Duplication

necro scan detects copy-paste clones — duplicated code, even when the copy renamed its variables. Clones appear in a Duplication section (and a duplication array in --json).

Duplication (1 clone)
  43 tokens duplicated: src/a.ts:1-6, src/b.ts:1-6

How it works

necro tokenizes each file with the same tree-sitter parser the complexity detectors use — no jscpd or other external tool. Before matching, it normalizes the tokens:

identifiers collapse to a single ID token,
literal values collapse to LIT,
comments and whitespace are dropped,
keywords, operators, and punctuation keep their kind.

It then finds the longest normalized token sequences shared by two or more places. Because names and literals are erased, a block copied and then renamed (a Type-2 clone) still matches:

function add(a, b) { const sum = a + b; return sum; }   // ← reported as
function plus(x, y) { const total = x + y; return total; } //   one clone group

Each clone is reported once, as its maximal region — not as a swarm of overlapping fragments.

Tuning

A sequence must be at least minTokens long to count (default 50). Lower it to catch smaller clones, raise it to focus on large ones:

{
  "duplication": { "minTokens": 80 }
}