code-gauge 10.1.0 → 10.2.1

package/dist/duplication.d.ts

@@ -48,7 +48,7 @@ export interface TokenRange {
     endLine: number;
 }
 /** A contiguous run of matched tokens; gapped (merged) duplicates carry several per occurrence. */
-interface TokenSegment {
+export interface TokenSegment {
     startTokenIndex: number;
     endTokenIndex: number;
 }
@@ -63,8 +63,21 @@ interface DuplicateCandidate {
     endLine: number;
 }
 export interface CountedOccurrence {
-    /** Matched token runs; more than one once gapped groups are merged. */
+    /**
+     * The token runs the occurrence covers, matched throughout except in a near-miss copy (see
+     * `matchedRuns`); more than one once gapped groups are merged.
+     */
     segments: TokenSegment[];
+    /**
+     * The token runs of a near-miss copy that its partners match, within its segments;
+     * absent for an exact or gapped copy, whose segments are matched throughout.
+     */
+    matchedRuns?: TokenSegment[];
+    /**
+     * The exact fragments coalesced into a near-miss copy: each line they reach is duplicated,
+     * whatever else of the copy shares the line.
+     */
+    exactSegments?: TokenSegment[];
     /**
      * Set on occurrences whose span another reported group already counts: a retained group's
      * occurrences that a partial gapped merge also paired into a merged group, and cross-file copies
@@ -158,6 +171,11 @@ export declare function lcsLength(a: Int32Array, b: Int32Array): number;
  * every `b` compared with it.
  */
 export declare function createLcsLengthCounter(a: Int32Array): (b: Int32Array) => number;
+/**
+ * Which symbols of each sequence a longest common subsequence pairs, by Hirschberg's divide and
+ * conquer over the bit-parallel recurrence, which alone yields lengths but no pairing.
+ */
+export declare function markLcs(a: Int32Array, b: Int32Array): [Uint8Array, Uint8Array];
 /**
  * Merges duplicate groups separated by a small token gap into one gapped (Type-3) clone group: a
  * copy edited in one spot splits into two exact groups whose occurrences sit side by side in the
@@ -190,9 +208,12 @@ export declare function mergeAdjacentGroups<T extends CountedOccurrence>(groups:
  * content — so no token span contributes to the count twice.
  */
 export declare function countRedundantFragments(group: CountedOccurrence[]): number;
-/** Adds the 1-based code lines the segment's matched tokens cover; shared with cross-file coverage. */
-export declare function collectSegmentLines(segment: {
-    startTokenIndex: number;
-    endTokenIndex: number;
-}, tokens: Token[], codeLineNumbers: Set<number> | undefined, duplicatedLines: Set<number>): void;
+/**
+ * The 1-based code lines the occurrence's matched tokens cover, given its segments and matched runs
+ * as indexes into `tokens`. A line of a near-miss copy counts when its partners match more than half
+ * of its tokens: a line the copy adds or rewrites is not duplicated, while a longest common
+ * subsequence still pairs a stray token of it (a bracket, a separator) with one of the partner.
+ * A line one of `exactSegments` reaches counts in any case.
+ */
+export declare function collectMatchedLines(segments: TokenSegment[], matchedRuns: TokenSegment[] | undefined, tokens: Token[], codeLineNumbers: Set<number> | undefined, exactSegments?: TokenSegment[]): Set<number>;
 export {};

package/dist/duplication.js

@@ -1,2 +1,2 @@
-import{clamp as e}from"@willbooster/shared-lib";const t={minTokens:40,maxGapTokens:30,minSimilarityPercent:70};function n(e){return{minTokens:r(e?.minTokens,t.minTokens),maxGapTokens:r(e?.maxGapTokens,t.maxGapTokens),minSimilarityPercent:r(e?.minSimilarityPercent,t.minSimilarityPercent)}}function r(t,n){return t===void 0||Number.isNaN(t)?n:e(Math.trunc(t),0,4294967295)}function i(e,t){return e*5>=t}function a(e){let t=new Int32Array(e.length+1);for(let[n,r]of e.entries())t[n+1]=(t[n]??0)+(r.literalHash===void 0?0:1);return t}function o(e,t,n){let r=[],i=[],a=[];for(let[t,n]of e.entries())for(let e of n.containers)i.push(t),a.push(e);let o=t=>e[i[t]??0],l=/* @__PURE__ */ new Map,u=a.map((e,n)=>c(o(n)?.tokens??[],e,t));for(let[e,t]of u.entries()){let n=i[e]??0;for(let[r,i]of t.windowKeysByStart.entries())for(let t of i){if(t===void 0)continue;let i=l.get(t);i?(i.count+=1,i.containerIndex!==e&&(i.containerIndex=-1),i.contextIndex!==n&&(i.contextIndex=-1),i.minStart=Math.min(i.minStart,r),i.maxStart=Math.max(i.maxStart,r)):l.set(t,{count:1,containerIndex:e,contextIndex:n,minStart:r,maxStart:r})}}let d=(e,t)=>{if(e===void 0)return!1;let r=l.get(e);return r===void 0||r.count<2?!1:n?r.contextIndex===-1:r.containerIndex===-1||r.maxStart-r.minStart>=t},p=e=>{let t=u[e.containerIndex]?.statementHashes??[],n=t[e.start];for(let r=e.start+1;r<e.start+e.length;r+=1)if(t[r]!==n)return!0;return!1},m=[];for(let[e,t]of u.entries())for(let[n,r]of t.windowKeysByStart.entries())for(let[i,a]of r.entries()){if(!d(a,i)||!p({containerIndex:e,start:n,length:i}))continue;let r=t.windowKeysByStart[n]?.[i+1],o=t.windowKeysByStart[n-1]?.[i+1];d(r,i+1)||d(o,i+1)||m.push({containerIndex:e,start:n,length:i})}let h=new Set(m.map(s)),g=m;for(;g.length>0;){let e=[];for(let t of g){let n=a[t.containerIndex],s=n?.[t.start],c=n?.[t.start+t.length-1],l=o(t.containerIndex);if(!s||!c||!l)continue;let u=`s:${_(l.tokens,l.literalCountPrefix,s.startTokenIndex,c.endTokenIndex)}`;r.push({candidate:f(u,s.startTokenIndex,c.endTokenIndex,s,c),contextIndex:i[t.containerIndex]??0}),e.push(t)}g=[];for(let t of e)for(let e of[t.start,t.start+1]){let n={containerIndex:t.containerIndex,start:e,length:t.length-1},r=u[t.containerIndex]?.windowKeysByStart[e]?.[n.length];!h.has(s(n))&&d(r,n.length)&&p(n)&&(h.add(s(n)),g.push(n))}}return r}function s(e){return`${e.containerIndex}:${e.start}:${e.length}`}function c(e,t,n){let r=t.map(t=>v(e,t.startTokenIndex,t.endTokenIndex)),i=[];for(let e=0;e<t.length;e+=1){let a=[],o=5381,s=0,c=Math.min(t.length,e+100);for(let i=e;i<c;i+=1){let c=t[i],l=r[i];if(!c||l===void 0)break;o=S(o,l),s+=c.endTokenIndex-c.startTokenIndex;let u=i-e+1;a[u]=u>=2&&s>=n?S(o,u):void 0}i.push(a)}return{windowKeysByStart:i,statementHashes:r}}function l(e,t){return u(e)(t)}function u(e){let t=e.length+31>>>5,n=/* @__PURE__ */ new Map;for(let[r,i]of e.entries()){let e=n.get(i);e||(e=new Uint32Array(t),n.set(i,e));let a=r>>>5;e[a]=(e[a]??0)|1<<(r&31)}let r=new Uint32Array(t);return e=>{r.fill(0);for(let i of e){let e=n.get(i),a=1,o=0;for(let n=0;n<t;n+=1){let t=r[n]??0,i=((e?.[n]??0)|t)>>>0,s=(t<<1|a)>>>0;a=t>>>31;let c=i-s-o;o=+(c<0),r[n]=i&~c}}let i=0;for(let e of r)i+=d(e);return i}}function d(e){let t=e-(e>>>1&1431655765);return t=(t&858993459)+(t>>>2&858993459),Math.imul(t+(t>>>4)&252645135,16843009)>>>24&255}function f(e,t,n,r,i){return{fingerprint:e,tokenCount:n-t,startTokenIndex:t,endTokenIndex:n,startIndex:r.startIndex,endIndex:i.endIndex,startLine:r.startLine,endLine:i.endLine}}const p=[],m=[];function h(e){let t=p[e];return t===void 0&&(t=b(`$${e}`),p[e]=t),t}function g(e){let t=m[e];return t===void 0&&(t=x(`$${e}`),m[e]=t),t}function _(e,t,n,r){let[a,o]=y(e,n,r,i((t[r]??0)-(t[n]??0),r-n));return`${a}:${o}:${r-n}`}function v(e,t,n){let[r,i]=y(e,t,n,!1);return r^Math.imul(i,31)}function y(e,t,n,r){let i=/* @__PURE__ */ new Map,a=5381,o=52711;for(let s=t;s<n;s+=1){let t=e[s];if(!t)continue;let n,c;if(t.kind===`id`){let e=i.get(t.text);e===void 0&&(e=i.size,i.set(t.text,e)),n=h(e),c=g(e)}else n=t.textHash,c=t.textHash2;a=Math.imul(a,31)+n|0,o=Math.imul(o,37)^c,r&&t.literalHash!==void 0&&t.literalHash2!==void 0&&(a=Math.imul(a,31)+t.literalHash|0,o=Math.imul(o,37)^t.literalHash2)}return[a,o]}function b(e){let t=5381;for(let n=0;n<e.length;n+=1)t=Math.imul(t,33)^e.charCodeAt(n);return t}function x(e){let t=-2128831035;for(let n=0;n<e.length;n+=1)t=Math.imul(t^e.charCodeAt(n),16777619);return t}function S(e,t){return Math.imul(e,31)+t}function C(e,t,n=()=>!0){if(t<=0||e.length<2)return e;e.sort(E);for(let r=!0;r;){r=!1;let i=w(e,t);for(let a=0;a<e.length&&!r;a+=1)for(let o of i[a]??[]){let i=e[a],s=e[o];if(!i||!s)continue;let c=D(i,s,t,n),l=c??D(s,i,t,n);if(!l)continue;let u=c?l.firstReplaced:l.secondReplaced,d=c?l.secondReplaced:l.firstReplaced;u&&d?(e[a]=l.merged,e.splice(o,1)):d?e[o]=l.merged:u?e[a]=l.merged:e.push(l.merged);for(let e of l.pairedRetained)e.spanCountedElsewhere=!0;e.sort(E),r=!0;break}}return e}function w(e,t){let n=[];for(let[t,r]of e.entries())for(let e of r)n.push({startTokenIndex:e.startTokenIndex,groupIndex:t});n.sort((e,t)=>e.startTokenIndex-t.startTokenIndex);let r=e.map(()=>/* @__PURE__ */ new Set);for(let[i,a]of e.entries())for(let e of a)for(let a=T(n,e.endTokenIndex);a<n.length&&(n[a]?.startTokenIndex??0)<=e.endTokenIndex+t;a+=1){let e=n[a]?.groupIndex??i;e!==i&&r[Math.min(e,i)]?.add(Math.max(e,i))}return r.map(e=>[...e].toSorted((e,t)=>e-t))}function T(e,t){let n=0,r=e.length;for(;n<r;){let i=n+r>>>1;(e[i]?.startTokenIndex??0)<t?n=i+1:r=i}return n}function E(e,t){let n=e[0],r=t[0];return(n?.startTokenIndex??0)-(r?.startTokenIndex??0)||(n?.endTokenIndex??0)-(r?.endTokenIndex??0)}function D(e,t,n,r){let[i,a]=O(e),[o,s]=O(t),c=[],l=0,u=-1;for(let e of o){for(;l<i.length;){let t=i[l];if(t&&t.endTokenIndex+n<e.startTokenIndex)l+=1;else break}let t=i[l];t&&t.endTokenIndex<=e.startTokenIndex&&t.startTokenIndex>=u&&(c.push([t,e]),u=e.endTokenIndex,l+=1)}let d=c.length===a,f=c.length===s;if(c.length<2||!d&&!f)return;let p=d&&!e.some(e=>e.nestedInLargerGroup),m=f&&!t.some(e=>e.nestedInLargerGroup),h=c.map(([e,t])=>({...e,spanCountedElsewhere:void 0,nestedInLargerGroup:void 0,segments:[...e.segments,...t.segments],tokenCount:e.tokenCount+t.tokenCount,endTokenIndex:t.endTokenIndex,endIndex:t.endIndex,endLine:t.endLine}));if(r(h))return{merged:h,firstReplaced:p,secondReplaced:m,pairedRetained:[...p?[]:c.map(([e])=>e),...m?[]:c.map(([,e])=>e)]}}function O(e){let t=[],n=0;for(let r of e)r.nestedInLargerGroup||(n+=1),!r.spanCountedElsewhere&&!r.nestedInLargerGroup&&t.push(r);return[t,n]}function k(e){let t=0,n=0,r=!1;for(let i of e){if(i.spanCountedElsewhere){r=!0;continue}t+=i.segments.length,n=Math.max(n,i.segments.length)}return r?t:t-n}function A(e,t,n,r){for(let i=e.startTokenIndex;i<e.endTokenIndex;i+=1){let e=t[i];for(let t=e?.startRow??0;t<=(e?.endRow??-1);t+=1)(!n||n.has(t+1))&&r.add(t+1)}}export{a as buildLiteralCountPrefix,A as collectSegmentLines,o as collectSequenceWindowCandidates,k as countRedundantFragments,u as createLcsLengthCounter,t as defaultDuplicationOptions,l as lcsLength,C as mergeAdjacentGroups,n as resolveDuplicationOptions};
+import{clamp as e}from"@willbooster/shared-lib";const t={minTokens:40,maxGapTokens:30,minSimilarityPercent:70};function n(e){return{minTokens:r(e?.minTokens,t.minTokens),maxGapTokens:r(e?.maxGapTokens,t.maxGapTokens),minSimilarityPercent:r(e?.minSimilarityPercent,t.minSimilarityPercent)}}function r(t,n){return t===void 0||Number.isNaN(t)?n:e(Math.trunc(t),0,4294967295)}function i(e,t){return e*5>=t}function a(e){let t=new Int32Array(e.length+1);for(let[n,r]of e.entries())t[n+1]=(t[n]??0)+(r.literalHash===void 0?0:1);return t}function o(e,t,n){let r=[],i=[],a=[];for(let[t,n]of e.entries())for(let e of n.containers)i.push(t),a.push(e);let o=t=>e[i[t]??0],l=/* @__PURE__ */ new Map,u=a.map((e,n)=>c(o(n)?.tokens??[],e,t));for(let[e,t]of u.entries()){let n=i[e]??0;for(let[r,i]of t.windowKeysByStart.entries())for(let t of i){if(t===void 0)continue;let i=l.get(t);i?(i.count+=1,i.containerIndex!==e&&(i.containerIndex=-1),i.contextIndex!==n&&(i.contextIndex=-1),i.minStart=Math.min(i.minStart,r),i.maxStart=Math.max(i.maxStart,r)):l.set(t,{count:1,containerIndex:e,contextIndex:n,minStart:r,maxStart:r})}}let d=(e,t)=>{if(e===void 0)return!1;let r=l.get(e);return r===void 0||r.count<2?!1:n?r.contextIndex===-1:r.containerIndex===-1||r.maxStart-r.minStart>=t},f=e=>{let t=u[e.containerIndex]?.statementHashes??[],n=t[e.start];for(let r=e.start+1;r<e.start+e.length;r+=1)if(t[r]!==n)return!0;return!1},p=[];for(let[e,t]of u.entries())for(let[n,r]of t.windowKeysByStart.entries())for(let[i,a]of r.entries()){if(!d(a,i)||!f({containerIndex:e,start:n,length:i}))continue;let r=t.windowKeysByStart[n]?.[i+1],o=t.windowKeysByStart[n-1]?.[i+1];d(r,i+1)||d(o,i+1)||p.push({containerIndex:e,start:n,length:i})}let m=new Set(p.map(s)),h=p;for(;h.length>0;){let e=[];for(let t of h){let n=a[t.containerIndex],s=n?.[t.start],c=n?.[t.start+t.length-1],l=o(t.containerIndex);if(!s||!c||!l)continue;let u=`s:${x(l.tokens,l.literalCountPrefix,s.startTokenIndex,c.endTokenIndex)}`;r.push({candidate:g(u,s.startTokenIndex,c.endTokenIndex,s,c),contextIndex:i[t.containerIndex]??0}),e.push(t)}h=[];for(let t of e)for(let e of[t.start,t.start+1]){let n={containerIndex:t.containerIndex,start:e,length:t.length-1},r=u[t.containerIndex]?.windowKeysByStart[e]?.[n.length];!m.has(s(n))&&d(r,n.length)&&f(n)&&(m.add(s(n)),h.push(n))}}return r}function s(e){return`${e.containerIndex}:${e.start}:${e.length}`}function c(e,t,n){let r=t.map(t=>S(e,t.startTokenIndex,t.endTokenIndex)),i=[];for(let e=0;e<t.length;e+=1){let a=[],o=5381,s=0,c=Math.min(t.length,e+100);for(let i=e;i<c;i+=1){let c=t[i],l=r[i];if(!c||l===void 0)break;o=E(o,l),s+=c.endTokenIndex-c.startTokenIndex;let u=i-e+1;a[u]=u>=2&&s>=n?E(o,u):void 0}i.push(a)}return{windowKeysByStart:i,statementHashes:r}}function l(e,t){return u(e)(t)}function u(e){let t=d(e);return e=>{let n=0;for(let r of t(e))n+=h(r);return n}}function d(e){let t=e.length+31>>>5,n=/* @__PURE__ */ new Map;for(let[r,i]of e.entries()){let e=n.get(i);e||(e=new Uint32Array(t),n.set(i,e));let a=r>>>5;e[a]=(e[a]??0)|1<<(r&31)}let r=new Uint32Array(t);return e=>{r.fill(0);for(let i of e){let e=n.get(i),a=1,o=0;for(let n=0;n<t;n+=1){let t=r[n]??0,i=((e?.[n]??0)|t)>>>0,s=(t<<1|a)>>>0;a=t>>>31;let c=i-s-o;o=+(c<0),r[n]=i&~c}}return r}}function f(e,t){let n=new Uint8Array(e.length),r=new Uint8Array(t.length);return p(e,t,n,r),[n,r]}function p(e,t,n,r){let i=0;for(;i<e.length&&i<t.length&&e[i]===t[i];)i+=1;let a=0;for(;a<e.length-i&&a<t.length-i&&e[e.length-1-a]===t[t.length-1-a];)a+=1;let o=e.length-a,s=t.length-a;n.fill(1,0,i).fill(1,o),r.fill(1,0,i).fill(1,s);let c=e.subarray(i,o),l=t.subarray(i,s),u=n.subarray(i,o),d=r.subarray(i,s);if(c.length===0||l.length===0)return;if(c.length===1){let e=l.indexOf(c[0]??0);e!==-1&&(u[0]=1,d[e]=1);return}let f=c.length>>>1,h=m(c.subarray(0,f),l),g=m(c.subarray(f).toReversed(),l.toReversed()),_=0,v=-1;for(let e=0;e<=l.length;e+=1){let t=(h[e]??0)+(g[l.length-e]??0);t>v&&(v=t,_=e)}p(c.subarray(0,f),l.subarray(0,_),u.subarray(0,f),d.subarray(0,_)),p(c.subarray(f),l.subarray(_),u.subarray(f),d.subarray(_))}function m(e,t){let n=d(t)(e),r=new Int32Array(t.length+1);for(let e=0;e<t.length;e+=1)r[e+1]=(r[e]??0)+((n[e>>>5]??0)>>>(e&31)&1);return r}function h(e){let t=e-(e>>>1&1431655765);return t=(t&858993459)+(t>>>2&858993459),Math.imul(t+(t>>>4)&252645135,16843009)>>>24&255}function g(e,t,n,r,i){return{fingerprint:e,tokenCount:n-t,startTokenIndex:t,endTokenIndex:n,startIndex:r.startIndex,endIndex:i.endIndex,startLine:r.startLine,endLine:i.endLine}}const _=[],v=[];function y(e){let t=_[e];return t===void 0&&(t=w(`$${e}`),_[e]=t),t}function b(e){let t=v[e];return t===void 0&&(t=T(`$${e}`),v[e]=t),t}function x(e,t,n,r){let[a,o]=C(e,n,r,i((t[r]??0)-(t[n]??0),r-n));return`${a}:${o}:${r-n}`}function S(e,t,n){let[r,i]=C(e,t,n,!1);return r^Math.imul(i,31)}function C(e,t,n,r){let i=/* @__PURE__ */ new Map,a=5381,o=52711;for(let s=t;s<n;s+=1){let t=e[s];if(!t)continue;let n,c;if(t.kind===`id`){let e=i.get(t.text);e===void 0&&(e=i.size,i.set(t.text,e)),n=y(e),c=b(e)}else n=t.textHash,c=t.textHash2;a=Math.imul(a,31)+n|0,o=Math.imul(o,37)^c,r&&t.literalHash!==void 0&&t.literalHash2!==void 0&&(a=Math.imul(a,31)+t.literalHash|0,o=Math.imul(o,37)^t.literalHash2)}return[a,o]}function w(e){let t=5381;for(let n=0;n<e.length;n+=1)t=Math.imul(t,33)^e.charCodeAt(n);return t}function T(e){let t=-2128831035;for(let n=0;n<e.length;n+=1)t=Math.imul(t^e.charCodeAt(n),16777619);return t}function E(e,t){return Math.imul(e,31)+t}function D(e,t,n=()=>!0){if(t<=0||e.length<2)return e;e.sort(A);for(let r=!0;r;){r=!1;let i=O(e,t);for(let a=0;a<e.length&&!r;a+=1)for(let o of i[a]??[]){let i=e[a],s=e[o];if(!i||!s)continue;let c=j(i,s,t,n),l=c??j(s,i,t,n);if(!l)continue;let u=c?l.firstReplaced:l.secondReplaced,d=c?l.secondReplaced:l.firstReplaced;u&&d?(e[a]=l.merged,e.splice(o,1)):d?e[o]=l.merged:u?e[a]=l.merged:e.push(l.merged);for(let e of l.pairedRetained)e.spanCountedElsewhere=!0;e.sort(A),r=!0;break}}return e}function O(e,t){let n=[];for(let[t,r]of e.entries())for(let e of r)n.push({startTokenIndex:e.startTokenIndex,groupIndex:t});n.sort((e,t)=>e.startTokenIndex-t.startTokenIndex);let r=e.map(()=>/* @__PURE__ */ new Set);for(let[i,a]of e.entries())for(let e of a)for(let a=k(n,e.endTokenIndex);a<n.length&&(n[a]?.startTokenIndex??0)<=e.endTokenIndex+t;a+=1){let e=n[a]?.groupIndex??i;e!==i&&r[Math.min(e,i)]?.add(Math.max(e,i))}return r.map(e=>[...e].toSorted((e,t)=>e-t))}function k(e,t){let n=0,r=e.length;for(;n<r;){let i=n+r>>>1;(e[i]?.startTokenIndex??0)<t?n=i+1:r=i}return n}function A(e,t){let n=e[0],r=t[0];return(n?.startTokenIndex??0)-(r?.startTokenIndex??0)||(n?.endTokenIndex??0)-(r?.endTokenIndex??0)}function j(e,t,n,r){let[i,a]=M(e),[o,s]=M(t),c=[],l=0,u=-1;for(let e of o){for(;l<i.length;){let t=i[l];if(t&&t.endTokenIndex+n<e.startTokenIndex)l+=1;else break}let t=i[l];t&&t.endTokenIndex<=e.startTokenIndex&&t.startTokenIndex>=u&&(c.push([t,e]),u=e.endTokenIndex,l+=1)}let d=c.length===a,f=c.length===s;if(c.length<2||!d&&!f)return;let p=d&&!e.some(e=>e.nestedInLargerGroup),m=f&&!t.some(e=>e.nestedInLargerGroup),h=c.map(([e,t])=>({...e,spanCountedElsewhere:void 0,nestedInLargerGroup:void 0,matchedRuns:void 0,segments:[...e.segments,...t.segments],tokenCount:e.tokenCount+t.tokenCount,endTokenIndex:t.endTokenIndex,endIndex:t.endIndex,endLine:t.endLine}));if(r(h))return{merged:h,firstReplaced:p,secondReplaced:m,pairedRetained:[...p?[]:c.map(([e])=>e),...m?[]:c.map(([,e])=>e)]}}function M(e){let t=[],n=0;for(let r of e)r.nestedInLargerGroup||(n+=1),!r.spanCountedElsewhere&&!r.nestedInLargerGroup&&t.push(r);return[t,n]}function N(e){let t=0,n=0,r=!1;for(let i of e){if(i.spanCountedElsewhere){r=!0;continue}t+=i.segments.length,n=Math.max(n,i.segments.length)}return r?t:t-n}function P(e,t,n,r,i=[]){let a=/* @__PURE__ */ new Set;for(let e of i)for(let t=e.startTokenIndex;t<e.endTokenIndex;t+=1){let e=n[t];for(let t=e?.startRow??0;t<=(e?.endRow??-1);t+=1)a.add(t+1)}let o=/* @__PURE__ */ new Map,s=/* @__PURE__ */ new Map,c=0;for(let r of e)for(let e=r.startTokenIndex;e<r.endTokenIndex;e+=1){for(;t&&(t[c]?.endTokenIndex??1/0)<=e;)c+=1;let r=!t||(t[c]?.startTokenIndex??1/0)<=e,i=n[e];for(let e=i?.startRow??0;e<=(i?.endRow??-1);e+=1)s.set(e+1,(s.get(e+1)??0)+1),o.set(e+1,(o.get(e+1)??0)+ +!!r)}let l=/* @__PURE__ */ new Set;for(let[e,t]of s)(!r||r.has(e))&&((o.get(e)??0)*2>t||a.has(e))&&l.add(e);return l}export{a as buildLiteralCountPrefix,P as collectMatchedLines,o as collectSequenceWindowCandidates,N as countRedundantFragments,u as createLcsLengthCounter,t as defaultDuplicationOptions,l as lcsLength,f as markLcs,D as mergeAdjacentGroups,n as resolveDuplicationOptions};
 //# sourceMappingURL=duplication.js.map

package/dist/duplication.js.map

@@ -1 +1 @@
-{"version":3,"file":"duplication.js","names":[],"sources":["../src/duplication.ts"],"sourcesContent":["import { clamp } from '@willbooster/shared-lib';\nimport type { DuplicationOptions } from './types.js';\n\n/**\n * Project-level duplication machinery operating on normalized token streams. Tokenization itself\n * (parsing, identifier anonymization, literal normalization) happens in the Rust addon, which\n * serializes each file's Token stream and statement structure; the helpers here match statement\n * windows across files, merge gap-adjacent groups, and count duplicated lines over that data.\n */\n\nexport const defaultDuplicationOptions: Required<DuplicationOptions> = {\n  minTokens: 40,\n  maxGapTokens: 30,\n  minSimilarityPercent: 70,\n};\n\n/**\n * Fills defaults for absent settings, applying the same normalization as the native boundary's\n * clampToU32 — NaN (e.g. `Number(unsetEnvVariable)`) counts as absent, and other values truncate\n * and clamp to [0, u32::MAX] — so the TypeScript half of cross-file matching cannot diverge from\n * the natively collected candidates on such input.\n */\nexport function resolveDuplicationOptions(options?: DuplicationOptions): Required<DuplicationOptions> {\n  return {\n    minTokens: resolveOption(options?.minTokens, defaultDuplicationOptions.minTokens),\n    maxGapTokens: resolveOption(options?.maxGapTokens, defaultDuplicationOptions.maxGapTokens),\n    minSimilarityPercent: resolveOption(options?.minSimilarityPercent, defaultDuplicationOptions.minSimilarityPercent),\n  };\n}\n\nfunction resolveOption(value: number | undefined, fallback: number): number {\n  return value === undefined || Number.isNaN(value) ? fallback : clamp(Math.trunc(value), 0, 0xFF_FF_FF_FF);\n}\n\n/** Minimum consecutive statements for a statement-sequence duplicate candidate. */\nconst minSequenceStatementCount = 2;\n/**\n * Caps the window length so statement-sequence enumeration stays linear in the statement count.\n * Heterogeneous clones longer than the cap are reported as capped windows (a deliberate\n * conservative undercount trading completeness for bounded discovery cost).\n */\nconst maxSequenceStatementCount = 100;\n\n/**\n * A region whose normalized tokens are at least 20% literal values is data-like (a lookup table, a\n * constant list, a value-mapping switch), not logic: literal values re-enter its fingerprint so\n * tables that merely share their shape stop counting as copy-paste. Compared in integer math\n * (5 * literals >= total) so the TypeScript and native sides cannot disagree on the boundary.\n */\nfunction isLiteralDense(literalCount: number, tokenCount: number): boolean {\n  return literalCount * 5 >= tokenCount;\n}\n\nexport interface Token {\n  /** Normalization target: identifiers to anonymize, literal kind tags, or the raw token text. */\n  kind: 'id' | 'text';\n  text: string;\n  /**\n   * Two INDEPENDENT hashes of `text` (djb2 and FNV-1a), precomputed so fingerprinting nested\n   * regions never re-hashes a token. Feeding the same per-token hash to both fingerprint\n   * accumulators would collapse the key to 32 effective bits: one djb2 collision between two\n   * token texts would then equate whole regions.\n   */\n  textHash: number;\n  textHash2: number;\n  /** Hash pair of a value-carrying literal's value, folded into data-like region fingerprints. */\n  literalHash?: number;\n  literalHash2?: number;\n  /**\n   * True for verbatim-kept NAMES (member/callee/type names, named grammar leaves): together with\n   * value-carrying literals these are the content-bearing tokens the near-miss content gate\n   * counts. Keywords, operators, and punctuation come from unnamed nodes and stay false.\n   */\n  isName?: boolean;\n  /** 0-based source rows the token occupies, so line coverage counts only matched-token lines. */\n  startRow: number;\n  endRow: number;\n}\n\n/** A token span with its source position; plain data so cross-file matching can retain it. */\nexport interface TokenRange {\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A contiguous run of matched tokens; gapped (merged) duplicates carry several per occurrence. */\ninterface TokenSegment {\n  startTokenIndex: number;\n  endTokenIndex: number;\n}\n\ninterface DuplicateCandidate {\n  fingerprint: string;\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\nexport interface CountedOccurrence {\n  /** Matched token runs; more than one once gapped groups are merged. */\n  segments: TokenSegment[];\n  /**\n   * Set on occurrences whose span another reported group already counts: a retained group's\n   * occurrences that a partial gapped merge also paired into a merged group, and cross-file copies\n   * nested inside a larger group's region. Block counting must not count them again.\n   */\n  spanCountedElsewhere?: boolean;\n  /**\n   * Set on cross-file copies nested inside a larger group's region (they also set\n   * `spanCountedElsewhere`). They never pair in gapped merging, and they do not keep their group's\n   * standalone copies from merging, which they are not copies of. They do keep the merged group\n   * from taking their group's place, since only the original group reports the nesting.\n   */\n  nestedInLargerGroup?: boolean;\n  /** Sum of segment token counts (the gap tokens are not matched content). */\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A duplicate region found in one file, exported for cross-file matching by fingerprint. */\nexport interface CrossFileDuplicateCandidate {\n  /** Content key: equal fingerprints mean equal normalized token sequences (up to hash collision). */\n  fingerprint: string;\n  tokenCount: number;\n  /** Token positions within the owning file, for cross-file gapped (Type-3) merging. */\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/**\n * One file's contribution to cross-file clone detection: catalogued candidates plus the normalized\n * token stream and statement structure, so the project-level pass can match partial statement runs\n * (windows that a single file cannot know repeat elsewhere) and merge gap-adjacent groups.\n */\nexport interface CrossFileDuplicationFileData {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  /**\n   * The mutually disjoint blocks (at least `minTokens` long, not literal-dense) compared for\n   * cross-file near-miss (Type-3) clones. Optional for backward compatibility; without it, the\n   * file takes part in exact and gapped cross-file matching only.\n   */\n  nearMissBlocks?: TokenRange[];\n  /**\n   * 1-based lines that are neither blank nor comment-only, so cross-file line coverage counts only\n   * code lines (blank rows inside multi-row tokens such as template literals carry no content).\n   * Optional for backward compatibility; without it, every matched-token row counts.\n   */\n  codeLineNumbers?: Set<number>;\n}\n\n/** literalCountPrefix[i] = value-carrying literal tokens in tokens[0..i), for O(1) density checks. */\nexport function buildLiteralCountPrefix(tokens: Token[]): Int32Array {\n  const prefix = new Int32Array(tokens.length + 1);\n  for (const [index, token] of tokens.entries()) {\n    prefix[index + 1] = (prefix[index] ?? 0) + (token.literalHash === undefined ? 0 : 1);\n  }\n  return prefix;\n}\n\ninterface WindowOccurrences {\n  count: number;\n  /** -1 once occurrences span more than one container. */\n  containerIndex: number;\n  /** -1 once occurrences span more than one context (file). */\n  contextIndex: number;\n  minStart: number;\n  maxStart: number;\n}\n\ninterface SequenceWindow {\n  containerIndex: number;\n  start: number;\n  length: number;\n}\n\n/** One file's token stream and statement containers, as a window-matching context. */\nexport interface SequenceWindowContext {\n  tokens: Token[];\n  literalCountPrefix: Int32Array;\n  containers: TokenRange[][];\n}\n\nexport interface ContextualSequenceCandidate {\n  candidate: DuplicateCandidate;\n  contextIndex: number;\n}\n\n/**\n * Enumerates runs of consecutive sibling statements over one or more contexts (files). Every\n * container statement participates; only the window length is capped, so enumeration stays linear\n * in the statement count. Windows are grouped by a cheap rolling hash of per-statement\n * fingerprints, and only locally maximal repeated windows — those whose one-statement extensions\n * stop repeating — become candidates with an exact (window-consistent) fingerprint. Without the\n * maximality filter a degenerate file of near-identical statements would fingerprint every\n * sub-window of every repeated region. With `requireMultipleContexts` a window only counts as\n * repeated when its occurrences span at least two contexts (CPD-style cross-file matching): a\n * repeat confined to one file is that file's own concern, and emitting it here would flood the\n * project-level selection with unusable single-file groups.\n */\nexport function collectSequenceWindowCandidates(\n  contexts: SequenceWindowContext[],\n  minTokens: number,\n  requireMultipleContexts: boolean\n): ContextualSequenceCandidate[] {\n  const candidates: ContextualSequenceCandidate[] = [];\n  const contextIndexByContainer: number[] = [];\n  const containers: TokenRange[][] = [];\n  for (const [contextIndex, context] of contexts.entries()) {\n    for (const statements of context.containers) {\n      contextIndexByContainer.push(contextIndex);\n      containers.push(statements);\n    }\n  }\n  const contextAt = (containerIndex: number): SequenceWindowContext | undefined =>\n    contexts[contextIndexByContainer[containerIndex] ?? 0];\n  const occurrencesByWindowKey = new Map<number, WindowOccurrences>();\n  const containerWindows = containers.map((statements, containerIndex) =>\n    enumerateContainerWindows(contextAt(containerIndex)?.tokens ?? [], statements, minTokens)\n  );\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    const contextIndex = contextIndexByContainer[containerIndex] ?? 0;\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const windowKey of row) {\n        if (windowKey === undefined) {\n          continue;\n        }\n        const occurrences = occurrencesByWindowKey.get(windowKey);\n        if (occurrences) {\n          occurrences.count += 1;\n          if (occurrences.containerIndex !== containerIndex) {\n            occurrences.containerIndex = -1;\n          }\n          if (occurrences.contextIndex !== contextIndex) {\n            occurrences.contextIndex = -1;\n          }\n          occurrences.minStart = Math.min(occurrences.minStart, start);\n          occurrences.maxStart = Math.max(occurrences.maxStart, start);\n        } else {\n          occurrencesByWindowKey.set(windowKey, {\n            count: 1,\n            containerIndex,\n            contextIndex,\n            minStart: start,\n            maxStart: start,\n          });\n        }\n      }\n    }\n  }\n\n  // A window only \"repeats\" when two of its occurrences can coexist without overlapping: sliding\n  // matches inside a homogeneous run (start spread smaller than the window length) can never both\n  // be counted and must neither qualify a window nor dominate its sub-windows. Cross-context\n  // matching instead requires occurrences in two contexts, which coexist by construction.\n  const repeats = (windowKey: number | undefined, length: number): boolean => {\n    if (windowKey === undefined) {\n      return false;\n    }\n    const occurrences = occurrencesByWindowKey.get(windowKey);\n    if (occurrences === undefined || occurrences.count < 2) {\n      return false;\n    }\n    if (requireMultipleContexts) {\n      return occurrences.contextIndex === -1;\n    }\n    return occurrences.containerIndex === -1 || occurrences.maxStart - occurrences.minStart >= length;\n  };\n\n  // A window whose statements all share one normalized shape (sixteen `let x = 0;` declarations,\n  // a constant table) is a homogeneous preamble, not a copy-paste: requiring two distinct\n  // per-statement shapes keeps such runs out of duplicate groups and the duplication ratio.\n  const hasDistinctStatements = (window: SequenceWindow): boolean => {\n    const hashes = containerWindows[window.containerIndex]?.statementHashes ?? [];\n    const firstHash = hashes[window.start];\n    for (let index = window.start + 1; index < window.start + window.length; index += 1) {\n      if (hashes[index] !== firstHash) {\n        return true;\n      }\n    }\n    return false;\n  };\n\n  const maximalWindows: SequenceWindow[] = [];\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const [length, windowKey] of row.entries()) {\n        if (!repeats(windowKey, length) || !hasDistinctStatements({ containerIndex, start, length })) {\n          continue;\n        }\n        // Dominated windows are skipped: the one-statement extension also repeats, so a larger\n        // candidate covering this window exists.\n        const extendedRight = windows.windowKeysByStart[start]?.[length + 1];\n        const extendedLeft = windows.windowKeysByStart[start - 1]?.[length + 1];\n        if (repeats(extendedRight, length + 1) || repeats(extendedLeft, length + 1)) {\n          continue;\n        }\n        maximalWindows.push({ containerIndex, start, length });\n      }\n    }\n  }\n\n  // The rolling hash anonymizes identifiers per statement, so a window can look repeated coarsely\n  // while its exact (window-consistent) fingerprints differ, and a longer window's match can\n  // dominate sub-windows that other copies still need (three copies where only two extend one\n  // statement further). Every emitted window therefore exposes its repeating, unvisited\n  // sub-windows; `visited` bounds the worklist and lengths strictly decrease, so it terminates.\n  const visited = new Set(maximalWindows.map(windowId));\n  let frontier = maximalWindows;\n  while (frontier.length > 0) {\n    const emitted: SequenceWindow[] = [];\n    for (const window of frontier) {\n      const statements = containers[window.containerIndex];\n      const first = statements?.[window.start];\n      const last = statements?.[window.start + window.length - 1];\n      const context = contextAt(window.containerIndex);\n      if (!first || !last || !context) {\n        continue;\n      }\n      const fingerprint = `s:${fingerprintKey(context.tokens, context.literalCountPrefix, first.startTokenIndex, last.endTokenIndex)}`;\n      candidates.push({\n        candidate: toCandidate(fingerprint, first.startTokenIndex, last.endTokenIndex, first, last),\n        contextIndex: contextIndexByContainer[window.containerIndex] ?? 0,\n      });\n      emitted.push(window);\n    }\n    frontier = [];\n    for (const window of emitted) {\n      for (const start of [window.start, window.start + 1]) {\n        const subWindow = { containerIndex: window.containerIndex, start, length: window.length - 1 };\n        const subWindowKey = containerWindows[window.containerIndex]?.windowKeysByStart[start]?.[subWindow.length];\n        if (\n          visited.has(windowId(subWindow)) ||\n          !repeats(subWindowKey, subWindow.length) ||\n          !hasDistinctStatements(subWindow)\n        ) {\n          continue;\n        }\n        visited.add(windowId(subWindow));\n        frontier.push(subWindow);\n      }\n    }\n  }\n  return candidates;\n}\n\nfunction windowId(window: SequenceWindow): string {\n  return `${window.containerIndex}:${window.start}:${window.length}`;\n}\n\ninterface ContainerWindows {\n  /** windowKeysByStart[start][length] is the rolling-hash key of the window, or undefined if it is below the size thresholds. */\n  windowKeysByStart: (number | undefined)[][];\n  /** Per-statement fingerprint hashes, for the distinct-shape requirement on windows. */\n  statementHashes: number[];\n}\n\nfunction enumerateContainerWindows(tokens: Token[], statements: TokenRange[], minTokens: number): ContainerWindows {\n  const statementHashes = statements.map((statement) =>\n    fingerprintHash(tokens, statement.startTokenIndex, statement.endTokenIndex)\n  );\n  const windowKeysByStart: (number | undefined)[][] = [];\n  for (let start = 0; start < statements.length; start += 1) {\n    const row: (number | undefined)[] = [];\n    let hash = 5381;\n    let tokenCount = 0;\n    const maxEnd = Math.min(statements.length, start + maxSequenceStatementCount);\n    for (let end = start; end < maxEnd; end += 1) {\n      const statement = statements[end];\n      const statementHash = statementHashes[end];\n      if (!statement || statementHash === undefined) {\n        break;\n      }\n      hash = combineHashes(hash, statementHash);\n      tokenCount += statement.endTokenIndex - statement.startTokenIndex;\n      const statementCount = end - start + 1;\n      row[statementCount] =\n        statementCount >= minSequenceStatementCount && tokenCount >= minTokens\n          ? combineHashes(hash, statementCount)\n          : undefined;\n    }\n    windowKeysByStart.push(row);\n  }\n  return { windowKeysByStart, statementHashes };\n}\n\n/**\n * Longest-common-subsequence LENGTH of two symbol sequences via the Allison–Dix bit-parallel\n * recurrence (O(|a|/32 · |b|) words): per symbol of `b`, `x = match | v` and\n * `v = x & ~(x - ((v << 1) | 1))` over multi-word bit vectors; the set bits of `v` count the LCS.\n */\nexport function lcsLength(a: Int32Array, b: Int32Array): number {\n  return createLcsLengthCounter(a)(b);\n}\n\n/**\n * lcsLength against a fixed `a`, whose per-symbol position masks are built once and reused for\n * every `b` compared with it.\n */\nexport function createLcsLengthCounter(a: Int32Array): (b: Int32Array) => number {\n  const wordCount = (a.length + 31) >>> 5;\n  const positionMasks = new Map<number, Uint32Array>();\n  for (const [index, symbol] of a.entries()) {\n    let mask = positionMasks.get(symbol);\n    if (!mask) {\n      mask = new Uint32Array(wordCount);\n      positionMasks.set(symbol, mask);\n    }\n    const word = index >>> 5;\n    // The Uint32Array store wraps the signed int32 bit pattern to unsigned. The `?? 0` guards in\n    // this function are required by noUncheckedIndexedAccess (typed-array reads type as\n    // `number | undefined`), not redundancy: every index is in bounds.\n    mask[word] = (mask[word] ?? 0) | (1 << (index & 31));\n  }\n\n  const v = new Uint32Array(wordCount);\n  return (b) => {\n    v.fill(0);\n    for (const symbol of b) {\n      const matchMask = positionMasks.get(symbol);\n      // `(v << 1) | 1` shifts a carry bit across words; subtraction borrows across words.\n      let shiftCarry = 1;\n      let borrow = 0;\n      for (let word = 0; word < wordCount; word += 1) {\n        const previous = v[word] ?? 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- `>>> 0` reinterprets the signed int32 bit pattern as unsigned so the borrow subtraction below compares magnitudes; Math.trunc would keep it negative.\n        const x = ((matchMask?.[word] ?? 0) | previous) >>> 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- same unsigned reinterpretation as `x`.\n        const shifted = ((previous << 1) | shiftCarry) >>> 0;\n        shiftCarry = previous >>> 31;\n        const difference = x - shifted - borrow;\n        borrow = difference < 0 ? 1 : 0;\n        // The Uint32Array store wraps the signed int32 bit pattern to unsigned.\n        v[word] = x & ~difference;\n      }\n    }\n\n    let length = 0;\n    for (const word of v) {\n      length += popCount(word);\n    }\n    return length;\n  };\n}\n\nfunction popCount(value: number): number {\n  let count = value - ((value >>> 1) & 0x55_55_55_55);\n  count = (count & 0x33_33_33_33) + ((count >>> 2) & 0x33_33_33_33);\n  return (Math.imul((count + (count >>> 4)) & 0x0F_0F_0F_0F, 0x01_01_01_01) >>> 24) & 0xFF;\n}\n\nfunction toCandidate(\n  fingerprint: string,\n  startTokenIndex: number,\n  endTokenIndex: number,\n  first: TokenRange,\n  last: TokenRange\n): DuplicateCandidate {\n  return {\n    fingerprint,\n    tokenCount: endTokenIndex - startTokenIndex,\n    startTokenIndex,\n    endTokenIndex,\n    startIndex: first.startIndex,\n    endIndex: last.endIndex,\n    startLine: first.startLine,\n    endLine: last.endLine,\n  };\n}\n\n/** Caches of hashText/hashText2 over '$0', '$1', ... so anonymized identifiers hash without allocating. */\nconst anonymizedIndexHashes: number[] = [];\nconst anonymizedIndexHashes2: number[] = [];\n\nfunction anonymizedIndexHash(index: number): number {\n  let hash = anonymizedIndexHashes[index];\n  if (hash === undefined) {\n    hash = hashText(`$${index}`);\n    anonymizedIndexHashes[index] = hash;\n  }\n  return hash;\n}\n\nfunction anonymizedIndexHash2(index: number): number {\n  let hash = anonymizedIndexHashes2[index];\n  if (hash === undefined) {\n    hash = hashText2(`$${index}`);\n    anonymizedIndexHashes2[index] = hash;\n  }\n  return hash;\n}\n\n/**\n * Content key of a token range: two independent 32-bit hashes over the normalized token sequence\n * (identifiers anonymized consistently by first-occurrence order) plus the token count. Regions\n * with equal keys are treated as equal content; a collision would need both 32-bit hashes and the\n * length to coincide, which is negligible for a metrics report. The format and arithmetic match\n * fingerprint_key in native/src/duplication.rs exactly, so window candidates fingerprinted here\n * group together with the per-file candidates the addon catalogues.\n */\nfunction fingerprintKey(\n  tokens: Token[],\n  literalCountPrefix: Int32Array,\n  startTokenIndex: number,\n  endTokenIndex: number\n): string {\n  const literalCount = (literalCountPrefix[endTokenIndex] ?? 0) - (literalCountPrefix[startTokenIndex] ?? 0);\n  const literalDense = isLiteralDense(literalCount, endTokenIndex - startTokenIndex);\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, literalDense);\n  return `${primary}:${secondary}:${endTokenIndex - startTokenIndex}`;\n}\n\n/**\n * A single 32-bit summary of a range for the coarse rolling-hash phase. Deliberately\n * density-agnostic: density is a property of the final candidate REGION, and folding literal\n * values into per-statement hashes would make a dense statement inside a logic-heavy window\n * (`const weights = [1, 2, 3];`) block the window from ever being enumerated. The coarse phase\n * over-approximates on shape alone; the exact region fingerprint still applies the density rule.\n */\nfunction fingerprintHash(tokens: Token[], startTokenIndex: number, endTokenIndex: number): number {\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, false);\n  // XOR already coerces to int32, matching the native side's i32 arithmetic.\n  return primary ^ Math.imul(secondary, 31);\n}\n\nfunction fingerprintHashPair(\n  tokens: Token[],\n  startTokenIndex: number,\n  endTokenIndex: number,\n  foldLiteralValues: boolean\n): [number, number] {\n  const indexByIdentifier = new Map<string, number>();\n  let primary = 5381;\n  let secondary = 52_711;\n  for (let index = startTokenIndex; index < endTokenIndex; index += 1) {\n    const token = tokens[index];\n    if (!token) {\n      continue;\n    }\n    // Each accumulator consumes its own independent per-token hash: sharing one would collapse\n    // the key to 32 effective bits (a single djb2 collision would equate whole regions).\n    let part: number;\n    let part2: number;\n    if (token.kind === 'id') {\n      let identifierIndex = indexByIdentifier.get(token.text);\n      if (identifierIndex === undefined) {\n        identifierIndex = indexByIdentifier.size;\n        indexByIdentifier.set(token.text, identifierIndex);\n      }\n      part = anonymizedIndexHash(identifierIndex);\n      part2 = anonymizedIndexHash2(identifierIndex);\n    } else {\n      part = token.textHash;\n      part2 = token.textHash2;\n    }\n    // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n    primary = (Math.imul(primary, 31) + part) | 0;\n    secondary = Math.imul(secondary, 37) ^ part2;\n    if (foldLiteralValues && token.literalHash !== undefined && token.literalHash2 !== undefined) {\n      // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n      primary = (Math.imul(primary, 31) + token.literalHash) | 0;\n      secondary = Math.imul(secondary, 37) ^ token.literalHash2;\n    }\n  }\n  return [primary, secondary];\n}\n\n/** djb2-style hash; XOR keeps the value in signed 32-bit range, which is fine for a grouping key. */\nfunction hashText(text: string): number {\n  let hash = 5381;\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- djb2 hashes UTF-16 code units; codePointAt would hash surrogate pairs twice (full code point, then the lone low surrogate).\n    hash = Math.imul(hash, 33) ^ text.charCodeAt(index);\n  }\n  return hash;\n}\n\n/** FNV-1a over UTF-16 code units: independent of hashText so the two accumulators never share input. */\nfunction hashText2(text: string): number {\n  let hash = -2_128_831_035; // 2166136261 as int32 (the FNV-1a offset basis)\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- hashes UTF-16 code units like hashText.\n    hash = Math.imul(hash ^ text.charCodeAt(index), 16_777_619);\n  }\n  return hash;\n}\n\nfunction combineHashes(hash: number, value: number): number {\n  return Math.imul(hash, 31) + value;\n}\n\n/**\n * Merges duplicate groups separated by a small token gap into one gapped (Type-3) clone group: a\n * copy edited in one spot splits into two exact groups whose occurrences sit side by side in the\n * same order. Occurrences are paired greedily in source order; a merge happens when the pairing\n * fully pairs at least one group with at least two pairs. Equal-cardinality groups whose\n * occurrences all pair merge into one group as before. When cardinalities differ (a fragment also\n * occurs standalone: prefix ×3, suffix ×2), the fully-paired group is subsumed into the merged\n * gapped group while the other group is RETAINED with ALL its occurrences: dropping the leftover\n * would lose duplicated-line coverage, and reporting it alone would make a single-occurrence group\n * (contradicting duplicateBlockGroupCount's \"appears more than once\" meaning). Line coverage\n * unions ranges, so the overlap between the retained exact group and the merged group is harmless.\n * Merging repeats to a fixpoint so a clone edited in several spots still reassembles; it\n * terminates because a merge either removes the input groups it replaces or marks their paired\n * occurrences as counted elsewhere, and only unmarked occurrences of remaining groups pair, so the\n * number of pairable occurrences strictly decreases with every merge. Gap tokens are not matched\n * content: line coverage and sizes count only the matched segments. Generic so cross-file merging\n * can thread file identity through occurrences.\n */\nexport function mergeAdjacentGroups<T extends CountedOccurrence>(\n  groups: T[][],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean = () => true\n): T[][] {\n  if (maxGapTokens <= 0 || groups.length < 2) {\n    return groups;\n  }\n  // Deterministic processing order (mirrored by the native side): by first occurrence position.\n  groups.sort(compareGroups);\n  for (let restart = true; restart;) {\n    restart = false;\n    const partnersByGroup = collectGapAdjacentPartners(groups, maxGapTokens);\n    for (let leftIndex = 0; leftIndex < groups.length && !restart; leftIndex += 1) {\n      for (const rightIndex of partnersByGroup[leftIndex] ?? []) {\n        const left = groups[leftIndex];\n        const right = groups[rightIndex];\n        if (!left || !right) {\n          continue;\n        }\n        const forward = mergeGroups(left, right, maxGapTokens, isReportableGroup);\n        const result = forward ?? mergeGroups(right, left, maxGapTokens, isReportableGroup);\n        if (!result) {\n          continue;\n        }\n        const leftReplaced = forward ? result.firstReplaced : result.secondReplaced;\n        const rightReplaced = forward ? result.secondReplaced : result.firstReplaced;\n        if (leftReplaced && rightReplaced) {\n          groups[leftIndex] = result.merged;\n          groups.splice(rightIndex, 1);\n        } else if (rightReplaced) {\n          groups[rightIndex] = result.merged;\n        } else if (leftReplaced) {\n          groups[leftIndex] = result.merged;\n        } else {\n          // Both groups stay (each is only partly paired, or reports nested copies of its own), so\n          // the merged group joins them instead of taking a place.\n          groups.push(result.merged);\n        }\n        // A group that stays keeps ALL its occurrences (line coverage must not shrink, and a\n        // reported group must keep >= 2 occurrences), so its paired occurrences now also live\n        // inside the merged group's occurrences: mark them so duplicateBlockCount counts each\n        // token span once, and so the same pair cannot merge again.\n        for (const occurrence of result.pairedRetained) {\n          occurrence.spanCountedElsewhere = true;\n        }\n        groups.sort(compareGroups);\n        restart = true;\n        break;\n      }\n    }\n  }\n  return groups;\n}\n\n/**\n * Per group index, the ascending indexes of later groups that mergeGroups can pair with it in\n * either direction: an occurrence of one starts within `maxGapTokens` after an occurrence of the\n * other ends. Every other pair fails mergeGroups, so skipping it leaves the fixpoint unchanged\n * while each restart stops costing a comparison per pair of groups.\n */\nfunction collectGapAdjacentPartners(groups: CountedOccurrence[][], maxGapTokens: number): number[][] {\n  const starts: { startTokenIndex: number; groupIndex: number }[] = [];\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      starts.push({ startTokenIndex: occurrence.startTokenIndex, groupIndex });\n    }\n  }\n  starts.sort((left, right) => left.startTokenIndex - right.startTokenIndex);\n  const partners = groups.map(() => new Set<number>());\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      for (\n        let index = lowerBoundByStart(starts, occurrence.endTokenIndex);\n        index < starts.length && (starts[index]?.startTokenIndex ?? 0) <= occurrence.endTokenIndex + maxGapTokens;\n        index += 1\n      ) {\n        const other = starts[index]?.groupIndex ?? groupIndex;\n        if (other !== groupIndex) {\n          partners[Math.min(other, groupIndex)]?.add(Math.max(other, groupIndex));\n        }\n      }\n    }\n  }\n  return partners.map((set) => [...set].toSorted((left, right) => left - right));\n}\n\nfunction lowerBoundByStart(sorted: { startTokenIndex: number }[], target: number): number {\n  let low = 0;\n  let high = sorted.length;\n  while (low < high) {\n    const middle = (low + high) >>> 1;\n    if ((sorted[middle]?.startTokenIndex ?? 0) < target) {\n      low = middle + 1;\n    } else {\n      high = middle;\n    }\n  }\n  return low;\n}\n\nfunction compareGroups(left: CountedOccurrence[], right: CountedOccurrence[]): number {\n  const leftFirst = left[0];\n  const rightFirst = right[0];\n  return (\n    (leftFirst?.startTokenIndex ?? 0) - (rightFirst?.startTokenIndex ?? 0) ||\n    (leftFirst?.endTokenIndex ?? 0) - (rightFirst?.endTokenIndex ?? 0)\n  );\n}\n\ninterface MergeResult<T> {\n  merged: T[];\n  /**\n   * Whether the merged group takes the respective input group's place: its standalone occurrences\n   * were all paired and it holds no nested copies that only it can report.\n   */\n  firstReplaced: boolean;\n  secondReplaced: boolean;\n  /** The occurrences of groups that stay, which the merged group's spans now also cover. */\n  pairedRetained: T[];\n}\n\n/**\n * Pairs `second` occurrences with gap-preceding `first` occurrences, greedily in source order:\n * each trailing occurrence takes the earliest unused leading occurrence within the gap, and a\n * pair's leading must start at or after the previous pair's trailing end so merged spans never\n * overlap. A merge needs at least two pairs (a merged group must still mean \"appears more than\n * once\") and must fully consume at least one group; for equal cardinalities this reduces to the\n * strict all-pairs merge, so pre-partial-merge behavior is unchanged there.\n */\nfunction mergeGroups<T extends CountedOccurrence>(\n  first: T[],\n  second: T[],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean\n): MergeResult<T> | undefined {\n  // Occurrences a previous partial merge already paired into a merged group must not pair again:\n  // their spans already live inside that merged group, so re-pairing them would assemble a second,\n  // competing merged group instead of letting the existing merged group extend (and would count\n  // the same span twice). Consumption is still judged against the FULL group, so a group holding\n  // shared occurrences is never subsumed away; nested copies, which the merged span would not\n  // cover anyway, are left out of that judgment so they cannot veto a merge of the standalone\n  // copies.\n  const [leadings, firstLength] = pairableOccurrences(first);\n  const [trailings, secondLength] = pairableOccurrences(second);\n  const pairs: [T, T][] = [];\n  let leadingIndex = 0;\n  let previousTrailingEnd = -1;\n  for (const trailing of trailings) {\n    // Leadings ending too far before this trailing can never pair a later (even farther) one.\n    while (leadingIndex < leadings.length) {\n      const leading = leadings[leadingIndex];\n      if (leading && leading.endTokenIndex + maxGapTokens < trailing.startTokenIndex) {\n        leadingIndex += 1;\n      } else {\n        break;\n      }\n    }\n    const leading = leadings[leadingIndex];\n    if (\n      leading &&\n      leading.endTokenIndex <= trailing.startTokenIndex &&\n      leading.startTokenIndex >= previousTrailingEnd\n    ) {\n      pairs.push([leading, trailing]);\n      previousTrailingEnd = trailing.endTokenIndex;\n      leadingIndex += 1;\n    }\n  }\n  const firstFullyPaired = pairs.length === firstLength;\n  const secondFullyPaired = pairs.length === secondLength;\n  if (pairs.length < 2 || (!firstFullyPaired && !secondFullyPaired)) {\n    return undefined;\n  }\n  // A group holding nested copies stays even when all its standalone copies pair: the merged\n  // group's content is larger than what those copies matched, so only the original group can\n  // report which files share the matched fragment.\n  const firstReplaced = firstFullyPaired && !first.some((occurrence) => occurrence.nestedInLargerGroup);\n  const secondReplaced = secondFullyPaired && !second.some((occurrence) => occurrence.nestedInLargerGroup);\n  const merged = pairs.map(([leading, trailing]) => ({\n    ...leading,\n    // A merged occurrence is a fresh span combination; it never inherits shared-span marks.\n    spanCountedElsewhere: undefined,\n    nestedInLargerGroup: undefined,\n    segments: [...leading.segments, ...trailing.segments],\n    tokenCount: leading.tokenCount + trailing.tokenCount,\n    endTokenIndex: trailing.endTokenIndex,\n    endIndex: trailing.endIndex,\n    endLine: trailing.endLine,\n  }));\n  // Only occurrences of one file pair (file offsets exceed the gap), so a merged group spans the\n  // files its pairs sit in: with nested copies left out of pairing, that can be fewer files than\n  // the input groups covered, and a merged group that is no longer reportable must not form.\n  if (!isReportableGroup(merged)) {\n    return undefined;\n  }\n  const pairedRetained = [\n    ...(firstReplaced ? [] : pairs.map(([leading]) => leading)),\n    ...(secondReplaced ? [] : pairs.map(([, trailing]) => trailing)),\n  ];\n  return { merged, firstReplaced, secondReplaced, pairedRetained };\n}\n\n/** One pass over a group: its pairable occurrences and its non-nested occurrence count. */\nfunction pairableOccurrences<T extends CountedOccurrence>(group: T[]): [T[], number] {\n  const pairable: T[] = [];\n  let length = 0;\n  for (const occurrence of group) {\n    if (!occurrence.nestedInLargerGroup) {\n      length += 1;\n    }\n    // Either flag keeps an occurrence out of pairing: its span is already counted elsewhere, or it\n    // is a nested copy of content the merged span would not cover.\n    if (!occurrence.spanCountedElsewhere && !occurrence.nestedInLargerGroup) {\n      pairable.push(occurrence);\n    }\n  }\n  return [pairable, length];\n}\n\n/**\n * Redundant copies one group adds to duplicateBlockCount. Each redundant occurrence contributes\n * one count per matched fragment, so merging a gapped clone's fragments into one group does not\n * halve duplicateBlockCount: an edited two-fragment pair still counts 2, exactly as its unmerged\n * fragments did. Occurrence shapes can differ within one group (a\n * gap-merged exact pair plus an appended whole-block near-miss copy), so every occurrence's\n * fragments are summed and one representative — the largest — is deducted, keeping the count\n * independent of source order. Occurrences a partial gapped merge also paired into a merged group\n * are skipped: their spans are already counted there, and such a retained group deducts no\n * representative of its own — the merged group's representative already stands for the shared\n * content — so no token span contributes to the count twice.\n */\nexport function countRedundantFragments(group: CountedOccurrence[]): number {\n  let fragmentCount = 0;\n  let maxFragmentCount = 0;\n  let hasSharedOccurrence = false;\n  for (const occurrence of group) {\n    if (occurrence.spanCountedElsewhere) {\n      hasSharedOccurrence = true;\n      continue;\n    }\n    fragmentCount += occurrence.segments.length;\n    maxFragmentCount = Math.max(maxFragmentCount, occurrence.segments.length);\n  }\n  return hasSharedOccurrence ? fragmentCount : fragmentCount - maxFragmentCount;\n}\n\n/** Adds the 1-based code lines the segment's matched tokens cover; shared with cross-file coverage. */\nexport function collectSegmentLines(\n  segment: { startTokenIndex: number; endTokenIndex: number },\n  tokens: Token[],\n  codeLineNumbers: Set<number> | undefined,\n  duplicatedLines: Set<number>\n): void {\n  for (let index = segment.startTokenIndex; index < segment.endTokenIndex; index += 1) {\n    const token = tokens[index];\n    for (let row = token?.startRow ?? 0; row <= (token?.endRow ?? -1); row += 1) {\n      if (!codeLineNumbers || codeLineNumbers.has(row + 1)) {\n        duplicatedLines.add(row + 1);\n      }\n    }\n  }\n}\n"],"mappings":"gDAUA,MAAa,EAA0D,CACrE,UAAW,GACX,aAAc,GACd,qBAAsB,EACxB,EAQA,SAAgB,EAA0B,EAA4D,CACpG,MAAO,CACL,UAAW,EAAc,GAAS,UAAW,EAA0B,SAAS,EAChF,aAAc,EAAc,GAAS,aAAc,EAA0B,YAAY,EACzF,qBAAsB,EAAc,GAAS,qBAAsB,EAA0B,oBAAoB,CACnH,CACF,CAEA,SAAS,EAAc,EAA2B,EAA0B,CAC1E,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAAI,EAAW,EAAM,KAAK,MAAM,CAAK,EAAG,EAAG,UAAa,CAC1G,CAiBA,SAAS,EAAe,EAAsB,EAA6B,CACzE,OAAO,EAAe,GAAK,CAC7B,CAuHA,SAAgB,EAAwB,EAA6B,CACnE,IAAM,EAAS,IAAI,WAAW,EAAO,OAAS,CAAC,EAC/C,IAAK,GAAM,CAAC,EAAO,KAAU,EAAO,QAAQ,EAC1C,EAAO,EAAQ,IAAM,EAAO,IAAU,IAAM,EAAM,cAAgB,IAAA,GAAY,EAAI,GAEpF,OAAO,CACT,CA0CA,SAAgB,EACd,EACA,EACA,EAC+B,CAC/B,IAAM,EAA4C,CAAC,EAC7C,EAAoC,CAAC,EACrC,EAA6B,CAAC,EACpC,IAAK,GAAM,CAAC,EAAc,KAAY,EAAS,QAAQ,EACrD,IAAK,IAAM,KAAc,EAAQ,WAC/B,EAAwB,KAAK,CAAY,EACzC,EAAW,KAAK,CAAU,EAG9B,IAAM,EAAa,GACjB,EAAS,EAAwB,IAAmB,GAChD,kBAAyB,IAAI,IAC7B,EAAmB,EAAW,KAAK,EAAY,IACnD,EAA0B,EAAU,CAAc,CAAC,EAAE,QAAU,CAAC,EAAG,EAAY,CAAS,CAC1F,EACA,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAAG,CAClE,IAAM,EAAe,EAAwB,IAAmB,EAChE,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,IAAM,KAAa,EAAK,CAC3B,GAAI,IAAc,IAAA,GAChB,SAEF,IAAM,EAAc,EAAuB,IAAI,CAAS,EACpD,GACF,EAAY,OAAS,EACjB,EAAY,iBAAmB,IACjC,EAAY,eAAiB,IAE3B,EAAY,eAAiB,IAC/B,EAAY,aAAe,IAE7B,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,EAC3D,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,GAE3D,EAAuB,IAAI,EAAW,CACpC,MAAO,EACP,iBACA,eACA,SAAU,EACV,SAAU,CACZ,CAAC,CAEL,CAEJ,CAMA,IAAM,GAAW,EAA+B,IAA4B,CAC1E,GAAI,IAAc,IAAA,GAChB,MAAO,GAET,IAAM,EAAc,EAAuB,IAAI,CAAS,EAOxD,OANI,IAAgB,IAAA,IAAa,EAAY,MAAQ,EAC5C,GAEL,EACK,EAAY,eAAiB,GAE/B,EAAY,iBAAmB,IAAM,EAAY,SAAW,EAAY,UAAY,CAC7F,EAKM,EAAyB,GAAoC,CACjE,IAAM,EAAS,EAAiB,EAAO,eAAe,EAAE,iBAAmB,CAAC,EACtE,EAAY,EAAO,EAAO,OAChC,IAAK,IAAI,EAAQ,EAAO,MAAQ,EAAG,EAAQ,EAAO,MAAQ,EAAO,OAAQ,GAAS,EAChF,GAAI,EAAO,KAAW,EACpB,MAAO,GAGX,MAAO,EACT,EAEM,EAAmC,CAAC,EAC1C,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAC/D,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,GAAM,CAAC,EAAQ,KAAc,EAAI,QAAQ,EAAG,CAC/C,GAAI,CAAC,EAAQ,EAAW,CAAM,GAAK,CAAC,EAAsB,CAAE,iBAAgB,QAAO,QAAO,CAAC,EACzF,SAIF,IAAM,EAAgB,EAAQ,kBAAkB,EAAM,GAAG,EAAS,GAC5D,EAAe,EAAQ,kBAAkB,EAAQ,EAAE,GAAG,EAAS,GACjE,EAAQ,EAAe,EAAS,CAAC,GAAK,EAAQ,EAAc,EAAS,CAAC,GAG1E,EAAe,KAAK,CAAE,iBAAgB,QAAO,QAAO,CAAC,CACvD,CASJ,IAAM,EAAU,IAAI,IAAI,EAAe,IAAI,CAAQ,CAAC,EAChD,EAAW,EACf,KAAO,EAAS,OAAS,GAAG,CAC1B,IAAM,EAA4B,CAAC,EACnC,IAAK,IAAM,KAAU,EAAU,CAC7B,IAAM,EAAa,EAAW,EAAO,gBAC/B,EAAQ,IAAa,EAAO,OAC5B,EAAO,IAAa,EAAO,MAAQ,EAAO,OAAS,GACnD,EAAU,EAAU,EAAO,cAAc,EAC/C,GAAI,CAAC,GAAS,CAAC,GAAQ,CAAC,EACtB,SAEF,IAAM,EAAc,KAAK,EAAe,EAAQ,OAAQ,EAAQ,mBAAoB,EAAM,gBAAiB,EAAK,aAAa,IAC7H,EAAW,KAAK,CACd,UAAW,EAAY,EAAa,EAAM,gBAAiB,EAAK,cAAe,EAAO,CAAI,EAC1F,aAAc,EAAwB,EAAO,iBAAmB,CAClE,CAAC,EACD,EAAQ,KAAK,CAAM,CACrB,CACA,EAAW,CAAC,EACZ,IAAK,IAAM,KAAU,EACnB,IAAK,IAAM,IAAS,CAAC,EAAO,MAAO,EAAO,MAAQ,CAAC,EAAG,CACpD,IAAM,EAAY,CAAE,eAAgB,EAAO,eAAgB,QAAO,OAAQ,EAAO,OAAS,CAAE,EACtF,EAAe,EAAiB,EAAO,eAAe,EAAE,kBAAkB,EAAM,GAAG,EAAU,QAEjG,GAAQ,IAAI,EAAS,CAAS,CAAC,GAC9B,EAAQ,EAAc,EAAU,MAAM,GACtC,EAAsB,CAAS,IAIlC,EAAQ,IAAI,EAAS,CAAS,CAAC,EAC/B,EAAS,KAAK,CAAS,EACzB,CAEJ,CACA,OAAO,CACT,CAEA,SAAS,EAAS,EAAgC,CAChD,MAAO,GAAG,EAAO,eAAe,GAAG,EAAO,MAAM,GAAG,EAAO,QAC5D,CASA,SAAS,EAA0B,EAAiB,EAA0B,EAAqC,CACjH,IAAM,EAAkB,EAAW,IAAK,GACtC,EAAgB,EAAQ,EAAU,gBAAiB,EAAU,aAAa,CAC5E,EACM,EAA8C,CAAC,EACrD,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAW,OAAQ,GAAS,EAAG,CACzD,IAAM,EAA8B,CAAC,EACjC,EAAO,KACP,EAAa,EACX,EAAS,KAAK,IAAI,EAAW,OAAQ,EAAQ,GAAyB,EAC5E,IAAK,IAAI,EAAM,EAAO,EAAM,EAAQ,GAAO,EAAG,CAC5C,IAAM,EAAY,EAAW,GACvB,EAAgB,EAAgB,GACtC,GAAI,CAAC,GAAa,IAAkB,IAAA,GAClC,MAEF,EAAO,EAAc,EAAM,CAAa,EACxC,GAAc,EAAU,cAAgB,EAAU,gBAClD,IAAM,EAAiB,EAAM,EAAQ,EACrC,EAAI,GACF,GAAkB,GAA6B,GAAc,EACzD,EAAc,EAAM,CAAc,EAClC,IAAA,EACR,CACA,EAAkB,KAAK,CAAG,CAC5B,CACA,MAAO,CAAE,oBAAmB,iBAAgB,CAC9C,CAOA,SAAgB,EAAU,EAAe,EAAuB,CAC9D,OAAO,EAAuB,CAAC,CAAC,CAAC,CAAC,CACpC,CAMA,SAAgB,EAAuB,EAA0C,CAC/E,IAAM,EAAa,EAAE,OAAS,KAAQ,EAChC,kBAAgB,IAAI,IAC1B,IAAK,GAAM,CAAC,EAAO,KAAW,EAAE,QAAQ,EAAG,CACzC,IAAI,EAAO,EAAc,IAAI,CAAM,EAC9B,IACH,EAAO,IAAI,YAAY,CAAS,EAChC,EAAc,IAAI,EAAQ,CAAI,GAEhC,IAAM,EAAO,IAAU,EAIvB,EAAK,IAAS,EAAK,IAAS,GAAM,IAAM,EAAQ,GAClD,CAEA,IAAM,EAAI,IAAI,YAAY,CAAS,EACnC,MAAQ,IAAM,CACZ,EAAE,KAAK,CAAC,EACR,IAAK,IAAM,KAAU,EAAG,CACtB,IAAM,EAAY,EAAc,IAAI,CAAM,EAEtC,EAAa,EACb,EAAS,EACb,IAAK,IAAI,EAAO,EAAG,EAAO,EAAW,GAAQ,EAAG,CAC9C,IAAM,EAAW,EAAE,IAAS,EAEtB,IAAM,IAAY,IAAS,GAAK,KAAc,EAE9C,GAAY,GAAY,EAAK,KAAgB,EACnD,EAAa,IAAa,GAC1B,IAAM,EAAa,EAAI,EAAU,EACjC,EAAS,IAAa,GAEtB,EAAE,GAAQ,EAAI,CAAC,CACjB,CACF,CAEA,IAAI,EAAS,EACb,IAAK,IAAM,KAAQ,EACjB,GAAU,EAAS,CAAI,EAEzB,OAAO,CACT,CACF,CAEA,SAAS,EAAS,EAAuB,CACvC,IAAI,EAAQ,GAAU,IAAU,EAAK,YAErC,MADA,IAAS,EAAQ,YAAmB,IAAU,EAAK,WAC3C,KAAK,KAAM,GAAS,IAAU,GAAM,UAAe,QAAa,IAAM,GAAM,GACtF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACoB,CACpB,MAAO,CACL,cACA,WAAY,EAAgB,EAC5B,kBACA,gBACA,WAAY,EAAM,WAClB,SAAU,EAAK,SACf,UAAW,EAAM,UACjB,QAAS,EAAK,OAChB,CACF,CAGA,MAAM,EAAkC,CAAC,EACnC,EAAmC,CAAC,EAE1C,SAAS,EAAoB,EAAuB,CAClD,IAAI,EAAO,EAAsB,GAKjC,OAJI,IAAS,IAAA,KACX,EAAO,EAAS,IAAI,GAAO,EAC3B,EAAsB,GAAS,GAE1B,CACT,CAEA,SAAS,EAAqB,EAAuB,CACnD,IAAI,EAAO,EAAuB,GAKlC,OAJI,IAAS,IAAA,KACX,EAAO,EAAU,IAAI,GAAO,EAC5B,EAAuB,GAAS,GAE3B,CACT,CAUA,SAAS,EACP,EACA,EACA,EACA,EACQ,CAGR,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EADrD,GADC,EAAmB,IAAkB,IAAM,EAAmB,IAAoB,GACtD,EAAgB,CACkC,CAAC,EACrG,MAAO,GAAG,EAAQ,GAAG,EAAU,GAAG,EAAgB,GACpD,CASA,SAAS,EAAgB,EAAiB,EAAyB,EAA+B,CAChG,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EAAe,EAAK,EAE9F,OAAO,EAAU,KAAK,KAAK,EAAW,EAAE,CAC1C,CAEA,SAAS,EACP,EACA,EACA,EACA,EACkB,CAClB,IAAM,kBAAoB,IAAI,IAC1B,EAAU,KACV,EAAY,MAChB,IAAK,IAAI,EAAQ,EAAiB,EAAQ,EAAe,GAAS,EAAG,CACnE,IAAM,EAAQ,EAAO,GACrB,GAAI,CAAC,EACH,SAIF,IAAI,EACA,EACJ,GAAI,EAAM,OAAS,KAAM,CACvB,IAAI,EAAkB,EAAkB,IAAI,EAAM,IAAI,EAClD,IAAoB,IAAA,KACtB,EAAkB,EAAkB,KACpC,EAAkB,IAAI,EAAM,KAAM,CAAe,GAEnD,EAAO,EAAoB,CAAe,EAC1C,EAAQ,EAAqB,CAAe,CAC9C,KACE,GAAO,EAAM,SACb,EAAQ,EAAM,UAGhB,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAQ,EAC5C,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EACnC,GAAqB,EAAM,cAAgB,IAAA,IAAa,EAAM,eAAiB,IAAA,KAEjF,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAM,YAAe,EACzD,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EAAM,aAEjD,CACA,MAAO,CAAC,EAAS,CAAS,CAC5B,CAGA,SAAS,EAAS,EAAsB,CACtC,IAAI,EAAO,KACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAM,EAAE,EAAI,EAAK,WAAW,CAAK,EAEpD,OAAO,CACT,CAGA,SAAS,EAAU,EAAsB,CACvC,IAAI,EAAO,YACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAO,EAAK,WAAW,CAAK,EAAG,QAAU,EAE5D,OAAO,CACT,CAEA,SAAS,EAAc,EAAc,EAAuB,CAC1D,OAAO,KAAK,KAAK,EAAM,EAAE,EAAI,CAC/B,CAoBA,SAAgB,EACd,EACA,EACA,MAAmD,GAC5C,CACP,GAAI,GAAgB,GAAK,EAAO,OAAS,EACvC,OAAO,EAGT,EAAO,KAAK,CAAa,EACzB,IAAK,IAAI,EAAU,GAAM,GAAU,CACjC,EAAU,GACV,IAAM,EAAkB,EAA2B,EAAQ,CAAY,EACvE,IAAK,IAAI,EAAY,EAAG,EAAY,EAAO,QAAU,CAAC,EAAS,GAAa,EAC1E,IAAK,IAAM,KAAc,EAAgB,IAAc,CAAC,EAAG,CACzD,IAAM,EAAO,EAAO,GACd,EAAQ,EAAO,GACrB,GAAI,CAAC,GAAQ,CAAC,EACZ,SAEF,IAAM,EAAU,EAAY,EAAM,EAAO,EAAc,CAAiB,EAClE,EAAS,GAAW,EAAY,EAAO,EAAM,EAAc,CAAiB,EAClF,GAAI,CAAC,EACH,SAEF,IAAM,EAAe,EAAU,EAAO,cAAgB,EAAO,eACvD,EAAgB,EAAU,EAAO,eAAiB,EAAO,cAC3D,GAAgB,GAClB,EAAO,GAAa,EAAO,OAC3B,EAAO,OAAO,EAAY,CAAC,GAClB,EACT,EAAO,GAAc,EAAO,OACnB,EACT,EAAO,GAAa,EAAO,OAI3B,EAAO,KAAK,EAAO,MAAM,EAM3B,IAAK,IAAM,KAAc,EAAO,eAC9B,EAAW,qBAAuB,GAEpC,EAAO,KAAK,CAAa,EACzB,EAAU,GACV,KACF,CAEJ,CACA,OAAO,CACT,CAQA,SAAS,EAA2B,EAA+B,EAAkC,CACnG,IAAM,EAA4D,CAAC,EACnE,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,EAAO,KAAK,CAAE,gBAAiB,EAAW,gBAAiB,YAAW,CAAC,EAG3E,EAAO,MAAM,EAAM,IAAU,EAAK,gBAAkB,EAAM,eAAe,EACzE,IAAM,EAAW,EAAO,wBAAU,IAAI,GAAa,EACnD,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,IACE,IAAI,EAAQ,EAAkB,EAAQ,EAAW,aAAa,EAC9D,EAAQ,EAAO,SAAW,EAAO,EAAM,EAAE,iBAAmB,IAAM,EAAW,cAAgB,EAC7F,GAAS,EACT,CACA,IAAM,EAAQ,EAAO,EAAM,EAAE,YAAc,EACvC,IAAU,GACZ,EAAS,KAAK,IAAI,EAAO,CAAU,EAAE,EAAE,IAAI,KAAK,IAAI,EAAO,CAAU,CAAC,CAE1E,CAGJ,OAAO,EAAS,IAAK,GAAQ,CAAC,GAAG,CAAG,CAAC,CAAC,UAAU,EAAM,IAAU,EAAO,CAAK,CAAC,CAC/E,CAEA,SAAS,EAAkB,EAAuC,EAAwB,CACxF,IAAI,EAAM,EACN,EAAO,EAAO,OAClB,KAAO,EAAM,GAAM,CACjB,IAAM,EAAU,EAAM,IAAU,GAC3B,EAAO,EAAO,EAAE,iBAAmB,GAAK,EAC3C,EAAM,EAAS,EAEf,EAAO,CAEX,CACA,OAAO,CACT,CAEA,SAAS,EAAc,EAA2B,EAAoC,CACpF,IAAM,EAAY,EAAK,GACjB,EAAa,EAAM,GACzB,OACG,GAAW,iBAAmB,IAAM,GAAY,iBAAmB,KACnE,GAAW,eAAiB,IAAM,GAAY,eAAiB,EAEpE,CAsBA,SAAS,EACP,EACA,EACA,EACA,EAC4B,CAQ5B,GAAM,CAAC,EAAU,GAAe,EAAoB,CAAK,EACnD,CAAC,EAAW,GAAgB,EAAoB,CAAM,EACtD,EAAkB,CAAC,EACrB,EAAe,EACf,EAAsB,GAC1B,IAAK,IAAM,KAAY,EAAW,CAEhC,KAAO,EAAe,EAAS,QAAQ,CACrC,IAAM,EAAU,EAAS,GACzB,GAAI,GAAW,EAAQ,cAAgB,EAAe,EAAS,gBAC7D,GAAgB,OAEhB,KAEJ,CACA,IAAM,EAAU,EAAS,GAEvB,GACA,EAAQ,eAAiB,EAAS,iBAClC,EAAQ,iBAAmB,IAE3B,EAAM,KAAK,CAAC,EAAS,CAAQ,CAAC,EAC9B,EAAsB,EAAS,cAC/B,GAAgB,EAEpB,CACA,IAAM,EAAmB,EAAM,SAAW,EACpC,EAAoB,EAAM,SAAW,EAC3C,GAAI,EAAM,OAAS,GAAM,CAAC,GAAoB,CAAC,EAC7C,OAKF,IAAM,EAAgB,GAAoB,CAAC,EAAM,KAAM,GAAe,EAAW,mBAAmB,EAC9F,EAAiB,GAAqB,CAAC,EAAO,KAAM,GAAe,EAAW,mBAAmB,EACjG,EAAS,EAAM,KAAK,CAAC,EAAS,MAAe,CACjD,GAAG,EAEH,qBAAsB,IAAA,GACtB,oBAAqB,IAAA,GACrB,SAAU,CAAC,GAAG,EAAQ,SAAU,GAAG,EAAS,QAAQ,EACpD,WAAY,EAAQ,WAAa,EAAS,WAC1C,cAAe,EAAS,cACxB,SAAU,EAAS,SACnB,QAAS,EAAS,OACpB,EAAE,EAIG,KAAkB,CAAM,EAO7B,MAAO,CAAE,SAAQ,gBAAe,iBAAgB,eAAA,CAH9C,GAAI,EAAgB,CAAC,EAAI,EAAM,KAAK,CAAC,KAAa,CAAO,EACzD,GAAI,EAAiB,CAAC,EAAI,EAAM,KAAK,EAAG,KAAc,CAAQ,CAEH,CAAE,CACjE,CAGA,SAAS,EAAiD,EAA2B,CACnF,IAAM,EAAgB,CAAC,EACnB,EAAS,EACb,IAAK,IAAM,KAAc,EAClB,EAAW,sBACd,GAAU,GAIR,CAAC,EAAW,sBAAwB,CAAC,EAAW,qBAClD,EAAS,KAAK,CAAU,EAG5B,MAAO,CAAC,EAAU,CAAM,CAC1B,CAcA,SAAgB,EAAwB,EAAoC,CAC1E,IAAI,EAAgB,EAChB,EAAmB,EACnB,EAAsB,GAC1B,IAAK,IAAM,KAAc,EAAO,CAC9B,GAAI,EAAW,qBAAsB,CACnC,EAAsB,GACtB,QACF,CACA,GAAiB,EAAW,SAAS,OACrC,EAAmB,KAAK,IAAI,EAAkB,EAAW,SAAS,MAAM,CAC1E,CACA,OAAO,EAAsB,EAAgB,EAAgB,CAC/D,CAGA,SAAgB,EACd,EACA,EACA,EACA,EACM,CACN,IAAK,IAAI,EAAQ,EAAQ,gBAAiB,EAAQ,EAAQ,cAAe,GAAS,EAAG,CACnF,IAAM,EAAQ,EAAO,GACrB,IAAK,IAAI,EAAM,GAAO,UAAY,EAAG,IAAQ,GAAO,QAAU,IAAK,GAAO,GACpE,CAAC,GAAmB,EAAgB,IAAI,EAAM,CAAC,IACjD,EAAgB,IAAI,EAAM,CAAC,CAGjC,CACF"}
+{"version":3,"file":"duplication.js","names":[],"sources":["../src/duplication.ts"],"sourcesContent":["import { clamp } from '@willbooster/shared-lib';\nimport type { DuplicationOptions } from './types.js';\n\n/**\n * Project-level duplication machinery operating on normalized token streams. Tokenization itself\n * (parsing, identifier anonymization, literal normalization) happens in the Rust addon, which\n * serializes each file's Token stream and statement structure; the helpers here match statement\n * windows across files, merge gap-adjacent groups, and count duplicated lines over that data.\n */\n\nexport const defaultDuplicationOptions: Required<DuplicationOptions> = {\n  minTokens: 40,\n  maxGapTokens: 30,\n  minSimilarityPercent: 70,\n};\n\n/**\n * Fills defaults for absent settings, applying the same normalization as the native boundary's\n * clampToU32 — NaN (e.g. `Number(unsetEnvVariable)`) counts as absent, and other values truncate\n * and clamp to [0, u32::MAX] — so the TypeScript half of cross-file matching cannot diverge from\n * the natively collected candidates on such input.\n */\nexport function resolveDuplicationOptions(options?: DuplicationOptions): Required<DuplicationOptions> {\n  return {\n    minTokens: resolveOption(options?.minTokens, defaultDuplicationOptions.minTokens),\n    maxGapTokens: resolveOption(options?.maxGapTokens, defaultDuplicationOptions.maxGapTokens),\n    minSimilarityPercent: resolveOption(options?.minSimilarityPercent, defaultDuplicationOptions.minSimilarityPercent),\n  };\n}\n\nfunction resolveOption(value: number | undefined, fallback: number): number {\n  return value === undefined || Number.isNaN(value) ? fallback : clamp(Math.trunc(value), 0, 0xFF_FF_FF_FF);\n}\n\n/** Minimum consecutive statements for a statement-sequence duplicate candidate. */\nconst minSequenceStatementCount = 2;\n/**\n * Caps the window length so statement-sequence enumeration stays linear in the statement count.\n * Heterogeneous clones longer than the cap are reported as capped windows (a deliberate\n * conservative undercount trading completeness for bounded discovery cost).\n */\nconst maxSequenceStatementCount = 100;\n\n/**\n * A region whose normalized tokens are at least 20% literal values is data-like (a lookup table, a\n * constant list, a value-mapping switch), not logic: literal values re-enter its fingerprint so\n * tables that merely share their shape stop counting as copy-paste. Compared in integer math\n * (5 * literals >= total) so the TypeScript and native sides cannot disagree on the boundary.\n */\nfunction isLiteralDense(literalCount: number, tokenCount: number): boolean {\n  return literalCount * 5 >= tokenCount;\n}\n\nexport interface Token {\n  /** Normalization target: identifiers to anonymize, literal kind tags, or the raw token text. */\n  kind: 'id' | 'text';\n  text: string;\n  /**\n   * Two INDEPENDENT hashes of `text` (djb2 and FNV-1a), precomputed so fingerprinting nested\n   * regions never re-hashes a token. Feeding the same per-token hash to both fingerprint\n   * accumulators would collapse the key to 32 effective bits: one djb2 collision between two\n   * token texts would then equate whole regions.\n   */\n  textHash: number;\n  textHash2: number;\n  /** Hash pair of a value-carrying literal's value, folded into data-like region fingerprints. */\n  literalHash?: number;\n  literalHash2?: number;\n  /**\n   * True for verbatim-kept NAMES (member/callee/type names, named grammar leaves): together with\n   * value-carrying literals these are the content-bearing tokens the near-miss content gate\n   * counts. Keywords, operators, and punctuation come from unnamed nodes and stay false.\n   */\n  isName?: boolean;\n  /** 0-based source rows the token occupies, so line coverage counts only matched-token lines. */\n  startRow: number;\n  endRow: number;\n}\n\n/** A token span with its source position; plain data so cross-file matching can retain it. */\nexport interface TokenRange {\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A contiguous run of matched tokens; gapped (merged) duplicates carry several per occurrence. */\nexport interface TokenSegment {\n  startTokenIndex: number;\n  endTokenIndex: number;\n}\n\ninterface DuplicateCandidate {\n  fingerprint: string;\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\nexport interface CountedOccurrence {\n  /**\n   * The token runs the occurrence covers, matched throughout except in a near-miss copy (see\n   * `matchedRuns`); more than one once gapped groups are merged.\n   */\n  segments: TokenSegment[];\n  /**\n   * The token runs of a near-miss copy that its partners match, within its segments;\n   * absent for an exact or gapped copy, whose segments are matched throughout.\n   */\n  matchedRuns?: TokenSegment[];\n  /**\n   * The exact fragments coalesced into a near-miss copy: each line they reach is duplicated,\n   * whatever else of the copy shares the line.\n   */\n  exactSegments?: TokenSegment[];\n  /**\n   * Set on occurrences whose span another reported group already counts: a retained group's\n   * occurrences that a partial gapped merge also paired into a merged group, and cross-file copies\n   * nested inside a larger group's region. Block counting must not count them again.\n   */\n  spanCountedElsewhere?: boolean;\n  /**\n   * Set on cross-file copies nested inside a larger group's region (they also set\n   * `spanCountedElsewhere`). They never pair in gapped merging, and they do not keep their group's\n   * standalone copies from merging, which they are not copies of. They do keep the merged group\n   * from taking their group's place, since only the original group reports the nesting.\n   */\n  nestedInLargerGroup?: boolean;\n  /** Sum of segment token counts (the gap tokens are not matched content). */\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A duplicate region found in one file, exported for cross-file matching by fingerprint. */\nexport interface CrossFileDuplicateCandidate {\n  /** Content key: equal fingerprints mean equal normalized token sequences (up to hash collision). */\n  fingerprint: string;\n  tokenCount: number;\n  /** Token positions within the owning file, for cross-file gapped (Type-3) merging. */\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/**\n * One file's contribution to cross-file clone detection: catalogued candidates plus the normalized\n * token stream and statement structure, so the project-level pass can match partial statement runs\n * (windows that a single file cannot know repeat elsewhere) and merge gap-adjacent groups.\n */\nexport interface CrossFileDuplicationFileData {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  /**\n   * The mutually disjoint blocks (at least `minTokens` long, not literal-dense) compared for\n   * cross-file near-miss (Type-3) clones. Optional for backward compatibility; without it, the\n   * file takes part in exact and gapped cross-file matching only.\n   */\n  nearMissBlocks?: TokenRange[];\n  /**\n   * 1-based lines that are neither blank nor comment-only, so cross-file line coverage counts only\n   * code lines (blank rows inside multi-row tokens such as template literals carry no content).\n   * Optional for backward compatibility; without it, every matched-token row counts.\n   */\n  codeLineNumbers?: Set<number>;\n}\n\n/** literalCountPrefix[i] = value-carrying literal tokens in tokens[0..i), for O(1) density checks. */\nexport function buildLiteralCountPrefix(tokens: Token[]): Int32Array {\n  const prefix = new Int32Array(tokens.length + 1);\n  for (const [index, token] of tokens.entries()) {\n    prefix[index + 1] = (prefix[index] ?? 0) + (token.literalHash === undefined ? 0 : 1);\n  }\n  return prefix;\n}\n\ninterface WindowOccurrences {\n  count: number;\n  /** -1 once occurrences span more than one container. */\n  containerIndex: number;\n  /** -1 once occurrences span more than one context (file). */\n  contextIndex: number;\n  minStart: number;\n  maxStart: number;\n}\n\ninterface SequenceWindow {\n  containerIndex: number;\n  start: number;\n  length: number;\n}\n\n/** One file's token stream and statement containers, as a window-matching context. */\nexport interface SequenceWindowContext {\n  tokens: Token[];\n  literalCountPrefix: Int32Array;\n  containers: TokenRange[][];\n}\n\nexport interface ContextualSequenceCandidate {\n  candidate: DuplicateCandidate;\n  contextIndex: number;\n}\n\n/**\n * Enumerates runs of consecutive sibling statements over one or more contexts (files). Every\n * container statement participates; only the window length is capped, so enumeration stays linear\n * in the statement count. Windows are grouped by a cheap rolling hash of per-statement\n * fingerprints, and only locally maximal repeated windows — those whose one-statement extensions\n * stop repeating — become candidates with an exact (window-consistent) fingerprint. Without the\n * maximality filter a degenerate file of near-identical statements would fingerprint every\n * sub-window of every repeated region. With `requireMultipleContexts` a window only counts as\n * repeated when its occurrences span at least two contexts (CPD-style cross-file matching): a\n * repeat confined to one file is that file's own concern, and emitting it here would flood the\n * project-level selection with unusable single-file groups.\n */\nexport function collectSequenceWindowCandidates(\n  contexts: SequenceWindowContext[],\n  minTokens: number,\n  requireMultipleContexts: boolean\n): ContextualSequenceCandidate[] {\n  const candidates: ContextualSequenceCandidate[] = [];\n  const contextIndexByContainer: number[] = [];\n  const containers: TokenRange[][] = [];\n  for (const [contextIndex, context] of contexts.entries()) {\n    for (const statements of context.containers) {\n      contextIndexByContainer.push(contextIndex);\n      containers.push(statements);\n    }\n  }\n  const contextAt = (containerIndex: number): SequenceWindowContext | undefined =>\n    contexts[contextIndexByContainer[containerIndex] ?? 0];\n  const occurrencesByWindowKey = new Map<number, WindowOccurrences>();\n  const containerWindows = containers.map((statements, containerIndex) =>\n    enumerateContainerWindows(contextAt(containerIndex)?.tokens ?? [], statements, minTokens)\n  );\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    const contextIndex = contextIndexByContainer[containerIndex] ?? 0;\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const windowKey of row) {\n        if (windowKey === undefined) {\n          continue;\n        }\n        const occurrences = occurrencesByWindowKey.get(windowKey);\n        if (occurrences) {\n          occurrences.count += 1;\n          if (occurrences.containerIndex !== containerIndex) {\n            occurrences.containerIndex = -1;\n          }\n          if (occurrences.contextIndex !== contextIndex) {\n            occurrences.contextIndex = -1;\n          }\n          occurrences.minStart = Math.min(occurrences.minStart, start);\n          occurrences.maxStart = Math.max(occurrences.maxStart, start);\n        } else {\n          occurrencesByWindowKey.set(windowKey, {\n            count: 1,\n            containerIndex,\n            contextIndex,\n            minStart: start,\n            maxStart: start,\n          });\n        }\n      }\n    }\n  }\n\n  // A window only \"repeats\" when two of its occurrences can coexist without overlapping: sliding\n  // matches inside a homogeneous run (start spread smaller than the window length) can never both\n  // be counted and must neither qualify a window nor dominate its sub-windows. Cross-context\n  // matching instead requires occurrences in two contexts, which coexist by construction.\n  const repeats = (windowKey: number | undefined, length: number): boolean => {\n    if (windowKey === undefined) {\n      return false;\n    }\n    const occurrences = occurrencesByWindowKey.get(windowKey);\n    if (occurrences === undefined || occurrences.count < 2) {\n      return false;\n    }\n    if (requireMultipleContexts) {\n      return occurrences.contextIndex === -1;\n    }\n    return occurrences.containerIndex === -1 || occurrences.maxStart - occurrences.minStart >= length;\n  };\n\n  // A window whose statements all share one normalized shape (sixteen `let x = 0;` declarations,\n  // a constant table) is a homogeneous preamble, not a copy-paste: requiring two distinct\n  // per-statement shapes keeps such runs out of duplicate groups and the duplication ratio.\n  const hasDistinctStatements = (window: SequenceWindow): boolean => {\n    const hashes = containerWindows[window.containerIndex]?.statementHashes ?? [];\n    const firstHash = hashes[window.start];\n    for (let index = window.start + 1; index < window.start + window.length; index += 1) {\n      if (hashes[index] !== firstHash) {\n        return true;\n      }\n    }\n    return false;\n  };\n\n  const maximalWindows: SequenceWindow[] = [];\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const [length, windowKey] of row.entries()) {\n        if (!repeats(windowKey, length) || !hasDistinctStatements({ containerIndex, start, length })) {\n          continue;\n        }\n        // Dominated windows are skipped: the one-statement extension also repeats, so a larger\n        // candidate covering this window exists.\n        const extendedRight = windows.windowKeysByStart[start]?.[length + 1];\n        const extendedLeft = windows.windowKeysByStart[start - 1]?.[length + 1];\n        if (repeats(extendedRight, length + 1) || repeats(extendedLeft, length + 1)) {\n          continue;\n        }\n        maximalWindows.push({ containerIndex, start, length });\n      }\n    }\n  }\n\n  // The rolling hash anonymizes identifiers per statement, so a window can look repeated coarsely\n  // while its exact (window-consistent) fingerprints differ, and a longer window's match can\n  // dominate sub-windows that other copies still need (three copies where only two extend one\n  // statement further). Every emitted window therefore exposes its repeating, unvisited\n  // sub-windows; `visited` bounds the worklist and lengths strictly decrease, so it terminates.\n  const visited = new Set(maximalWindows.map(windowId));\n  let frontier = maximalWindows;\n  while (frontier.length > 0) {\n    const emitted: SequenceWindow[] = [];\n    for (const window of frontier) {\n      const statements = containers[window.containerIndex];\n      const first = statements?.[window.start];\n      const last = statements?.[window.start + window.length - 1];\n      const context = contextAt(window.containerIndex);\n      if (!first || !last || !context) {\n        continue;\n      }\n      const fingerprint = `s:${fingerprintKey(context.tokens, context.literalCountPrefix, first.startTokenIndex, last.endTokenIndex)}`;\n      candidates.push({\n        candidate: toCandidate(fingerprint, first.startTokenIndex, last.endTokenIndex, first, last),\n        contextIndex: contextIndexByContainer[window.containerIndex] ?? 0,\n      });\n      emitted.push(window);\n    }\n    frontier = [];\n    for (const window of emitted) {\n      for (const start of [window.start, window.start + 1]) {\n        const subWindow = { containerIndex: window.containerIndex, start, length: window.length - 1 };\n        const subWindowKey = containerWindows[window.containerIndex]?.windowKeysByStart[start]?.[subWindow.length];\n        if (\n          visited.has(windowId(subWindow)) ||\n          !repeats(subWindowKey, subWindow.length) ||\n          !hasDistinctStatements(subWindow)\n        ) {\n          continue;\n        }\n        visited.add(windowId(subWindow));\n        frontier.push(subWindow);\n      }\n    }\n  }\n  return candidates;\n}\n\nfunction windowId(window: SequenceWindow): string {\n  return `${window.containerIndex}:${window.start}:${window.length}`;\n}\n\ninterface ContainerWindows {\n  /** windowKeysByStart[start][length] is the rolling-hash key of the window, or undefined if it is below the size thresholds. */\n  windowKeysByStart: (number | undefined)[][];\n  /** Per-statement fingerprint hashes, for the distinct-shape requirement on windows. */\n  statementHashes: number[];\n}\n\nfunction enumerateContainerWindows(tokens: Token[], statements: TokenRange[], minTokens: number): ContainerWindows {\n  const statementHashes = statements.map((statement) =>\n    fingerprintHash(tokens, statement.startTokenIndex, statement.endTokenIndex)\n  );\n  const windowKeysByStart: (number | undefined)[][] = [];\n  for (let start = 0; start < statements.length; start += 1) {\n    const row: (number | undefined)[] = [];\n    let hash = 5381;\n    let tokenCount = 0;\n    const maxEnd = Math.min(statements.length, start + maxSequenceStatementCount);\n    for (let end = start; end < maxEnd; end += 1) {\n      const statement = statements[end];\n      const statementHash = statementHashes[end];\n      if (!statement || statementHash === undefined) {\n        break;\n      }\n      hash = combineHashes(hash, statementHash);\n      tokenCount += statement.endTokenIndex - statement.startTokenIndex;\n      const statementCount = end - start + 1;\n      row[statementCount] =\n        statementCount >= minSequenceStatementCount && tokenCount >= minTokens\n          ? combineHashes(hash, statementCount)\n          : undefined;\n    }\n    windowKeysByStart.push(row);\n  }\n  return { windowKeysByStart, statementHashes };\n}\n\n/**\n * Longest-common-subsequence LENGTH of two symbol sequences via the Allison–Dix bit-parallel\n * recurrence (O(|a|/32 · |b|) words): per symbol of `b`, `x = match | v` and\n * `v = x & ~(x - ((v << 1) | 1))` over multi-word bit vectors; the set bits of `v` count the LCS.\n */\nexport function lcsLength(a: Int32Array, b: Int32Array): number {\n  return createLcsLengthCounter(a)(b);\n}\n\n/**\n * lcsLength against a fixed `a`, whose per-symbol position masks are built once and reused for\n * every `b` compared with it.\n */\nexport function createLcsLengthCounter(a: Int32Array): (b: Int32Array) => number {\n  const growthBits = createLcsGrowthBits(a);\n  return (b) => {\n    let length = 0;\n    for (const word of growthBits(b)) {\n      length += popCount(word);\n    }\n    return length;\n  };\n}\n\n/**\n * Per sequence `b`, one bit per position of `a`, set where the longest common subsequence of `b`\n * with the prefix of `a` ending there is longer than with the prefix before it, so the set bits up\n * to a position count that length. The returned words are overwritten by the next call.\n */\nfunction createLcsGrowthBits(a: Int32Array): (b: Int32Array) => Uint32Array {\n  const wordCount = (a.length + 31) >>> 5;\n  const positionMasks = new Map<number, Uint32Array>();\n  for (const [index, symbol] of a.entries()) {\n    let mask = positionMasks.get(symbol);\n    if (!mask) {\n      mask = new Uint32Array(wordCount);\n      positionMasks.set(symbol, mask);\n    }\n    const word = index >>> 5;\n    // The Uint32Array store wraps the signed int32 bit pattern to unsigned. The `?? 0` guards in\n    // this function are required by noUncheckedIndexedAccess (typed-array reads type as\n    // `number | undefined`), not redundancy: every index is in bounds.\n    mask[word] = (mask[word] ?? 0) | (1 << (index & 31));\n  }\n\n  const v = new Uint32Array(wordCount);\n  return (b) => {\n    v.fill(0);\n    for (const symbol of b) {\n      const matchMask = positionMasks.get(symbol);\n      // `(v << 1) | 1` shifts a carry bit across words; subtraction borrows across words.\n      let shiftCarry = 1;\n      let borrow = 0;\n      for (let word = 0; word < wordCount; word += 1) {\n        const previous = v[word] ?? 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- `>>> 0` reinterprets the signed int32 bit pattern as unsigned so the borrow subtraction below compares magnitudes; Math.trunc would keep it negative.\n        const x = ((matchMask?.[word] ?? 0) | previous) >>> 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- same unsigned reinterpretation as `x`.\n        const shifted = ((previous << 1) | shiftCarry) >>> 0;\n        shiftCarry = previous >>> 31;\n        const difference = x - shifted - borrow;\n        borrow = difference < 0 ? 1 : 0;\n        // The Uint32Array store wraps the signed int32 bit pattern to unsigned.\n        v[word] = x & ~difference;\n      }\n    }\n    return v;\n  };\n}\n\n/**\n * Which symbols of each sequence a longest common subsequence pairs, by Hirschberg's divide and\n * conquer over the bit-parallel recurrence, which alone yields lengths but no pairing.\n */\nexport function markLcs(a: Int32Array, b: Int32Array): [Uint8Array, Uint8Array] {\n  const aMatched = new Uint8Array(a.length);\n  const bMatched = new Uint8Array(b.length);\n  markLcsInto(a, b, aMatched, bMatched);\n  return [aMatched, bMatched];\n}\n\nfunction markLcsInto(a: Int32Array, b: Int32Array, aMatched: Uint8Array, bMatched: Uint8Array): void {\n  // A common prefix or suffix belongs to some longest common subsequence.\n  let prefix = 0;\n  while (prefix < a.length && prefix < b.length && a[prefix] === b[prefix]) {\n    prefix += 1;\n  }\n  let suffix = 0;\n  while (\n    suffix < a.length - prefix &&\n    suffix < b.length - prefix &&\n    a[a.length - 1 - suffix] === b[b.length - 1 - suffix]\n  ) {\n    suffix += 1;\n  }\n  const aEnd = a.length - suffix;\n  const bEnd = b.length - suffix;\n  aMatched.fill(1, 0, prefix).fill(1, aEnd);\n  bMatched.fill(1, 0, prefix).fill(1, bEnd);\n  const aRest = a.subarray(prefix, aEnd);\n  const bRest = b.subarray(prefix, bEnd);\n  const aRestMatched = aMatched.subarray(prefix, aEnd);\n  const bRestMatched = bMatched.subarray(prefix, bEnd);\n  if (aRest.length === 0 || bRest.length === 0) {\n    return;\n  }\n  if (aRest.length === 1) {\n    const position = bRest.indexOf(aRest[0] ?? 0);\n    if (position !== -1) {\n      aRestMatched[0] = 1;\n      bRestMatched[position] = 1;\n    }\n    return;\n  }\n  const middle = aRest.length >>> 1;\n  const forward = lcsPrefixLengths(aRest.subarray(0, middle), bRest);\n  const backward = lcsPrefixLengths(aRest.subarray(middle).toReversed(), bRest.toReversed());\n  let split = 0;\n  let best = -1;\n  for (let candidate = 0; candidate <= bRest.length; candidate += 1) {\n    const length = (forward[candidate] ?? 0) + (backward[bRest.length - candidate] ?? 0);\n    if (length > best) {\n      best = length;\n      split = candidate;\n    }\n  }\n  markLcsInto(\n    aRest.subarray(0, middle),\n    bRest.subarray(0, split),\n    aRestMatched.subarray(0, middle),\n    bRestMatched.subarray(0, split)\n  );\n  markLcsInto(\n    aRest.subarray(middle),\n    bRest.subarray(split),\n    aRestMatched.subarray(middle),\n    bRestMatched.subarray(split)\n  );\n}\n\n/** The longest-common-subsequence length of `a` with every prefix of `b`. */\nfunction lcsPrefixLengths(a: Int32Array, b: Int32Array): Int32Array {\n  const bits = createLcsGrowthBits(b)(a);\n  const lengths = new Int32Array(b.length + 1);\n  for (let index = 0; index < b.length; index += 1) {\n    lengths[index + 1] = (lengths[index] ?? 0) + (((bits[index >>> 5] ?? 0) >>> (index & 31)) & 1);\n  }\n  return lengths;\n}\n\nfunction popCount(value: number): number {\n  let count = value - ((value >>> 1) & 0x55_55_55_55);\n  count = (count & 0x33_33_33_33) + ((count >>> 2) & 0x33_33_33_33);\n  return (Math.imul((count + (count >>> 4)) & 0x0F_0F_0F_0F, 0x01_01_01_01) >>> 24) & 0xFF;\n}\n\nfunction toCandidate(\n  fingerprint: string,\n  startTokenIndex: number,\n  endTokenIndex: number,\n  first: TokenRange,\n  last: TokenRange\n): DuplicateCandidate {\n  return {\n    fingerprint,\n    tokenCount: endTokenIndex - startTokenIndex,\n    startTokenIndex,\n    endTokenIndex,\n    startIndex: first.startIndex,\n    endIndex: last.endIndex,\n    startLine: first.startLine,\n    endLine: last.endLine,\n  };\n}\n\n/** Caches of hashText/hashText2 over '$0', '$1', ... so anonymized identifiers hash without allocating. */\nconst anonymizedIndexHashes: number[] = [];\nconst anonymizedIndexHashes2: number[] = [];\n\nfunction anonymizedIndexHash(index: number): number {\n  let hash = anonymizedIndexHashes[index];\n  if (hash === undefined) {\n    hash = hashText(`$${index}`);\n    anonymizedIndexHashes[index] = hash;\n  }\n  return hash;\n}\n\nfunction anonymizedIndexHash2(index: number): number {\n  let hash = anonymizedIndexHashes2[index];\n  if (hash === undefined) {\n    hash = hashText2(`$${index}`);\n    anonymizedIndexHashes2[index] = hash;\n  }\n  return hash;\n}\n\n/**\n * Content key of a token range: two independent 32-bit hashes over the normalized token sequence\n * (identifiers anonymized consistently by first-occurrence order) plus the token count. Regions\n * with equal keys are treated as equal content; a collision would need both 32-bit hashes and the\n * length to coincide, which is negligible for a metrics report. The format and arithmetic match\n * fingerprint_key in native/src/duplication.rs exactly, so window candidates fingerprinted here\n * group together with the per-file candidates the addon catalogues.\n */\nfunction fingerprintKey(\n  tokens: Token[],\n  literalCountPrefix: Int32Array,\n  startTokenIndex: number,\n  endTokenIndex: number\n): string {\n  const literalCount = (literalCountPrefix[endTokenIndex] ?? 0) - (literalCountPrefix[startTokenIndex] ?? 0);\n  const literalDense = isLiteralDense(literalCount, endTokenIndex - startTokenIndex);\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, literalDense);\n  return `${primary}:${secondary}:${endTokenIndex - startTokenIndex}`;\n}\n\n/**\n * A single 32-bit summary of a range for the coarse rolling-hash phase. Deliberately\n * density-agnostic: density is a property of the final candidate REGION, and folding literal\n * values into per-statement hashes would make a dense statement inside a logic-heavy window\n * (`const weights = [1, 2, 3];`) block the window from ever being enumerated. The coarse phase\n * over-approximates on shape alone; the exact region fingerprint still applies the density rule.\n */\nfunction fingerprintHash(tokens: Token[], startTokenIndex: number, endTokenIndex: number): number {\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, false);\n  // XOR already coerces to int32, matching the native side's i32 arithmetic.\n  return primary ^ Math.imul(secondary, 31);\n}\n\nfunction fingerprintHashPair(\n  tokens: Token[],\n  startTokenIndex: number,\n  endTokenIndex: number,\n  foldLiteralValues: boolean\n): [number, number] {\n  const indexByIdentifier = new Map<string, number>();\n  let primary = 5381;\n  let secondary = 52_711;\n  for (let index = startTokenIndex; index < endTokenIndex; index += 1) {\n    const token = tokens[index];\n    if (!token) {\n      continue;\n    }\n    // Each accumulator consumes its own independent per-token hash: sharing one would collapse\n    // the key to 32 effective bits (a single djb2 collision would equate whole regions).\n    let part: number;\n    let part2: number;\n    if (token.kind === 'id') {\n      let identifierIndex = indexByIdentifier.get(token.text);\n      if (identifierIndex === undefined) {\n        identifierIndex = indexByIdentifier.size;\n        indexByIdentifier.set(token.text, identifierIndex);\n      }\n      part = anonymizedIndexHash(identifierIndex);\n      part2 = anonymizedIndexHash2(identifierIndex);\n    } else {\n      part = token.textHash;\n      part2 = token.textHash2;\n    }\n    // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n    primary = (Math.imul(primary, 31) + part) | 0;\n    secondary = Math.imul(secondary, 37) ^ part2;\n    if (foldLiteralValues && token.literalHash !== undefined && token.literalHash2 !== undefined) {\n      // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n      primary = (Math.imul(primary, 31) + token.literalHash) | 0;\n      secondary = Math.imul(secondary, 37) ^ token.literalHash2;\n    }\n  }\n  return [primary, secondary];\n}\n\n/** djb2-style hash; XOR keeps the value in signed 32-bit range, which is fine for a grouping key. */\nfunction hashText(text: string): number {\n  let hash = 5381;\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- djb2 hashes UTF-16 code units; codePointAt would hash surrogate pairs twice (full code point, then the lone low surrogate).\n    hash = Math.imul(hash, 33) ^ text.charCodeAt(index);\n  }\n  return hash;\n}\n\n/** FNV-1a over UTF-16 code units: independent of hashText so the two accumulators never share input. */\nfunction hashText2(text: string): number {\n  let hash = -2_128_831_035; // 2166136261 as int32 (the FNV-1a offset basis)\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- hashes UTF-16 code units like hashText.\n    hash = Math.imul(hash ^ text.charCodeAt(index), 16_777_619);\n  }\n  return hash;\n}\n\nfunction combineHashes(hash: number, value: number): number {\n  return Math.imul(hash, 31) + value;\n}\n\n/**\n * Merges duplicate groups separated by a small token gap into one gapped (Type-3) clone group: a\n * copy edited in one spot splits into two exact groups whose occurrences sit side by side in the\n * same order. Occurrences are paired greedily in source order; a merge happens when the pairing\n * fully pairs at least one group with at least two pairs. Equal-cardinality groups whose\n * occurrences all pair merge into one group as before. When cardinalities differ (a fragment also\n * occurs standalone: prefix ×3, suffix ×2), the fully-paired group is subsumed into the merged\n * gapped group while the other group is RETAINED with ALL its occurrences: dropping the leftover\n * would lose duplicated-line coverage, and reporting it alone would make a single-occurrence group\n * (contradicting duplicateBlockGroupCount's \"appears more than once\" meaning). Line coverage\n * unions ranges, so the overlap between the retained exact group and the merged group is harmless.\n * Merging repeats to a fixpoint so a clone edited in several spots still reassembles; it\n * terminates because a merge either removes the input groups it replaces or marks their paired\n * occurrences as counted elsewhere, and only unmarked occurrences of remaining groups pair, so the\n * number of pairable occurrences strictly decreases with every merge. Gap tokens are not matched\n * content: line coverage and sizes count only the matched segments. Generic so cross-file merging\n * can thread file identity through occurrences.\n */\nexport function mergeAdjacentGroups<T extends CountedOccurrence>(\n  groups: T[][],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean = () => true\n): T[][] {\n  if (maxGapTokens <= 0 || groups.length < 2) {\n    return groups;\n  }\n  // Deterministic processing order (mirrored by the native side): by first occurrence position.\n  groups.sort(compareGroups);\n  for (let restart = true; restart;) {\n    restart = false;\n    const partnersByGroup = collectGapAdjacentPartners(groups, maxGapTokens);\n    for (let leftIndex = 0; leftIndex < groups.length && !restart; leftIndex += 1) {\n      for (const rightIndex of partnersByGroup[leftIndex] ?? []) {\n        const left = groups[leftIndex];\n        const right = groups[rightIndex];\n        if (!left || !right) {\n          continue;\n        }\n        const forward = mergeGroups(left, right, maxGapTokens, isReportableGroup);\n        const result = forward ?? mergeGroups(right, left, maxGapTokens, isReportableGroup);\n        if (!result) {\n          continue;\n        }\n        const leftReplaced = forward ? result.firstReplaced : result.secondReplaced;\n        const rightReplaced = forward ? result.secondReplaced : result.firstReplaced;\n        if (leftReplaced && rightReplaced) {\n          groups[leftIndex] = result.merged;\n          groups.splice(rightIndex, 1);\n        } else if (rightReplaced) {\n          groups[rightIndex] = result.merged;\n        } else if (leftReplaced) {\n          groups[leftIndex] = result.merged;\n        } else {\n          // Both groups stay (each is only partly paired, or reports nested copies of its own), so\n          // the merged group joins them instead of taking a place.\n          groups.push(result.merged);\n        }\n        // A group that stays keeps ALL its occurrences (line coverage must not shrink, and a\n        // reported group must keep >= 2 occurrences), so its paired occurrences now also live\n        // inside the merged group's occurrences: mark them so duplicateBlockCount counts each\n        // token span once, and so the same pair cannot merge again.\n        for (const occurrence of result.pairedRetained) {\n          occurrence.spanCountedElsewhere = true;\n        }\n        groups.sort(compareGroups);\n        restart = true;\n        break;\n      }\n    }\n  }\n  return groups;\n}\n\n/**\n * Per group index, the ascending indexes of later groups that mergeGroups can pair with it in\n * either direction: an occurrence of one starts within `maxGapTokens` after an occurrence of the\n * other ends. Every other pair fails mergeGroups, so skipping it leaves the fixpoint unchanged\n * while each restart stops costing a comparison per pair of groups.\n */\nfunction collectGapAdjacentPartners(groups: CountedOccurrence[][], maxGapTokens: number): number[][] {\n  const starts: { startTokenIndex: number; groupIndex: number }[] = [];\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      starts.push({ startTokenIndex: occurrence.startTokenIndex, groupIndex });\n    }\n  }\n  starts.sort((left, right) => left.startTokenIndex - right.startTokenIndex);\n  const partners = groups.map(() => new Set<number>());\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      for (\n        let index = lowerBoundByStart(starts, occurrence.endTokenIndex);\n        index < starts.length && (starts[index]?.startTokenIndex ?? 0) <= occurrence.endTokenIndex + maxGapTokens;\n        index += 1\n      ) {\n        const other = starts[index]?.groupIndex ?? groupIndex;\n        if (other !== groupIndex) {\n          partners[Math.min(other, groupIndex)]?.add(Math.max(other, groupIndex));\n        }\n      }\n    }\n  }\n  return partners.map((set) => [...set].toSorted((left, right) => left - right));\n}\n\nfunction lowerBoundByStart(sorted: { startTokenIndex: number }[], target: number): number {\n  let low = 0;\n  let high = sorted.length;\n  while (low < high) {\n    const middle = (low + high) >>> 1;\n    if ((sorted[middle]?.startTokenIndex ?? 0) < target) {\n      low = middle + 1;\n    } else {\n      high = middle;\n    }\n  }\n  return low;\n}\n\nfunction compareGroups(left: CountedOccurrence[], right: CountedOccurrence[]): number {\n  const leftFirst = left[0];\n  const rightFirst = right[0];\n  return (\n    (leftFirst?.startTokenIndex ?? 0) - (rightFirst?.startTokenIndex ?? 0) ||\n    (leftFirst?.endTokenIndex ?? 0) - (rightFirst?.endTokenIndex ?? 0)\n  );\n}\n\ninterface MergeResult<T> {\n  merged: T[];\n  /**\n   * Whether the merged group takes the respective input group's place: its standalone occurrences\n   * were all paired and it holds no nested copies that only it can report.\n   */\n  firstReplaced: boolean;\n  secondReplaced: boolean;\n  /** The occurrences of groups that stay, which the merged group's spans now also cover. */\n  pairedRetained: T[];\n}\n\n/**\n * Pairs `second` occurrences with gap-preceding `first` occurrences, greedily in source order:\n * each trailing occurrence takes the earliest unused leading occurrence within the gap, and a\n * pair's leading must start at or after the previous pair's trailing end so merged spans never\n * overlap. A merge needs at least two pairs (a merged group must still mean \"appears more than\n * once\") and must fully consume at least one group; for equal cardinalities this reduces to the\n * strict all-pairs merge, so pre-partial-merge behavior is unchanged there.\n */\nfunction mergeGroups<T extends CountedOccurrence>(\n  first: T[],\n  second: T[],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean\n): MergeResult<T> | undefined {\n  // Occurrences a previous partial merge already paired into a merged group must not pair again:\n  // their spans already live inside that merged group, so re-pairing them would assemble a second,\n  // competing merged group instead of letting the existing merged group extend (and would count\n  // the same span twice). Consumption is still judged against the FULL group, so a group holding\n  // shared occurrences is never subsumed away; nested copies, which the merged span would not\n  // cover anyway, are left out of that judgment so they cannot veto a merge of the standalone\n  // copies.\n  const [leadings, firstLength] = pairableOccurrences(first);\n  const [trailings, secondLength] = pairableOccurrences(second);\n  const pairs: [T, T][] = [];\n  let leadingIndex = 0;\n  let previousTrailingEnd = -1;\n  for (const trailing of trailings) {\n    // Leadings ending too far before this trailing can never pair a later (even farther) one.\n    while (leadingIndex < leadings.length) {\n      const leading = leadings[leadingIndex];\n      if (leading && leading.endTokenIndex + maxGapTokens < trailing.startTokenIndex) {\n        leadingIndex += 1;\n      } else {\n        break;\n      }\n    }\n    const leading = leadings[leadingIndex];\n    if (\n      leading &&\n      leading.endTokenIndex <= trailing.startTokenIndex &&\n      leading.startTokenIndex >= previousTrailingEnd\n    ) {\n      pairs.push([leading, trailing]);\n      previousTrailingEnd = trailing.endTokenIndex;\n      leadingIndex += 1;\n    }\n  }\n  const firstFullyPaired = pairs.length === firstLength;\n  const secondFullyPaired = pairs.length === secondLength;\n  if (pairs.length < 2 || (!firstFullyPaired && !secondFullyPaired)) {\n    return undefined;\n  }\n  // A group holding nested copies stays even when all its standalone copies pair: the merged\n  // group's content is larger than what those copies matched, so only the original group can\n  // report which files share the matched fragment.\n  const firstReplaced = firstFullyPaired && !first.some((occurrence) => occurrence.nestedInLargerGroup);\n  const secondReplaced = secondFullyPaired && !second.some((occurrence) => occurrence.nestedInLargerGroup);\n  const merged = pairs.map(([leading, trailing]) => ({\n    ...leading,\n    // A merged occurrence is a fresh span combination; it never inherits shared-span marks, and\n    // its segments are matched throughout.\n    spanCountedElsewhere: undefined,\n    nestedInLargerGroup: undefined,\n    matchedRuns: undefined,\n    segments: [...leading.segments, ...trailing.segments],\n    tokenCount: leading.tokenCount + trailing.tokenCount,\n    endTokenIndex: trailing.endTokenIndex,\n    endIndex: trailing.endIndex,\n    endLine: trailing.endLine,\n  }));\n  // Only occurrences of one file pair (file offsets exceed the gap), so a merged group spans the\n  // files its pairs sit in: with nested copies left out of pairing, that can be fewer files than\n  // the input groups covered, and a merged group that is no longer reportable must not form.\n  if (!isReportableGroup(merged)) {\n    return undefined;\n  }\n  const pairedRetained = [\n    ...(firstReplaced ? [] : pairs.map(([leading]) => leading)),\n    ...(secondReplaced ? [] : pairs.map(([, trailing]) => trailing)),\n  ];\n  return { merged, firstReplaced, secondReplaced, pairedRetained };\n}\n\n/** One pass over a group: its pairable occurrences and its non-nested occurrence count. */\nfunction pairableOccurrences<T extends CountedOccurrence>(group: T[]): [T[], number] {\n  const pairable: T[] = [];\n  let length = 0;\n  for (const occurrence of group) {\n    if (!occurrence.nestedInLargerGroup) {\n      length += 1;\n    }\n    // Either flag keeps an occurrence out of pairing: its span is already counted elsewhere, or it\n    // is a nested copy of content the merged span would not cover.\n    if (!occurrence.spanCountedElsewhere && !occurrence.nestedInLargerGroup) {\n      pairable.push(occurrence);\n    }\n  }\n  return [pairable, length];\n}\n\n/**\n * Redundant copies one group adds to duplicateBlockCount. Each redundant occurrence contributes\n * one count per matched fragment, so merging a gapped clone's fragments into one group does not\n * halve duplicateBlockCount: an edited two-fragment pair still counts 2, exactly as its unmerged\n * fragments did. Occurrence shapes can differ within one group (a\n * gap-merged exact pair plus an appended whole-block near-miss copy), so every occurrence's\n * fragments are summed and one representative — the largest — is deducted, keeping the count\n * independent of source order. Occurrences a partial gapped merge also paired into a merged group\n * are skipped: their spans are already counted there, and such a retained group deducts no\n * representative of its own — the merged group's representative already stands for the shared\n * content — so no token span contributes to the count twice.\n */\nexport function countRedundantFragments(group: CountedOccurrence[]): number {\n  let fragmentCount = 0;\n  let maxFragmentCount = 0;\n  let hasSharedOccurrence = false;\n  for (const occurrence of group) {\n    if (occurrence.spanCountedElsewhere) {\n      hasSharedOccurrence = true;\n      continue;\n    }\n    fragmentCount += occurrence.segments.length;\n    maxFragmentCount = Math.max(maxFragmentCount, occurrence.segments.length);\n  }\n  return hasSharedOccurrence ? fragmentCount : fragmentCount - maxFragmentCount;\n}\n\n/**\n * The 1-based code lines the occurrence's matched tokens cover, given its segments and matched runs\n * as indexes into `tokens`. A line of a near-miss copy counts when its partners match more than half\n * of its tokens: a line the copy adds or rewrites is not duplicated, while a longest common\n * subsequence still pairs a stray token of it (a bracket, a separator) with one of the partner.\n * A line one of `exactSegments` reaches counts in any case.\n */\nexport function collectMatchedLines(\n  segments: TokenSegment[],\n  matchedRuns: TokenSegment[] | undefined,\n  tokens: Token[],\n  codeLineNumbers: Set<number> | undefined,\n  exactSegments: TokenSegment[] = []\n): Set<number> {\n  const exactLines = new Set<number>();\n  for (const segment of exactSegments) {\n    for (let index = segment.startTokenIndex; index < segment.endTokenIndex; index += 1) {\n      const token = tokens[index];\n      for (let row = token?.startRow ?? 0; row <= (token?.endRow ?? -1); row += 1) {\n        exactLines.add(row + 1);\n      }\n    }\n  }\n  const matchedCountByLine = new Map<number, number>();\n  const tokenCountByLine = new Map<number, number>();\n  let runIndex = 0;\n  for (const segment of segments) {\n    for (let index = segment.startTokenIndex; index < segment.endTokenIndex; index += 1) {\n      while (matchedRuns && (matchedRuns[runIndex]?.endTokenIndex ?? Infinity) <= index) {\n        runIndex += 1;\n      }\n      const matched = !matchedRuns || (matchedRuns[runIndex]?.startTokenIndex ?? Infinity) <= index;\n      const token = tokens[index];\n      for (let row = token?.startRow ?? 0; row <= (token?.endRow ?? -1); row += 1) {\n        tokenCountByLine.set(row + 1, (tokenCountByLine.get(row + 1) ?? 0) + 1);\n        matchedCountByLine.set(row + 1, (matchedCountByLine.get(row + 1) ?? 0) + (matched ? 1 : 0));\n      }\n    }\n  }\n  const lines = new Set<number>();\n  for (const [line, tokenCount] of tokenCountByLine) {\n    if (\n      (!codeLineNumbers || codeLineNumbers.has(line)) &&\n      ((matchedCountByLine.get(line) ?? 0) * 2 > tokenCount || exactLines.has(line))\n    ) {\n      lines.add(line);\n    }\n  }\n  return lines;\n}\n"],"mappings":"gDAUA,MAAa,EAA0D,CACrE,UAAW,GACX,aAAc,GACd,qBAAsB,EACxB,EAQA,SAAgB,EAA0B,EAA4D,CACpG,MAAO,CACL,UAAW,EAAc,GAAS,UAAW,EAA0B,SAAS,EAChF,aAAc,EAAc,GAAS,aAAc,EAA0B,YAAY,EACzF,qBAAsB,EAAc,GAAS,qBAAsB,EAA0B,oBAAoB,CACnH,CACF,CAEA,SAAS,EAAc,EAA2B,EAA0B,CAC1E,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAAI,EAAW,EAAM,KAAK,MAAM,CAAK,EAAG,EAAG,UAAa,CAC1G,CAiBA,SAAS,EAAe,EAAsB,EAA6B,CACzE,OAAO,EAAe,GAAK,CAC7B,CAoIA,SAAgB,EAAwB,EAA6B,CACnE,IAAM,EAAS,IAAI,WAAW,EAAO,OAAS,CAAC,EAC/C,IAAK,GAAM,CAAC,EAAO,KAAU,EAAO,QAAQ,EAC1C,EAAO,EAAQ,IAAM,EAAO,IAAU,IAAM,EAAM,cAAgB,IAAA,GAAY,EAAI,GAEpF,OAAO,CACT,CA0CA,SAAgB,EACd,EACA,EACA,EAC+B,CAC/B,IAAM,EAA4C,CAAC,EAC7C,EAAoC,CAAC,EACrC,EAA6B,CAAC,EACpC,IAAK,GAAM,CAAC,EAAc,KAAY,EAAS,QAAQ,EACrD,IAAK,IAAM,KAAc,EAAQ,WAC/B,EAAwB,KAAK,CAAY,EACzC,EAAW,KAAK,CAAU,EAG9B,IAAM,EAAa,GACjB,EAAS,EAAwB,IAAmB,GAChD,kBAAyB,IAAI,IAC7B,EAAmB,EAAW,KAAK,EAAY,IACnD,EAA0B,EAAU,CAAc,CAAC,EAAE,QAAU,CAAC,EAAG,EAAY,CAAS,CAC1F,EACA,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAAG,CAClE,IAAM,EAAe,EAAwB,IAAmB,EAChE,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,IAAM,KAAa,EAAK,CAC3B,GAAI,IAAc,IAAA,GAChB,SAEF,IAAM,EAAc,EAAuB,IAAI,CAAS,EACpD,GACF,EAAY,OAAS,EACjB,EAAY,iBAAmB,IACjC,EAAY,eAAiB,IAE3B,EAAY,eAAiB,IAC/B,EAAY,aAAe,IAE7B,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,EAC3D,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,GAE3D,EAAuB,IAAI,EAAW,CACpC,MAAO,EACP,iBACA,eACA,SAAU,EACV,SAAU,CACZ,CAAC,CAEL,CAEJ,CAMA,IAAM,GAAW,EAA+B,IAA4B,CAC1E,GAAI,IAAc,IAAA,GAChB,MAAO,GAET,IAAM,EAAc,EAAuB,IAAI,CAAS,EAOxD,OANI,IAAgB,IAAA,IAAa,EAAY,MAAQ,EAC5C,GAEL,EACK,EAAY,eAAiB,GAE/B,EAAY,iBAAmB,IAAM,EAAY,SAAW,EAAY,UAAY,CAC7F,EAKM,EAAyB,GAAoC,CACjE,IAAM,EAAS,EAAiB,EAAO,eAAe,EAAE,iBAAmB,CAAC,EACtE,EAAY,EAAO,EAAO,OAChC,IAAK,IAAI,EAAQ,EAAO,MAAQ,EAAG,EAAQ,EAAO,MAAQ,EAAO,OAAQ,GAAS,EAChF,GAAI,EAAO,KAAW,EACpB,MAAO,GAGX,MAAO,EACT,EAEM,EAAmC,CAAC,EAC1C,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAC/D,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,GAAM,CAAC,EAAQ,KAAc,EAAI,QAAQ,EAAG,CAC/C,GAAI,CAAC,EAAQ,EAAW,CAAM,GAAK,CAAC,EAAsB,CAAE,iBAAgB,QAAO,QAAO,CAAC,EACzF,SAIF,IAAM,EAAgB,EAAQ,kBAAkB,EAAM,GAAG,EAAS,GAC5D,EAAe,EAAQ,kBAAkB,EAAQ,EAAE,GAAG,EAAS,GACjE,EAAQ,EAAe,EAAS,CAAC,GAAK,EAAQ,EAAc,EAAS,CAAC,GAG1E,EAAe,KAAK,CAAE,iBAAgB,QAAO,QAAO,CAAC,CACvD,CASJ,IAAM,EAAU,IAAI,IAAI,EAAe,IAAI,CAAQ,CAAC,EAChD,EAAW,EACf,KAAO,EAAS,OAAS,GAAG,CAC1B,IAAM,EAA4B,CAAC,EACnC,IAAK,IAAM,KAAU,EAAU,CAC7B,IAAM,EAAa,EAAW,EAAO,gBAC/B,EAAQ,IAAa,EAAO,OAC5B,EAAO,IAAa,EAAO,MAAQ,EAAO,OAAS,GACnD,EAAU,EAAU,EAAO,cAAc,EAC/C,GAAI,CAAC,GAAS,CAAC,GAAQ,CAAC,EACtB,SAEF,IAAM,EAAc,KAAK,EAAe,EAAQ,OAAQ,EAAQ,mBAAoB,EAAM,gBAAiB,EAAK,aAAa,IAC7H,EAAW,KAAK,CACd,UAAW,EAAY,EAAa,EAAM,gBAAiB,EAAK,cAAe,EAAO,CAAI,EAC1F,aAAc,EAAwB,EAAO,iBAAmB,CAClE,CAAC,EACD,EAAQ,KAAK,CAAM,CACrB,CACA,EAAW,CAAC,EACZ,IAAK,IAAM,KAAU,EACnB,IAAK,IAAM,IAAS,CAAC,EAAO,MAAO,EAAO,MAAQ,CAAC,EAAG,CACpD,IAAM,EAAY,CAAE,eAAgB,EAAO,eAAgB,QAAO,OAAQ,EAAO,OAAS,CAAE,EACtF,EAAe,EAAiB,EAAO,eAAe,EAAE,kBAAkB,EAAM,GAAG,EAAU,QAEjG,GAAQ,IAAI,EAAS,CAAS,CAAC,GAC9B,EAAQ,EAAc,EAAU,MAAM,GACtC,EAAsB,CAAS,IAIlC,EAAQ,IAAI,EAAS,CAAS,CAAC,EAC/B,EAAS,KAAK,CAAS,EACzB,CAEJ,CACA,OAAO,CACT,CAEA,SAAS,EAAS,EAAgC,CAChD,MAAO,GAAG,EAAO,eAAe,GAAG,EAAO,MAAM,GAAG,EAAO,QAC5D,CASA,SAAS,EAA0B,EAAiB,EAA0B,EAAqC,CACjH,IAAM,EAAkB,EAAW,IAAK,GACtC,EAAgB,EAAQ,EAAU,gBAAiB,EAAU,aAAa,CAC5E,EACM,EAA8C,CAAC,EACrD,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAW,OAAQ,GAAS,EAAG,CACzD,IAAM,EAA8B,CAAC,EACjC,EAAO,KACP,EAAa,EACX,EAAS,KAAK,IAAI,EAAW,OAAQ,EAAQ,GAAyB,EAC5E,IAAK,IAAI,EAAM,EAAO,EAAM,EAAQ,GAAO,EAAG,CAC5C,IAAM,EAAY,EAAW,GACvB,EAAgB,EAAgB,GACtC,GAAI,CAAC,GAAa,IAAkB,IAAA,GAClC,MAEF,EAAO,EAAc,EAAM,CAAa,EACxC,GAAc,EAAU,cAAgB,EAAU,gBAClD,IAAM,EAAiB,EAAM,EAAQ,EACrC,EAAI,GACF,GAAkB,GAA6B,GAAc,EACzD,EAAc,EAAM,CAAc,EAClC,IAAA,EACR,CACA,EAAkB,KAAK,CAAG,CAC5B,CACA,MAAO,CAAE,oBAAmB,iBAAgB,CAC9C,CAOA,SAAgB,EAAU,EAAe,EAAuB,CAC9D,OAAO,EAAuB,CAAC,CAAC,CAAC,CAAC,CACpC,CAMA,SAAgB,EAAuB,EAA0C,CAC/E,IAAM,EAAa,EAAoB,CAAC,EACxC,MAAQ,IAAM,CACZ,IAAI,EAAS,EACb,IAAK,IAAM,KAAQ,EAAW,CAAC,EAC7B,GAAU,EAAS,CAAI,EAEzB,OAAO,CACT,CACF,CAOA,SAAS,EAAoB,EAA+C,CAC1E,IAAM,EAAa,EAAE,OAAS,KAAQ,EAChC,kBAAgB,IAAI,IAC1B,IAAK,GAAM,CAAC,EAAO,KAAW,EAAE,QAAQ,EAAG,CACzC,IAAI,EAAO,EAAc,IAAI,CAAM,EAC9B,IACH,EAAO,IAAI,YAAY,CAAS,EAChC,EAAc,IAAI,EAAQ,CAAI,GAEhC,IAAM,EAAO,IAAU,EAIvB,EAAK,IAAS,EAAK,IAAS,GAAM,IAAM,EAAQ,GAClD,CAEA,IAAM,EAAI,IAAI,YAAY,CAAS,EACnC,MAAQ,IAAM,CACZ,EAAE,KAAK,CAAC,EACR,IAAK,IAAM,KAAU,EAAG,CACtB,IAAM,EAAY,EAAc,IAAI,CAAM,EAEtC,EAAa,EACb,EAAS,EACb,IAAK,IAAI,EAAO,EAAG,EAAO,EAAW,GAAQ,EAAG,CAC9C,IAAM,EAAW,EAAE,IAAS,EAEtB,IAAM,IAAY,IAAS,GAAK,KAAc,EAE9C,GAAY,GAAY,EAAK,KAAgB,EACnD,EAAa,IAAa,GAC1B,IAAM,EAAa,EAAI,EAAU,EACjC,EAAS,IAAa,GAEtB,EAAE,GAAQ,EAAI,CAAC,CACjB,CACF,CACA,OAAO,CACT,CACF,CAMA,SAAgB,EAAQ,EAAe,EAAyC,CAC9E,IAAM,EAAW,IAAI,WAAW,EAAE,MAAM,EAClC,EAAW,IAAI,WAAW,EAAE,MAAM,EAExC,OADA,EAAY,EAAG,EAAG,EAAU,CAAQ,EAC7B,CAAC,EAAU,CAAQ,CAC5B,CAEA,SAAS,EAAY,EAAe,EAAe,EAAsB,EAA4B,CAEnG,IAAI,EAAS,EACb,KAAO,EAAS,EAAE,QAAU,EAAS,EAAE,QAAU,EAAE,KAAY,EAAE,IAC/D,GAAU,EAEZ,IAAI,EAAS,EACb,KACE,EAAS,EAAE,OAAS,GACpB,EAAS,EAAE,OAAS,GACpB,EAAE,EAAE,OAAS,EAAI,KAAY,EAAE,EAAE,OAAS,EAAI,IAE9C,GAAU,EAEZ,IAAM,EAAO,EAAE,OAAS,EAClB,EAAO,EAAE,OAAS,EACxB,EAAS,KAAK,EAAG,EAAG,CAAM,CAAC,CAAC,KAAK,EAAG,CAAI,EACxC,EAAS,KAAK,EAAG,EAAG,CAAM,CAAC,CAAC,KAAK,EAAG,CAAI,EACxC,IAAM,EAAQ,EAAE,SAAS,EAAQ,CAAI,EAC/B,EAAQ,EAAE,SAAS,EAAQ,CAAI,EAC/B,EAAe,EAAS,SAAS,EAAQ,CAAI,EAC7C,EAAe,EAAS,SAAS,EAAQ,CAAI,EACnD,GAAI,EAAM,SAAW,GAAK,EAAM,SAAW,EACzC,OAEF,GAAI,EAAM,SAAW,EAAG,CACtB,IAAM,EAAW,EAAM,QAAQ,EAAM,IAAM,CAAC,EACxC,IAAa,KACf,EAAa,GAAK,EAClB,EAAa,GAAY,GAE3B,MACF,CACA,IAAM,EAAS,EAAM,SAAW,EAC1B,EAAU,EAAiB,EAAM,SAAS,EAAG,CAAM,EAAG,CAAK,EAC3D,EAAW,EAAiB,EAAM,SAAS,CAAM,CAAC,CAAC,WAAW,EAAG,EAAM,WAAW,CAAC,EACrF,EAAQ,EACR,EAAO,GACX,IAAK,IAAI,EAAY,EAAG,GAAa,EAAM,OAAQ,GAAa,EAAG,CACjE,IAAM,GAAU,EAAQ,IAAc,IAAM,EAAS,EAAM,OAAS,IAAc,GAC9E,EAAS,IACX,EAAO,EACP,EAAQ,EAEZ,CACA,EACE,EAAM,SAAS,EAAG,CAAM,EACxB,EAAM,SAAS,EAAG,CAAK,EACvB,EAAa,SAAS,EAAG,CAAM,EAC/B,EAAa,SAAS,EAAG,CAAK,CAChC,EACA,EACE,EAAM,SAAS,CAAM,EACrB,EAAM,SAAS,CAAK,EACpB,EAAa,SAAS,CAAM,EAC5B,EAAa,SAAS,CAAK,CAC7B,CACF,CAGA,SAAS,EAAiB,EAAe,EAA2B,CAClE,IAAM,EAAO,EAAoB,CAAC,CAAC,CAAC,CAAC,EAC/B,EAAU,IAAI,WAAW,EAAE,OAAS,CAAC,EAC3C,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAE,OAAQ,GAAS,EAC7C,EAAQ,EAAQ,IAAM,EAAQ,IAAU,KAAQ,EAAK,IAAU,IAAM,MAAQ,EAAQ,IAAO,GAE9F,OAAO,CACT,CAEA,SAAS,EAAS,EAAuB,CACvC,IAAI,EAAQ,GAAU,IAAU,EAAK,YAErC,MADA,IAAS,EAAQ,YAAmB,IAAU,EAAK,WAC3C,KAAK,KAAM,GAAS,IAAU,GAAM,UAAe,QAAa,IAAM,GAAM,GACtF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACoB,CACpB,MAAO,CACL,cACA,WAAY,EAAgB,EAC5B,kBACA,gBACA,WAAY,EAAM,WAClB,SAAU,EAAK,SACf,UAAW,EAAM,UACjB,QAAS,EAAK,OAChB,CACF,CAGA,MAAM,EAAkC,CAAC,EACnC,EAAmC,CAAC,EAE1C,SAAS,EAAoB,EAAuB,CAClD,IAAI,EAAO,EAAsB,GAKjC,OAJI,IAAS,IAAA,KACX,EAAO,EAAS,IAAI,GAAO,EAC3B,EAAsB,GAAS,GAE1B,CACT,CAEA,SAAS,EAAqB,EAAuB,CACnD,IAAI,EAAO,EAAuB,GAKlC,OAJI,IAAS,IAAA,KACX,EAAO,EAAU,IAAI,GAAO,EAC5B,EAAuB,GAAS,GAE3B,CACT,CAUA,SAAS,EACP,EACA,EACA,EACA,EACQ,CAGR,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EADrD,GADC,EAAmB,IAAkB,IAAM,EAAmB,IAAoB,GACtD,EAAgB,CACkC,CAAC,EACrG,MAAO,GAAG,EAAQ,GAAG,EAAU,GAAG,EAAgB,GACpD,CASA,SAAS,EAAgB,EAAiB,EAAyB,EAA+B,CAChG,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EAAe,EAAK,EAE9F,OAAO,EAAU,KAAK,KAAK,EAAW,EAAE,CAC1C,CAEA,SAAS,EACP,EACA,EACA,EACA,EACkB,CAClB,IAAM,kBAAoB,IAAI,IAC1B,EAAU,KACV,EAAY,MAChB,IAAK,IAAI,EAAQ,EAAiB,EAAQ,EAAe,GAAS,EAAG,CACnE,IAAM,EAAQ,EAAO,GACrB,GAAI,CAAC,EACH,SAIF,IAAI,EACA,EACJ,GAAI,EAAM,OAAS,KAAM,CACvB,IAAI,EAAkB,EAAkB,IAAI,EAAM,IAAI,EAClD,IAAoB,IAAA,KACtB,EAAkB,EAAkB,KACpC,EAAkB,IAAI,EAAM,KAAM,CAAe,GAEnD,EAAO,EAAoB,CAAe,EAC1C,EAAQ,EAAqB,CAAe,CAC9C,KACE,GAAO,EAAM,SACb,EAAQ,EAAM,UAGhB,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAQ,EAC5C,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EACnC,GAAqB,EAAM,cAAgB,IAAA,IAAa,EAAM,eAAiB,IAAA,KAEjF,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAM,YAAe,EACzD,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EAAM,aAEjD,CACA,MAAO,CAAC,EAAS,CAAS,CAC5B,CAGA,SAAS,EAAS,EAAsB,CACtC,IAAI,EAAO,KACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAM,EAAE,EAAI,EAAK,WAAW,CAAK,EAEpD,OAAO,CACT,CAGA,SAAS,EAAU,EAAsB,CACvC,IAAI,EAAO,YACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAO,EAAK,WAAW,CAAK,EAAG,QAAU,EAE5D,OAAO,CACT,CAEA,SAAS,EAAc,EAAc,EAAuB,CAC1D,OAAO,KAAK,KAAK,EAAM,EAAE,EAAI,CAC/B,CAoBA,SAAgB,EACd,EACA,EACA,MAAmD,GAC5C,CACP,GAAI,GAAgB,GAAK,EAAO,OAAS,EACvC,OAAO,EAGT,EAAO,KAAK,CAAa,EACzB,IAAK,IAAI,EAAU,GAAM,GAAU,CACjC,EAAU,GACV,IAAM,EAAkB,EAA2B,EAAQ,CAAY,EACvE,IAAK,IAAI,EAAY,EAAG,EAAY,EAAO,QAAU,CAAC,EAAS,GAAa,EAC1E,IAAK,IAAM,KAAc,EAAgB,IAAc,CAAC,EAAG,CACzD,IAAM,EAAO,EAAO,GACd,EAAQ,EAAO,GACrB,GAAI,CAAC,GAAQ,CAAC,EACZ,SAEF,IAAM,EAAU,EAAY,EAAM,EAAO,EAAc,CAAiB,EAClE,EAAS,GAAW,EAAY,EAAO,EAAM,EAAc,CAAiB,EAClF,GAAI,CAAC,EACH,SAEF,IAAM,EAAe,EAAU,EAAO,cAAgB,EAAO,eACvD,EAAgB,EAAU,EAAO,eAAiB,EAAO,cAC3D,GAAgB,GAClB,EAAO,GAAa,EAAO,OAC3B,EAAO,OAAO,EAAY,CAAC,GAClB,EACT,EAAO,GAAc,EAAO,OACnB,EACT,EAAO,GAAa,EAAO,OAI3B,EAAO,KAAK,EAAO,MAAM,EAM3B,IAAK,IAAM,KAAc,EAAO,eAC9B,EAAW,qBAAuB,GAEpC,EAAO,KAAK,CAAa,EACzB,EAAU,GACV,KACF,CAEJ,CACA,OAAO,CACT,CAQA,SAAS,EAA2B,EAA+B,EAAkC,CACnG,IAAM,EAA4D,CAAC,EACnE,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,EAAO,KAAK,CAAE,gBAAiB,EAAW,gBAAiB,YAAW,CAAC,EAG3E,EAAO,MAAM,EAAM,IAAU,EAAK,gBAAkB,EAAM,eAAe,EACzE,IAAM,EAAW,EAAO,wBAAU,IAAI,GAAa,EACnD,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,IACE,IAAI,EAAQ,EAAkB,EAAQ,EAAW,aAAa,EAC9D,EAAQ,EAAO,SAAW,EAAO,EAAM,EAAE,iBAAmB,IAAM,EAAW,cAAgB,EAC7F,GAAS,EACT,CACA,IAAM,EAAQ,EAAO,EAAM,EAAE,YAAc,EACvC,IAAU,GACZ,EAAS,KAAK,IAAI,EAAO,CAAU,EAAE,EAAE,IAAI,KAAK,IAAI,EAAO,CAAU,CAAC,CAE1E,CAGJ,OAAO,EAAS,IAAK,GAAQ,CAAC,GAAG,CAAG,CAAC,CAAC,UAAU,EAAM,IAAU,EAAO,CAAK,CAAC,CAC/E,CAEA,SAAS,EAAkB,EAAuC,EAAwB,CACxF,IAAI,EAAM,EACN,EAAO,EAAO,OAClB,KAAO,EAAM,GAAM,CACjB,IAAM,EAAU,EAAM,IAAU,GAC3B,EAAO,EAAO,EAAE,iBAAmB,GAAK,EAC3C,EAAM,EAAS,EAEf,EAAO,CAEX,CACA,OAAO,CACT,CAEA,SAAS,EAAc,EAA2B,EAAoC,CACpF,IAAM,EAAY,EAAK,GACjB,EAAa,EAAM,GACzB,OACG,GAAW,iBAAmB,IAAM,GAAY,iBAAmB,KACnE,GAAW,eAAiB,IAAM,GAAY,eAAiB,EAEpE,CAsBA,SAAS,EACP,EACA,EACA,EACA,EAC4B,CAQ5B,GAAM,CAAC,EAAU,GAAe,EAAoB,CAAK,EACnD,CAAC,EAAW,GAAgB,EAAoB,CAAM,EACtD,EAAkB,CAAC,EACrB,EAAe,EACf,EAAsB,GAC1B,IAAK,IAAM,KAAY,EAAW,CAEhC,KAAO,EAAe,EAAS,QAAQ,CACrC,IAAM,EAAU,EAAS,GACzB,GAAI,GAAW,EAAQ,cAAgB,EAAe,EAAS,gBAC7D,GAAgB,OAEhB,KAEJ,CACA,IAAM,EAAU,EAAS,GAEvB,GACA,EAAQ,eAAiB,EAAS,iBAClC,EAAQ,iBAAmB,IAE3B,EAAM,KAAK,CAAC,EAAS,CAAQ,CAAC,EAC9B,EAAsB,EAAS,cAC/B,GAAgB,EAEpB,CACA,IAAM,EAAmB,EAAM,SAAW,EACpC,EAAoB,EAAM,SAAW,EAC3C,GAAI,EAAM,OAAS,GAAM,CAAC,GAAoB,CAAC,EAC7C,OAKF,IAAM,EAAgB,GAAoB,CAAC,EAAM,KAAM,GAAe,EAAW,mBAAmB,EAC9F,EAAiB,GAAqB,CAAC,EAAO,KAAM,GAAe,EAAW,mBAAmB,EACjG,EAAS,EAAM,KAAK,CAAC,EAAS,MAAe,CACjD,GAAG,EAGH,qBAAsB,IAAA,GACtB,oBAAqB,IAAA,GACrB,YAAa,IAAA,GACb,SAAU,CAAC,GAAG,EAAQ,SAAU,GAAG,EAAS,QAAQ,EACpD,WAAY,EAAQ,WAAa,EAAS,WAC1C,cAAe,EAAS,cACxB,SAAU,EAAS,SACnB,QAAS,EAAS,OACpB,EAAE,EAIG,KAAkB,CAAM,EAO7B,MAAO,CAAE,SAAQ,gBAAe,iBAAgB,eAAA,CAH9C,GAAI,EAAgB,CAAC,EAAI,EAAM,KAAK,CAAC,KAAa,CAAO,EACzD,GAAI,EAAiB,CAAC,EAAI,EAAM,KAAK,EAAG,KAAc,CAAQ,CAEH,CAAE,CACjE,CAGA,SAAS,EAAiD,EAA2B,CACnF,IAAM,EAAgB,CAAC,EACnB,EAAS,EACb,IAAK,IAAM,KAAc,EAClB,EAAW,sBACd,GAAU,GAIR,CAAC,EAAW,sBAAwB,CAAC,EAAW,qBAClD,EAAS,KAAK,CAAU,EAG5B,MAAO,CAAC,EAAU,CAAM,CAC1B,CAcA,SAAgB,EAAwB,EAAoC,CAC1E,IAAI,EAAgB,EAChB,EAAmB,EACnB,EAAsB,GAC1B,IAAK,IAAM,KAAc,EAAO,CAC9B,GAAI,EAAW,qBAAsB,CACnC,EAAsB,GACtB,QACF,CACA,GAAiB,EAAW,SAAS,OACrC,EAAmB,KAAK,IAAI,EAAkB,EAAW,SAAS,MAAM,CAC1E,CACA,OAAO,EAAsB,EAAgB,EAAgB,CAC/D,CASA,SAAgB,EACd,EACA,EACA,EACA,EACA,EAAgC,CAAC,EACpB,CACb,IAAM,kBAAa,IAAI,IACvB,IAAK,IAAM,KAAW,EACpB,IAAK,IAAI,EAAQ,EAAQ,gBAAiB,EAAQ,EAAQ,cAAe,GAAS,EAAG,CACnF,IAAM,EAAQ,EAAO,GACrB,IAAK,IAAI,EAAM,GAAO,UAAY,EAAG,IAAQ,GAAO,QAAU,IAAK,GAAO,EACxE,EAAW,IAAI,EAAM,CAAC,CAE1B,CAEF,IAAM,kBAAqB,IAAI,IACzB,kBAAmB,IAAI,IACzB,EAAW,EACf,IAAK,IAAM,KAAW,EACpB,IAAK,IAAI,EAAQ,EAAQ,gBAAiB,EAAQ,EAAQ,cAAe,GAAS,EAAG,CACnF,KAAO,IAAgB,EAAY,EAAS,EAAE,eAAiB,MAAa,GAC1E,GAAY,EAEd,IAAM,EAAU,CAAC,IAAgB,EAAY,EAAS,EAAE,iBAAmB,MAAa,EAClF,EAAQ,EAAO,GACrB,IAAK,IAAI,EAAM,GAAO,UAAY,EAAG,IAAQ,GAAO,QAAU,IAAK,GAAO,EACxE,EAAiB,IAAI,EAAM,GAAI,EAAiB,IAAI,EAAM,CAAC,GAAK,GAAK,CAAC,EACtE,EAAmB,IAAI,EAAM,GAAI,EAAmB,IAAI,EAAM,CAAC,GAAK,IAAM,IAAgB,CAE9F,CAEF,IAAM,kBAAQ,IAAI,IAClB,IAAK,GAAM,CAAC,EAAM,KAAe,GAE5B,CAAC,GAAmB,EAAgB,IAAI,CAAI,MAC3C,EAAmB,IAAI,CAAI,GAAK,GAAK,EAAI,GAAc,EAAW,IAAI,CAAI,IAE5E,EAAM,IAAI,CAAI,EAGlB,OAAO,CACT"}

package/dist/thresholdCheck.cjs

@@ -1,2 +1,2 @@
-"use strict";const e=require("./_virtual/_rolldown/runtime.cjs"),t=require("./thresholds.cjs");let n=require("node:assert");n=e.__toESM(n,1);const r=[`file`,`function`,`duplication`],i=new Set(t.functionThresholds.filter(({includesNestedFunctions:e})=>e).map(e=>t.metricNameOf(e)));function a(e,n,r){let i=g(n.crossFile,new Set(e.map(({file:e})=>e))),a=[],l=0;for(let o of e){let{file:e,metrics:d,hunks:m}=o,h=r(d.language),g=m?.length===0?[]:p(t.fileThresholds,d,h);g.length>0&&a.push(f({file:e,startLine:1,endLine:d.lines.total},`file`,g));let v=s(o);l+=v.length;let y=v.map(e=>({fn:e,exceeded:p(t.functionThresholds,e,h)})).filter(({exceeded:e})=>e.length>0);for(let{fn:t,exceeded:n}of y){let r=n.filter(e=>!c(e,t,y));r.length>0&&a.push(u(e,t,r))}a.push(..._(o,i.get(e)??[],h,n))}return{violations:a.toSorted(o),checkedFunctionCount:l}}function o(e,n){return t.levels.indexOf(n.level)-t.levels.indexOf(e.level)||C(e.file,n.file)||e.startLine-n.startLine||r.indexOf(e.kind)-r.indexOf(n.kind)||e.endLine-n.endLine}function s({metrics:e,hunks:t}){let n=e.language===`python`;return t?e.functions.filter(e=>t.some(t=>h(t,e.startLine,e.endLine,n))):e.functions}function c(e,t,n){return i.has(e.metric)?n.some(n=>n.fn!==t&&l(n.fn,t)&&n.exceeded.some(({metric:t})=>t===e.metric)):!1}function l(e,t){let n=e.startLine<t.startLine||e.startLine===t.startLine&&e.startColumn<=t.startColumn,r=e.endLine>t.endLine||e.endLine===t.endLine&&e.endColumn>=t.endColumn;return n&&r}function u(e,t,n){let{startLine:r,endLine:i}=t,a=n.some(({metric:e})=>e===`functionCognitiveComplexity`);return{...f({file:e,startLine:r,endLine:i},`function`,n),name:d(t),...a&&t.cognitiveBlocks.length>0&&{largestBlocks:t.cognitiveBlocks}}}function d({name:e=`<anonymous>`,containerName:t}){return t?`${t}.${e}`:e}function f(e,t,n){return{kind:t,level:n.some(e=>e.level===`error`)?`error`:`warning`,...e,exceeded:n}}function p(e,t,n){return e.flatMap(e=>m(e,t,n,(e,t)=>e>t))}function m(e,n,r,i){let a=e.measure(n),o=t.levels.findLast(t=>i(a,r[t][e.key]));return o?[{metric:t.metricNameOf(e),value:a,level:o,limit:r[o][e.key]}]:[]}function h(e,t,n,r=!1){return e.headCount===0?e.headStart>=t&&(e.headStart<n||r&&e.headStart===n):e.headStart<=n&&e.headStart+e.headCount>t}function g(e,t){let n=/* @__PURE__ */ new Map;for(let r of e?.groups??[])for(let e of r.files)if(t.has(e)){let t=n.get(e)??[];t.push(r),n.set(e,t)}return n}function _({file:e,metrics:n,hunks:r},i,a,{minSimilarityPercent:o}){let s=e=>m(t.duplicationThreshold,{duplicatedLineCount:e},a,(e,t)=>e>=t),c=[...n.duplication.duplicateBlockGroups.map(t=>t.map(t=>({...t,file:e}))),...i.map(e=>e.occurrences.map(e=>v(e)))],l=t=>t.file===e&&s(t.lineNumbers.length).length>0&&(!r||r.some(e=>e.headCount>0&&h(e,t.startLine,t.endLine))),u=/* @__PURE__ */ new Map;for(let e of c)for(let t of e.filter(e=>l(e)))u.set(t,e);return b([...u.keys()]).map(({merged:e,sources:t})=>{let n=n=>b(t.flatMap(e=>{let r=Math.max(...t.filter(t=>x(t,e)).map(e=>e.tokenCount));return(u.get(e)??[]).filter(e=>e.tokenCount*100>=r*n)})).filter(t=>!S(t.merged,e)).toSorted((e,t)=>y(t.sources)-y(e.sources)).map(e=>e.merged),r=n(o);return{...f(e,`duplication`,s(new Set(t.flatMap(e=>e.lineNumbers)).size)),partners:r.length>0?r:n(0)}})}function v(e){let{lineNumbers:t}=e;return n.default.ok(t,`No matched lines for ${e.file}, a scanned file.`),{...e,lineNumbers:t}}function y(e){return e.reduce((e,t)=>e+t.tokenCount,0)}function b(e){let t=[],n=e.toSorted((e,t)=>C(e.file,t.file)||e.startLine-t.startLine);for(let e of n){let n=t.at(-1);if(n&&S(n.merged,e))n.merged.endLine=Math.max(n.merged.endLine,e.endLine),n.sources.push(e);else{let{file:n,startLine:r,endLine:i}=e;t.push({merged:{file:n,startLine:r,endLine:i},sources:[e]})}}return t}function x(e,t){return e.file===t.file&&e.startLine<=t.startLine&&t.endLine<=e.endLine}function S(e,t){return e.file===t.file&&e.startLine<=t.endLine&&t.startLine<=e.endLine}function C(e,t){return e<t?-1:+(e>t)}exports.checkThresholds=a,exports.qualifyFunctionName=d;
+"use strict";const e=require("./_virtual/_rolldown/runtime.cjs"),t=require("./thresholds.cjs");let n=require("node:assert");n=e.__toESM(n,1);const r=[`file`,`function`,`duplication`],i=new Set(t.functionThresholds.filter(({includesNestedFunctions:e})=>e).map(e=>t.metricNameOf(e)));function a(e,n,r){let i=g(n.crossFile,new Set(e.map(({file:e})=>e))),a=[],l=0;for(let o of e){let{file:e,metrics:d,hunks:m}=o,h=r(d.language),g=m?.length===0?[]:p(t.fileThresholds,d,h);g.length>0&&a.push(f({file:e,startLine:1,endLine:d.lines.total},`file`,g));let v=s(o);l+=v.length;let y=v.map(e=>({fn:e,exceeded:p(t.functionThresholds,e,h)})).filter(({exceeded:e})=>e.length>0);for(let{fn:t,exceeded:n}of y){let r=n.filter(e=>!c(e,t,y));r.length>0&&a.push(u(e,t,r))}a.push(..._(o,i.get(e)??[],h,n))}return{violations:a.toSorted(o),checkedFunctionCount:l}}function o(e,n){return t.levels.indexOf(n.level)-t.levels.indexOf(e.level)||C(e.file,n.file)||e.startLine-n.startLine||r.indexOf(e.kind)-r.indexOf(n.kind)||e.endLine-n.endLine}function s({metrics:e,hunks:t}){let n=e.language===`python`;return t?e.functions.filter(e=>t.some(t=>h(t,e.startLine,e.endLine,n))):e.functions}function c(e,t,n){return i.has(e.metric)?n.some(n=>n.fn!==t&&l(n.fn,t)&&n.exceeded.some(({metric:t})=>t===e.metric)):!1}function l(e,t){let n=e.startLine<t.startLine||e.startLine===t.startLine&&e.startColumn<=t.startColumn,r=e.endLine>t.endLine||e.endLine===t.endLine&&e.endColumn>=t.endColumn;return n&&r}function u(e,t,n){let{startLine:r,endLine:i}=t,a=n.some(({metric:e})=>e===`functionCognitiveComplexity`);return{...f({file:e,startLine:r,endLine:i},`function`,n),name:d(t),...a&&t.cognitiveBlocks.length>0&&{largestBlocks:t.cognitiveBlocks}}}function d({name:e=`<anonymous>`,containerName:t}){return t?`${t}.${e}`:e}function f(e,t,n){return{kind:t,level:n.some(e=>e.level===`error`)?`error`:`warning`,...e,exceeded:n}}function p(e,t,n){return e.flatMap(e=>m(e,t,n,(e,t)=>e>t))}function m(e,n,r,i){let a=e.measure(n),o=t.levels.findLast(t=>i(a,r[t][e.key]));return o?[{metric:t.metricNameOf(e),label:e.label,value:a,level:o,limit:r[o][e.key]}]:[]}function h(e,t,n,r=!1){return e.headCount===0?e.headStart>=t&&(e.headStart<n||r&&e.headStart===n):e.headStart<=n&&e.headStart+e.headCount>t}function g(e,t){let n=/* @__PURE__ */ new Map;for(let r of e?.groups??[])for(let e of r.files)if(t.has(e)){let t=n.get(e)??[];t.push(r),n.set(e,t)}return n}function _({file:e,metrics:n,hunks:r},i,a,{minSimilarityPercent:o}){let s=e=>m(t.duplicationThreshold,{duplicatedLineCount:e},a,(e,t)=>e>=t),c=[...n.duplication.duplicateBlockGroups.map(t=>t.map(t=>({...t,file:e}))),...i.map(e=>e.occurrences.map(e=>v(e)))],l=t=>t.file===e&&s(t.lineNumbers.length).length>0&&(!r||r.some(e=>e.headCount>0&&h(e,t.startLine,t.endLine))),u=new Map(c.flatMap(e=>e.map(t=>[t,e]))),d=(e,t)=>u.get(e)===u.get(t)&&!x(e,t)&&!x(t,e);return b(c.flat().filter(e=>l(e)),d).map(({merged:e,sources:t})=>{let n=e=>b(t.flatMap(n=>{let r=Math.max(...t.filter(e=>x(e,n)).map(e=>e.tokenCount));return(u.get(n)??[]).filter(n=>n.tokenCount*100>=r*e&&!t.some(e=>x(e,n)))}),d).toSorted((e,t)=>y(t.sources)-y(e.sources)).map(e=>e.merged),r=n(o);return{...f(e,`duplication`,s(new Set(t.flatMap(e=>e.lineNumbers)).size)),partners:r.length>0?r:n(0)}})}function v(e){let{lineNumbers:t}=e;return n.default.ok(t,`No matched lines for ${e.file}, a scanned file.`),{...e,lineNumbers:t}}function y(e){return e.reduce((e,t)=>e+t.tokenCount,0)}function b(e,t){let n=[],r=e.toSorted((e,t)=>C(e.file,t.file)||e.startLine-t.startLine);for(let e of r){let r=n.at(-1);if(r&&S(r.merged,e)&&(r.sources.some(t=>x(t,e))||!r.sources.some(n=>t(n,e))))r.merged.endLine=Math.max(r.merged.endLine,e.endLine),r.sources.push(e);else{let{file:t,startLine:r,endLine:i}=e;n.push({merged:{file:t,startLine:r,endLine:i},sources:[e]})}}return n}function x(e,t){return e.file===t.file&&e.startLine<=t.startLine&&t.endLine<=e.endLine}function S(e,t){return e.file===t.file&&e.startLine<=t.endLine&&t.startLine<=e.endLine}function C(e,t){return e<t?-1:+(e>t)}exports.checkThresholds=a,exports.qualifyFunctionName=d;
 //# sourceMappingURL=thresholdCheck.cjs.map

package/dist/thresholdCheck.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"thresholdCheck.cjs","names":["functionThresholds","metricNameOf","fileThresholds","levels","duplicationThreshold"],"sources":["../src/thresholdCheck.ts"],"sourcesContent":["import assert from 'node:assert';\nimport type {\n  CrossFileDuplicateBlockGroup,\n  CrossFileDuplicateOccurrence,\n  CrossFileDuplicationMetrics,\n} from './crossFileDuplication.js';\nimport type { LineHunk } from './git.js';\nimport {\n  duplicationThreshold,\n  fileThresholds,\n  functionThresholds,\n  levels,\n  metricNameOf,\n  type Level,\n  type LimitsByLevel,\n  type Threshold,\n} from './thresholds.js';\nimport type { CodeMetrics, CognitiveBlock, FunctionMetrics } from './types.js';\n\n/** A measured file to check, under the path cross-file duplication knows it by. */\nexport interface CheckedFile {\n  file: string;\n  metrics: CodeMetrics;\n  /** The change to the file; when set, only what the change touches is checked. */\n  hunks?: readonly LineHunk[];\n}\n\n/** The duplication of the measured files, in which duplicated blocks and their copies are found. */\nexport interface CheckedDuplication {\n  crossFile?: CrossFileDuplicationMetrics;\n  /** The detection's `minSimilarityPercent`, the share of a block a copy of it holds at least. */\n  minSimilarityPercent: number;\n}\n\nexport interface ExceededLimit {\n  metric: string;\n  value: number;\n  /** The most severe level whose limit the value violates. */\n  level: Level;\n  /** That level's limit: the value is allowed up to it; a duplicated block is allowed below it. */\n  limit: number;\n}\n\nexport interface BlockLocation {\n  file: string;\n  startLine: number;\n  endLine: number;\n}\n\n/** One copy of a clone group. */\ninterface CloneOccurrence extends BlockLocation {\n  /**\n   * Its code lines matched in another copy, whose number is the length of a duplicated block. Its\n   * line span would also count the lines between the matched parts of one copy, which the copies\n   * do not share; a block matched as a whole by a near-miss copy holds all its code lines.\n   */\n  lineNumbers: readonly number[];\n  tokenCount: number;\n}\n\nexport interface Violation extends BlockLocation {\n  kind: 'function' | 'file' | 'duplication';\n  /** The most severe level among `exceeded`. */\n  level: Level;\n  /** The function's name, qualified with the class it is a member of, for kind `function`. */\n  name?: string;\n  exceeded: ExceededLimit[];\n  /**\n   * The parts adding the most to the cognitive complexity, largest first, for a function that\n   * exceeds a limit of it and has such parts: operator sequences and jumps outside any branching\n   * construct add to the complexity without being one.\n   */\n  largestBlocks?: CognitiveBlock[];\n  /** The other copies of the block, most matched tokens first, for kind `duplication`. */\n  partners?: BlockLocation[];\n}\n\nexport interface CheckResult {\n  violations: Violation[];\n  checkedFunctionCount: number;\n}\n\nconst violationKinds: readonly Violation['kind'][] = ['file', 'function', 'duplication'];\n\nconst cognitiveComplexityMetric = 'functionCognitiveComplexity';\n\nconst nestedInclusiveMetrics = new Set(\n  functionThresholds\n    .filter(({ includesNestedFunctions }) => includesNestedFunctions)\n    .map((threshold) => metricNameOf(threshold))\n);\n\n/** Every limit the files exceed, errors first, then ordered by path, then line. */\nexport function checkThresholds(\n  files: readonly CheckedFile[],\n  duplication: CheckedDuplication,\n  limitsOf: (language: string) => LimitsByLevel\n): CheckResult {\n  const crossFileGroupsByFile = indexGroupsByFile(duplication.crossFile, new Set(files.map(({ file }) => file)));\n  const violations: Violation[] = [];\n  let checkedFunctionCount = 0;\n  for (const checkedFile of files) {\n    const { file, metrics, hunks } = checkedFile;\n    const limits = limitsOf(metrics.language);\n    // A rename or a mode change lists the file as changed without touching any of its lines.\n    const fileExceeded = hunks?.length === 0 ? [] : collectExceeded(fileThresholds, metrics, limits);\n    if (fileExceeded.length > 0) {\n      violations.push(toViolation({ file, startLine: 1, endLine: metrics.lines.total }, 'file', fileExceeded));\n    }\n    const checkedFunctions = listCheckedFunctions(checkedFile);\n    checkedFunctionCount += checkedFunctions.length;\n    const exceedingFunctions = checkedFunctions\n      .map((fn) => ({ fn, exceeded: collectExceeded(functionThresholds, fn, limits) }))\n      .filter(({ exceeded }) => exceeded.length > 0);\n    for (const { fn, exceeded } of exceedingFunctions) {\n      const ownExceeded = exceeded.filter((limit) => !isReportedByEnclosingFunction(limit, fn, exceedingFunctions));\n      if (ownExceeded.length > 0) {\n        violations.push(toFunctionViolation(file, fn, ownExceeded));\n      }\n    }\n    violations.push(\n      ...collectDuplicationViolations(checkedFile, crossFileGroupsByFile.get(file) ?? [], limits, duplication)\n    );\n  }\n  return { violations: violations.toSorted(compareViolations), checkedFunctionCount };\n}\n\n// The kind is compared before the end line because a file violation's end line is corrected after\n// this ordering (to the last line that exists), and the order must not depend on that correction.\nfunction compareViolations(left: Violation, right: Violation): number {\n  return (\n    levels.indexOf(right.level) - levels.indexOf(left.level) ||\n    compareStrings(left.file, right.file) ||\n    left.startLine - right.startLine ||\n    violationKinds.indexOf(left.kind) - violationKinds.indexOf(right.kind) ||\n    left.endLine - right.endLine\n  );\n}\n\n/** The file's functions or, for a changed file, those the change touches. */\nfunction listCheckedFunctions({ metrics, hunks }: CheckedFile): FunctionMetrics[] {\n  const endsByIndentation = metrics.language === 'python';\n  return hunks\n    ? metrics.functions.filter((fn) =>\n        hunks.some((hunk) => touchesSpan(hunk, fn.startLine, fn.endLine, endsByIndentation))\n      )\n    : metrics.functions;\n}\n\n/**\n * Whether a function enclosing `fn` exceeds the same limit with a value that covers `fn`, so\n * reporting it for `fn` too would count the same code twice. The enclosing value is at least that\n * of `fn` and both share the file's limits, so its level is never milder.\n */\nfunction isReportedByEnclosingFunction(\n  limit: ExceededLimit,\n  fn: FunctionMetrics,\n  exceedingFunctions: readonly { fn: FunctionMetrics; exceeded: ExceededLimit[] }[]\n): boolean {\n  if (!nestedInclusiveMetrics.has(limit.metric)) {\n    return false;\n  }\n  return exceedingFunctions.some(\n    (outer) => outer.fn !== fn && encloses(outer.fn, fn) && outer.exceeded.some(({ metric }) => metric === limit.metric)\n  );\n}\n\nfunction encloses(outer: FunctionMetrics, inner: FunctionMetrics): boolean {\n  const startsBefore =\n    outer.startLine < inner.startLine ||\n    (outer.startLine === inner.startLine && outer.startColumn <= inner.startColumn);\n  const endsAfter =\n    outer.endLine > inner.endLine || (outer.endLine === inner.endLine && outer.endColumn >= inner.endColumn);\n  return startsBefore && endsAfter;\n}\n\nfunction toFunctionViolation(file: string, fn: FunctionMetrics, exceeded: ExceededLimit[]): Violation {\n  const { startLine, endLine } = fn;\n  const exceedsCognitiveComplexity = exceeded.some(({ metric }) => metric === cognitiveComplexityMetric);\n  return {\n    ...toViolation({ file, startLine, endLine }, 'function', exceeded),\n    name: qualifyFunctionName(fn),\n    ...(exceedsCognitiveComplexity && fn.cognitiveBlocks.length > 0 && { largestBlocks: fn.cognitiveBlocks }),\n  };\n}\n\n/** How reports name a function: with the type or module it is a member of, so that it can be found. */\nexport function qualifyFunctionName({ name = '<anonymous>', containerName }: FunctionMetrics): string {\n  return containerName ? `${containerName}.${name}` : name;\n}\n\nfunction toViolation(location: BlockLocation, kind: Violation['kind'], exceeded: ExceededLimit[]): Violation {\n  const level = exceeded.some((limit) => limit.level === 'error') ? 'error' : 'warning';\n  return { kind, level, ...location, exceeded };\n}\n\nfunction collectExceeded<Subject>(\n  thresholds: readonly Threshold<Subject>[],\n  subject: Subject,\n  limits: LimitsByLevel\n): ExceededLimit[] {\n  return thresholds.flatMap((threshold) => findExceeded(threshold, subject, limits, (value, limit) => value > limit));\n}\n\n/** The most severe violated limit of the threshold, if any. */\nfunction findExceeded<Subject>(\n  threshold: Threshold<Subject>,\n  subject: Subject,\n  limits: LimitsByLevel,\n  violates: (value: number, limit: number) => boolean\n): ExceededLimit[] {\n  const value = threshold.measure(subject);\n  const level = levels.findLast((candidate) => violates(value, limits[candidate][threshold.key] as number));\n  return level\n    ? [{ metric: metricNameOf(threshold), value, level, limit: limits[level][threshold.key] as number }]\n    : [];\n}\n\n/**\n * Whether the hunk changes the head lines [startLine, endLine]. A hunk without head lines (a pure\n * deletion) sits between its headStart line and the next one. Where indentation ends a span, its\n * last line is a statement rather than a closing delimiter, so a deletion right after that line may\n * have removed the span's tail; `endsByIndentation` counts it as touching, since the head alone\n * cannot tell it from a deletion after the span.\n */\nfunction touchesSpan(hunk: LineHunk, startLine: number, endLine: number, endsByIndentation = false): boolean {\n  return hunk.headCount === 0\n    ? hunk.headStart >= startLine && (hunk.headStart < endLine || (endsByIndentation && hunk.headStart === endLine))\n    : hunk.headStart <= endLine && hunk.headStart + hunk.headCount > startLine;\n}\n\n/** The cross-file groups with an occurrence in each of the given files. */\nfunction indexGroupsByFile(\n  crossFileDuplication: CrossFileDuplicationMetrics | undefined,\n  files: ReadonlySet<string>\n): Map<string, CrossFileDuplicateBlockGroup[]> {\n  const groupsByFile = new Map<string, CrossFileDuplicateBlockGroup[]>();\n  for (const group of crossFileDuplication?.groups ?? []) {\n    for (const file of group.files) {\n      if (files.has(file)) {\n        const groups = groupsByFile.get(file) ?? [];\n        groups.push(group);\n        groupsByFile.set(file, groups);\n      }\n    }\n  }\n  return groupsByFile;\n}\n\n/**\n * One violation per duplicated region of the file: its clone occurrences, within-file and\n * cross-file, that hold at least the duplicated lines of a level (with `hunks`, only those\n * overlapping added lines), with overlapping occurrences merged into one region.\n */\nfunction collectDuplicationViolations(\n  { file, metrics, hunks }: CheckedFile,\n  crossFileGroups: readonly CrossFileDuplicateBlockGroup[],\n  limits: LimitsByLevel,\n  { minSimilarityPercent }: CheckedDuplication\n): Violation[] {\n  const findExceededLines = (duplicatedLineCount: number): ExceededLimit[] =>\n    findExceeded(duplicationThreshold, { duplicatedLineCount }, limits, (value, limit) => value >= limit);\n  const groups: CloneOccurrence[][] = [\n    ...metrics.duplication.duplicateBlockGroups.map((group) => group.map((occurrence) => ({ ...occurrence, file }))),\n    ...crossFileGroups.map((group) => group.occurrences.map((occurrence) => toCloneOccurrence(occurrence))),\n  ];\n  const isReported = (block: CloneOccurrence): boolean =>\n    block.file === file &&\n    findExceededLines(block.lineNumbers.length).length > 0 &&\n    (!hunks || hunks.some((hunk) => hunk.headCount > 0 && touchesSpan(hunk, block.startLine, block.endLine)));\n  const groupByBlock = new Map<CloneOccurrence, CloneOccurrence[]>();\n  for (const group of groups) {\n    for (const block of group.filter((occurrence) => isReported(occurrence))) {\n      groupByBlock.set(block, group);\n    }\n  }\n  return mergeOverlapping([...groupByBlock.keys()]).map(({ merged, sources }) => {\n    const listCopies = (minTokenPercent: number): BlockLocation[] =>\n      mergeOverlapping(\n        sources.flatMap((source) => {\n          // The region merges every clone that overlaps it, so a copy of an occurrence lying\n          // within a larger one is measured against the larger one.\n          const tokenCount = Math.max(\n            ...sources.filter((outer) => contains(outer, source)).map((outer) => outer.tokenCount)\n          );\n          return (groupByBlock.get(source) ?? []).filter(\n            (copy) => copy.tokenCount * 100 >= tokenCount * minTokenPercent\n          );\n        })\n      )\n        // The region's own occurrences are among the copies of the groups it belongs to.\n        .filter((partner) => !overlaps(partner.merged, merged))\n        // The largest say the most about what to share.\n        .toSorted((left, right) => sumTokens(right.sources) - sumTokens(left.sources))\n        .map((partner) => partner.merged);\n    // A group joins a block with every place repeating a part of it: a function another file\n    // copies whole is grouped with each file holding a few of its lines. Such a fragment is not a\n    // copy of the block: sharing it would not remove the block, and a report listing it reads as\n    // one more copy. Sizes are compared in tokens, since the same code may be laid out over any\n    // number of lines.\n    const copies = listCopies(minSimilarityPercent);\n    return {\n      ...toViolation(\n        merged,\n        'duplication',\n        // Occurrences that overlap share lines, which count once.\n        findExceededLines(new Set(sources.flatMap((source) => source.lineNumbers)).size)\n      ),\n      // A group does not always hold the copy a block was matched with as a whole: within a file,\n      // a copy that already belongs to an exact group is listed as its part in that group. The\n      // fragments then are all that locates the block's copies.\n      partners: copies.length > 0 ? copies : listCopies(0),\n    };\n  });\n}\n\nfunction toCloneOccurrence(occurrence: CrossFileDuplicateOccurrence): CloneOccurrence {\n  const { lineNumbers } = occurrence;\n  assert.ok(lineNumbers, `No matched lines for ${occurrence.file}, a scanned file.`);\n  return { ...occurrence, lineNumbers };\n}\n\nfunction sumTokens(occurrences: readonly CloneOccurrence[]): number {\n  return occurrences.reduce((sum, occurrence) => sum + occurrence.tokenCount, 0);\n}\n\n/** The locations with those overlapping in the same file merged, ordered by file, then line. */\nfunction mergeOverlapping(\n  locations: readonly CloneOccurrence[]\n): { merged: BlockLocation; sources: CloneOccurrence[] }[] {\n  const regions: { merged: BlockLocation; sources: CloneOccurrence[] }[] = [];\n  const ordered = locations.toSorted(\n    (left, right) => compareStrings(left.file, right.file) || left.startLine - right.startLine\n  );\n  for (const location of ordered) {\n    const last = regions.at(-1);\n    if (last && overlaps(last.merged, location)) {\n      last.merged.endLine = Math.max(last.merged.endLine, location.endLine);\n      last.sources.push(location);\n    } else {\n      const { file, startLine, endLine } = location;\n      regions.push({ merged: { file, startLine, endLine }, sources: [location] });\n    }\n  }\n  return regions;\n}\n\n/** Whether `outer` spans every line of `inner`; a block contains itself. */\nfunction contains(outer: BlockLocation, inner: BlockLocation): boolean {\n  return outer.file === inner.file && outer.startLine <= inner.startLine && inner.endLine <= outer.endLine;\n}\n\nfunction overlaps(left: BlockLocation, right: BlockLocation): boolean {\n  return left.file === right.file && left.startLine <= right.endLine && right.startLine <= left.endLine;\n}\n\n/** Code-unit order, so reports do not depend on the locale. */\nfunction compareStrings(left: string, right: string): number {\n  return left < right ? -1 : left > right ? 1 : 0;\n}\n"],"mappings":"6IAkFA,MAAM,EAA+C,CAAC,OAAQ,WAAY,aAAa,EAIjF,EAAyB,IAAI,IACjCA,EAAAA,mBACG,QAAQ,CAAE,6BAA8B,CAAuB,CAAC,CAChE,IAAK,GAAcC,EAAAA,aAAa,CAAS,CAAC,CAC/C,EAGA,SAAgB,EACd,EACA,EACA,EACa,CACb,IAAM,EAAwB,EAAkB,EAAY,UAAW,IAAI,IAAI,EAAM,KAAK,CAAE,UAAW,CAAI,CAAC,CAAC,EACvG,EAA0B,CAAC,EAC7B,EAAuB,EAC3B,IAAK,IAAM,KAAe,EAAO,CAC/B,GAAM,CAAE,OAAM,UAAS,SAAU,EAC3B,EAAS,EAAS,EAAQ,QAAQ,EAElC,EAAe,GAAO,SAAW,EAAI,CAAC,EAAI,EAAgBC,EAAAA,eAAgB,EAAS,CAAM,EAC3F,EAAa,OAAS,GACxB,EAAW,KAAK,EAAY,CAAE,OAAM,UAAW,EAAG,QAAS,EAAQ,MAAM,KAAM,EAAG,OAAQ,CAAY,CAAC,EAEzG,IAAM,EAAmB,EAAqB,CAAW,EACzD,GAAwB,EAAiB,OACzC,IAAM,EAAqB,EACxB,IAAK,IAAQ,CAAE,KAAI,SAAU,EAAgBF,EAAAA,mBAAoB,EAAI,CAAM,CAAE,EAAE,CAAC,CAChF,QAAQ,CAAE,cAAe,EAAS,OAAS,CAAC,EAC/C,IAAK,GAAM,CAAE,KAAI,cAAc,EAAoB,CACjD,IAAM,EAAc,EAAS,OAAQ,GAAU,CAAC,EAA8B,EAAO,EAAI,CAAkB,CAAC,EACxG,EAAY,OAAS,GACvB,EAAW,KAAK,EAAoB,EAAM,EAAI,CAAW,CAAC,CAE9D,CACA,EAAW,KACT,GAAG,EAA6B,EAAa,EAAsB,IAAI,CAAI,GAAK,CAAC,EAAG,EAAQ,CAAW,CACzG,CACF,CACA,MAAO,CAAE,WAAY,EAAW,SAAS,CAAiB,EAAG,sBAAqB,CACpF,CAIA,SAAS,EAAkB,EAAiB,EAA0B,CACpE,OACEG,EAAAA,OAAO,QAAQ,EAAM,KAAK,EAAIA,EAAAA,OAAO,QAAQ,EAAK,KAAK,GACvD,EAAe,EAAK,KAAM,EAAM,IAAI,GACpC,EAAK,UAAY,EAAM,WACvB,EAAe,QAAQ,EAAK,IAAI,EAAI,EAAe,QAAQ,EAAM,IAAI,GACrE,EAAK,QAAU,EAAM,OAEzB,CAGA,SAAS,EAAqB,CAAE,UAAS,SAAyC,CAChF,IAAM,EAAoB,EAAQ,WAAa,SAC/C,OAAO,EACH,EAAQ,UAAU,OAAQ,GACxB,EAAM,KAAM,GAAS,EAAY,EAAM,EAAG,UAAW,EAAG,QAAS,CAAiB,CAAC,CACrF,EACA,EAAQ,SACd,CAOA,SAAS,EACP,EACA,EACA,EACS,CAIT,OAHK,EAAuB,IAAI,EAAM,MAAM,EAGrC,EAAmB,KACvB,GAAU,EAAM,KAAO,GAAM,EAAS,EAAM,GAAI,CAAE,GAAK,EAAM,SAAS,MAAM,CAAE,YAAa,IAAW,EAAM,MAAM,CACrH,EAJS,EAKX,CAEA,SAAS,EAAS,EAAwB,EAAiC,CACzE,IAAM,EACJ,EAAM,UAAY,EAAM,WACvB,EAAM,YAAc,EAAM,WAAa,EAAM,aAAe,EAAM,YAC/D,EACJ,EAAM,QAAU,EAAM,SAAY,EAAM,UAAY,EAAM,SAAW,EAAM,WAAa,EAAM,UAChG,OAAO,GAAgB,CACzB,CAEA,SAAS,EAAoB,EAAc,EAAqB,EAAsC,CACpG,GAAM,CAAE,YAAW,WAAY,EACzB,EAA6B,EAAS,MAAM,CAAE,YAAa,IAAW,6BAAyB,EACrG,MAAO,CACL,GAAG,EAAY,CAAE,OAAM,YAAW,SAAQ,EAAG,WAAY,CAAQ,EACjE,KAAM,EAAoB,CAAE,EAC5B,GAAI,GAA8B,EAAG,gBAAgB,OAAS,GAAK,CAAE,cAAe,EAAG,eAAgB,CACzG,CACF,CAGA,SAAgB,EAAoB,CAAE,OAAO,cAAe,iBAA0C,CACpG,OAAO,EAAgB,GAAG,EAAc,GAAG,IAAS,CACtD,CAEA,SAAS,EAAY,EAAyB,EAAyB,EAAsC,CAE3G,MAAO,CAAE,OAAM,MADD,EAAS,KAAM,GAAU,EAAM,QAAU,OAAO,EAAI,QAAU,UACtD,GAAG,EAAU,UAAS,CAC9C,CAEA,SAAS,EACP,EACA,EACA,EACiB,CACjB,OAAO,EAAW,QAAS,GAAc,EAAa,EAAW,EAAS,GAAS,EAAO,IAAU,EAAQ,CAAK,CAAC,CACpH,CAGA,SAAS,EACP,EACA,EACA,EACA,EACiB,CACjB,IAAM,EAAQ,EAAU,QAAQ,CAAO,EACjC,EAAQA,EAAAA,OAAO,SAAU,GAAc,EAAS,EAAO,EAAO,EAAU,CAAC,EAAU,IAAc,CAAC,EACxG,OAAO,EACH,CAAC,CAAE,OAAQF,EAAAA,aAAa,CAAS,EAAG,QAAO,QAAO,MAAO,EAAO,EAAM,CAAC,EAAU,IAAe,CAAC,EACjG,CAAC,CACP,CASA,SAAS,EAAY,EAAgB,EAAmB,EAAiB,EAAoB,GAAgB,CAC3G,OAAO,EAAK,YAAc,EACtB,EAAK,WAAa,IAAc,EAAK,UAAY,GAAY,GAAqB,EAAK,YAAc,GACrG,EAAK,WAAa,GAAW,EAAK,UAAY,EAAK,UAAY,CACrE,CAGA,SAAS,EACP,EACA,EAC6C,CAC7C,IAAM,kBAAe,IAAI,IACzB,IAAK,IAAM,KAAS,GAAsB,QAAU,CAAC,EACnD,IAAK,IAAM,KAAQ,EAAM,MACvB,GAAI,EAAM,IAAI,CAAI,EAAG,CACnB,IAAM,EAAS,EAAa,IAAI,CAAI,GAAK,CAAC,EAC1C,EAAO,KAAK,CAAK,EACjB,EAAa,IAAI,EAAM,CAAM,CAC/B,CAGJ,OAAO,CACT,CAOA,SAAS,EACP,CAAE,OAAM,UAAS,SACjB,EACA,EACA,CAAE,wBACW,CACb,IAAM,EAAqB,GACzB,EAAaG,EAAAA,qBAAsB,CAAE,qBAAoB,EAAG,GAAS,EAAO,IAAU,GAAS,CAAK,EAChG,EAA8B,CAClC,GAAG,EAAQ,YAAY,qBAAqB,IAAK,GAAU,EAAM,IAAK,IAAgB,CAAE,GAAG,EAAY,MAAK,EAAE,CAAC,EAC/G,GAAG,EAAgB,IAAK,GAAU,EAAM,YAAY,IAAK,GAAe,EAAkB,CAAU,CAAC,CAAC,CACxG,EACM,EAAc,GAClB,EAAM,OAAS,GACf,EAAkB,EAAM,YAAY,MAAM,CAAC,CAAC,OAAS,IACpD,CAAC,GAAS,EAAM,KAAM,GAAS,EAAK,UAAY,GAAK,EAAY,EAAM,EAAM,UAAW,EAAM,OAAO,CAAC,GACnG,kBAAe,IAAI,IACzB,IAAK,IAAM,KAAS,EAClB,IAAK,IAAM,KAAS,EAAM,OAAQ,GAAe,EAAW,CAAU,CAAC,EACrE,EAAa,IAAI,EAAO,CAAK,EAGjC,OAAO,EAAiB,CAAC,GAAG,EAAa,KAAK,CAAC,CAAC,CAAC,CAAC,KAAK,CAAE,SAAQ,aAAc,CAC7E,IAAM,EAAc,GAClB,EACE,EAAQ,QAAS,GAAW,CAG1B,IAAM,EAAa,KAAK,IACtB,GAAG,EAAQ,OAAQ,GAAU,EAAS,EAAO,CAAM,CAAC,CAAC,CAAC,IAAK,GAAU,EAAM,UAAU,CACvF,EACA,OAAQ,EAAa,IAAI,CAAM,GAAK,CAAC,EAAA,CAAG,OACrC,GAAS,EAAK,WAAa,KAAO,EAAa,CAClD,CACF,CAAC,CACH,CAAC,CAEE,OAAQ,GAAY,CAAC,EAAS,EAAQ,OAAQ,CAAM,CAAC,CAAC,CAEtD,UAAU,EAAM,IAAU,EAAU,EAAM,OAAO,EAAI,EAAU,EAAK,OAAO,CAAC,CAAC,CAC7E,IAAK,GAAY,EAAQ,MAAM,EAM9B,EAAS,EAAW,CAAoB,EAC9C,MAAO,CACL,GAAG,EACD,EACA,cAEA,EAAkB,IAAI,IAAI,EAAQ,QAAS,GAAW,EAAO,WAAW,CAAC,CAAC,CAAC,IAAI,CACjF,EAIA,SAAU,EAAO,OAAS,EAAI,EAAS,EAAW,CAAC,CACrD,CACF,CAAC,CACH,CAEA,SAAS,EAAkB,EAA2D,CACpF,GAAM,CAAE,eAAgB,EAExB,OADA,EAAA,QAAO,GAAG,EAAa,wBAAwB,EAAW,KAAK,kBAAkB,EAC1E,CAAE,GAAG,EAAY,aAAY,CACtC,CAEA,SAAS,EAAU,EAAiD,CAClE,OAAO,EAAY,QAAQ,EAAK,IAAe,EAAM,EAAW,WAAY,CAAC,CAC/E,CAGA,SAAS,EACP,EACyD,CACzD,IAAM,EAAmE,CAAC,EACpE,EAAU,EAAU,UACvB,EAAM,IAAU,EAAe,EAAK,KAAM,EAAM,IAAI,GAAK,EAAK,UAAY,EAAM,SACnF,EACA,IAAK,IAAM,KAAY,EAAS,CAC9B,IAAM,EAAO,EAAQ,GAAG,EAAE,EAC1B,GAAI,GAAQ,EAAS,EAAK,OAAQ,CAAQ,EACxC,EAAK,OAAO,QAAU,KAAK,IAAI,EAAK,OAAO,QAAS,EAAS,OAAO,EACpE,EAAK,QAAQ,KAAK,CAAQ,MACrB,CACL,GAAM,CAAE,OAAM,YAAW,WAAY,EACrC,EAAQ,KAAK,CAAE,OAAQ,CAAE,OAAM,YAAW,SAAQ,EAAG,QAAS,CAAC,CAAQ,CAAE,CAAC,CAC5E,CACF,CACA,OAAO,CACT,CAGA,SAAS,EAAS,EAAsB,EAA+B,CACrE,OAAO,EAAM,OAAS,EAAM,MAAQ,EAAM,WAAa,EAAM,WAAa,EAAM,SAAW,EAAM,OACnG,CAEA,SAAS,EAAS,EAAqB,EAA+B,CACpE,OAAO,EAAK,OAAS,EAAM,MAAQ,EAAK,WAAa,EAAM,SAAW,EAAM,WAAa,EAAK,OAChG,CAGA,SAAS,EAAe,EAAc,EAAuB,CAC3D,OAAO,EAAO,EAAQ,GAAK,IAAO,EACpC"}
+{"version":3,"file":"thresholdCheck.cjs","names":["functionThresholds","metricNameOf","fileThresholds","levels","duplicationThreshold"],"sources":["../src/thresholdCheck.ts"],"sourcesContent":["import assert from 'node:assert';\nimport type {\n  CrossFileDuplicateBlockGroup,\n  CrossFileDuplicateOccurrence,\n  CrossFileDuplicationMetrics,\n} from './crossFileDuplication.js';\nimport type { LineHunk } from './git.js';\nimport {\n  duplicationThreshold,\n  fileThresholds,\n  functionThresholds,\n  levels,\n  metricNameOf,\n  type Level,\n  type LimitsByLevel,\n  type Threshold,\n} from './thresholds.js';\nimport type { CodeMetrics, CognitiveBlock, FunctionMetrics } from './types.js';\n\n/** A measured file to check, under the path cross-file duplication knows it by. */\nexport interface CheckedFile {\n  file: string;\n  metrics: CodeMetrics;\n  /** The change to the file; when set, only what the change touches is checked. */\n  hunks?: readonly LineHunk[];\n}\n\n/** The duplication of the measured files, in which duplicated blocks and their copies are found. */\nexport interface CheckedDuplication {\n  crossFile?: CrossFileDuplicationMetrics;\n  /** The detection's `minSimilarityPercent`, the share of a block a copy of it holds at least. */\n  minSimilarityPercent: number;\n}\n\nexport interface ExceededLimit {\n  metric: string;\n  /** What reports call the metric (`cognitive complexity`, `duplicated lines`). */\n  label: string;\n  value: number;\n  /** The most severe level whose limit the value violates. */\n  level: Level;\n  /** That level's limit: the value is allowed up to it; a duplicated block is allowed below it. */\n  limit: number;\n}\n\nexport interface BlockLocation {\n  file: string;\n  startLine: number;\n  endLine: number;\n}\n\n/** One copy of a clone group. */\ninterface CloneOccurrence extends BlockLocation {\n  /**\n   * Its code lines matched in another copy, whose number is the length of a duplicated block. Its\n   * line span would also count the lines the copies do not share.\n   */\n  lineNumbers: readonly number[];\n  tokenCount: number;\n}\n\nexport interface Violation extends BlockLocation {\n  kind: 'function' | 'file' | 'duplication';\n  /** The most severe level among `exceeded`. */\n  level: Level;\n  /** The function's name, qualified with the class it is a member of, for kind `function`. */\n  name?: string;\n  exceeded: ExceededLimit[];\n  /**\n   * The parts adding the most to the cognitive complexity, largest first, for a function that\n   * exceeds a limit of it and has such parts: operator sequences and jumps outside any branching\n   * construct add to the complexity without being one.\n   */\n  largestBlocks?: CognitiveBlock[];\n  /** The other copies of the block, most matched tokens first, for kind `duplication`. */\n  partners?: BlockLocation[];\n}\n\nexport interface CheckResult {\n  violations: Violation[];\n  checkedFunctionCount: number;\n}\n\nconst violationKinds: readonly Violation['kind'][] = ['file', 'function', 'duplication'];\n\nconst cognitiveComplexityMetric = 'functionCognitiveComplexity';\n\nconst nestedInclusiveMetrics = new Set(\n  functionThresholds\n    .filter(({ includesNestedFunctions }) => includesNestedFunctions)\n    .map((threshold) => metricNameOf(threshold))\n);\n\n/** Every limit the files exceed, errors first, then ordered by path, then line. */\nexport function checkThresholds(\n  files: readonly CheckedFile[],\n  duplication: CheckedDuplication,\n  limitsOf: (language: string) => LimitsByLevel\n): CheckResult {\n  const crossFileGroupsByFile = indexGroupsByFile(duplication.crossFile, new Set(files.map(({ file }) => file)));\n  const violations: Violation[] = [];\n  let checkedFunctionCount = 0;\n  for (const checkedFile of files) {\n    const { file, metrics, hunks } = checkedFile;\n    const limits = limitsOf(metrics.language);\n    // A rename or a mode change lists the file as changed without touching any of its lines.\n    const fileExceeded = hunks?.length === 0 ? [] : collectExceeded(fileThresholds, metrics, limits);\n    if (fileExceeded.length > 0) {\n      violations.push(toViolation({ file, startLine: 1, endLine: metrics.lines.total }, 'file', fileExceeded));\n    }\n    const checkedFunctions = listCheckedFunctions(checkedFile);\n    checkedFunctionCount += checkedFunctions.length;\n    const exceedingFunctions = checkedFunctions\n      .map((fn) => ({ fn, exceeded: collectExceeded(functionThresholds, fn, limits) }))\n      .filter(({ exceeded }) => exceeded.length > 0);\n    for (const { fn, exceeded } of exceedingFunctions) {\n      const ownExceeded = exceeded.filter((limit) => !isReportedByEnclosingFunction(limit, fn, exceedingFunctions));\n      if (ownExceeded.length > 0) {\n        violations.push(toFunctionViolation(file, fn, ownExceeded));\n      }\n    }\n    violations.push(\n      ...collectDuplicationViolations(checkedFile, crossFileGroupsByFile.get(file) ?? [], limits, duplication)\n    );\n  }\n  return { violations: violations.toSorted(compareViolations), checkedFunctionCount };\n}\n\n// The kind is compared before the end line because a file violation's end line is corrected after\n// this ordering (to the last line that exists), and the order must not depend on that correction.\nfunction compareViolations(left: Violation, right: Violation): number {\n  return (\n    levels.indexOf(right.level) - levels.indexOf(left.level) ||\n    compareStrings(left.file, right.file) ||\n    left.startLine - right.startLine ||\n    violationKinds.indexOf(left.kind) - violationKinds.indexOf(right.kind) ||\n    left.endLine - right.endLine\n  );\n}\n\n/** The file's functions or, for a changed file, those the change touches. */\nfunction listCheckedFunctions({ metrics, hunks }: CheckedFile): FunctionMetrics[] {\n  const endsByIndentation = metrics.language === 'python';\n  return hunks\n    ? metrics.functions.filter((fn) =>\n        hunks.some((hunk) => touchesSpan(hunk, fn.startLine, fn.endLine, endsByIndentation))\n      )\n    : metrics.functions;\n}\n\n/**\n * Whether a function enclosing `fn` exceeds the same limit with a value that covers `fn`, so\n * reporting it for `fn` too would count the same code twice. The enclosing value is at least that\n * of `fn` and both share the file's limits, so its level is never milder.\n */\nfunction isReportedByEnclosingFunction(\n  limit: ExceededLimit,\n  fn: FunctionMetrics,\n  exceedingFunctions: readonly { fn: FunctionMetrics; exceeded: ExceededLimit[] }[]\n): boolean {\n  if (!nestedInclusiveMetrics.has(limit.metric)) {\n    return false;\n  }\n  return exceedingFunctions.some(\n    (outer) => outer.fn !== fn && encloses(outer.fn, fn) && outer.exceeded.some(({ metric }) => metric === limit.metric)\n  );\n}\n\nfunction encloses(outer: FunctionMetrics, inner: FunctionMetrics): boolean {\n  const startsBefore =\n    outer.startLine < inner.startLine ||\n    (outer.startLine === inner.startLine && outer.startColumn <= inner.startColumn);\n  const endsAfter =\n    outer.endLine > inner.endLine || (outer.endLine === inner.endLine && outer.endColumn >= inner.endColumn);\n  return startsBefore && endsAfter;\n}\n\nfunction toFunctionViolation(file: string, fn: FunctionMetrics, exceeded: ExceededLimit[]): Violation {\n  const { startLine, endLine } = fn;\n  const exceedsCognitiveComplexity = exceeded.some(({ metric }) => metric === cognitiveComplexityMetric);\n  return {\n    ...toViolation({ file, startLine, endLine }, 'function', exceeded),\n    name: qualifyFunctionName(fn),\n    ...(exceedsCognitiveComplexity && fn.cognitiveBlocks.length > 0 && { largestBlocks: fn.cognitiveBlocks }),\n  };\n}\n\n/** How reports name a function: with the type or module it is a member of, so that it can be found. */\nexport function qualifyFunctionName({ name = '<anonymous>', containerName }: FunctionMetrics): string {\n  return containerName ? `${containerName}.${name}` : name;\n}\n\nfunction toViolation(location: BlockLocation, kind: Violation['kind'], exceeded: ExceededLimit[]): Violation {\n  const level = exceeded.some((limit) => limit.level === 'error') ? 'error' : 'warning';\n  return { kind, level, ...location, exceeded };\n}\n\nfunction collectExceeded<Subject>(\n  thresholds: readonly Threshold<Subject>[],\n  subject: Subject,\n  limits: LimitsByLevel\n): ExceededLimit[] {\n  return thresholds.flatMap((threshold) => findExceeded(threshold, subject, limits, (value, limit) => value > limit));\n}\n\n/** The most severe violated limit of the threshold, if any. */\nfunction findExceeded<Subject>(\n  threshold: Threshold<Subject>,\n  subject: Subject,\n  limits: LimitsByLevel,\n  violates: (value: number, limit: number) => boolean\n): ExceededLimit[] {\n  const value = threshold.measure(subject);\n  const level = levels.findLast((candidate) => violates(value, limits[candidate][threshold.key] as number));\n  return level\n    ? [\n        {\n          metric: metricNameOf(threshold),\n          label: threshold.label,\n          value,\n          level,\n          limit: limits[level][threshold.key] as number,\n        },\n      ]\n    : [];\n}\n\n/**\n * Whether the hunk changes the head lines [startLine, endLine]. A hunk without head lines (a pure\n * deletion) sits between its headStart line and the next one. Where indentation ends a span, its\n * last line is a statement rather than a closing delimiter, so a deletion right after that line may\n * have removed the span's tail; `endsByIndentation` counts it as touching, since the head alone\n * cannot tell it from a deletion after the span.\n */\nfunction touchesSpan(hunk: LineHunk, startLine: number, endLine: number, endsByIndentation = false): boolean {\n  return hunk.headCount === 0\n    ? hunk.headStart >= startLine && (hunk.headStart < endLine || (endsByIndentation && hunk.headStart === endLine))\n    : hunk.headStart <= endLine && hunk.headStart + hunk.headCount > startLine;\n}\n\n/** The cross-file groups with an occurrence in each of the given files. */\nfunction indexGroupsByFile(\n  crossFileDuplication: CrossFileDuplicationMetrics | undefined,\n  files: ReadonlySet<string>\n): Map<string, CrossFileDuplicateBlockGroup[]> {\n  const groupsByFile = new Map<string, CrossFileDuplicateBlockGroup[]>();\n  for (const group of crossFileDuplication?.groups ?? []) {\n    for (const file of group.files) {\n      if (files.has(file)) {\n        const groups = groupsByFile.get(file) ?? [];\n        groups.push(group);\n        groupsByFile.set(file, groups);\n      }\n    }\n  }\n  return groupsByFile;\n}\n\n/**\n * One violation per duplicated region of the file: its clone occurrences, within-file and\n * cross-file, that hold at least the duplicated lines of a level (with `hunks`, only those\n * overlapping added lines), with overlapping occurrences merged into one region unless they are\n * copies of each other.\n */\nfunction collectDuplicationViolations(\n  { file, metrics, hunks }: CheckedFile,\n  crossFileGroups: readonly CrossFileDuplicateBlockGroup[],\n  limits: LimitsByLevel,\n  { minSimilarityPercent }: CheckedDuplication\n): Violation[] {\n  const findExceededLines = (duplicatedLineCount: number): ExceededLimit[] =>\n    findExceeded(duplicationThreshold, { duplicatedLineCount }, limits, (value, limit) => value >= limit);\n  const groups: CloneOccurrence[][] = [\n    ...metrics.duplication.duplicateBlockGroups.map((group) => group.map((occurrence) => ({ ...occurrence, file }))),\n    ...crossFileGroups.map((group) => group.occurrences.map((occurrence) => toCloneOccurrence(occurrence))),\n  ];\n  const isReported = (block: CloneOccurrence): boolean =>\n    block.file === file &&\n    findExceededLines(block.lineNumbers.length).length > 0 &&\n    (!hunks || hunks.some((hunk) => hunk.headCount > 0 && touchesSpan(hunk, block.startLine, block.endLine)));\n  const groupOf = new Map(groups.flatMap((group) => group.map((occurrence) => [occurrence, group] as const)));\n  // Copies of one group that share a line (the branches of an `else if` chain, the entries of a\n  // list) stay apart: merged, they would be one block without a copy. One lying within another is\n  // a repetition inside that block, not a block beside it.\n  const areCopies = (left: CloneOccurrence, right: CloneOccurrence): boolean =>\n    groupOf.get(left) === groupOf.get(right) && !contains(left, right) && !contains(right, left);\n  const reportedBlocks = groups.flat().filter((occurrence) => isReported(occurrence));\n  return mergeOverlapping(reportedBlocks, areCopies).map(({ merged, sources }) => {\n    const listCopies = (minTokenPercent: number): BlockLocation[] =>\n      mergeOverlapping(\n        sources.flatMap((source) => {\n          // The region merges every clone that overlaps it, so a copy of an occurrence lying\n          // within a larger one is measured against the larger one.\n          const tokenCount = Math.max(\n            ...sources.filter((outer) => contains(outer, source)).map((outer) => outer.tokenCount)\n          );\n          return (groupOf.get(source) ?? []).filter(\n            (copy) =>\n              copy.tokenCount * 100 >= tokenCount * minTokenPercent &&\n              // The region's own occurrences are among the copies of the groups it belongs to, and\n              // so may be an occurrence within one of them that is not reported itself: a\n              // repetition inside the block, not another place.\n              !sources.some((outer) => contains(outer, copy))\n          );\n        }),\n        areCopies\n      )\n        // The largest say the most about what to share.\n        .toSorted((left, right) => sumTokens(right.sources) - sumTokens(left.sources))\n        .map((partner) => partner.merged);\n    // A group joins a block with every place repeating a part of it: a function another file\n    // copies whole is grouped with each file holding a few of its lines. Such a fragment is not a\n    // copy of the block: sharing it would not remove the block, and a report listing it reads as\n    // one more copy. Sizes are compared in tokens, since the same code may be laid out over any\n    // number of lines.\n    const copies = listCopies(minSimilarityPercent);\n    return {\n      ...toViolation(\n        merged,\n        'duplication',\n        // Occurrences that overlap share lines, which count once.\n        findExceededLines(new Set(sources.flatMap((source) => source.lineNumbers)).size)\n      ),\n      // A block whose parts were matched with different places has no copy of its size. The\n      // places then are all that locates what it repeats.\n      partners: copies.length > 0 ? copies : listCopies(0),\n    };\n  });\n}\n\nfunction toCloneOccurrence(occurrence: CrossFileDuplicateOccurrence): CloneOccurrence {\n  const { lineNumbers } = occurrence;\n  assert.ok(lineNumbers, `No matched lines for ${occurrence.file}, a scanned file.`);\n  return { ...occurrence, lineNumbers };\n}\n\nfunction sumTokens(occurrences: readonly CloneOccurrence[]): number {\n  return occurrences.reduce((sum, occurrence) => sum + occurrence.tokenCount, 0);\n}\n\n/**\n * The locations with those overlapping in the same file merged, except those `areCopies` holds\n * for, ordered by file, then line.\n */\nfunction mergeOverlapping(\n  locations: readonly CloneOccurrence[],\n  areCopies: (left: CloneOccurrence, right: CloneOccurrence) => boolean\n): { merged: BlockLocation; sources: CloneOccurrence[] }[] {\n  const regions: { merged: BlockLocation; sources: CloneOccurrence[] }[] = [];\n  const ordered = locations.toSorted(\n    (left, right) => compareStrings(left.file, right.file) || left.startLine - right.startLine\n  );\n  for (const location of ordered) {\n    const last = regions.at(-1);\n    // A block containing the location holds it whatever it is a copy of.\n    const joins =\n      last &&\n      overlaps(last.merged, location) &&\n      (last.sources.some((source) => contains(source, location)) ||\n        !last.sources.some((source) => areCopies(source, location)));\n    if (joins) {\n      last.merged.endLine = Math.max(last.merged.endLine, location.endLine);\n      last.sources.push(location);\n    } else {\n      const { file, startLine, endLine } = location;\n      regions.push({ merged: { file, startLine, endLine }, sources: [location] });\n    }\n  }\n  return regions;\n}\n\n/** Whether `outer` spans every line of `inner`; a block contains itself. */\nfunction contains(outer: BlockLocation, inner: BlockLocation): boolean {\n  return outer.file === inner.file && outer.startLine <= inner.startLine && inner.endLine <= outer.endLine;\n}\n\nfunction overlaps(left: BlockLocation, right: BlockLocation): boolean {\n  return left.file === right.file && left.startLine <= right.endLine && right.startLine <= left.endLine;\n}\n\n/** Code-unit order, so reports do not depend on the locale. */\nfunction compareStrings(left: string, right: string): number {\n  return left < right ? -1 : left > right ? 1 : 0;\n}\n"],"mappings":"6IAmFA,MAAM,EAA+C,CAAC,OAAQ,WAAY,aAAa,EAIjF,EAAyB,IAAI,IACjCA,EAAAA,mBACG,QAAQ,CAAE,6BAA8B,CAAuB,CAAC,CAChE,IAAK,GAAcC,EAAAA,aAAa,CAAS,CAAC,CAC/C,EAGA,SAAgB,EACd,EACA,EACA,EACa,CACb,IAAM,EAAwB,EAAkB,EAAY,UAAW,IAAI,IAAI,EAAM,KAAK,CAAE,UAAW,CAAI,CAAC,CAAC,EACvG,EAA0B,CAAC,EAC7B,EAAuB,EAC3B,IAAK,IAAM,KAAe,EAAO,CAC/B,GAAM,CAAE,OAAM,UAAS,SAAU,EAC3B,EAAS,EAAS,EAAQ,QAAQ,EAElC,EAAe,GAAO,SAAW,EAAI,CAAC,EAAI,EAAgBC,EAAAA,eAAgB,EAAS,CAAM,EAC3F,EAAa,OAAS,GACxB,EAAW,KAAK,EAAY,CAAE,OAAM,UAAW,EAAG,QAAS,EAAQ,MAAM,KAAM,EAAG,OAAQ,CAAY,CAAC,EAEzG,IAAM,EAAmB,EAAqB,CAAW,EACzD,GAAwB,EAAiB,OACzC,IAAM,EAAqB,EACxB,IAAK,IAAQ,CAAE,KAAI,SAAU,EAAgBF,EAAAA,mBAAoB,EAAI,CAAM,CAAE,EAAE,CAAC,CAChF,QAAQ,CAAE,cAAe,EAAS,OAAS,CAAC,EAC/C,IAAK,GAAM,CAAE,KAAI,cAAc,EAAoB,CACjD,IAAM,EAAc,EAAS,OAAQ,GAAU,CAAC,EAA8B,EAAO,EAAI,CAAkB,CAAC,EACxG,EAAY,OAAS,GACvB,EAAW,KAAK,EAAoB,EAAM,EAAI,CAAW,CAAC,CAE9D,CACA,EAAW,KACT,GAAG,EAA6B,EAAa,EAAsB,IAAI,CAAI,GAAK,CAAC,EAAG,EAAQ,CAAW,CACzG,CACF,CACA,MAAO,CAAE,WAAY,EAAW,SAAS,CAAiB,EAAG,sBAAqB,CACpF,CAIA,SAAS,EAAkB,EAAiB,EAA0B,CACpE,OACEG,EAAAA,OAAO,QAAQ,EAAM,KAAK,EAAIA,EAAAA,OAAO,QAAQ,EAAK,KAAK,GACvD,EAAe,EAAK,KAAM,EAAM,IAAI,GACpC,EAAK,UAAY,EAAM,WACvB,EAAe,QAAQ,EAAK,IAAI,EAAI,EAAe,QAAQ,EAAM,IAAI,GACrE,EAAK,QAAU,EAAM,OAEzB,CAGA,SAAS,EAAqB,CAAE,UAAS,SAAyC,CAChF,IAAM,EAAoB,EAAQ,WAAa,SAC/C,OAAO,EACH,EAAQ,UAAU,OAAQ,GACxB,EAAM,KAAM,GAAS,EAAY,EAAM,EAAG,UAAW,EAAG,QAAS,CAAiB,CAAC,CACrF,EACA,EAAQ,SACd,CAOA,SAAS,EACP,EACA,EACA,EACS,CAIT,OAHK,EAAuB,IAAI,EAAM,MAAM,EAGrC,EAAmB,KACvB,GAAU,EAAM,KAAO,GAAM,EAAS,EAAM,GAAI,CAAE,GAAK,EAAM,SAAS,MAAM,CAAE,YAAa,IAAW,EAAM,MAAM,CACrH,EAJS,EAKX,CAEA,SAAS,EAAS,EAAwB,EAAiC,CACzE,IAAM,EACJ,EAAM,UAAY,EAAM,WACvB,EAAM,YAAc,EAAM,WAAa,EAAM,aAAe,EAAM,YAC/D,EACJ,EAAM,QAAU,EAAM,SAAY,EAAM,UAAY,EAAM,SAAW,EAAM,WAAa,EAAM,UAChG,OAAO,GAAgB,CACzB,CAEA,SAAS,EAAoB,EAAc,EAAqB,EAAsC,CACpG,GAAM,CAAE,YAAW,WAAY,EACzB,EAA6B,EAAS,MAAM,CAAE,YAAa,IAAW,6BAAyB,EACrG,MAAO,CACL,GAAG,EAAY,CAAE,OAAM,YAAW,SAAQ,EAAG,WAAY,CAAQ,EACjE,KAAM,EAAoB,CAAE,EAC5B,GAAI,GAA8B,EAAG,gBAAgB,OAAS,GAAK,CAAE,cAAe,EAAG,eAAgB,CACzG,CACF,CAGA,SAAgB,EAAoB,CAAE,OAAO,cAAe,iBAA0C,CACpG,OAAO,EAAgB,GAAG,EAAc,GAAG,IAAS,CACtD,CAEA,SAAS,EAAY,EAAyB,EAAyB,EAAsC,CAE3G,MAAO,CAAE,OAAM,MADD,EAAS,KAAM,GAAU,EAAM,QAAU,OAAO,EAAI,QAAU,UACtD,GAAG,EAAU,UAAS,CAC9C,CAEA,SAAS,EACP,EACA,EACA,EACiB,CACjB,OAAO,EAAW,QAAS,GAAc,EAAa,EAAW,EAAS,GAAS,EAAO,IAAU,EAAQ,CAAK,CAAC,CACpH,CAGA,SAAS,EACP,EACA,EACA,EACA,EACiB,CACjB,IAAM,EAAQ,EAAU,QAAQ,CAAO,EACjC,EAAQA,EAAAA,OAAO,SAAU,GAAc,EAAS,EAAO,EAAO,EAAU,CAAC,EAAU,IAAc,CAAC,EACxG,OAAO,EACH,CACE,CACE,OAAQF,EAAAA,aAAa,CAAS,EAC9B,MAAO,EAAU,MACjB,QACA,QACA,MAAO,EAAO,EAAM,CAAC,EAAU,IACjC,CACF,EACA,CAAC,CACP,CASA,SAAS,EAAY,EAAgB,EAAmB,EAAiB,EAAoB,GAAgB,CAC3G,OAAO,EAAK,YAAc,EACtB,EAAK,WAAa,IAAc,EAAK,UAAY,GAAY,GAAqB,EAAK,YAAc,GACrG,EAAK,WAAa,GAAW,EAAK,UAAY,EAAK,UAAY,CACrE,CAGA,SAAS,EACP,EACA,EAC6C,CAC7C,IAAM,kBAAe,IAAI,IACzB,IAAK,IAAM,KAAS,GAAsB,QAAU,CAAC,EACnD,IAAK,IAAM,KAAQ,EAAM,MACvB,GAAI,EAAM,IAAI,CAAI,EAAG,CACnB,IAAM,EAAS,EAAa,IAAI,CAAI,GAAK,CAAC,EAC1C,EAAO,KAAK,CAAK,EACjB,EAAa,IAAI,EAAM,CAAM,CAC/B,CAGJ,OAAO,CACT,CAQA,SAAS,EACP,CAAE,OAAM,UAAS,SACjB,EACA,EACA,CAAE,wBACW,CACb,IAAM,EAAqB,GACzB,EAAaG,EAAAA,qBAAsB,CAAE,qBAAoB,EAAG,GAAS,EAAO,IAAU,GAAS,CAAK,EAChG,EAA8B,CAClC,GAAG,EAAQ,YAAY,qBAAqB,IAAK,GAAU,EAAM,IAAK,IAAgB,CAAE,GAAG,EAAY,MAAK,EAAE,CAAC,EAC/G,GAAG,EAAgB,IAAK,GAAU,EAAM,YAAY,IAAK,GAAe,EAAkB,CAAU,CAAC,CAAC,CACxG,EACM,EAAc,GAClB,EAAM,OAAS,GACf,EAAkB,EAAM,YAAY,MAAM,CAAC,CAAC,OAAS,IACpD,CAAC,GAAS,EAAM,KAAM,GAAS,EAAK,UAAY,GAAK,EAAY,EAAM,EAAM,UAAW,EAAM,OAAO,CAAC,GACnG,EAAU,IAAI,IAAI,EAAO,QAAS,GAAU,EAAM,IAAK,GAAe,CAAC,EAAY,CAAK,CAAU,CAAC,CAAC,EAIpG,GAAa,EAAuB,IACxC,EAAQ,IAAI,CAAI,IAAM,EAAQ,IAAI,CAAK,GAAK,CAAC,EAAS,EAAM,CAAK,GAAK,CAAC,EAAS,EAAO,CAAI,EAE7F,OAAO,EADgB,EAAO,KAAK,CAAC,CAAC,OAAQ,GAAe,EAAW,CAAU,CAC5C,EAAG,CAAS,CAAC,CAAC,KAAK,CAAE,SAAQ,aAAc,CAC9E,IAAM,EAAc,GAClB,EACE,EAAQ,QAAS,GAAW,CAG1B,IAAM,EAAa,KAAK,IACtB,GAAG,EAAQ,OAAQ,GAAU,EAAS,EAAO,CAAM,CAAC,CAAC,CAAC,IAAK,GAAU,EAAM,UAAU,CACvF,EACA,OAAQ,EAAQ,IAAI,CAAM,GAAK,CAAC,EAAA,CAAG,OAChC,GACC,EAAK,WAAa,KAAO,EAAa,GAItC,CAAC,EAAQ,KAAM,GAAU,EAAS,EAAO,CAAI,CAAC,CAClD,CACF,CAAC,EACD,CACF,CAAC,CAEE,UAAU,EAAM,IAAU,EAAU,EAAM,OAAO,EAAI,EAAU,EAAK,OAAO,CAAC,CAAC,CAC7E,IAAK,GAAY,EAAQ,MAAM,EAM9B,EAAS,EAAW,CAAoB,EAC9C,MAAO,CACL,GAAG,EACD,EACA,cAEA,EAAkB,IAAI,IAAI,EAAQ,QAAS,GAAW,EAAO,WAAW,CAAC,CAAC,CAAC,IAAI,CACjF,EAGA,SAAU,EAAO,OAAS,EAAI,EAAS,EAAW,CAAC,CACrD,CACF,CAAC,CACH,CAEA,SAAS,EAAkB,EAA2D,CACpF,GAAM,CAAE,eAAgB,EAExB,OADA,EAAA,QAAO,GAAG,EAAa,wBAAwB,EAAW,KAAK,kBAAkB,EAC1E,CAAE,GAAG,EAAY,aAAY,CACtC,CAEA,SAAS,EAAU,EAAiD,CAClE,OAAO,EAAY,QAAQ,EAAK,IAAe,EAAM,EAAW,WAAY,CAAC,CAC/E,CAMA,SAAS,EACP,EACA,EACyD,CACzD,IAAM,EAAmE,CAAC,EACpE,EAAU,EAAU,UACvB,EAAM,IAAU,EAAe,EAAK,KAAM,EAAM,IAAI,GAAK,EAAK,UAAY,EAAM,SACnF,EACA,IAAK,IAAM,KAAY,EAAS,CAC9B,IAAM,EAAO,EAAQ,GAAG,EAAE,EAO1B,GAJE,GACA,EAAS,EAAK,OAAQ,CAAQ,IAC7B,EAAK,QAAQ,KAAM,GAAW,EAAS,EAAQ,CAAQ,CAAC,GACvD,CAAC,EAAK,QAAQ,KAAM,GAAW,EAAU,EAAQ,CAAQ,CAAC,GAE5D,EAAK,OAAO,QAAU,KAAK,IAAI,EAAK,OAAO,QAAS,EAAS,OAAO,EACpE,EAAK,QAAQ,KAAK,CAAQ,MACrB,CACL,GAAM,CAAE,OAAM,YAAW,WAAY,EACrC,EAAQ,KAAK,CAAE,OAAQ,CAAE,OAAM,YAAW,SAAQ,EAAG,QAAS,CAAC,CAAQ,CAAE,CAAC,CAC5E,CACF,CACA,OAAO,CACT,CAGA,SAAS,EAAS,EAAsB,EAA+B,CACrE,OAAO,EAAM,OAAS,EAAM,MAAQ,EAAM,WAAa,EAAM,WAAa,EAAM,SAAW,EAAM,OACnG,CAEA,SAAS,EAAS,EAAqB,EAA+B,CACpE,OAAO,EAAK,OAAS,EAAM,MAAQ,EAAK,WAAa,EAAM,SAAW,EAAM,WAAa,EAAK,OAChG,CAGA,SAAS,EAAe,EAAc,EAAuB,CAC3D,OAAO,EAAO,EAAQ,GAAK,IAAO,EACpC"}

package/dist/thresholdCheck.d.ts

@@ -17,6 +17,8 @@ export interface CheckedDuplication {
 }
 export interface ExceededLimit {
     metric: string;
+    /** What reports call the metric (`cognitive complexity`, `duplicated lines`). */
+    label: string;
     value: number;
     /** The most severe level whose limit the value violates. */
     level: Level;

package/dist/thresholdCheck.js

@@ -1,2 +1,2 @@
-import{duplicationThreshold as e,fileThresholds as t,functionThresholds as n,levels as r,metricNameOf as i}from"./thresholds.js";import a from"node:assert";const o=[`file`,`function`,`duplication`],s=new Set(n.filter(({includesNestedFunctions:e})=>e).map(e=>i(e)));function c(e,r,i){let a=y(r.crossFile,new Set(e.map(({file:e})=>e))),o=[],s=0;for(let c of e){let{file:e,metrics:l,hunks:f}=c,m=i(l.language),_=f?.length===0?[]:g(t,l,m);_.length>0&&o.push(h({file:e,startLine:1,endLine:l.lines.total},`file`,_));let v=u(c);s+=v.length;let y=v.map(e=>({fn:e,exceeded:g(n,e,m)})).filter(({exceeded:e})=>e.length>0);for(let{fn:t,exceeded:n}of y){let r=n.filter(e=>!d(e,t,y));r.length>0&&o.push(p(e,t,r))}o.push(...b(c,a.get(e)??[],m,r))}return{violations:o.toSorted(l),checkedFunctionCount:s}}function l(e,t){return r.indexOf(t.level)-r.indexOf(e.level)||E(e.file,t.file)||e.startLine-t.startLine||o.indexOf(e.kind)-o.indexOf(t.kind)||e.endLine-t.endLine}function u({metrics:e,hunks:t}){let n=e.language===`python`;return t?e.functions.filter(e=>t.some(t=>v(t,e.startLine,e.endLine,n))):e.functions}function d(e,t,n){return s.has(e.metric)?n.some(n=>n.fn!==t&&f(n.fn,t)&&n.exceeded.some(({metric:t})=>t===e.metric)):!1}function f(e,t){let n=e.startLine<t.startLine||e.startLine===t.startLine&&e.startColumn<=t.startColumn,r=e.endLine>t.endLine||e.endLine===t.endLine&&e.endColumn>=t.endColumn;return n&&r}function p(e,t,n){let{startLine:r,endLine:i}=t,a=n.some(({metric:e})=>e===`functionCognitiveComplexity`);return{...h({file:e,startLine:r,endLine:i},`function`,n),name:m(t),...a&&t.cognitiveBlocks.length>0&&{largestBlocks:t.cognitiveBlocks}}}function m({name:e=`<anonymous>`,containerName:t}){return t?`${t}.${e}`:e}function h(e,t,n){return{kind:t,level:n.some(e=>e.level===`error`)?`error`:`warning`,...e,exceeded:n}}function g(e,t,n){return e.flatMap(e=>_(e,t,n,(e,t)=>e>t))}function _(e,t,n,a){let o=e.measure(t),s=r.findLast(t=>a(o,n[t][e.key]));return s?[{metric:i(e),value:o,level:s,limit:n[s][e.key]}]:[]}function v(e,t,n,r=!1){return e.headCount===0?e.headStart>=t&&(e.headStart<n||r&&e.headStart===n):e.headStart<=n&&e.headStart+e.headCount>t}function y(e,t){let n=/* @__PURE__ */ new Map;for(let r of e?.groups??[])for(let e of r.files)if(t.has(e)){let t=n.get(e)??[];t.push(r),n.set(e,t)}return n}function b({file:t,metrics:n,hunks:r},i,a,{minSimilarityPercent:o}){let s=t=>_(e,{duplicatedLineCount:t},a,(e,t)=>e>=t),c=[...n.duplication.duplicateBlockGroups.map(e=>e.map(e=>({...e,file:t}))),...i.map(e=>e.occurrences.map(e=>x(e)))],l=e=>e.file===t&&s(e.lineNumbers.length).length>0&&(!r||r.some(t=>t.headCount>0&&v(t,e.startLine,e.endLine))),u=/* @__PURE__ */ new Map;for(let e of c)for(let t of e.filter(e=>l(e)))u.set(t,e);return C([...u.keys()]).map(({merged:e,sources:t})=>{let n=n=>C(t.flatMap(e=>{let r=Math.max(...t.filter(t=>w(t,e)).map(e=>e.tokenCount));return(u.get(e)??[]).filter(e=>e.tokenCount*100>=r*n)})).filter(t=>!T(t.merged,e)).toSorted((e,t)=>S(t.sources)-S(e.sources)).map(e=>e.merged),r=n(o);return{...h(e,`duplication`,s(new Set(t.flatMap(e=>e.lineNumbers)).size)),partners:r.length>0?r:n(0)}})}function x(e){let{lineNumbers:t}=e;return a.ok(t,`No matched lines for ${e.file}, a scanned file.`),{...e,lineNumbers:t}}function S(e){return e.reduce((e,t)=>e+t.tokenCount,0)}function C(e){let t=[],n=e.toSorted((e,t)=>E(e.file,t.file)||e.startLine-t.startLine);for(let e of n){let n=t.at(-1);if(n&&T(n.merged,e))n.merged.endLine=Math.max(n.merged.endLine,e.endLine),n.sources.push(e);else{let{file:n,startLine:r,endLine:i}=e;t.push({merged:{file:n,startLine:r,endLine:i},sources:[e]})}}return t}function w(e,t){return e.file===t.file&&e.startLine<=t.startLine&&t.endLine<=e.endLine}function T(e,t){return e.file===t.file&&e.startLine<=t.endLine&&t.startLine<=e.endLine}function E(e,t){return e<t?-1:+(e>t)}export{c as checkThresholds,m as qualifyFunctionName};
+import{duplicationThreshold as e,fileThresholds as t,functionThresholds as n,levels as r,metricNameOf as i}from"./thresholds.js";import a from"node:assert";const o=[`file`,`function`,`duplication`],s=new Set(n.filter(({includesNestedFunctions:e})=>e).map(e=>i(e)));function c(e,r,i){let a=y(r.crossFile,new Set(e.map(({file:e})=>e))),o=[],s=0;for(let c of e){let{file:e,metrics:l,hunks:f}=c,m=i(l.language),_=f?.length===0?[]:g(t,l,m);_.length>0&&o.push(h({file:e,startLine:1,endLine:l.lines.total},`file`,_));let v=u(c);s+=v.length;let y=v.map(e=>({fn:e,exceeded:g(n,e,m)})).filter(({exceeded:e})=>e.length>0);for(let{fn:t,exceeded:n}of y){let r=n.filter(e=>!d(e,t,y));r.length>0&&o.push(p(e,t,r))}o.push(...b(c,a.get(e)??[],m,r))}return{violations:o.toSorted(l),checkedFunctionCount:s}}function l(e,t){return r.indexOf(t.level)-r.indexOf(e.level)||E(e.file,t.file)||e.startLine-t.startLine||o.indexOf(e.kind)-o.indexOf(t.kind)||e.endLine-t.endLine}function u({metrics:e,hunks:t}){let n=e.language===`python`;return t?e.functions.filter(e=>t.some(t=>v(t,e.startLine,e.endLine,n))):e.functions}function d(e,t,n){return s.has(e.metric)?n.some(n=>n.fn!==t&&f(n.fn,t)&&n.exceeded.some(({metric:t})=>t===e.metric)):!1}function f(e,t){let n=e.startLine<t.startLine||e.startLine===t.startLine&&e.startColumn<=t.startColumn,r=e.endLine>t.endLine||e.endLine===t.endLine&&e.endColumn>=t.endColumn;return n&&r}function p(e,t,n){let{startLine:r,endLine:i}=t,a=n.some(({metric:e})=>e===`functionCognitiveComplexity`);return{...h({file:e,startLine:r,endLine:i},`function`,n),name:m(t),...a&&t.cognitiveBlocks.length>0&&{largestBlocks:t.cognitiveBlocks}}}function m({name:e=`<anonymous>`,containerName:t}){return t?`${t}.${e}`:e}function h(e,t,n){return{kind:t,level:n.some(e=>e.level===`error`)?`error`:`warning`,...e,exceeded:n}}function g(e,t,n){return e.flatMap(e=>_(e,t,n,(e,t)=>e>t))}function _(e,t,n,a){let o=e.measure(t),s=r.findLast(t=>a(o,n[t][e.key]));return s?[{metric:i(e),label:e.label,value:o,level:s,limit:n[s][e.key]}]:[]}function v(e,t,n,r=!1){return e.headCount===0?e.headStart>=t&&(e.headStart<n||r&&e.headStart===n):e.headStart<=n&&e.headStart+e.headCount>t}function y(e,t){let n=/* @__PURE__ */ new Map;for(let r of e?.groups??[])for(let e of r.files)if(t.has(e)){let t=n.get(e)??[];t.push(r),n.set(e,t)}return n}function b({file:t,metrics:n,hunks:r},i,a,{minSimilarityPercent:o}){let s=t=>_(e,{duplicatedLineCount:t},a,(e,t)=>e>=t),c=[...n.duplication.duplicateBlockGroups.map(e=>e.map(e=>({...e,file:t}))),...i.map(e=>e.occurrences.map(e=>x(e)))],l=e=>e.file===t&&s(e.lineNumbers.length).length>0&&(!r||r.some(t=>t.headCount>0&&v(t,e.startLine,e.endLine))),u=new Map(c.flatMap(e=>e.map(t=>[t,e]))),d=(e,t)=>u.get(e)===u.get(t)&&!w(e,t)&&!w(t,e);return C(c.flat().filter(e=>l(e)),d).map(({merged:e,sources:t})=>{let n=e=>C(t.flatMap(n=>{let r=Math.max(...t.filter(e=>w(e,n)).map(e=>e.tokenCount));return(u.get(n)??[]).filter(n=>n.tokenCount*100>=r*e&&!t.some(e=>w(e,n)))}),d).toSorted((e,t)=>S(t.sources)-S(e.sources)).map(e=>e.merged),r=n(o);return{...h(e,`duplication`,s(new Set(t.flatMap(e=>e.lineNumbers)).size)),partners:r.length>0?r:n(0)}})}function x(e){let{lineNumbers:t}=e;return a.ok(t,`No matched lines for ${e.file}, a scanned file.`),{...e,lineNumbers:t}}function S(e){return e.reduce((e,t)=>e+t.tokenCount,0)}function C(e,t){let n=[],r=e.toSorted((e,t)=>E(e.file,t.file)||e.startLine-t.startLine);for(let e of r){let r=n.at(-1);if(r&&T(r.merged,e)&&(r.sources.some(t=>w(t,e))||!r.sources.some(n=>t(n,e))))r.merged.endLine=Math.max(r.merged.endLine,e.endLine),r.sources.push(e);else{let{file:t,startLine:r,endLine:i}=e;n.push({merged:{file:t,startLine:r,endLine:i},sources:[e]})}}return n}function w(e,t){return e.file===t.file&&e.startLine<=t.startLine&&t.endLine<=e.endLine}function T(e,t){return e.file===t.file&&e.startLine<=t.endLine&&t.startLine<=e.endLine}function E(e,t){return e<t?-1:+(e>t)}export{c as checkThresholds,m as qualifyFunctionName};
 //# sourceMappingURL=thresholdCheck.js.map