sdxclient 0.10.0

A Python client library for interacting with the AtlanticWave-SDX L2VPN API

SDXClient

A thin Python client that wraps the SDX HTTP routes, stages L2VPN payload metadata, guides endpoint selection, and keeps responses consistent.

• Returns dict everywhere: {"status_code": int, "data": Any, "error": Optional[str]}
• No user-facing tracebacks; friendly errors only.

---

Key Points

• Uses FABRIC authentication (token loaded via fablib; no token argument needed).
• Provides a guided flow: set first endpoint → set second endpoint → preview → create.
• One unified method:
  • set_endpoint(endpoint_position="first", ...)
  • set_endpoint(endpoint_position="second", ...)
• VLAN choice is derived from /device_info; supports preferring untagged.
• Consistent return shape across all methods:

Initialization

from sdxclient import SDXClient
client = SDXClient(timeout=6.0)            # tries FABRIC token

Return Shape (always)

{
  "status_code": int,
  "data": Any,
  "error": Optional[str]
}

• status_code == 0 → client-side validation / network failure
• status_code >= 400 → server-side error (with best-effort error message)

---

Requirements

• Python 3.9 – 3.12
• requests
• fabrictestbed-extensions (for automatic FABRIC token loading)
• A valid FABRIC token (fablib locates it automatically)

---

Core Read APIs (pass-through)

client.get_topology()
client.get_available_ports(search=None, filter=None, limit=None, fields=None, format="html|json")
client.get_all_vlans_available()
client.get_port_vlans_available(port_id)

Notes:

• filter wins over search.
• format="html" returns HTML table (string). format="json" returns structured JSON.

Selection Flow (endpoints)

client.clear_selection()
client.set_endpoint(
    endpoint_position="first" | "second",
    filter=None, search=None,
    port_id=None, vlan=None,
    prefer_untagged=False
)
client.get_selected_endpoints()

Rules:

• Use port_id for precise selection, or filter/search to resolve a single row.
• If multiple candidates match, you get an ambiguity error with a short candidate list.
• VLAN availability is checked inline against /available_vlans.

L2VPN Payload: Stage → Preview → Create

1) Stage payload metadata once

client.set_l2vpn_payload(
    name="project-alpha-l2vpn",
    notifications=["[e-mail]", "[e-mail]"]
)

• Validates name (non-empty, ≤ 50 chars).
• Validates notifications (1–10 emails; str input is normalized to a list).

2) Build payload locally (no network)

payload_result = client.get_l2vpn_payload()
# -> {"name": "...", "notifications": [...], "endpoints": [{...}, {...}]}

• Requires both endpoints to be selected and metadata to be staged.
• Re-validates endpoint shape and rejects same-VLAN per policy (see “VLAN Rules”).
• No availability calls here.

3) Preview payload with availability checks

preview_result = client.preview_l2vpn_payload()
# Re-checks VLAN availability via /available_vlans for both endpoints.

4) Create L2VPN (uses previewed payload)

create_result = client.create_l2vpn_from_selection()

• Uses the staged name/notifications + selected endpoints.
• Performs preview internally first (includes availability checks).
• No need to pass name/notifications again here.

CRUD Mirrors

• client.get_l2vpns(**query)
• client.get_l2vpn(service_id)
• client.update_l2vpn(service_id, **fields)
• client.delete_l2vpn(service_id)

Minimal End-to-End Example

from sdxclient import SDXClient
client = SDXClient()

1) Start fresh

client.clear_selection()

2) Choose endpoints

client.set_endpoint(endpoint_position="first",  filter="SW17:7")
client.set_endpoint(endpoint_position="second", filter="SW17:27", prefer_untagged=True)

3) Stage metadata once

client.set_l2vpn_payload(
    name="proj-alpha-mia-test",
    notifications="[e-mail]"
)

4) Optional: local payload (no network)

local_payload = client.get_l2vpn_payload()

5) Preview with availability checks

preview = client.preview_l2vpn_payload()

6) Create

created = client.create_l2vpn_from_selection()

Error Examples (friendly)

client.get_port_vlans_available(None)
# {'status_code': 0, 'data': None, 'error': 'missing required parameter(s): port_id'}

client.set_endpoint(endpoint_position="first", filter="too-broad")
# {'status_code': 0, 'data': {'candidates': [...]}, 'error': 'ambiguous filter/search matched ...'}

client.get_l2vpn_payload()   # without staging metadata
# {'status_code': 0, 'data': None, 'error': 'missing L2VPN name (set via set_l2vpn_payload)'}

VLAN Validation Rules for L2VPN

Allowed forms

• Numeric VLANs: 1..4095
• Ranges: A:B where 1 ≤ A < B ≤ 4095
• Keywords: any, untagged
• (all is not allowed)

Mixing rules

• Numeric ↔ Range (e.g., 200 with 100:300)
• any ↔ untagged
• any/untagged mixed with numeric/range

Rejection rules

• all mixed with anything (since all is not allowed at all)
• Same numeric VLAN twice (e.g., 200 & 200)
• Same exact range twice (e.g., 100:200 & 100:200)

Tips

• Prefer format="json" for machine workflows; parse ports from /available_ports.

• If you pass both filter and search, filter is used.

• “No usable VLAN found” often means the port advertises none or they’re all in use.

• If you want to skip availability checks, use get_l2vpn_payload() and post it yourself; otherwise use create_l2vpn_from_selection() to keep checks on.

Troubleshooting

• 401 / missing token → set a valid token or ensure FABRIC token is discoverable.
• Empty listings → loosen your filter/search; inspect /topology.
• Availability errors → verify get_port_vlans_available(port_id) shows a range that contains your requested VLAN token.

Changelog (relevant)

• Added set_l2vpn_payload(name, notifications): stage metadata once.
• Added get_l2vpn_payload(): build local payload without network.
• create_l2vpn_from_selection() no longer requires passing name/notifications; it uses staged metadata and runs a preview internally.

Install

Production (PyPI)

pip install sdxclient

or a specific version

pip install sdxclient==0.10.0

Test (TestPyPI pre-releases)

• Prefer exact RC version when installing from TestPyPI

pip install --index-url https://test.pypi.org/simple/
--extra-index-url https://pypi.org/simple
sdxclient==0.10.0rc1

• or allow any pre-release (if available) pip install --index-url https://test.pypi.org/simple/
  --extra-index-url https://pypi.org/simple
  --pre sdxclient

Runtime configuration (test vs prod API)

• sdxclient reads the base URL from the environment:

Test (default if unset in config.py)

export SDX_BASE_URL=http://190.103.184.194

Production

export SDX_BASE_URL=https://sdxapi.atlanticwave-sdx.ai

• If SDX_BASE_URL isn’t set, the package falls back to the default defined in sdxclient/config.py.

Releasing (maintainers)

• This project uses GitHub Actions + Trusted Publishing.

Branch/PR flow

• Open PRs as usual. CI will build only (no publish).

• On merge to main, if the version is a pre-release (contains rc), CI publishes to TestPyPI.

Test release (TestPyPI)

• Bump version in sdx-client/pyproject.toml to an RC:

• version = "X.Y.ZrcN"

Merge the PR to main or run the workflow manually:

• Actions → “Build & Publish (TestPyPI → PyPI)” → Run workflow

• target: testpypi

Install for testing:

pip install --index-url https://test.pypi.org/simple/
--extra-index-url https://pypi.org/simple
sdxclient==X.Y.ZrcN

Production release (PyPI)

Bump version to a final:

• version = "X.Y.Z"

Tag and push the tag:

• git tag vX.Y.Z
• git push origin vX.Y.Z
• (Or run the workflow manually with target: pypi.)

Install from PyPI:

pip install sdxclient==X.Y.Z

Rules the workflow enforces

• TestPyPI requires version to contain rc.
• PyPI requires version without rc.
• Publishing uses OIDC (no API tokens).
• Manual run (optional)

From GitHub → Actions → “Build & Publish (TestPyPI → PyPI)” → Run workflow:

• target: testpypi (expects rc in version)
• target: pypi (expects final version)
• Pick the branch that contains your version bump.