Scan Options (ScanOptions)

中文

The CLI and the Node.js API share the same scan options. CLI flags are mapped into a ScanOptions struct passed into the native core.

Defaults follow Rust ScanOptions::default(); --help also shows some defaults.

Directories & ignore rules

ignoreDirs / --ignore-dir

Ignores specific directory names (matches by path segment). Commonly used to skip dependencies and build outputs.

Default includes (partial):

• .git, node_modules, target, dist, build, out, .next, .turbo, .cache

Repeatable in CLI:

dup-code-check --ignore-dir vendor --ignore-dir .venv .

respectGitignore / --no-gitignore

Default true: respects .gitignore rules (and uses git to accelerate file collection when available).

Disable:

dup-code-check --no-gitignore .

Re-enable (mostly useful in scripts):

dup-code-check --gitignore .

Notes:

• even when .gitignore is disabled, ignoreDirs still applies
• when scanning inside a Git repo, ignore rules include .gitignore, .git/info/exclude, and global Git ignores

followSymlinks / --follow-symlinks

Default false (don’t follow symlinks). Enable to scan symlinked dirs/files:

dup-code-check --follow-symlinks .

In monorepos or build outputs with many symlinks, enable carefully to avoid exploding scan scope or cycles.

Scan budgets

Budgets help control scan cost, especially in CI.

maxFiles / --max-files

Stop scanning after reading/processing n files. When the limit is reached, the scan ends early and scanStats.skippedBudgetMaxFiles becomes non-zero.

With --strict, hitting maxFiles is treated as an “incomplete scan” and will fail.

maxTotalBytes / --max-total-bytes

Total bytes budget: if reading a file would make scannedBytes + fileSize > maxTotalBytes, that file is skipped and counted in scanStats.skippedBudgetMaxTotalBytes.

Unlike maxFiles (which stops scanning once the limit is reached), maxTotalBytes continues scanning but may skip many files.

maxFileSize / --max-file-size

Skips files larger than n bytes (default 10 MiB). Counted in scanStats.skippedTooLarge.

maxNormalizedChars / --max-normalized-chars

Stops scanning once the total stored normalized code characters would exceed n. When hit, the scan ends early and scanStats.skippedBudgetMaxNormalizedChars becomes non-zero.

Used by --code-spans and --report (text-based detectors).

With --strict, hitting maxNormalizedChars is treated as an “incomplete scan” and will fail.

maxTokens / --max-tokens

Stops scanning once the total stored tokens would exceed n (report mode). When hit, the scan ends early and scanStats.skippedBudgetMaxTokens becomes non-zero.

With --strict, hitting maxTokens is treated as an “incomplete scan” and will fail.

In --report mode, if maxNormalizedChars / maxTokens are unset, defaults are derived from maxTotalBytes to bound memory use.

Detector thresholds

minMatchLen / --min-match-len

Affects:

• --code-spans minimum normalized length
• --report codeSpanDuplicates
• --report lineSpanDuplicates filtering (prevents tiny line fragments from being treated as duplicates)

Default 50.

Must be >= 1. Core APIs reject 0 with an InvalidInput error.

minTokenLen / --min-token-len

Affects token/block detectors in report mode:

• tokenSpanDuplicates
• blockDuplicates
• astSubtreeDuplicates
• similarBlocksMinhash
• similarBlocksSimhash

Default 50.

Must be >= 1. Core APIs reject 0 with an InvalidInput error.

similarityThreshold / --similarity-threshold

Similarity detectors (MinHash/SimHash). Default 0.85 (range 0..1).

Core APIs validate this range and reject invalid values.

simhashMaxDistance / --simhash-max-distance

SimHash maximum Hamming distance (default 3, range 0..64).

Core APIs validate this range and reject invalid values.

Output controls (only for --report)

maxReportItems / --max-report-items

Maximum items per report section (default 200).

• larger values: more complete, but larger output and higher memory/time
• 0: outputs an empty report (fast way to “disable report”)

Cross-root only

crossRepoOnly / --cross-repo-only

When true, only output groups spanning >= 2 roots (for both file duplicates and span duplicates).