development:

git clone [e-mail]:AISquare-Studio/pipe.git && cd pipe
pip install -e ".[dev]"

Note: Plugin discovery via entry_points requires the package to be installed (pip install -e .).

Connectors

Each connector is its own independently published pip package under connectors/<name>/. The user-facing tier is encoded in the extras bundles — folder location does not determine tier.

Connector	Package	In [popular]	In [full]
Local filesystem	aisquare-pipe-local	✓	✓
Dropbox	aisquare-pipe-dropbox	✓	✓
OneDrive	aisquare-pipe-onedrive	✓	✓
Salesforce	aisquare-pipe-salesforce		✓
DocuSign	aisquare-pipe-docusign		✓

Third-party connectors published as aisquare-pipe-<name> (or any package declaring the aisquare_pipe.connectors entry-point group) are auto-discovered after install.

Quick Start

from aisquare.pipe import Pipeline
from aisquare.pipe.testing.mock_connectors import MockSource, MockSink

source = MockSource(count=5)
sink = MockSink()

result = Pipeline(source=source, sink=sink).run({})

print(f"Transferred {result.success_count} envelopes")
print(f"Sink received: {[e.data for e in sink.received]}")

Architecture

┌─────────────┐    ┌──────────────┐    ┌─────────────┐
│   Source     │───>│ DataEnvelope │───>│    Sink     │
│  Connector   │    │              │    │  Connector   │
│              │    │ content_type │    │              │
│  pull() ─────┼──> │ data         │ ──>│──── push()  │
│              │    │ metadata     │    │              │
└─────────────┘    └──────────────┘    └─────────────┘
       │                  │                    │
       │          ┌───────┴────────┐           │
       │          │  TypeMatcher   │           │
       │          │  ┌──────────┐  │           │
       │          │  │Converter │  │           │
       │          │  └──────────┘  │           │
       │          └────────────────┘           │
       │                                       │
       └──────── Pipeline.run() ───────────────┘

CLI

pipe list                            # list installed connectors
pipe describe mock-source            # show connector details
pipe check mock-source mock-sink     # check type compatibility
pipe run --source mock-source --sink mock-sink --config config.json
pipe new-connector my-service        # scaffold a new connector plugin

Building a Connector

See CONTRIBUTING.md for the full guide. In short:

1. pipe new-connector my-service — scaffolds a plugin project
2. Implement SourceConnector.pull() and/or SinkConnector.push()
3. Register via entry_points in your pyproject.toml
4. Run connector_compliance_suite(MyConnector) to validate

License

Apache License 2.0