CI Integration

中文

This page provides CI-ready recipes with stable output, controlled cost, and explicit exit code behavior.

Recommended command templates

1) Lowest cost: duplicate files only

dup-code-check --json --stats --strict .

Good for a low-cost gate: permission/traversal errors or budget aborts will fail; otherwise it outputs JSON for downstream checks.

2) Full report: run all detectors

dup-code-check --json --stats --strict --report .

Control output size with --max-report-items:

dup-code-check --json --stats --strict --report --max-report-items 100 .

3) Multiple repos/roots: only cross-root duplicates

dup-code-check --json --stats --strict --report --cross-repo-only /repoA /repoB

Persist outputs as CI artifacts

Save results and stats separately (avoid mixing stdout/stderr):

dup-code-check --json --stats --report . >dup-code-check.result.json 2>dup-code-check.stats.txt

In text mode, --stats prints to stderr. In JSON mode, --stats merges scanStats into stdout JSON.

How to fail the CI?

dup-code-check fails mainly in two cases:

1. runtime errors: invalid args, root does not exist / is not a directory, scan failures
2. --strict: scan was incomplete (permission / relativize / traversal / budget abort)

If you also want to fail when duplicates are found, add a separate step to check the JSON output, e.g.:

• fileDuplicates.length > 0 → fail
• codeSpanDuplicates.length > 0 → fail

(This is policy; define thresholds based on your team’s tolerance.)

Cost control tips

In large repos, consider enabling at least one of:

• --ignore-dir (skip dependencies/build outputs)
• --max-file-size (skip huge files)
• --max-files / --max-total-bytes (set budgets)

Use carefully:

• --follow-symlinks (may expand scan scope)

Exit code quick reference

See the “Exit codes” section in CLI Usage.