jscpd vs Rev-dep

jscpd is a copy-paste detector for many languages. rev-dep detects duplication too, and for a JS/TS codebase it is both a faster and a more useful alternative.

More useful because of what it reports. jscpd finds matching token ranges; rev-dep finds repeated code blocks and JSX elements - units the language already recognises, which means every finding comes with its own refactor: a duplicated function body becomes a shared function, a duplicated element becomes a component. It also matches copies whose variables were renamed, which is where most real copy-and-adapt duplication lives.

It is faster, and it runs in the same pass as the rest of rev-dep's checks - so on a project that already runs them, duplication detection adds no second traversal and no second tool to the pipeline.

At a glance

	jscpd	Rev-dep
Primary focus	copy-paste detection	dependency hygiene + architecture, duplication included
Runtime	Node	Go - single parallel pass, alongside twelve other checks
Languages	many, via per-language tokenizers	JS / TS / JSX / TSX, plus Vue and Svelte
Unit reported	token range, as a line/column span	whole code block or JSX element
Renamed copies	not matched	matched with --blind-identifiers
Reporters	console, json, html, xml, sonar and more	console, json (with a published schema)
CI baseline	duplication percentage threshold	snapshot of the exact duplications acknowledged
Other checks	none	twelve more, in the same run

The difference that matters: what gets reported

Both tools find the same duplication in the same file. They describe it differently:

jscpd     components/Card.tsx [3:19 - 13:13]
          components/Panel.tsx [3:20 - 13:13]

rev-dep   JSX block, 7 lines, depth 3
          src/components/Card.tsx:4:3
          src/components/Panel.tsx:4:3

jscpd's range starts at column 19 of line 3 and ends at column 13 of line 13 - it begins mid-statement and ends mid-statement, because a token window has no reason to stop where a language construct does. rev-dep reports the <section> element itself.

The practical difference is what you do next. A token range has to be read at both sites before you know where the extractable unit actually begins. A block is the unit: a duplicated function body becomes a shared function, a duplicated JSX element becomes a component.

The trade runs the other way too. rev-dep will not report a copy-paste that starts halfway through one function and ends halfway through another; jscpd will.

Renamed copies

jscpd compares tokens including identifier names, so a copy whose variables were renamed is not a clone. Its --mode settings (strict, mild, weak) change how whitespace and comments are treated, not names.

On a fixture with three copies of the same request helper - one of them renamed - jscpd reports 2 clones in every mode. rev-dep reports the same 2 by default, and 3 with --blind-identifiers, which is where most real copy-and-adapt duplication lives.

Where jscpd may still fit

• Other languages. jscpd covers Python, Java, C#, Go, CSS and many more. rev-dep is JS/TS only.
• Reporter formats. jscpd ships html, xml, sonar and others. rev-dep prints human output and JSON.
• Whole-file percentage budgets. jscpd's threshold is a duplication percentage; rev-dep's baseline is the exact set of acknowledged duplications, which is a different way to work.

Which should you choose?

• Need duplication detection across a polyglot repository, or an HTML report? jscpd.
• Working in JS/TS? rev-dep - faster, and every finding is a unit you can actually extract, renamed copies included, checked alongside your other structural rules in a single run.

Learn more in How duplicated code detection works.