fix(cli): clean errors for malformed inputs and honest dbt-gate exit codes

[kevincostner17]

Summary

• CLIs print tracebacks for malformed config/suite/manifest files; validate exits 1, not 2 #289: malformed input files printed tracebacks.
  • freshdata clean --config now reports these as a one-line freshdata: error: ... naming the file, with exit 1 (the existing convention for bad input files):
    • invalid YAML or JSON
    • a top level that isn't an object
    • a clean/enterprise section that isn't an object
    • unknown keys (a typo in a masking rule or clean option)
  • freshdata validate --suite/--contract with a non-object file now exits 2 ("could not load rules") instead of 1 ("validation failed"). ValidationSuite.from_dict rejects non-mappings with ValueError.
  • dbt-gate now catches (OSError, ValueError). Invalid JSON, a directory path, or a file that isn't a manifest now gives a one-line dbt-gate: error: ... with exit 1, like a missing manifest.
• freshdata clean exits 1 after writing output when stdout isn't UTF-8 (summary prints '→') #295: exit 1 when stdout isn't UTF-8. New _safe_print retries with errors="replace" on UnicodeEncodeError. It's used for the clean summary and notes and the validate verdict, so a cp1252/ascii stdout no longer turns a passed gate (outputs already written) into exit 1.
• dbt-gate --fail exits 0 when the manifest has no models or is the wrong file #296: dbt-gate --fail passed when nothing was gated.
  • gate_manifest raises ValueError when the file has no nodes mapping (e.g. run_results.json).
  • all_passed is false when models_processed == 0.
  • dbt-gate prints a "no models were gated" note to stderr, and --fail exits 1.
• dbt-gate counts ephemeral models as failures, so --fail always fails when they exist #249: ephemeral models were counted as failures. Models with config.materialized == "ephemeral" or config.enabled == false are skipped. They're listed under a new "skipped" summary key and not counted in models_processed or failed_models.

Behaviour changes:

• The dbt-gate summary has a new skipped key.
• all_passed is false when nothing was gated.
• validate exits 2 for non-object rule files.

Tests

• New tests/test_cli_malformed_inputs.py:
  • invalid, non-object and typo'd YAML/JSON configs
  • non-object sections and masking entries
  • unknown clean options
  • validate exits 2 for non-object --suite and --contract files
  • ValidationSuite.from_dict rejects non-mappings
  • clean and validate on an ascii stdout: a passed gate exits 0, a failed gate still exits 1
• tests/test_integrations/test_dbt.py:
  • malformed manifests and a directory path give one-line errors
  • a file that isn't a manifest raises ValueError
  • a zero-model manifest doesn't pass, and --fail exits 1
  • ephemeral and disabled models are skipped, with --fail exiting 0 when the real tables pass

Verification

• ruff check .: all checks passed
• mypy src/freshdata: no issues found in 202 source files
• pytest (-m "not online and not large"), py3.12 / pandas 2.3.3: 4272 passed, 6 skipped
• pytest, py3.9 / pandas 1.5.3: 4268 passed, 10 skipped
• All four issue reproductions checked before and after the fix.

Closes #289
Closes #295
Closes #296
Closes #249

[coderabbitai]

Important

• 🔍 Trigger review

This repository does not receive automatic reviews because it has fewer than 10 stars.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: 802cfc84-5529-4605-bbeb-43782b28dd30

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[github-actions]

FreshData benchmark report — performance

• freshdata: ?
• python: ?
• platform: ?

fixture	n_rows	n_cols	p50 s	p95 s	peak MB	repair %	false-repair %	preserve %	trust	monotonic	export %

Authored-code reduction (Metric 6)