tokei

Count lines of code

$ tokei

$ tokei [path/to/project]

$ tokei -e [vendor] -e [node_modules]

$ tokei --files

$ tokei -s [code]

$ tokei -o json

$ tokei -t [Rust,Python]

$ tokei --languages

tokei [path] [-e exclude] [-t types] [-s column] [-o format] [options]

tokei counts lines of code quickly and accurately. It identifies languages by file extension, categorizing lines as code, comments, or blank.
The tool is extremely fast, written in Rust with parallel file processing. Large codebases analyze in seconds. Results show per-language breakdown with totals.
Language detection supports over 150 languages with proper handling of multi-line comments, nested comments, and language-specific syntax.
File filtering respects .gitignore by default, excluding generated code and dependencies. Additional excludes target specific directories or patterns.
Individual file mode (--files) shows statistics per file rather than aggregated, useful for finding verbose files or tracking specific components.
Machine-readable output (JSON, YAML) enables integration with other tools, CI/CD pipelines, or tracking metrics over time.

-e, --exclude PATTERN

Exclude files/directories matching pattern.

Count only specified languages.

Sort by: files, lines, blanks, code, comments.

Output format: json, yaml, cbor.

Show individual file statistics.

Include hidden files.

Don't use .gitignore.

Compact output.

Print supported languages.

Set output width.

Verbose output.

Lines of code is an imperfect metric. Generated code may skew results. Multi-language files counted once. Extension-based detection may misidentify some files. Binary files are skipped.

tokei was created by Erin Power (Aaronepower) around 2016. The Japanese name means "clock/statistics." Written in Rust for performance, it became a popular alternative to cloc and other line counting tools. The project emphasizes speed and accuracy across many programming languages.

cloc(1), scc(1), wc(1), loc(1)