Auditing the scale and composition of a software project is essential when planning migrations, assessing technical debt, or calculating development costs. Running standard word-count utilities like wc -l counts every single newline character, including blank formatting lines, licensing headers, and documentation comments. To obtain accurate metrics on actual source code, developers rely on specialized code analyzers.
cloc (Count Lines of Code) is the industry-standard open-source command-line utility for inspecting software repositories. It automatically detects hundreds of programming languages, separates physical lines of executable source code from blank lines and comments, and can even process compressed archive files and Git commit diffs directly. While modern compiled utilities allow you to count lines of code with tokei or calculate code complexity with scc, cloc remains the most widely packaged and battle-tested standard across Linux distributions.
Installing cloc on Linux
Because cloc is written in self-contained Perl without complex external binary dependencies, it is available in official repositories across almost every Linux distribution.
Installing on Ubuntu, Debian, and Linux Mint
sudo apt update
sudo apt install cloc
Installing on Fedora, RHEL, and CentOS
sudo dnf install cloc
Installing on Arch Linux and Manjaro
sudo pacman -S cloc
Installing on openSUSE
sudo zypper install cloc
How to Use cloc in Linux Terminal
Using cloc is straightforward and requires minimal configuration for standard project directories.
1. Counting Lines in a Directory
To analyze all source files within your current working directory, run cloc followed by a period or directory path:
cloc .
The output generates an easy-to-read ASCII table broken down by programming language:
-------------------------------------------------------------------------------
Language files blank comment code
-------------------------------------------------------------------------------
TypeScript 142 2410 1890 14820
HTML 18 180 45 2310
CSS 12 210 80 1950
JSON 24 0 0 1240
Markdown 15 320 0 890
-------------------------------------------------------------------------------
SUM: 211 3120 2015 21210
-------------------------------------------------------------------------------
2. Excluding Vendor and Cache Folders
Modern projects typically contain external dependencies (such as node_modules, vendor, dist, or virtual environments) that artificially inflate line counts. Use the --exclude-dir flag to omit them:
cloc --exclude-dir=node_modules,dist,build,vendor .
3. Analyzing Specific File Types
If you only want metrics for specific programming languages, use the --include-lang flag:
cloc --include-lang=Python,SQL ./backend
4. Counting Code Inside Compressed Archives
A standout capability of cloc is its ability to inspect compressed archives without requiring manual extraction. You can pass tarballs or zip archives directly:
cloc linux-6.12.tar.xz
The utility temporarily unpacks the archive into memory, categorises the source files, and displays the report.
5. Comparing Code Diffs Between Git Commits
You can use cloc to audit how many lines of actual code (excluding comments) were added, deleted, or modified between two Git branches or commits:
cloc --git --diff main feature-branch
This report highlights the exact net code change introduced during a pull request or sprint cycle.
Exporting cloc Reports to JSON, CSV, and XML
If you are integrating cloc into continuous integration (CI/CD) pipelines or automated code-review dashboards, export the results in structured formats:
- Export to JSON:
cloc --json --out=code-report.json . - Export to CSV:
cloc --csv --out=code-report.csv . - Export to XML:
cloc --xml --out=code-report.xml .
cloc Command Options Quick Reference
| Command Option | Description |
|---|---|
--by-file |
Displays metrics for each individual file instead of aggregating by language |
--exclude-dir=DIR1,DIR2 |
Ignores specified directory names (e.g., test directories, dependencies) |
--include-lang=LANG |
Restricts analysis to specified language(s) |
--diff FILE1 FILE2 |
Calculates statistical differences between two directories or archives |
--show-lang |
Prints all recognized programming languages and file extensions |
Frequently Asked Questions
How does cloc distinguish comments from code?
cloc maintains extensive language grammar definition tables that recognize single-line comments (like // or #), multi-line docstrings (like /* ... */ or """ ... """), and embedded HTML/CSS tags, preventing comments from inflating physical code counts.
Can cloc handle custom file extensions?
Yes. If your team uses non-standard extensions (for example, mapping .inc to PHP), define the mapping using the --force-lang argument:
cloc --force-lang=PHP,inc .
Is cloc suitable for massive multi-gigabyte mono-repositories?
While cloc handles hundreds of thousands of files reliably, it runs as a single Perl process. For massive mono-repos where scanning speed is paramount, multithreaded alternatives like tokei or scc execute faster, while cloc excels at precision, language customization, and archive parsing.