[karlhillx]

GitHub Issue: #374

Description

import ncompare no longer loads xarray (with pandas) or openpyxl at import
time. xarray moved into get_root_dims, openpyxl into the Excel output path.
Module-level h5py and netCDF4 imports are untouched, so the #363
import-order question is unaffected.

Local test steps

• uv run pytest: 26 passed, 2 integration deselected (Python 3.11, macOS)
• uv run ruff check ncompare: clean
• python -X importtime -c "import ncompare": 214 ms to 58 ms warm;
  fixture CLI run 0.47 s to 0.23 s
• New regression test asserts xarray and openpyxl stay out of sys.modules
  after import ncompare (runs in a fresh interpreter, since the suite
  imports xarray)

Overview of integration done

Golden-file tests cover text, CSV, and XLSX output, so both deferred paths
run end to end and pass.

PR Acceptance Checklist

• Unit tests added/updated and passing.
• Integration testing (CI runs the Earthdata suite)
• CHANGELOG.md updated (entries added at release time)
• Documentation updated (if needed).

---

📚 Documentation preview 📚: https://ncompare--375.org.readthedocs.build/en/375/

[Copilot]

Copilot review overview

🟢 Approval recommended

No unresolved review issues were identified.

Review effort: Lite
Findings: None

What changed in this PR

Defers heavy xarray and openpyxl imports to reduce CLI startup time.

Changes:

• Lazily imports xarray for dimension inspection.
• Lazily imports openpyxl for Excel output.
• Adds a subprocess regression test for import behavior.

File	Description
tests/​test_imports.py	Verifies deferred imports remain unloaded.
ncompare/​printing.py	Defers Excel dependencies.
ncompare/​getters.py	Defers the xarray dependency.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.