code-gauge 4.3.1 → 4.5.0

package/dist/metrics.d.ts

@@ -9,6 +9,11 @@ export declare class TreeMeasurer {
     private readonly registry;
     getSupportedLanguages(): LanguageName[];
     measure(code: string, options: MeasureOptions): CodeMetrics;
+    /** Measures a file and collects its cross-file clone-detection contribution from one parse. */
+    measureWithCrossFileData(code: string, options: MeasureOptions): {
+        metrics: CodeMetrics;
+        crossFileData: CrossFileDuplicationFileData;
+    };
     /** Collects one file's duplicate candidates for cross-file clone detection. */
     collectDuplicationCandidates(code: string, options: MeasureOptions): CrossFileDuplicateCandidate[];
     /**
@@ -28,6 +33,11 @@ export declare class TreeMeasurer {
 export declare const defaultMeasurer: TreeMeasurer;
 export declare function measureCode(code: string, options: MeasureOptions): CodeMetrics;
 /** Standalone helper mirroring measureCode for the default measurer. */
+export declare function measureCodeWithCrossFileData(code: string, options: MeasureOptions): {
+    metrics: CodeMetrics;
+    crossFileData: CrossFileDuplicationFileData;
+};
+/** Standalone helper mirroring measureCode for the default measurer. */
 export declare function collectDuplicationCandidates(code: string, options: MeasureOptions): CrossFileDuplicateCandidate[];
 /** Standalone helper mirroring measureCode for the default measurer. */
 export declare function collectFunctionTokenSequences(code: string, options: MeasureOptions): Int32Array[];

package/dist/metrics.js

@@ -1,2 +1,2 @@
-import{createLanguageRegistry as e}from"./languages.js";import{collectCrossFileDataNative as t,collectFunctionTokenSequencesNative as n,measureCodeNative as r}from"./nativeMetrics.js";var i=class{registry=e();getSupportedLanguages(){return[...new Set([...this.registry.values()].map(e=>e.name))]}measure(e,t){let n=this.resolveLanguage(t.language),i=t.includeSyntaxTree??!1;return a(r(e,n.name,i,t.duplication),i)}collectDuplicationCandidates(e,t){return this.collectCrossFileDuplicationFileData(e,t).candidates}collectCrossFileDuplicationFileData(e,n){let r=this.resolveLanguage(n.language),i=t(e,r.name,n.duplication?.minTokens);return{candidates:i.candidates,tokens:i.tokens,containerStatements:i.containerStatements,codeLineNumbers:new Set(i.codeLineNumbers)}}collectFunctionTokenSequences(e,t){let r=this.resolveLanguage(t.language);return n(e,r.name)}resolveLanguage(e){let t=this.registry.get(e);if(!t)throw Error(`Unsupported language: ${e}`);return t}};function a(e,t){return{language:e.language,bytes:e.bytes,lines:e.lines,functions:e.functions.map(e=>({name:e.name,nodeType:e.nodeType,startLine:e.startLine,startColumn:e.startColumn,endLine:e.endLine,endColumn:e.endColumn,cyclomaticComplexity:e.cyclomaticComplexity,cognitiveComplexity:e.cognitiveComplexity,nestingDepth:e.nestingDepth,ncss:e.ncss,parameterCount:e.parameterCount,halstead:o(e.halsteadCounts),depDegree:e.depDegree})),cognitiveComplexity:e.cognitiveComplexity,maxCognitiveComplexity:e.maxCognitiveComplexity,nestingDepth:e.nestingDepth,ncssCount:e.ncssCount,duplication:e.duplication,halstead:o(e.halsteadCounts),syntaxTree:t?e.syntaxTree:void 0}}function o(e){let{distinctOperators:t,distinctOperands:n,totalOperators:r,totalOperands:i}=e,a=t+n,o=r+i,s=a===0?0:o*Math.log2(a);return{distinctOperators:t,distinctOperands:n,totalOperators:r,totalOperands:i,vocabulary:a,length:o,volume:s,effort:(n===0?0:t/2*(i/n))*s}}const s=new i;function c(e,t){return s.measure(e,t)}function l(e,t){return s.collectDuplicationCandidates(e,t)}function u(e,t){return s.collectFunctionTokenSequences(e,t)}function d(e,t){return s.collectCrossFileDuplicationFileData(e,t)}export{i as TreeMeasurer,d as collectCrossFileDuplicationFileData,l as collectDuplicationCandidates,u as collectFunctionTokenSequences,s as defaultMeasurer,c as measureCode};
+import{createLanguageRegistry as e}from"./languages.js";import{collectCrossFileDataNative as t,collectFunctionTokenSequencesNative as n,measureCodeNative as r}from"./nativeMetrics.js";var i=class{registry=e();getSupportedLanguages(){return[...new Set([...this.registry.values()].map(e=>e.name))]}measure(e,t){let n=this.resolveLanguage(t.language),i=t.includeSyntaxTree??!1;return o(r(e,n.name,i,t.duplication),i)}measureWithCrossFileData(e,t){let n=this.resolveLanguage(t.language),i=t.includeSyntaxTree??!1,s=r(e,n.name,i,t.duplication,!0);if(!s.crossFileData)throw Error(`The native addon returned no cross-file duplication data`);return{metrics:o(s,i),crossFileData:a(s.crossFileData)}}collectDuplicationCandidates(e,t){return this.collectCrossFileDuplicationFileData(e,t).candidates}collectCrossFileDuplicationFileData(e,n){let r=this.resolveLanguage(n.language);return a(t(e,r.name,n.duplication?.minTokens))}collectFunctionTokenSequences(e,t){let r=this.resolveLanguage(t.language);return n(e,r.name)}resolveLanguage(e){let t=this.registry.get(e);if(!t)throw Error(`Unsupported language: ${e}`);return t}};function a(e){return{candidates:e.candidates,tokens:e.tokens,containerStatements:e.containerStatements,nearMissBlocks:e.nearMissBlocks,codeLineNumbers:new Set(e.codeLineNumbers)}}function o(e,t){return{language:e.language,bytes:e.bytes,lines:e.lines,functions:e.functions.map(e=>({name:e.name,nodeType:e.nodeType,startLine:e.startLine,startColumn:e.startColumn,endLine:e.endLine,endColumn:e.endColumn,cyclomaticComplexity:e.cyclomaticComplexity,cognitiveComplexity:e.cognitiveComplexity,nestingDepth:e.nestingDepth,ncss:e.ncss,parameterCount:e.parameterCount,halstead:s(e.halsteadCounts),depDegree:e.depDegree})),cyclomaticComplexity:e.cyclomaticComplexity,cognitiveComplexity:e.cognitiveComplexity,maxCognitiveComplexity:e.maxCognitiveComplexity,nestingDepth:e.nestingDepth,ncssCount:e.ncssCount,duplication:e.duplication,halstead:s(e.halsteadCounts),syntaxTree:t?e.syntaxTree:void 0}}function s(e){let{distinctOperators:t,distinctOperands:n,totalOperators:r,totalOperands:i}=e,a=t+n,o=r+i,s=a===0?0:o*Math.log2(a);return{distinctOperators:t,distinctOperands:n,totalOperators:r,totalOperands:i,vocabulary:a,length:o,volume:s,effort:(n===0?0:t/2*(i/n))*s}}const c=new i;function l(e,t){return c.measure(e,t)}function u(e,t){return c.measureWithCrossFileData(e,t)}function d(e,t){return c.collectDuplicationCandidates(e,t)}function f(e,t){return c.collectFunctionTokenSequences(e,t)}function p(e,t){return c.collectCrossFileDuplicationFileData(e,t)}export{i as TreeMeasurer,p as collectCrossFileDuplicationFileData,d as collectDuplicationCandidates,f as collectFunctionTokenSequences,c as defaultMeasurer,l as measureCode,u as measureCodeWithCrossFileData};
 //# sourceMappingURL=metrics.js.map

package/dist/metrics.js.map

@@ -1 +1 @@
-{"version":3,"file":"metrics.js","names":[],"sources":["../src/metrics.ts"],"sourcesContent":["import type { CrossFileDuplicateCandidate, CrossFileDuplicationFileData } from './duplication.js';\nimport { createLanguageRegistry } from './languages.js';\nimport {\n  collectCrossFileDataNative,\n  collectFunctionTokenSequencesNative,\n  measureCodeNative,\n  type NativeHalsteadCounts,\n  type NativeMetricsPayload,\n} from './nativeMetrics.js';\nimport type { CodeMetrics, HalsteadMetrics, LanguageDefinition, LanguageName, MeasureOptions } from './types.js';\n\n/**\n * Measures code metrics for the built-in languages. Parsing and metric passes run in the bundled\n * Rust addon (tree-sitter); this class resolves language aliases, crosses the N-API boundary, and\n * derives the Halstead float metrics from the natively measured counts.\n */\nexport class TreeMeasurer {\n  private readonly registry = createLanguageRegistry();\n\n  getSupportedLanguages(): LanguageName[] {\n    return [...new Set([...this.registry.values()].map((language) => language.name))];\n  }\n\n  measure(code: string, options: MeasureOptions): CodeMetrics {\n    const language = this.resolveLanguage(options.language);\n    const includeSyntaxTree = options.includeSyntaxTree ?? false;\n    const payload = measureCodeNative(code, language.name, includeSyntaxTree, options.duplication);\n    return assembleNativeMetrics(payload, includeSyntaxTree);\n  }\n\n  /** Collects one file's duplicate candidates for cross-file clone detection. */\n  collectDuplicationCandidates(code: string, options: MeasureOptions): CrossFileDuplicateCandidate[] {\n    return this.collectCrossFileDuplicationFileData(code, options).candidates;\n  }\n\n  /**\n   * Collects one file's duplicate candidates, normalized tokens, and statement structure for\n   * cross-file clone detection with measureCrossFileDuplication.\n   */\n  collectCrossFileDuplicationFileData(code: string, options: MeasureOptions): CrossFileDuplicationFileData {\n    const language = this.resolveLanguage(options.language);\n    const payload = collectCrossFileDataNative(code, language.name, options.duplication?.minTokens);\n    return {\n      candidates: payload.candidates,\n      tokens: payload.tokens,\n      containerStatements: payload.containerStatements,\n      // Lets cross-file line coverage count only code lines, like within-file coverage.\n      codeLineNumbers: new Set(payload.codeLineNumbers),\n    };\n  }\n\n  /**\n   * Normalized token hash sequences of every function, index-parallel to the functions array of\n   * measure(): identifiers are anonymized by first occurrence within the function, literals by\n   * kind, and keywords/operators kept verbatim, so the regression gate can re-match renamed or\n   * moved functions across two revisions by token-LCS similarity.\n   */\n  collectFunctionTokenSequences(code: string, options: MeasureOptions): Int32Array[] {\n    const language = this.resolveLanguage(options.language);\n    return collectFunctionTokenSequencesNative(code, language.name);\n  }\n\n  private resolveLanguage(name: LanguageName): LanguageDefinition {\n    const language = this.registry.get(name);\n    if (!language) {\n      throw new Error(`Unsupported language: ${name}`);\n    }\n    return language;\n  }\n}\n\n/**\n * Completes a native measurement into CodeMetrics. The object is rebuilt field by field (rather\n * than spread from the parsed JSON) so the result has a stable shape, including\n * explicitly-undefined optional keys.\n */\nfunction assembleNativeMetrics(payload: NativeMetricsPayload, includeSyntaxTree: boolean): CodeMetrics {\n  return {\n    language: payload.language,\n    bytes: payload.bytes,\n    lines: payload.lines,\n    functions: payload.functions.map((fn) => ({\n      name: fn.name,\n      nodeType: fn.nodeType,\n      startLine: fn.startLine,\n      startColumn: fn.startColumn,\n      endLine: fn.endLine,\n      endColumn: fn.endColumn,\n      cyclomaticComplexity: fn.cyclomaticComplexity,\n      cognitiveComplexity: fn.cognitiveComplexity,\n      nestingDepth: fn.nestingDepth,\n      ncss: fn.ncss,\n      parameterCount: fn.parameterCount,\n      halstead: deriveHalsteadMetrics(fn.halsteadCounts),\n      depDegree: fn.depDegree,\n    })),\n    cognitiveComplexity: payload.cognitiveComplexity,\n    maxCognitiveComplexity: payload.maxCognitiveComplexity,\n    nestingDepth: payload.nestingDepth,\n    ncssCount: payload.ncssCount,\n    duplication: payload.duplication,\n    halstead: deriveHalsteadMetrics(payload.halsteadCounts),\n    syntaxTree: includeSyntaxTree ? payload.syntaxTree : undefined,\n  };\n}\n\nfunction deriveHalsteadMetrics(counts: NativeHalsteadCounts): HalsteadMetrics {\n  const { distinctOperators, distinctOperands, totalOperators, totalOperands } = counts;\n  const vocabulary = distinctOperators + distinctOperands;\n  const length = totalOperators + totalOperands;\n  const volume = vocabulary === 0 ? 0 : length * Math.log2(vocabulary);\n  const difficulty = distinctOperands === 0 ? 0 : (distinctOperators / 2) * (totalOperands / distinctOperands);\n  const effort = difficulty * volume;\n\n  return {\n    distinctOperators,\n    distinctOperands,\n    totalOperators,\n    totalOperands,\n    vocabulary,\n    length,\n    volume,\n    effort,\n  };\n}\n\nexport const defaultMeasurer = new TreeMeasurer();\n\nexport function measureCode(code: string, options: MeasureOptions): CodeMetrics {\n  return defaultMeasurer.measure(code, options);\n}\n\n/** Standalone helper mirroring measureCode for the default measurer. */\nexport function collectDuplicationCandidates(code: string, options: MeasureOptions): CrossFileDuplicateCandidate[] {\n  return defaultMeasurer.collectDuplicationCandidates(code, options);\n}\n\n/** Standalone helper mirroring measureCode for the default measurer. */\nexport function collectFunctionTokenSequences(code: string, options: MeasureOptions): Int32Array[] {\n  return defaultMeasurer.collectFunctionTokenSequences(code, options);\n}\n\n/** Standalone helper mirroring measureCode for the default measurer. */\nexport function collectCrossFileDuplicationFileData(\n  code: string,\n  options: MeasureOptions\n): CrossFileDuplicationFileData {\n  return defaultMeasurer.collectCrossFileDuplicationFileData(code, options);\n}\n"],"mappings":"wLAgBA,IAAa,EAAb,KAA0B,CACxB,SAA4B,EAAuB,EAEnD,uBAAwC,CACtC,MAAO,CAAC,GAAG,IAAI,IAAI,CAAC,GAAG,KAAK,SAAS,OAAO,CAAC,CAAC,CAAC,IAAK,GAAa,EAAS,IAAI,CAAC,CAAC,CAClF,CAEA,QAAQ,EAAc,EAAsC,CAC1D,IAAM,EAAW,KAAK,gBAAgB,EAAQ,QAAQ,EAChD,EAAoB,EAAQ,mBAAqB,GAEvD,OAAO,EADS,EAAkB,EAAM,EAAS,KAAM,EAAmB,EAAQ,WAC/C,EAAG,CAAiB,CACzD,CAGA,6BAA6B,EAAc,EAAwD,CACjG,OAAO,KAAK,oCAAoC,EAAM,CAAO,CAAC,CAAC,UACjE,CAMA,oCAAoC,EAAc,EAAuD,CACvG,IAAM,EAAW,KAAK,gBAAgB,EAAQ,QAAQ,EAChD,EAAU,EAA2B,EAAM,EAAS,KAAM,EAAQ,aAAa,SAAS,EAC9F,MAAO,CACL,WAAY,EAAQ,WACpB,OAAQ,EAAQ,OAChB,oBAAqB,EAAQ,oBAE7B,gBAAiB,IAAI,IAAI,EAAQ,eAAe,CAClD,CACF,CAQA,8BAA8B,EAAc,EAAuC,CACjF,IAAM,EAAW,KAAK,gBAAgB,EAAQ,QAAQ,EACtD,OAAO,EAAoC,EAAM,EAAS,IAAI,CAChE,CAEA,gBAAwB,EAAwC,CAC9D,IAAM,EAAW,KAAK,SAAS,IAAI,CAAI,EACvC,GAAI,CAAC,EACH,MAAU,MAAM,yBAAyB,GAAM,EAEjD,OAAO,CACT,CACF,EAOA,SAAS,EAAsB,EAA+B,EAAyC,CACrG,MAAO,CACL,SAAU,EAAQ,SAClB,MAAO,EAAQ,MACf,MAAO,EAAQ,MACf,UAAW,EAAQ,UAAU,IAAK,IAAQ,CACxC,KAAM,EAAG,KACT,SAAU,EAAG,SACb,UAAW,EAAG,UACd,YAAa,EAAG,YAChB,QAAS,EAAG,QACZ,UAAW,EAAG,UACd,qBAAsB,EAAG,qBACzB,oBAAqB,EAAG,oBACxB,aAAc,EAAG,aACjB,KAAM,EAAG,KACT,eAAgB,EAAG,eACnB,SAAU,EAAsB,EAAG,cAAc,EACjD,UAAW,EAAG,SAChB,EAAE,EACF,oBAAqB,EAAQ,oBAC7B,uBAAwB,EAAQ,uBAChC,aAAc,EAAQ,aACtB,UAAW,EAAQ,UACnB,YAAa,EAAQ,YACrB,SAAU,EAAsB,EAAQ,cAAc,EACtD,WAAY,EAAoB,EAAQ,WAAa,IAAA,EACvD,CACF,CAEA,SAAS,EAAsB,EAA+C,CAC5E,GAAM,CAAE,oBAAmB,mBAAkB,iBAAgB,iBAAkB,EACzE,EAAa,EAAoB,EACjC,EAAS,EAAiB,EAC1B,EAAS,IAAe,EAAI,EAAI,EAAS,KAAK,KAAK,CAAU,EAInE,MAAO,CACL,oBACA,mBACA,iBACA,gBACA,aACA,SACA,SACA,QAXiB,IAAqB,EAAI,EAAK,EAAoB,GAAM,EAAgB,IAC/D,CAW5B,CACF,CAEA,MAAa,EAAkB,IAAI,EAEnC,SAAgB,EAAY,EAAc,EAAsC,CAC9E,OAAO,EAAgB,QAAQ,EAAM,CAAO,CAC9C,CAGA,SAAgB,EAA6B,EAAc,EAAwD,CACjH,OAAO,EAAgB,6BAA6B,EAAM,CAAO,CACnE,CAGA,SAAgB,EAA8B,EAAc,EAAuC,CACjG,OAAO,EAAgB,8BAA8B,EAAM,CAAO,CACpE,CAGA,SAAgB,EACd,EACA,EAC8B,CAC9B,OAAO,EAAgB,oCAAoC,EAAM,CAAO,CAC1E"}
+{"version":3,"file":"metrics.js","names":[],"sources":["../src/metrics.ts"],"sourcesContent":["import type { CrossFileDuplicateCandidate, CrossFileDuplicationFileData } from './duplication.js';\nimport { createLanguageRegistry } from './languages.js';\nimport {\n  collectCrossFileDataNative,\n  collectFunctionTokenSequencesNative,\n  measureCodeNative,\n  type NativeCrossFileDataPayload,\n  type NativeHalsteadCounts,\n  type NativeMetricsPayload,\n} from './nativeMetrics.js';\nimport type { CodeMetrics, HalsteadMetrics, LanguageDefinition, LanguageName, MeasureOptions } from './types.js';\n\n/**\n * Measures code metrics for the built-in languages. Parsing and metric passes run in the bundled\n * Rust addon (tree-sitter); this class resolves language aliases, crosses the N-API boundary, and\n * derives the Halstead float metrics from the natively measured counts.\n */\nexport class TreeMeasurer {\n  private readonly registry = createLanguageRegistry();\n\n  getSupportedLanguages(): LanguageName[] {\n    return [...new Set([...this.registry.values()].map((language) => language.name))];\n  }\n\n  measure(code: string, options: MeasureOptions): CodeMetrics {\n    const language = this.resolveLanguage(options.language);\n    const includeSyntaxTree = options.includeSyntaxTree ?? false;\n    const payload = measureCodeNative(code, language.name, includeSyntaxTree, options.duplication);\n    return assembleNativeMetrics(payload, includeSyntaxTree);\n  }\n\n  /** Measures a file and collects its cross-file clone-detection contribution from one parse. */\n  measureWithCrossFileData(\n    code: string,\n    options: MeasureOptions\n  ): { metrics: CodeMetrics; crossFileData: CrossFileDuplicationFileData } {\n    const language = this.resolveLanguage(options.language);\n    const includeSyntaxTree = options.includeSyntaxTree ?? false;\n    const payload = measureCodeNative(code, language.name, includeSyntaxTree, options.duplication, true);\n    if (!payload.crossFileData) {\n      throw new Error('The native addon returned no cross-file duplication data');\n    }\n    return {\n      metrics: assembleNativeMetrics(payload, includeSyntaxTree),\n      crossFileData: toCrossFileDuplicationFileData(payload.crossFileData),\n    };\n  }\n\n  /** Collects one file's duplicate candidates for cross-file clone detection. */\n  collectDuplicationCandidates(code: string, options: MeasureOptions): CrossFileDuplicateCandidate[] {\n    return this.collectCrossFileDuplicationFileData(code, options).candidates;\n  }\n\n  /**\n   * Collects one file's duplicate candidates, normalized tokens, and statement structure for\n   * cross-file clone detection with measureCrossFileDuplication.\n   */\n  collectCrossFileDuplicationFileData(code: string, options: MeasureOptions): CrossFileDuplicationFileData {\n    const language = this.resolveLanguage(options.language);\n    return toCrossFileDuplicationFileData(\n      collectCrossFileDataNative(code, language.name, options.duplication?.minTokens)\n    );\n  }\n\n  /**\n   * Normalized token hash sequences of every function, index-parallel to the functions array of\n   * measure(): identifiers are anonymized by first occurrence within the function, literals by\n   * kind, and keywords/operators kept verbatim, so the regression gate can re-match renamed or\n   * moved functions across two revisions by token-LCS similarity.\n   */\n  collectFunctionTokenSequences(code: string, options: MeasureOptions): Int32Array[] {\n    const language = this.resolveLanguage(options.language);\n    return collectFunctionTokenSequencesNative(code, language.name);\n  }\n\n  private resolveLanguage(name: LanguageName): LanguageDefinition {\n    const language = this.registry.get(name);\n    if (!language) {\n      throw new Error(`Unsupported language: ${name}`);\n    }\n    return language;\n  }\n}\n\nfunction toCrossFileDuplicationFileData(payload: NativeCrossFileDataPayload): CrossFileDuplicationFileData {\n  return {\n    candidates: payload.candidates,\n    tokens: payload.tokens,\n    containerStatements: payload.containerStatements,\n    nearMissBlocks: payload.nearMissBlocks,\n    // Lets cross-file line coverage count only code lines, like within-file coverage.\n    codeLineNumbers: new Set(payload.codeLineNumbers),\n  };\n}\n\n/**\n * Completes a native measurement into CodeMetrics. The object is rebuilt field by field (rather\n * than spread from the parsed JSON) so the result has a stable shape, including\n * explicitly-undefined optional keys.\n */\nfunction assembleNativeMetrics(payload: NativeMetricsPayload, includeSyntaxTree: boolean): CodeMetrics {\n  return {\n    language: payload.language,\n    bytes: payload.bytes,\n    lines: payload.lines,\n    functions: payload.functions.map((fn) => ({\n      name: fn.name,\n      nodeType: fn.nodeType,\n      startLine: fn.startLine,\n      startColumn: fn.startColumn,\n      endLine: fn.endLine,\n      endColumn: fn.endColumn,\n      cyclomaticComplexity: fn.cyclomaticComplexity,\n      cognitiveComplexity: fn.cognitiveComplexity,\n      nestingDepth: fn.nestingDepth,\n      ncss: fn.ncss,\n      parameterCount: fn.parameterCount,\n      halstead: deriveHalsteadMetrics(fn.halsteadCounts),\n      depDegree: fn.depDegree,\n    })),\n    cyclomaticComplexity: payload.cyclomaticComplexity,\n    cognitiveComplexity: payload.cognitiveComplexity,\n    maxCognitiveComplexity: payload.maxCognitiveComplexity,\n    nestingDepth: payload.nestingDepth,\n    ncssCount: payload.ncssCount,\n    duplication: payload.duplication,\n    halstead: deriveHalsteadMetrics(payload.halsteadCounts),\n    syntaxTree: includeSyntaxTree ? payload.syntaxTree : undefined,\n  };\n}\n\nfunction deriveHalsteadMetrics(counts: NativeHalsteadCounts): HalsteadMetrics {\n  const { distinctOperators, distinctOperands, totalOperators, totalOperands } = counts;\n  const vocabulary = distinctOperators + distinctOperands;\n  const length = totalOperators + totalOperands;\n  const volume = vocabulary === 0 ? 0 : length * Math.log2(vocabulary);\n  const difficulty = distinctOperands === 0 ? 0 : (distinctOperators / 2) * (totalOperands / distinctOperands);\n  const effort = difficulty * volume;\n\n  return {\n    distinctOperators,\n    distinctOperands,\n    totalOperators,\n    totalOperands,\n    vocabulary,\n    length,\n    volume,\n    effort,\n  };\n}\n\nexport const defaultMeasurer = new TreeMeasurer();\n\nexport function measureCode(code: string, options: MeasureOptions): CodeMetrics {\n  return defaultMeasurer.measure(code, options);\n}\n\n/** Standalone helper mirroring measureCode for the default measurer. */\nexport function measureCodeWithCrossFileData(\n  code: string,\n  options: MeasureOptions\n): { metrics: CodeMetrics; crossFileData: CrossFileDuplicationFileData } {\n  return defaultMeasurer.measureWithCrossFileData(code, options);\n}\n\n/** Standalone helper mirroring measureCode for the default measurer. */\nexport function collectDuplicationCandidates(code: string, options: MeasureOptions): CrossFileDuplicateCandidate[] {\n  return defaultMeasurer.collectDuplicationCandidates(code, options);\n}\n\n/** Standalone helper mirroring measureCode for the default measurer. */\nexport function collectFunctionTokenSequences(code: string, options: MeasureOptions): Int32Array[] {\n  return defaultMeasurer.collectFunctionTokenSequences(code, options);\n}\n\n/** Standalone helper mirroring measureCode for the default measurer. */\nexport function collectCrossFileDuplicationFileData(\n  code: string,\n  options: MeasureOptions\n): CrossFileDuplicationFileData {\n  return defaultMeasurer.collectCrossFileDuplicationFileData(code, options);\n}\n"],"mappings":"wLAiBA,IAAa,EAAb,KAA0B,CACxB,SAA4B,EAAuB,EAEnD,uBAAwC,CACtC,MAAO,CAAC,GAAG,IAAI,IAAI,CAAC,GAAG,KAAK,SAAS,OAAO,CAAC,CAAC,CAAC,IAAK,GAAa,EAAS,IAAI,CAAC,CAAC,CAClF,CAEA,QAAQ,EAAc,EAAsC,CAC1D,IAAM,EAAW,KAAK,gBAAgB,EAAQ,QAAQ,EAChD,EAAoB,EAAQ,mBAAqB,GAEvD,OAAO,EADS,EAAkB,EAAM,EAAS,KAAM,EAAmB,EAAQ,WAC/C,EAAG,CAAiB,CACzD,CAGA,yBACE,EACA,EACuE,CACvE,IAAM,EAAW,KAAK,gBAAgB,EAAQ,QAAQ,EAChD,EAAoB,EAAQ,mBAAqB,GACjD,EAAU,EAAkB,EAAM,EAAS,KAAM,EAAmB,EAAQ,YAAa,EAAI,EACnG,GAAI,CAAC,EAAQ,cACX,MAAU,MAAM,0DAA0D,EAE5E,MAAO,CACL,QAAS,EAAsB,EAAS,CAAiB,EACzD,cAAe,EAA+B,EAAQ,aAAa,CACrE,CACF,CAGA,6BAA6B,EAAc,EAAwD,CACjG,OAAO,KAAK,oCAAoC,EAAM,CAAO,CAAC,CAAC,UACjE,CAMA,oCAAoC,EAAc,EAAuD,CACvG,IAAM,EAAW,KAAK,gBAAgB,EAAQ,QAAQ,EACtD,OAAO,EACL,EAA2B,EAAM,EAAS,KAAM,EAAQ,aAAa,SAAS,CAChF,CACF,CAQA,8BAA8B,EAAc,EAAuC,CACjF,IAAM,EAAW,KAAK,gBAAgB,EAAQ,QAAQ,EACtD,OAAO,EAAoC,EAAM,EAAS,IAAI,CAChE,CAEA,gBAAwB,EAAwC,CAC9D,IAAM,EAAW,KAAK,SAAS,IAAI,CAAI,EACvC,GAAI,CAAC,EACH,MAAU,MAAM,yBAAyB,GAAM,EAEjD,OAAO,CACT,CACF,EAEA,SAAS,EAA+B,EAAmE,CACzG,MAAO,CACL,WAAY,EAAQ,WACpB,OAAQ,EAAQ,OAChB,oBAAqB,EAAQ,oBAC7B,eAAgB,EAAQ,eAExB,gBAAiB,IAAI,IAAI,EAAQ,eAAe,CAClD,CACF,CAOA,SAAS,EAAsB,EAA+B,EAAyC,CACrG,MAAO,CACL,SAAU,EAAQ,SAClB,MAAO,EAAQ,MACf,MAAO,EAAQ,MACf,UAAW,EAAQ,UAAU,IAAK,IAAQ,CACxC,KAAM,EAAG,KACT,SAAU,EAAG,SACb,UAAW,EAAG,UACd,YAAa,EAAG,YAChB,QAAS,EAAG,QACZ,UAAW,EAAG,UACd,qBAAsB,EAAG,qBACzB,oBAAqB,EAAG,oBACxB,aAAc,EAAG,aACjB,KAAM,EAAG,KACT,eAAgB,EAAG,eACnB,SAAU,EAAsB,EAAG,cAAc,EACjD,UAAW,EAAG,SAChB,EAAE,EACF,qBAAsB,EAAQ,qBAC9B,oBAAqB,EAAQ,oBAC7B,uBAAwB,EAAQ,uBAChC,aAAc,EAAQ,aACtB,UAAW,EAAQ,UACnB,YAAa,EAAQ,YACrB,SAAU,EAAsB,EAAQ,cAAc,EACtD,WAAY,EAAoB,EAAQ,WAAa,IAAA,EACvD,CACF,CAEA,SAAS,EAAsB,EAA+C,CAC5E,GAAM,CAAE,oBAAmB,mBAAkB,iBAAgB,iBAAkB,EACzE,EAAa,EAAoB,EACjC,EAAS,EAAiB,EAC1B,EAAS,IAAe,EAAI,EAAI,EAAS,KAAK,KAAK,CAAU,EAInE,MAAO,CACL,oBACA,mBACA,iBACA,gBACA,aACA,SACA,SACA,QAXiB,IAAqB,EAAI,EAAK,EAAoB,GAAM,EAAgB,IAC/D,CAW5B,CACF,CAEA,MAAa,EAAkB,IAAI,EAEnC,SAAgB,EAAY,EAAc,EAAsC,CAC9E,OAAO,EAAgB,QAAQ,EAAM,CAAO,CAC9C,CAGA,SAAgB,EACd,EACA,EACuE,CACvE,OAAO,EAAgB,yBAAyB,EAAM,CAAO,CAC/D,CAGA,SAAgB,EAA6B,EAAc,EAAwD,CACjH,OAAO,EAAgB,6BAA6B,EAAM,CAAO,CACnE,CAGA,SAAgB,EAA8B,EAAc,EAAuC,CACjG,OAAO,EAAgB,8BAA8B,EAAM,CAAO,CACpE,CAGA,SAAgB,EACd,EACA,EAC8B,CAC9B,OAAO,EAAgB,oCAAoC,EAAM,CAAO,CAC1E"}

package/dist/nativeMetrics.cjs

@@ -1,4 +1,4 @@
-"use strict";let e=require("node:module");function t(e,t,n,r){return JSON.parse(l().measureCodeNative(i(e),t,n,a(r?.minTokens),a(r?.maxGapTokens),a(r?.minSimilarityPercent)))}function n(e,t,n){return JSON.parse(l().collectCrossFileDataNative(i(e),t,a(n)))}function r(e,t){return JSON.parse(l().collectFunctionTokenSequencesNative(i(e),t)).map(e=>Int32Array.from(e))}function i(e){return e.isWellFormed()?e:e.toWellFormed()}function a(e){return e===void 0||Number.isNaN(e)?void 0:Math.min(Math.max(Math.trunc(e),0),4294967295)}var o=class extends Error{};let s,c;function l(){if(s)return s;if(c)throw c;let t=(0,e.createRequire)(require("url").pathToFileURL(__filename).href),n=[`code-gauge-${u()}`,`../native/code-gauge.node`],r=[];for(let e of n){let n;try{n=t(e)}catch(t){r.push(`  ${e}: ${t instanceof Error?t.message.split(`
-`)[0]:String(t)}`);continue}let i=n.payloadVersion?.();if(i!==5){r.push(`  ${e}: payload version ${i??`unknown`} does not match the expected 5; rebuild the addon with \`bun run build-native\``);continue}return s=n,n}throw c=new o(`The code-gauge native addon is not available for ${u()}. Build it with \`node scripts/buildNative.mjs\` in the code-gauge package directory (requires a Rust toolchain); when installing with npm, also allow install scripts for code-gauge so its postinstall build can run.\n${r.join(`
+"use strict";let e=require("node:module");function t(e,t,n,r,o=!1){return JSON.parse(l().measureCodeNative(i(e),t,n,a(r?.minTokens),a(r?.maxGapTokens),a(r?.minSimilarityPercent),o))}function n(e,t,n){return JSON.parse(l().collectCrossFileDataNative(i(e),t,a(n)))}function r(e,t){return JSON.parse(l().collectFunctionTokenSequencesNative(i(e),t)).map(e=>Int32Array.from(e))}function i(e){return e.isWellFormed()?e:e.toWellFormed()}function a(e){return e===void 0||Number.isNaN(e)?void 0:Math.min(Math.max(Math.trunc(e),0),4294967295)}var o=class extends Error{};let s,c;function l(){if(s)return s;if(c)throw c;let t=(0,e.createRequire)(require("url").pathToFileURL(__filename).href),n=[`code-gauge-${u()}`,`../native/code-gauge.node`],r=[];for(let e of n){let n;try{n=t(e)}catch(t){r.push(`  ${e}: ${t instanceof Error?t.message.split(`
+`)[0]:String(t)}`);continue}let i=n.payloadVersion?.();if(i!==7){r.push(`  ${e}: payload version ${i??`unknown`} does not match the expected 7; rebuild the addon with \`bun run build-native\``);continue}return s=n,n}throw c=new o(`The code-gauge native addon is not available for ${u()}. Build it with \`node scripts/buildNative.mjs\` in the code-gauge package directory (requires a Rust toolchain); when installing with npm, also allow install scripts for code-gauge so its postinstall build can run.\n${r.join(`
 `)}`),c}function u(){let e=`${process.platform}-${process.arch}`;return process.platform===`win32`?`${e}-msvc`:process.platform===`linux`?(process.report?.getReport())?.header?.glibcVersionRuntime?`${e}-gnu`:`${e}-musl`:e}exports.NativeAddonError=o,exports.collectCrossFileDataNative=n,exports.collectFunctionTokenSequencesNative=r,exports.measureCodeNative=t;
 //# sourceMappingURL=nativeMetrics.cjs.map

package/dist/nativeMetrics.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"nativeMetrics.cjs","names":["createRequire"],"sources":["../src/nativeMetrics.ts"],"sourcesContent":["import { createRequire } from 'node:module';\nimport type { CrossFileDuplicateCandidate, Token, TokenRange } from './duplication.js';\nimport type { CodeMetrics, DuplicationOptions, FunctionMetrics } from './types.js';\n\n/**\n * Halstead counts measured natively; the derived float metrics (volume, effort, ...) are\n * computed in TypeScript because they involve transcendental functions (log2) whose last-bit\n * results can differ between V8 and Rust's libm, and results must not depend on the Rust side's\n * libm build.\n */\nexport interface NativeHalsteadCounts {\n  distinctOperators: number;\n  distinctOperands: number;\n  totalOperators: number;\n  totalOperands: number;\n}\n\nexport interface NativeFunctionMetricsPayload extends Omit<FunctionMetrics, 'halstead'> {\n  halsteadCounts: NativeHalsteadCounts;\n}\n\nexport interface NativeMetricsPayload extends Omit<CodeMetrics, 'halstead' | 'functions' | 'syntaxTree'> {\n  functions: NativeFunctionMetricsPayload[];\n  halsteadCounts: NativeHalsteadCounts;\n  syntaxTree?: string;\n}\n\n/** One file's cross-file clone-detection contribution as serialized by the native addon. */\nexport interface NativeCrossFileDataPayload {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  /** 1-based lines that are neither blank nor comment-only, sorted ascending. */\n  codeLineNumbers: number[];\n}\n\ninterface NativeBinding {\n  measureCodeNative(\n    code: string,\n    language: string,\n    includeSyntaxTree: boolean,\n    minTokens?: number,\n    maxGapTokens?: number,\n    minSimilarityPercent?: number\n  ): string;\n  collectCrossFileDataNative(code: string, language: string, minTokens?: number): string;\n  collectFunctionTokenSequencesNative(code: string, language: string): string;\n  payloadVersion?(): number;\n}\n\n/**\n * Must equal `payload_version` in native/src/lib.rs. A previously built addon survives a\n * `git pull` untouched, so without this handshake it would silently return payloads missing\n * newer fields instead of failing with a clear rebuild message.\n */\nconst expectedPayloadVersion = 5;\n\n/** Measures one file via the native addon, returning the raw payload for assembly in metrics.ts. */\nexport function measureCodeNative(\n  code: string,\n  language: string,\n  includeSyntaxTree: boolean,\n  duplication?: DuplicationOptions\n): NativeMetricsPayload {\n  return JSON.parse(\n    loadBinding().measureCodeNative(\n      toWellFormed(code),\n      language,\n      includeSyntaxTree,\n      clampToU32(duplication?.minTokens),\n      clampToU32(duplication?.maxGapTokens),\n      clampToU32(duplication?.minSimilarityPercent)\n    )\n  ) as NativeMetricsPayload;\n}\n\n/** Collects one file's cross-file clone-detection contribution via the native addon. */\nexport function collectCrossFileDataNative(\n  code: string,\n  language: string,\n  minTokens?: number\n): NativeCrossFileDataPayload {\n  return JSON.parse(\n    loadBinding().collectCrossFileDataNative(toWellFormed(code), language, clampToU32(minTokens))\n  ) as NativeCrossFileDataPayload;\n}\n\n/** Collects normalized token hash sequences of every function via the native addon. */\nexport function collectFunctionTokenSequencesNative(code: string, language: string): Int32Array[] {\n  const sequences = JSON.parse(\n    loadBinding().collectFunctionTokenSequencesNative(toWellFormed(code), language)\n  ) as number[][];\n  return sequences.map((sequence) => Int32Array.from(sequence));\n}\n\n/**\n * Lone surrogates cannot cross the N-API boundary losslessly, so ill-formed strings (invalid\n * UTF-16 occasionally present in real-world files) are measured with U+FFFD replacements — the\n * same code units V8's own UTF-8 conversion would substitute.\n */\nfunction toWellFormed(code: string): string {\n  return code.isWellFormed() ? code : code.toWellFormed();\n}\n\n/**\n * The duplication settings cross the boundary as u32, whose JavaScript conversion wraps modulo\n * 2^32 (2 ** 32 would become 0 and match everything). The public API accepts any safe integer, so\n * out-of-range values clamp to [0, u32::MAX] — no source can hold 2^32 tokens, so a clamped\n * threshold behaves identically to the requested one. NaN (e.g. `Number(unsetEnvVariable)`)\n * survives clamping arithmetic and would also convert to 0, so it is treated as an absent setting\n * instead.\n */\nfunction clampToU32(value: number | undefined): number | undefined {\n  return value === undefined || Number.isNaN(value)\n    ? undefined\n    : Math.min(Math.max(Math.trunc(value), 0), 0xFF_FF_FF_FF);\n}\n\n/**\n * Raised when no usable native addon can be loaded. Every measurement fails identically until the\n * addon is built, so callers measuring many files (the CLI scan) treat it as fatal for the whole\n * run instead of recording one \"skipped\" entry per file.\n */\nexport class NativeAddonError extends Error {}\n\nlet cachedBinding: NativeBinding | undefined;\nlet cachedFailure: NativeAddonError | undefined;\n\nfunction loadBinding(): NativeBinding {\n  if (cachedBinding) {\n    return cachedBinding;\n  }\n  // The failure is memoized too: resolution (including platformTriplet's diagnostic-report call\n  // on Linux) would otherwise repeat for every measured file of an already-failing run.\n  if (cachedFailure) {\n    throw cachedFailure;\n  }\n  // Resolved relative to this file, so both src/ (tests) and dist/ (build) find the addon.\n  const requireNative = createRequire(import.meta.url);\n  const specifiers = [\n    // A prebuilt platform package, when published for this platform.\n    `code-gauge-${platformTriplet()}`,\n    // A locally built addon (`bun run build-native`).\n    '../native/code-gauge.node',\n  ];\n  const failures: string[] = [];\n  for (const specifier of specifiers) {\n    let binding: NativeBinding;\n    try {\n      binding = requireNative(specifier) as NativeBinding;\n    } catch (error) {\n      failures.push(`  ${specifier}: ${error instanceof Error ? error.message.split('\\n')[0] : String(error)}`);\n      continue;\n    }\n    const version = binding.payloadVersion?.();\n    if (version !== expectedPayloadVersion) {\n      failures.push(\n        `  ${specifier}: payload version ${version ?? 'unknown'} does not match the expected ` +\n          `${expectedPayloadVersion}; rebuild the addon with \\`bun run build-native\\``\n      );\n      continue;\n    }\n    cachedBinding = binding;\n    return binding;\n  }\n  cachedFailure = new NativeAddonError(\n    `The code-gauge native addon is not available for ${platformTriplet()}. Build it with ` +\n      '`node scripts/buildNative.mjs` in the code-gauge package directory (requires a Rust ' +\n      'toolchain); when installing with npm, also allow install scripts for code-gauge so its ' +\n      `postinstall build can run.\\n${failures.join('\\n')}`\n  );\n  throw cachedFailure;\n}\n\n/**\n * The platform-package suffix in the napi-rs naming convention: Linux targets are qualified by\n * libc ABI (`linux-x64-gnu` / `linux-x64-musl`) because a glibc-linked addon cannot load on\n * Alpine/musl, and Windows by toolchain ABI (`win32-x64-msvc`), matching what napi-rs tooling\n * generates. Must match scripts/installNative.mjs and the build-native workflow's target list.\n */\nfunction platformTriplet(): string {\n  const base = `${process.platform}-${process.arch}`;\n  if (process.platform === 'win32') {\n    return `${base}-msvc`;\n  }\n  if (process.platform !== 'linux') {\n    return base;\n  }\n  // Musl builds of Node report no glibc runtime version; see napi-rs's isMusl detection.\n  const report = process.report?.getReport() as { header?: { glibcVersionRuntime?: string } } | undefined;\n  return report?.header?.glibcVersionRuntime ? `${base}-gnu` : `${base}-musl`;\n}\n"],"mappings":"0CA0DA,SAAgB,EACd,EACA,EACA,EACA,EACsB,CACtB,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,kBACZ,EAAa,CAAI,EACjB,EACA,EACA,EAAW,GAAa,SAAS,EACjC,EAAW,GAAa,YAAY,EACpC,EAAW,GAAa,oBAAoB,CAC9C,CACF,CACF,CAGA,SAAgB,EACd,EACA,EACA,EAC4B,CAC5B,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,2BAA2B,EAAa,CAAI,EAAG,EAAU,EAAW,CAAS,CAAC,CAC9F,CACF,CAGA,SAAgB,EAAoC,EAAc,EAAgC,CAIhG,OAHkB,KAAK,MACrB,EAAY,CAAC,CAAC,oCAAoC,EAAa,CAAI,EAAG,CAAQ,CAEjE,CAAC,CAAC,IAAK,GAAa,WAAW,KAAK,CAAQ,CAAC,CAC9D,CAOA,SAAS,EAAa,EAAsB,CAC1C,OAAO,EAAK,aAAa,EAAI,EAAO,EAAK,aAAa,CACxD,CAUA,SAAS,EAAW,EAA+C,CACjE,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAC5C,IAAA,GACA,KAAK,IAAI,KAAK,IAAI,KAAK,MAAM,CAAK,EAAG,CAAC,EAAG,UAAa,CAC5D,CAOA,IAAa,EAAb,cAAsC,KAAM,CAAC,EAE7C,IAAI,EACA,EAEJ,SAAS,GAA6B,CACpC,GAAI,EACF,OAAO,EAIT,GAAI,EACF,MAAM,EAGR,IAAM,GAAA,EAAgBA,EAAAA,cAAAA,CAAAA,QAAAA,KAAAA,CAAAA,CAAAA,cAAAA,UAAAA,CAAAA,CAAAA,IAA6B,EAC7C,EAAa,CAEjB,cAAc,EAAgB,IAE9B,2BACF,EACM,EAAqB,CAAC,EAC5B,IAAK,IAAM,KAAa,EAAY,CAClC,IAAI,EACJ,GAAI,CACF,EAAU,EAAc,CAAS,CACnC,OAAS,EAAO,CACd,EAAS,KAAK,KAAK,EAAU,IAAI,aAAiB,MAAQ,EAAM,QAAQ,MAAM;CAAI,CAAC,CAAC,GAAK,OAAO,CAAK,GAAG,EACxG,QACF,CACA,IAAM,EAAU,EAAQ,iBAAiB,EACzC,GAAI,IAAY,EAAwB,CACtC,EAAS,KACP,KAAK,EAAU,oBAAoB,GAAW,UAAU,gFAE1D,EACA,QACF,CAEA,MADA,GAAgB,EACT,CACT,CAOA,KANA,GAAgB,IAAI,EAClB,oDAAoD,EAAgB,EAAE,2NAGrC,EAAS,KAAK;CAAI,GACrD,EACM,CACR,CAQA,SAAS,GAA0B,CACjC,IAAM,EAAO,GAAG,QAAQ,SAAS,GAAG,QAAQ,OAS5C,OARI,QAAQ,WAAa,QAChB,GAAG,EAAK,OAEb,QAAQ,WAAa,SAIV,QAAQ,QAAQ,UAAU,EAAA,EAC1B,QAAQ,oBAAsB,GAAG,EAAK,MAAQ,GAAG,EAAK,OAJ5D,CAKX"}
+{"version":3,"file":"nativeMetrics.cjs","names":["createRequire"],"sources":["../src/nativeMetrics.ts"],"sourcesContent":["import { createRequire } from 'node:module';\nimport type { CrossFileDuplicateCandidate, Token, TokenRange } from './duplication.js';\nimport type { CodeMetrics, DuplicationOptions, FunctionMetrics } from './types.js';\n\n/**\n * Halstead counts measured natively; the derived float metrics (volume, effort, ...) are\n * computed in TypeScript because they involve transcendental functions (log2) whose last-bit\n * results can differ between V8 and Rust's libm, and results must not depend on the Rust side's\n * libm build.\n */\nexport interface NativeHalsteadCounts {\n  distinctOperators: number;\n  distinctOperands: number;\n  totalOperators: number;\n  totalOperands: number;\n}\n\nexport interface NativeFunctionMetricsPayload extends Omit<FunctionMetrics, 'halstead'> {\n  halsteadCounts: NativeHalsteadCounts;\n}\n\nexport interface NativeMetricsPayload extends Omit<CodeMetrics, 'halstead' | 'functions' | 'syntaxTree'> {\n  functions: NativeFunctionMetricsPayload[];\n  halsteadCounts: NativeHalsteadCounts;\n  syntaxTree?: string;\n  crossFileData?: NativeCrossFileDataPayload;\n}\n\n/** One file's cross-file clone-detection contribution as serialized by the native addon. */\nexport interface NativeCrossFileDataPayload {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  nearMissBlocks: TokenRange[];\n  /** 1-based lines that are neither blank nor comment-only, sorted ascending. */\n  codeLineNumbers: number[];\n}\n\ninterface NativeBinding {\n  measureCodeNative(\n    code: string,\n    language: string,\n    includeSyntaxTree: boolean,\n    minTokens?: number,\n    maxGapTokens?: number,\n    minSimilarityPercent?: number,\n    includeCrossFileData?: boolean\n  ): string;\n  collectCrossFileDataNative(code: string, language: string, minTokens?: number): string;\n  collectFunctionTokenSequencesNative(code: string, language: string): string;\n  payloadVersion?(): number;\n}\n\n/**\n * Must equal `payload_version` in native/src/lib.rs. A previously built addon survives a\n * `git pull` untouched, so without this handshake it would silently return payloads missing\n * newer fields instead of failing with a clear rebuild message.\n */\nconst expectedPayloadVersion = 7;\n\n/**\n * Measures one file via the native addon, returning the raw payload for assembly in metrics.ts;\n * with `includeCrossFileData`, the payload also carries the file's cross-file contribution.\n */\nexport function measureCodeNative(\n  code: string,\n  language: string,\n  includeSyntaxTree: boolean,\n  duplication?: DuplicationOptions,\n  includeCrossFileData = false\n): NativeMetricsPayload {\n  return JSON.parse(\n    loadBinding().measureCodeNative(\n      toWellFormed(code),\n      language,\n      includeSyntaxTree,\n      clampToU32(duplication?.minTokens),\n      clampToU32(duplication?.maxGapTokens),\n      clampToU32(duplication?.minSimilarityPercent),\n      includeCrossFileData\n    )\n  ) as NativeMetricsPayload;\n}\n\n/** Collects one file's cross-file clone-detection contribution via the native addon. */\nexport function collectCrossFileDataNative(\n  code: string,\n  language: string,\n  minTokens?: number\n): NativeCrossFileDataPayload {\n  return JSON.parse(\n    loadBinding().collectCrossFileDataNative(toWellFormed(code), language, clampToU32(minTokens))\n  ) as NativeCrossFileDataPayload;\n}\n\n/** Collects normalized token hash sequences of every function via the native addon. */\nexport function collectFunctionTokenSequencesNative(code: string, language: string): Int32Array[] {\n  const sequences = JSON.parse(\n    loadBinding().collectFunctionTokenSequencesNative(toWellFormed(code), language)\n  ) as number[][];\n  return sequences.map((sequence) => Int32Array.from(sequence));\n}\n\n/**\n * Lone surrogates cannot cross the N-API boundary losslessly, so ill-formed strings (invalid\n * UTF-16 occasionally present in real-world files) are measured with U+FFFD replacements — the\n * same code units V8's own UTF-8 conversion would substitute.\n */\nfunction toWellFormed(code: string): string {\n  return code.isWellFormed() ? code : code.toWellFormed();\n}\n\n/**\n * The duplication settings cross the boundary as u32, whose JavaScript conversion wraps modulo\n * 2^32 (2 ** 32 would become 0 and match everything). The public API accepts any safe integer, so\n * out-of-range values clamp to [0, u32::MAX] — no source can hold 2^32 tokens, so a clamped\n * threshold behaves identically to the requested one. NaN (e.g. `Number(unsetEnvVariable)`)\n * survives clamping arithmetic and would also convert to 0, so it is treated as an absent setting\n * instead.\n */\nfunction clampToU32(value: number | undefined): number | undefined {\n  return value === undefined || Number.isNaN(value)\n    ? undefined\n    : Math.min(Math.max(Math.trunc(value), 0), 0xFF_FF_FF_FF);\n}\n\n/**\n * Raised when no usable native addon can be loaded. Every measurement fails identically until the\n * addon is built, so callers measuring many files (the CLI scan) treat it as fatal for the whole\n * run instead of recording one \"skipped\" entry per file.\n */\nexport class NativeAddonError extends Error {}\n\nlet cachedBinding: NativeBinding | undefined;\nlet cachedFailure: NativeAddonError | undefined;\n\nfunction loadBinding(): NativeBinding {\n  if (cachedBinding) {\n    return cachedBinding;\n  }\n  // The failure is memoized too: resolution (including platformTriplet's diagnostic-report call\n  // on Linux) would otherwise repeat for every measured file of an already-failing run.\n  if (cachedFailure) {\n    throw cachedFailure;\n  }\n  // Resolved relative to this file, so both src/ (tests) and dist/ (build) find the addon.\n  const requireNative = createRequire(import.meta.url);\n  const specifiers = [\n    // A prebuilt platform package, when published for this platform.\n    `code-gauge-${platformTriplet()}`,\n    // A locally built addon (`bun run build-native`).\n    '../native/code-gauge.node',\n  ];\n  const failures: string[] = [];\n  for (const specifier of specifiers) {\n    let binding: NativeBinding;\n    try {\n      binding = requireNative(specifier) as NativeBinding;\n    } catch (error) {\n      failures.push(`  ${specifier}: ${error instanceof Error ? error.message.split('\\n')[0] : String(error)}`);\n      continue;\n    }\n    const version = binding.payloadVersion?.();\n    if (version !== expectedPayloadVersion) {\n      failures.push(\n        `  ${specifier}: payload version ${version ?? 'unknown'} does not match the expected ` +\n          `${expectedPayloadVersion}; rebuild the addon with \\`bun run build-native\\``\n      );\n      continue;\n    }\n    cachedBinding = binding;\n    return binding;\n  }\n  cachedFailure = new NativeAddonError(\n    `The code-gauge native addon is not available for ${platformTriplet()}. Build it with ` +\n      '`node scripts/buildNative.mjs` in the code-gauge package directory (requires a Rust ' +\n      'toolchain); when installing with npm, also allow install scripts for code-gauge so its ' +\n      `postinstall build can run.\\n${failures.join('\\n')}`\n  );\n  throw cachedFailure;\n}\n\n/**\n * The platform-package suffix in the napi-rs naming convention: Linux targets are qualified by\n * libc ABI (`linux-x64-gnu` / `linux-x64-musl`) because a glibc-linked addon cannot load on\n * Alpine/musl, and Windows by toolchain ABI (`win32-x64-msvc`), matching what napi-rs tooling\n * generates. Must match scripts/installNative.mjs and the build-native workflow's target list.\n */\nfunction platformTriplet(): string {\n  const base = `${process.platform}-${process.arch}`;\n  if (process.platform === 'win32') {\n    return `${base}-msvc`;\n  }\n  if (process.platform !== 'linux') {\n    return base;\n  }\n  // Musl builds of Node report no glibc runtime version; see napi-rs's isMusl detection.\n  const report = process.report?.getReport() as { header?: { glibcVersionRuntime?: string } } | undefined;\n  return report?.header?.glibcVersionRuntime ? `${base}-gnu` : `${base}-musl`;\n}\n"],"mappings":"0CAgEA,SAAgB,EACd,EACA,EACA,EACA,EACA,EAAuB,GACD,CACtB,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,kBACZ,EAAa,CAAI,EACjB,EACA,EACA,EAAW,GAAa,SAAS,EACjC,EAAW,GAAa,YAAY,EACpC,EAAW,GAAa,oBAAoB,EAC5C,CACF,CACF,CACF,CAGA,SAAgB,EACd,EACA,EACA,EAC4B,CAC5B,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,2BAA2B,EAAa,CAAI,EAAG,EAAU,EAAW,CAAS,CAAC,CAC9F,CACF,CAGA,SAAgB,EAAoC,EAAc,EAAgC,CAIhG,OAHkB,KAAK,MACrB,EAAY,CAAC,CAAC,oCAAoC,EAAa,CAAI,EAAG,CAAQ,CAEjE,CAAC,CAAC,IAAK,GAAa,WAAW,KAAK,CAAQ,CAAC,CAC9D,CAOA,SAAS,EAAa,EAAsB,CAC1C,OAAO,EAAK,aAAa,EAAI,EAAO,EAAK,aAAa,CACxD,CAUA,SAAS,EAAW,EAA+C,CACjE,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAC5C,IAAA,GACA,KAAK,IAAI,KAAK,IAAI,KAAK,MAAM,CAAK,EAAG,CAAC,EAAG,UAAa,CAC5D,CAOA,IAAa,EAAb,cAAsC,KAAM,CAAC,EAE7C,IAAI,EACA,EAEJ,SAAS,GAA6B,CACpC,GAAI,EACF,OAAO,EAIT,GAAI,EACF,MAAM,EAGR,IAAM,GAAA,EAAgBA,EAAAA,cAAAA,CAAAA,QAAAA,KAAAA,CAAAA,CAAAA,cAAAA,UAAAA,CAAAA,CAAAA,IAA6B,EAC7C,EAAa,CAEjB,cAAc,EAAgB,IAE9B,2BACF,EACM,EAAqB,CAAC,EAC5B,IAAK,IAAM,KAAa,EAAY,CAClC,IAAI,EACJ,GAAI,CACF,EAAU,EAAc,CAAS,CACnC,OAAS,EAAO,CACd,EAAS,KAAK,KAAK,EAAU,IAAI,aAAiB,MAAQ,EAAM,QAAQ,MAAM;CAAI,CAAC,CAAC,GAAK,OAAO,CAAK,GAAG,EACxG,QACF,CACA,IAAM,EAAU,EAAQ,iBAAiB,EACzC,GAAI,IAAY,EAAwB,CACtC,EAAS,KACP,KAAK,EAAU,oBAAoB,GAAW,UAAU,gFAE1D,EACA,QACF,CAEA,MADA,GAAgB,EACT,CACT,CAOA,KANA,GAAgB,IAAI,EAClB,oDAAoD,EAAgB,EAAE,2NAGrC,EAAS,KAAK;CAAI,GACrD,EACM,CACR,CAQA,SAAS,GAA0B,CACjC,IAAM,EAAO,GAAG,QAAQ,SAAS,GAAG,QAAQ,OAS5C,OARI,QAAQ,WAAa,QAChB,GAAG,EAAK,OAEb,QAAQ,WAAa,SAIV,QAAQ,QAAQ,UAAU,EAAA,EAC1B,QAAQ,oBAAsB,GAAG,EAAK,MAAQ,GAAG,EAAK,OAJ5D,CAKX"}

package/dist/nativeMetrics.d.ts

@@ -19,17 +19,22 @@ export interface NativeMetricsPayload extends Omit<CodeMetrics, 'halstead' | 'fu
     functions: NativeFunctionMetricsPayload[];
     halsteadCounts: NativeHalsteadCounts;
     syntaxTree?: string;
+    crossFileData?: NativeCrossFileDataPayload;
 }
 /** One file's cross-file clone-detection contribution as serialized by the native addon. */
 export interface NativeCrossFileDataPayload {
     candidates: CrossFileDuplicateCandidate[];
     tokens: Token[];
     containerStatements: TokenRange[][];
+    nearMissBlocks: TokenRange[];
     /** 1-based lines that are neither blank nor comment-only, sorted ascending. */
     codeLineNumbers: number[];
 }
-/** Measures one file via the native addon, returning the raw payload for assembly in metrics.ts. */
-export declare function measureCodeNative(code: string, language: string, includeSyntaxTree: boolean, duplication?: DuplicationOptions): NativeMetricsPayload;
+/**
+ * Measures one file via the native addon, returning the raw payload for assembly in metrics.ts;
+ * with `includeCrossFileData`, the payload also carries the file's cross-file contribution.
+ */
+export declare function measureCodeNative(code: string, language: string, includeSyntaxTree: boolean, duplication?: DuplicationOptions, includeCrossFileData?: boolean): NativeMetricsPayload;
 /** Collects one file's cross-file clone-detection contribution via the native addon. */
 export declare function collectCrossFileDataNative(code: string, language: string, minTokens?: number): NativeCrossFileDataPayload;
 /** Collects normalized token hash sequences of every function via the native addon. */

package/dist/nativeMetrics.js

@@ -1,4 +1,4 @@
-import{createRequire as e}from"node:module";function t(e,t,n,r){return JSON.parse(l().measureCodeNative(i(e),t,n,a(r?.minTokens),a(r?.maxGapTokens),a(r?.minSimilarityPercent)))}function n(e,t,n){return JSON.parse(l().collectCrossFileDataNative(i(e),t,a(n)))}function r(e,t){return JSON.parse(l().collectFunctionTokenSequencesNative(i(e),t)).map(e=>Int32Array.from(e))}function i(e){return e.isWellFormed()?e:e.toWellFormed()}function a(e){return e===void 0||Number.isNaN(e)?void 0:Math.min(Math.max(Math.trunc(e),0),4294967295)}var o=class extends Error{};let s,c;function l(){if(s)return s;if(c)throw c;let t=e(import.meta.url),n=[`code-gauge-${u()}`,`../native/code-gauge.node`],r=[];for(let e of n){let n;try{n=t(e)}catch(t){r.push(`  ${e}: ${t instanceof Error?t.message.split(`
-`)[0]:String(t)}`);continue}let i=n.payloadVersion?.();if(i!==5){r.push(`  ${e}: payload version ${i??`unknown`} does not match the expected 5; rebuild the addon with \`bun run build-native\``);continue}return s=n,n}throw c=new o(`The code-gauge native addon is not available for ${u()}. Build it with \`node scripts/buildNative.mjs\` in the code-gauge package directory (requires a Rust toolchain); when installing with npm, also allow install scripts for code-gauge so its postinstall build can run.\n${r.join(`
+import{createRequire as e}from"node:module";function t(e,t,n,r,o=!1){return JSON.parse(l().measureCodeNative(i(e),t,n,a(r?.minTokens),a(r?.maxGapTokens),a(r?.minSimilarityPercent),o))}function n(e,t,n){return JSON.parse(l().collectCrossFileDataNative(i(e),t,a(n)))}function r(e,t){return JSON.parse(l().collectFunctionTokenSequencesNative(i(e),t)).map(e=>Int32Array.from(e))}function i(e){return e.isWellFormed()?e:e.toWellFormed()}function a(e){return e===void 0||Number.isNaN(e)?void 0:Math.min(Math.max(Math.trunc(e),0),4294967295)}var o=class extends Error{};let s,c;function l(){if(s)return s;if(c)throw c;let t=e(import.meta.url),n=[`code-gauge-${u()}`,`../native/code-gauge.node`],r=[];for(let e of n){let n;try{n=t(e)}catch(t){r.push(`  ${e}: ${t instanceof Error?t.message.split(`
+`)[0]:String(t)}`);continue}let i=n.payloadVersion?.();if(i!==7){r.push(`  ${e}: payload version ${i??`unknown`} does not match the expected 7; rebuild the addon with \`bun run build-native\``);continue}return s=n,n}throw c=new o(`The code-gauge native addon is not available for ${u()}. Build it with \`node scripts/buildNative.mjs\` in the code-gauge package directory (requires a Rust toolchain); when installing with npm, also allow install scripts for code-gauge so its postinstall build can run.\n${r.join(`
 `)}`),c}function u(){let e=`${process.platform}-${process.arch}`;return process.platform===`win32`?`${e}-msvc`:process.platform===`linux`?(process.report?.getReport())?.header?.glibcVersionRuntime?`${e}-gnu`:`${e}-musl`:e}export{o as NativeAddonError,n as collectCrossFileDataNative,r as collectFunctionTokenSequencesNative,t as measureCodeNative};
 //# sourceMappingURL=nativeMetrics.js.map

package/dist/nativeMetrics.js.map

@@ -1 +1 @@
-{"version":3,"file":"nativeMetrics.js","names":[],"sources":["../src/nativeMetrics.ts"],"sourcesContent":["import { createRequire } from 'node:module';\nimport type { CrossFileDuplicateCandidate, Token, TokenRange } from './duplication.js';\nimport type { CodeMetrics, DuplicationOptions, FunctionMetrics } from './types.js';\n\n/**\n * Halstead counts measured natively; the derived float metrics (volume, effort, ...) are\n * computed in TypeScript because they involve transcendental functions (log2) whose last-bit\n * results can differ between V8 and Rust's libm, and results must not depend on the Rust side's\n * libm build.\n */\nexport interface NativeHalsteadCounts {\n  distinctOperators: number;\n  distinctOperands: number;\n  totalOperators: number;\n  totalOperands: number;\n}\n\nexport interface NativeFunctionMetricsPayload extends Omit<FunctionMetrics, 'halstead'> {\n  halsteadCounts: NativeHalsteadCounts;\n}\n\nexport interface NativeMetricsPayload extends Omit<CodeMetrics, 'halstead' | 'functions' | 'syntaxTree'> {\n  functions: NativeFunctionMetricsPayload[];\n  halsteadCounts: NativeHalsteadCounts;\n  syntaxTree?: string;\n}\n\n/** One file's cross-file clone-detection contribution as serialized by the native addon. */\nexport interface NativeCrossFileDataPayload {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  /** 1-based lines that are neither blank nor comment-only, sorted ascending. */\n  codeLineNumbers: number[];\n}\n\ninterface NativeBinding {\n  measureCodeNative(\n    code: string,\n    language: string,\n    includeSyntaxTree: boolean,\n    minTokens?: number,\n    maxGapTokens?: number,\n    minSimilarityPercent?: number\n  ): string;\n  collectCrossFileDataNative(code: string, language: string, minTokens?: number): string;\n  collectFunctionTokenSequencesNative(code: string, language: string): string;\n  payloadVersion?(): number;\n}\n\n/**\n * Must equal `payload_version` in native/src/lib.rs. A previously built addon survives a\n * `git pull` untouched, so without this handshake it would silently return payloads missing\n * newer fields instead of failing with a clear rebuild message.\n */\nconst expectedPayloadVersion = 5;\n\n/** Measures one file via the native addon, returning the raw payload for assembly in metrics.ts. */\nexport function measureCodeNative(\n  code: string,\n  language: string,\n  includeSyntaxTree: boolean,\n  duplication?: DuplicationOptions\n): NativeMetricsPayload {\n  return JSON.parse(\n    loadBinding().measureCodeNative(\n      toWellFormed(code),\n      language,\n      includeSyntaxTree,\n      clampToU32(duplication?.minTokens),\n      clampToU32(duplication?.maxGapTokens),\n      clampToU32(duplication?.minSimilarityPercent)\n    )\n  ) as NativeMetricsPayload;\n}\n\n/** Collects one file's cross-file clone-detection contribution via the native addon. */\nexport function collectCrossFileDataNative(\n  code: string,\n  language: string,\n  minTokens?: number\n): NativeCrossFileDataPayload {\n  return JSON.parse(\n    loadBinding().collectCrossFileDataNative(toWellFormed(code), language, clampToU32(minTokens))\n  ) as NativeCrossFileDataPayload;\n}\n\n/** Collects normalized token hash sequences of every function via the native addon. */\nexport function collectFunctionTokenSequencesNative(code: string, language: string): Int32Array[] {\n  const sequences = JSON.parse(\n    loadBinding().collectFunctionTokenSequencesNative(toWellFormed(code), language)\n  ) as number[][];\n  return sequences.map((sequence) => Int32Array.from(sequence));\n}\n\n/**\n * Lone surrogates cannot cross the N-API boundary losslessly, so ill-formed strings (invalid\n * UTF-16 occasionally present in real-world files) are measured with U+FFFD replacements — the\n * same code units V8's own UTF-8 conversion would substitute.\n */\nfunction toWellFormed(code: string): string {\n  return code.isWellFormed() ? code : code.toWellFormed();\n}\n\n/**\n * The duplication settings cross the boundary as u32, whose JavaScript conversion wraps modulo\n * 2^32 (2 ** 32 would become 0 and match everything). The public API accepts any safe integer, so\n * out-of-range values clamp to [0, u32::MAX] — no source can hold 2^32 tokens, so a clamped\n * threshold behaves identically to the requested one. NaN (e.g. `Number(unsetEnvVariable)`)\n * survives clamping arithmetic and would also convert to 0, so it is treated as an absent setting\n * instead.\n */\nfunction clampToU32(value: number | undefined): number | undefined {\n  return value === undefined || Number.isNaN(value)\n    ? undefined\n    : Math.min(Math.max(Math.trunc(value), 0), 0xFF_FF_FF_FF);\n}\n\n/**\n * Raised when no usable native addon can be loaded. Every measurement fails identically until the\n * addon is built, so callers measuring many files (the CLI scan) treat it as fatal for the whole\n * run instead of recording one \"skipped\" entry per file.\n */\nexport class NativeAddonError extends Error {}\n\nlet cachedBinding: NativeBinding | undefined;\nlet cachedFailure: NativeAddonError | undefined;\n\nfunction loadBinding(): NativeBinding {\n  if (cachedBinding) {\n    return cachedBinding;\n  }\n  // The failure is memoized too: resolution (including platformTriplet's diagnostic-report call\n  // on Linux) would otherwise repeat for every measured file of an already-failing run.\n  if (cachedFailure) {\n    throw cachedFailure;\n  }\n  // Resolved relative to this file, so both src/ (tests) and dist/ (build) find the addon.\n  const requireNative = createRequire(import.meta.url);\n  const specifiers = [\n    // A prebuilt platform package, when published for this platform.\n    `code-gauge-${platformTriplet()}`,\n    // A locally built addon (`bun run build-native`).\n    '../native/code-gauge.node',\n  ];\n  const failures: string[] = [];\n  for (const specifier of specifiers) {\n    let binding: NativeBinding;\n    try {\n      binding = requireNative(specifier) as NativeBinding;\n    } catch (error) {\n      failures.push(`  ${specifier}: ${error instanceof Error ? error.message.split('\\n')[0] : String(error)}`);\n      continue;\n    }\n    const version = binding.payloadVersion?.();\n    if (version !== expectedPayloadVersion) {\n      failures.push(\n        `  ${specifier}: payload version ${version ?? 'unknown'} does not match the expected ` +\n          `${expectedPayloadVersion}; rebuild the addon with \\`bun run build-native\\``\n      );\n      continue;\n    }\n    cachedBinding = binding;\n    return binding;\n  }\n  cachedFailure = new NativeAddonError(\n    `The code-gauge native addon is not available for ${platformTriplet()}. Build it with ` +\n      '`node scripts/buildNative.mjs` in the code-gauge package directory (requires a Rust ' +\n      'toolchain); when installing with npm, also allow install scripts for code-gauge so its ' +\n      `postinstall build can run.\\n${failures.join('\\n')}`\n  );\n  throw cachedFailure;\n}\n\n/**\n * The platform-package suffix in the napi-rs naming convention: Linux targets are qualified by\n * libc ABI (`linux-x64-gnu` / `linux-x64-musl`) because a glibc-linked addon cannot load on\n * Alpine/musl, and Windows by toolchain ABI (`win32-x64-msvc`), matching what napi-rs tooling\n * generates. Must match scripts/installNative.mjs and the build-native workflow's target list.\n */\nfunction platformTriplet(): string {\n  const base = `${process.platform}-${process.arch}`;\n  if (process.platform === 'win32') {\n    return `${base}-msvc`;\n  }\n  if (process.platform !== 'linux') {\n    return base;\n  }\n  // Musl builds of Node report no glibc runtime version; see napi-rs's isMusl detection.\n  const report = process.report?.getReport() as { header?: { glibcVersionRuntime?: string } } | undefined;\n  return report?.header?.glibcVersionRuntime ? `${base}-gnu` : `${base}-musl`;\n}\n"],"mappings":"4CA0DA,SAAgB,EACd,EACA,EACA,EACA,EACsB,CACtB,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,kBACZ,EAAa,CAAI,EACjB,EACA,EACA,EAAW,GAAa,SAAS,EACjC,EAAW,GAAa,YAAY,EACpC,EAAW,GAAa,oBAAoB,CAC9C,CACF,CACF,CAGA,SAAgB,EACd,EACA,EACA,EAC4B,CAC5B,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,2BAA2B,EAAa,CAAI,EAAG,EAAU,EAAW,CAAS,CAAC,CAC9F,CACF,CAGA,SAAgB,EAAoC,EAAc,EAAgC,CAIhG,OAHkB,KAAK,MACrB,EAAY,CAAC,CAAC,oCAAoC,EAAa,CAAI,EAAG,CAAQ,CAEjE,CAAC,CAAC,IAAK,GAAa,WAAW,KAAK,CAAQ,CAAC,CAC9D,CAOA,SAAS,EAAa,EAAsB,CAC1C,OAAO,EAAK,aAAa,EAAI,EAAO,EAAK,aAAa,CACxD,CAUA,SAAS,EAAW,EAA+C,CACjE,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAC5C,IAAA,GACA,KAAK,IAAI,KAAK,IAAI,KAAK,MAAM,CAAK,EAAG,CAAC,EAAG,UAAa,CAC5D,CAOA,IAAa,EAAb,cAAsC,KAAM,CAAC,EAE7C,IAAI,EACA,EAEJ,SAAS,GAA6B,CACpC,GAAI,EACF,OAAO,EAIT,GAAI,EACF,MAAM,EAGR,IAAM,EAAgB,EAAc,YAAY,GAAG,EAC7C,EAAa,CAEjB,cAAc,EAAgB,IAE9B,2BACF,EACM,EAAqB,CAAC,EAC5B,IAAK,IAAM,KAAa,EAAY,CAClC,IAAI,EACJ,GAAI,CACF,EAAU,EAAc,CAAS,CACnC,OAAS,EAAO,CACd,EAAS,KAAK,KAAK,EAAU,IAAI,aAAiB,MAAQ,EAAM,QAAQ,MAAM;CAAI,CAAC,CAAC,GAAK,OAAO,CAAK,GAAG,EACxG,QACF,CACA,IAAM,EAAU,EAAQ,iBAAiB,EACzC,GAAI,IAAY,EAAwB,CACtC,EAAS,KACP,KAAK,EAAU,oBAAoB,GAAW,UAAU,gFAE1D,EACA,QACF,CAEA,MADA,GAAgB,EACT,CACT,CAOA,KANA,GAAgB,IAAI,EAClB,oDAAoD,EAAgB,EAAE,2NAGrC,EAAS,KAAK;CAAI,GACrD,EACM,CACR,CAQA,SAAS,GAA0B,CACjC,IAAM,EAAO,GAAG,QAAQ,SAAS,GAAG,QAAQ,OAS5C,OARI,QAAQ,WAAa,QAChB,GAAG,EAAK,OAEb,QAAQ,WAAa,SAIV,QAAQ,QAAQ,UAAU,EAAA,EAC1B,QAAQ,oBAAsB,GAAG,EAAK,MAAQ,GAAG,EAAK,OAJ5D,CAKX"}
+{"version":3,"file":"nativeMetrics.js","names":[],"sources":["../src/nativeMetrics.ts"],"sourcesContent":["import { createRequire } from 'node:module';\nimport type { CrossFileDuplicateCandidate, Token, TokenRange } from './duplication.js';\nimport type { CodeMetrics, DuplicationOptions, FunctionMetrics } from './types.js';\n\n/**\n * Halstead counts measured natively; the derived float metrics (volume, effort, ...) are\n * computed in TypeScript because they involve transcendental functions (log2) whose last-bit\n * results can differ between V8 and Rust's libm, and results must not depend on the Rust side's\n * libm build.\n */\nexport interface NativeHalsteadCounts {\n  distinctOperators: number;\n  distinctOperands: number;\n  totalOperators: number;\n  totalOperands: number;\n}\n\nexport interface NativeFunctionMetricsPayload extends Omit<FunctionMetrics, 'halstead'> {\n  halsteadCounts: NativeHalsteadCounts;\n}\n\nexport interface NativeMetricsPayload extends Omit<CodeMetrics, 'halstead' | 'functions' | 'syntaxTree'> {\n  functions: NativeFunctionMetricsPayload[];\n  halsteadCounts: NativeHalsteadCounts;\n  syntaxTree?: string;\n  crossFileData?: NativeCrossFileDataPayload;\n}\n\n/** One file's cross-file clone-detection contribution as serialized by the native addon. */\nexport interface NativeCrossFileDataPayload {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  nearMissBlocks: TokenRange[];\n  /** 1-based lines that are neither blank nor comment-only, sorted ascending. */\n  codeLineNumbers: number[];\n}\n\ninterface NativeBinding {\n  measureCodeNative(\n    code: string,\n    language: string,\n    includeSyntaxTree: boolean,\n    minTokens?: number,\n    maxGapTokens?: number,\n    minSimilarityPercent?: number,\n    includeCrossFileData?: boolean\n  ): string;\n  collectCrossFileDataNative(code: string, language: string, minTokens?: number): string;\n  collectFunctionTokenSequencesNative(code: string, language: string): string;\n  payloadVersion?(): number;\n}\n\n/**\n * Must equal `payload_version` in native/src/lib.rs. A previously built addon survives a\n * `git pull` untouched, so without this handshake it would silently return payloads missing\n * newer fields instead of failing with a clear rebuild message.\n */\nconst expectedPayloadVersion = 7;\n\n/**\n * Measures one file via the native addon, returning the raw payload for assembly in metrics.ts;\n * with `includeCrossFileData`, the payload also carries the file's cross-file contribution.\n */\nexport function measureCodeNative(\n  code: string,\n  language: string,\n  includeSyntaxTree: boolean,\n  duplication?: DuplicationOptions,\n  includeCrossFileData = false\n): NativeMetricsPayload {\n  return JSON.parse(\n    loadBinding().measureCodeNative(\n      toWellFormed(code),\n      language,\n      includeSyntaxTree,\n      clampToU32(duplication?.minTokens),\n      clampToU32(duplication?.maxGapTokens),\n      clampToU32(duplication?.minSimilarityPercent),\n      includeCrossFileData\n    )\n  ) as NativeMetricsPayload;\n}\n\n/** Collects one file's cross-file clone-detection contribution via the native addon. */\nexport function collectCrossFileDataNative(\n  code: string,\n  language: string,\n  minTokens?: number\n): NativeCrossFileDataPayload {\n  return JSON.parse(\n    loadBinding().collectCrossFileDataNative(toWellFormed(code), language, clampToU32(minTokens))\n  ) as NativeCrossFileDataPayload;\n}\n\n/** Collects normalized token hash sequences of every function via the native addon. */\nexport function collectFunctionTokenSequencesNative(code: string, language: string): Int32Array[] {\n  const sequences = JSON.parse(\n    loadBinding().collectFunctionTokenSequencesNative(toWellFormed(code), language)\n  ) as number[][];\n  return sequences.map((sequence) => Int32Array.from(sequence));\n}\n\n/**\n * Lone surrogates cannot cross the N-API boundary losslessly, so ill-formed strings (invalid\n * UTF-16 occasionally present in real-world files) are measured with U+FFFD replacements — the\n * same code units V8's own UTF-8 conversion would substitute.\n */\nfunction toWellFormed(code: string): string {\n  return code.isWellFormed() ? code : code.toWellFormed();\n}\n\n/**\n * The duplication settings cross the boundary as u32, whose JavaScript conversion wraps modulo\n * 2^32 (2 ** 32 would become 0 and match everything). The public API accepts any safe integer, so\n * out-of-range values clamp to [0, u32::MAX] — no source can hold 2^32 tokens, so a clamped\n * threshold behaves identically to the requested one. NaN (e.g. `Number(unsetEnvVariable)`)\n * survives clamping arithmetic and would also convert to 0, so it is treated as an absent setting\n * instead.\n */\nfunction clampToU32(value: number | undefined): number | undefined {\n  return value === undefined || Number.isNaN(value)\n    ? undefined\n    : Math.min(Math.max(Math.trunc(value), 0), 0xFF_FF_FF_FF);\n}\n\n/**\n * Raised when no usable native addon can be loaded. Every measurement fails identically until the\n * addon is built, so callers measuring many files (the CLI scan) treat it as fatal for the whole\n * run instead of recording one \"skipped\" entry per file.\n */\nexport class NativeAddonError extends Error {}\n\nlet cachedBinding: NativeBinding | undefined;\nlet cachedFailure: NativeAddonError | undefined;\n\nfunction loadBinding(): NativeBinding {\n  if (cachedBinding) {\n    return cachedBinding;\n  }\n  // The failure is memoized too: resolution (including platformTriplet's diagnostic-report call\n  // on Linux) would otherwise repeat for every measured file of an already-failing run.\n  if (cachedFailure) {\n    throw cachedFailure;\n  }\n  // Resolved relative to this file, so both src/ (tests) and dist/ (build) find the addon.\n  const requireNative = createRequire(import.meta.url);\n  const specifiers = [\n    // A prebuilt platform package, when published for this platform.\n    `code-gauge-${platformTriplet()}`,\n    // A locally built addon (`bun run build-native`).\n    '../native/code-gauge.node',\n  ];\n  const failures: string[] = [];\n  for (const specifier of specifiers) {\n    let binding: NativeBinding;\n    try {\n      binding = requireNative(specifier) as NativeBinding;\n    } catch (error) {\n      failures.push(`  ${specifier}: ${error instanceof Error ? error.message.split('\\n')[0] : String(error)}`);\n      continue;\n    }\n    const version = binding.payloadVersion?.();\n    if (version !== expectedPayloadVersion) {\n      failures.push(\n        `  ${specifier}: payload version ${version ?? 'unknown'} does not match the expected ` +\n          `${expectedPayloadVersion}; rebuild the addon with \\`bun run build-native\\``\n      );\n      continue;\n    }\n    cachedBinding = binding;\n    return binding;\n  }\n  cachedFailure = new NativeAddonError(\n    `The code-gauge native addon is not available for ${platformTriplet()}. Build it with ` +\n      '`node scripts/buildNative.mjs` in the code-gauge package directory (requires a Rust ' +\n      'toolchain); when installing with npm, also allow install scripts for code-gauge so its ' +\n      `postinstall build can run.\\n${failures.join('\\n')}`\n  );\n  throw cachedFailure;\n}\n\n/**\n * The platform-package suffix in the napi-rs naming convention: Linux targets are qualified by\n * libc ABI (`linux-x64-gnu` / `linux-x64-musl`) because a glibc-linked addon cannot load on\n * Alpine/musl, and Windows by toolchain ABI (`win32-x64-msvc`), matching what napi-rs tooling\n * generates. Must match scripts/installNative.mjs and the build-native workflow's target list.\n */\nfunction platformTriplet(): string {\n  const base = `${process.platform}-${process.arch}`;\n  if (process.platform === 'win32') {\n    return `${base}-msvc`;\n  }\n  if (process.platform !== 'linux') {\n    return base;\n  }\n  // Musl builds of Node report no glibc runtime version; see napi-rs's isMusl detection.\n  const report = process.report?.getReport() as { header?: { glibcVersionRuntime?: string } } | undefined;\n  return report?.header?.glibcVersionRuntime ? `${base}-gnu` : `${base}-musl`;\n}\n"],"mappings":"4CAgEA,SAAgB,EACd,EACA,EACA,EACA,EACA,EAAuB,GACD,CACtB,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,kBACZ,EAAa,CAAI,EACjB,EACA,EACA,EAAW,GAAa,SAAS,EACjC,EAAW,GAAa,YAAY,EACpC,EAAW,GAAa,oBAAoB,EAC5C,CACF,CACF,CACF,CAGA,SAAgB,EACd,EACA,EACA,EAC4B,CAC5B,OAAO,KAAK,MACV,EAAY,CAAC,CAAC,2BAA2B,EAAa,CAAI,EAAG,EAAU,EAAW,CAAS,CAAC,CAC9F,CACF,CAGA,SAAgB,EAAoC,EAAc,EAAgC,CAIhG,OAHkB,KAAK,MACrB,EAAY,CAAC,CAAC,oCAAoC,EAAa,CAAI,EAAG,CAAQ,CAEjE,CAAC,CAAC,IAAK,GAAa,WAAW,KAAK,CAAQ,CAAC,CAC9D,CAOA,SAAS,EAAa,EAAsB,CAC1C,OAAO,EAAK,aAAa,EAAI,EAAO,EAAK,aAAa,CACxD,CAUA,SAAS,EAAW,EAA+C,CACjE,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAC5C,IAAA,GACA,KAAK,IAAI,KAAK,IAAI,KAAK,MAAM,CAAK,EAAG,CAAC,EAAG,UAAa,CAC5D,CAOA,IAAa,EAAb,cAAsC,KAAM,CAAC,EAE7C,IAAI,EACA,EAEJ,SAAS,GAA6B,CACpC,GAAI,EACF,OAAO,EAIT,GAAI,EACF,MAAM,EAGR,IAAM,EAAgB,EAAc,YAAY,GAAG,EAC7C,EAAa,CAEjB,cAAc,EAAgB,IAE9B,2BACF,EACM,EAAqB,CAAC,EAC5B,IAAK,IAAM,KAAa,EAAY,CAClC,IAAI,EACJ,GAAI,CACF,EAAU,EAAc,CAAS,CACnC,OAAS,EAAO,CACd,EAAS,KAAK,KAAK,EAAU,IAAI,aAAiB,MAAQ,EAAM,QAAQ,MAAM;CAAI,CAAC,CAAC,GAAK,OAAO,CAAK,GAAG,EACxG,QACF,CACA,IAAM,EAAU,EAAQ,iBAAiB,EACzC,GAAI,IAAY,EAAwB,CACtC,EAAS,KACP,KAAK,EAAU,oBAAoB,GAAW,UAAU,gFAE1D,EACA,QACF,CAEA,MADA,GAAgB,EACT,CACT,CAOA,KANA,GAAgB,IAAI,EAClB,oDAAoD,EAAgB,EAAE,2NAGrC,EAAS,KAAK;CAAI,GACrD,EACM,CACR,CAQA,SAAS,GAA0B,CACjC,IAAM,EAAO,GAAG,QAAQ,SAAS,GAAG,QAAQ,OAS5C,OARI,QAAQ,WAAa,QAChB,GAAG,EAAK,OAEb,QAAQ,WAAa,SAIV,QAAQ,QAAQ,UAAU,EAAA,EAC1B,QAAQ,oBAAsB,GAAG,EAAK,MAAQ,GAAG,EAAK,OAJ5D,CAKX"}

package/dist/scan.cjs

@@ -1,2 +1,2 @@
-"use strict";const e=require("./_virtual/_rolldown/runtime.cjs"),t=require("./crossFileDuplication.cjs"),n=require("./languages.cjs"),r=require("./nativeMetrics.cjs"),i=require("./metrics.cjs");let a=require("node:path");a=e.__toESM(a,1);let o=require("node:fs/promises"),s=require("node:os");s=e.__toESM(s,1);const c=new Set([`.agents`,`.claude`,`.cursor`,`.git`,`.next`,`.playwright-cli`,`.tox`,`.tmp`,`.turbo`,`.venv`,`.yarn`,`__fixtures__`,`__generated__`,`__pycache__`,`coverage`,`dist`,`fixtures`,`generated`,`node_modules`,`obj`,`target`,`test-fixtures`,`vendor`,`venv`]),l=new Set([`__tests__`,`test`,`tests`,`spec`]),u=/(?:^test(?:[_-].*)?|\.(?:spec|test)|[_-](?:test|spec))\.[^.]+$/iu,d=/Tests?\.(?:java|kt|cs)$/u;function f(e){return e===`~`?s.default.homedir():e.startsWith(`~/`)?a.default.join(s.default.homedir(),e.slice(2)):a.default.resolve(e)}async function p(e){try{return(await(0,o.stat)(e)).isDirectory()?e:a.default.dirname(e)}catch{return a.default.dirname(e)}}async function m(e,t){let n=[],r=[],i=[],s=e;try{s=await(0,o.realpath)(e)}catch{}let c=a.default.dirname(s),l;try{l=await(0,o.stat)(s)}catch(e){let t=`${A(s,c)}: ${N(e)}`;return{displayRoot:c,files:n,errors:[t],warnings:i,fatalError:t}}if(l.isFile()){let e=a.default.dirname(s),o=k(s,t,!0);if(!o){let t=`${A(s,e)}: unsupported file type`;return{displayRoot:e,files:n,errors:[t],warnings:i,fatalError:t}}let c=_(t,n,r,i,e);try{await C(s,o,`single-file`,c,s)}catch(t){return g(t,e,n,r,i)}return{displayRoot:e,files:n,errors:r,warnings:i}}try{await b(s,_(t,n,r,i,s))}catch(e){return g(e,s,n,r,i)}return{displayRoot:s,files:n,errors:r,warnings:i}}async function h(e,t,n){let r=[],i=[],s=[],c=_(n,r,i,s,e);for(let l of t){let t=O(l,n)?k(l,n):void 0;if(!t)continue;let u=a.default.join(e,l);if(!(await(0,o.lstat)(u).catch(()=>{}))?.isSymbolicLink())try{await C(u,t,`directory`,c)}catch(t){return g(t,e,r,i,s)}}return{displayRoot:e,files:r,errors:i,warnings:s}}function g(e,t,n,i,a){if(!(e instanceof r.NativeAddonError))throw e;let o=N(e);return{displayRoot:t,files:n,errors:[...i,o],warnings:a,fatalError:o}}function _(e,t,n,r,i){return{options:e,files:t,errors:n,warnings:r,visitedDirectories:new Set,visitedFiles:new Set,rootDirectory:i}}async function v(e,t,n){try{return await e()}catch(e){n.errors.push(`${A(t,n.rootDirectory)}: ${N(e)}`);return}}async function y(e,t){let n=await v(()=>(0,o.realpath)(e),e,t);return n!==void 0&&D(n,t.rootDirectory)?n:void 0}async function b(e,t){let n=await y(e,t);if(n===void 0||t.visitedDirectories.has(n))return;t.visitedDirectories.add(n);let r=await v(()=>(0,o.readdir)(e,{withFileTypes:!0}),e,t);if(r!==void 0)for(let n of r){let r=a.default.join(e,n.name);if(n.isSymbolicLink()){await x(n.name,r,t);continue}if(n.isDirectory()){if(E(n.name,t.options))continue;await b(r,t);continue}n.isFile()&&await S(r,t)}}async function x(e,t,n){let r=await y(t,n);if(r===void 0)return;let i=await v(()=>(0,o.stat)(t),t,n);if(i!==void 0){if(i.isDirectory()){if(E(e,n.options)||E(a.default.basename(r),n.options))return;await b(t,n);return}i.isFile()&&await S(t,n,r,r)}}async function S(e,t,n=e,r){let i=k(n,t.options);i&&await C(e,i,`directory`,t,r)}async function C(e,t,n,a,s){try{let r=s??await(0,o.realpath)(e);if(a.visitedFiles.has(r))return;a.visitedFiles.add(r);let c=await(0,o.readFile)(e,`utf8`),l={language:t,duplication:a.options.duplication},u={file:e,metrics:i.measureCode(c,l)};if(n===`directory`)try{u.duplicationCandidates=i.collectCrossFileDuplicationFileData(c,l)}catch(t){a.warnings.push(`${A(e,a.rootDirectory)}: cross-file duplication candidates unavailable: ${N(t)}`)}a.files.push(u)}catch(t){if(t instanceof r.NativeAddonError)throw t;a.errors.push(`${A(e,a.rootDirectory)}: ${N(t)}`)}}function w(e,n){if(e.fatalError||e.files.length<2)return;let r=e.files.flatMap(({file:t,duplicationCandidates:n})=>n?[{file:A(t,e.displayRoot),...n}]:[]);r.length<2||(e.crossFileDuplication=t.measureCrossFileDuplication(r,n.duplication))}function T(e,t,n){let r=new Set(e?.duplication.duplicateLineNumbers),i=t&&Object.hasOwn(t.duplicateLineNumbersByFile,n)?t.duplicateLineNumbersByFile[n]??[]:[];for(let e of i)r.add(e);return r}function E(e,t){return c.has(e)?!0:!t.includeTests&&l.has(e)}function D(e,t){let n=a.default.relative(t,e);return n===``||n!==`..`&&!n.startsWith(`..${a.default.sep}`)&&!a.default.isAbsolute(n)}function O(e,t){let n=e.split(`/`);for(let e of n.slice(0,-1))if(c.has(e)||!t.includeTests&&l.has(e))return!1;return k(e,t)!==void 0}function k(e,t,r=!1){let i=e.toLowerCase();if(!(!r&&(i.endsWith(`.d.ts`)||i.endsWith(`.d.mts`)||i.endsWith(`.d.cts`)||i.endsWith(`.min.js`)||i.endsWith(`.pnp.cjs`)))&&(r||t.includeTests||!u.test(a.default.basename(e))&&!d.test(a.default.basename(e))))return n.detectLanguage(e)}function A(e,t){return a.default.relative(t,e)||a.default.basename(e)}function j(e){process.stdout.write(e)}function M(e){process.stderr.write(e)}function N(e){return e instanceof Error?e.message:String(e)}exports.addCrossFileDuplication=w,exports.collectDuplicatedLineNumbers=T,exports.configSearchDirectory=p,exports.formatError=N,exports.formatPath=A,exports.getLanguage=k,exports.isScannedPath=O,exports.resolveTarget=f,exports.scanListedFiles=h,exports.scanTarget=m,exports.writeStderr=M,exports.writeStdout=j;
+"use strict";const e=require("./_virtual/_rolldown/runtime.cjs"),t=require("./crossFileDuplication.cjs"),n=require("./languages.cjs"),r=require("./nativeMetrics.cjs"),i=require("./metrics.cjs");let a=require("node:path");a=e.__toESM(a,1);let o=require("node:fs/promises"),s=require("node:os");s=e.__toESM(s,1);const c=new Set([`.agents`,`.claude`,`.cursor`,`.git`,`.next`,`.playwright-cli`,`.tox`,`.tmp`,`.turbo`,`.venv`,`.yarn`,`__fixtures__`,`__generated__`,`__pycache__`,`coverage`,`dist`,`fixtures`,`generated`,`node_modules`,`obj`,`target`,`test-fixtures`,`vendor`,`venv`]),l=new Set([`__tests__`,`test`,`tests`,`spec`]),u=/(?:^test(?:[_-].*)?|\.(?:spec|test)|[_-](?:test|spec))\.[^.]+$/iu,d=/Tests?\.(?:java|kt|cs)$/u;function f(e){return e===`~`?s.default.homedir():e.startsWith(`~/`)?a.default.join(s.default.homedir(),e.slice(2)):a.default.resolve(e)}async function p(e){try{return(await(0,o.stat)(e)).isDirectory()?e:a.default.dirname(e)}catch{return a.default.dirname(e)}}async function m(e,t){let n=[],r=[],i=[],s=e;try{s=await(0,o.realpath)(e)}catch{}let c=a.default.dirname(s),l;try{l=await(0,o.stat)(s)}catch(e){let t=`${j(s,c)}: ${P(e)}`;return{displayRoot:c,files:n,errors:[t],warnings:i,fatalError:t}}if(l.isFile()){let e=a.default.dirname(s),o=A(s,t,!0);if(!o){let t=`${j(s,e)}: unsupported file type`;return{displayRoot:e,files:n,errors:[t],warnings:i,fatalError:t}}let c=_(t,n,r,i,e);try{await C(s,o,`single-file`,c,s)}catch(t){return g(t,e,n,r,i)}return{displayRoot:e,files:n,errors:r,warnings:i}}try{await b(s,_(t,n,r,i,s))}catch(e){return g(e,s,n,r,i)}return{displayRoot:s,files:n,errors:r,warnings:i}}async function h(e,t,n){let r=[],i=[],s=[],c=_(n,r,i,s,e);for(let l of t){let t=k(l,n)?A(l,n):void 0;if(!t)continue;let u=a.default.join(e,l);if(!(await(0,o.lstat)(u).catch(()=>{}))?.isSymbolicLink())try{await C(u,t,`directory`,c)}catch(t){return g(t,e,r,i,s)}}return{displayRoot:e,files:r,errors:i,warnings:s}}function g(e,t,n,i,a){if(!(e instanceof r.NativeAddonError))throw e;let o=P(e);return{displayRoot:t,files:n,errors:[...i,o],warnings:a,fatalError:o}}function _(e,t,n,r,i){return{options:e,files:t,errors:n,warnings:r,visitedDirectories:new Set,visitedFiles:new Set,rootDirectory:i}}async function v(e,t,n){try{return await e()}catch(e){n.errors.push(`${j(t,n.rootDirectory)}: ${P(e)}`);return}}async function y(e,t){let n=await v(()=>(0,o.realpath)(e),e,t);return n!==void 0&&O(n,t.rootDirectory)?n:void 0}async function b(e,t){let n=await y(e,t);if(n===void 0||t.visitedDirectories.has(n))return;t.visitedDirectories.add(n);let r=await v(()=>(0,o.readdir)(e,{withFileTypes:!0}),e,t);if(r!==void 0)for(let n of r){let r=a.default.join(e,n.name);if(n.isSymbolicLink()){await x(n.name,r,t);continue}if(n.isDirectory()){if(D(n.name,t.options))continue;await b(r,t);continue}n.isFile()&&await S(r,t)}}async function x(e,t,n){let r=await y(t,n);if(r===void 0)return;let i=await v(()=>(0,o.stat)(t),t,n);if(i!==void 0){if(i.isDirectory()){if(D(e,n.options)||D(a.default.basename(r),n.options))return;await b(t,n);return}i.isFile()&&await S(t,n,r,r)}}async function S(e,t,n=e,r){let i=A(n,t.options);i&&await C(e,i,`directory`,t,r)}async function C(e,t,n,a,s){try{let r=s??await(0,o.realpath)(e);if(a.visitedFiles.has(r))return;a.visitedFiles.add(r);let c=await(0,o.readFile)(e,`utf8`),l={language:t,duplication:a.options.duplication};if(n===`single-file`){a.files.push({file:e,metrics:i.measureCode(c,l)});return}let{metrics:u,crossFileData:d,crossFileError:f}=w(c,l);f!==void 0&&a.warnings.push(`${j(e,a.rootDirectory)}: cross-file duplication candidates unavailable: ${f}`),a.files.push({file:e,metrics:u,duplicationCandidates:d})}catch(t){if(t instanceof r.NativeAddonError)throw t;a.errors.push(`${j(e,a.rootDirectory)}: ${P(t)}`)}}function w(e,t){try{return i.measureCodeWithCrossFileData(e,t)}catch(n){if(n instanceof r.NativeAddonError)throw n;return{metrics:i.measureCode(e,t),crossFileError:P(n)}}}function T(e,n){if(e.fatalError||e.files.length<2)return;let r=e.files.flatMap(({file:t,duplicationCandidates:n})=>n?[{file:j(t,e.displayRoot),...n}]:[]);r.length<2||(e.crossFileDuplication=t.measureCrossFileDuplication(r,n.duplication))}function E(e,t,n){let r=new Set(e?.duplication.duplicateLineNumbers),i=t&&Object.hasOwn(t.duplicateLineNumbersByFile,n)?t.duplicateLineNumbersByFile[n]??[]:[];for(let e of i)r.add(e);return r}function D(e,t){return c.has(e)?!0:!t.includeTests&&l.has(e)}function O(e,t){let n=a.default.relative(t,e);return n===``||n!==`..`&&!n.startsWith(`..${a.default.sep}`)&&!a.default.isAbsolute(n)}function k(e,t){let n=e.split(`/`);for(let e of n.slice(0,-1))if(c.has(e)||!t.includeTests&&l.has(e))return!1;return A(e,t)!==void 0}function A(e,t,r=!1){let i=e.toLowerCase();if(!(!r&&(i.endsWith(`.d.ts`)||i.endsWith(`.d.mts`)||i.endsWith(`.d.cts`)||i.endsWith(`.min.js`)||i.endsWith(`.pnp.cjs`)))&&(r||t.includeTests||!u.test(a.default.basename(e))&&!d.test(a.default.basename(e))))return n.detectLanguage(e)}function j(e,t){return a.default.relative(t,e)||a.default.basename(e)}function M(e){process.stdout.write(e)}function N(e){process.stderr.write(e)}function P(e){return e instanceof Error?e.message:String(e)}exports.addCrossFileDuplication=T,exports.collectDuplicatedLineNumbers=E,exports.configSearchDirectory=p,exports.formatError=P,exports.formatPath=j,exports.getLanguage=A,exports.isScannedPath=k,exports.measureWithCrossFileData=w,exports.resolveTarget=f,exports.scanListedFiles=h,exports.scanTarget=m,exports.writeStderr=N,exports.writeStdout=M;
 //# sourceMappingURL=scan.cjs.map

package/dist/scan.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"scan.cjs","names":["os","path","stat","realpath","lstat","NativeAddonError","readdir","readFile","measureCode","collectCrossFileDuplicationFileData","measureCrossFileDuplication","detectLanguage"],"sources":["../src/scan.ts"],"sourcesContent":["import { lstat, readdir, readFile, realpath, stat } from 'node:fs/promises';\nimport os from 'node:os';\nimport path from 'node:path';\nimport { measureCrossFileDuplication, type CrossFileDuplicationMetrics } from './crossFileDuplication.js';\nimport type { CrossFileDuplicationFileData } from './duplication.js';\nimport { detectLanguage } from './languages.js';\nimport { collectCrossFileDuplicationFileData, measureCode } from './metrics.js';\nimport { NativeAddonError } from './nativeMetrics.js';\nimport type { CodeMetrics, DuplicationOptions, LanguageName } from './types.js';\n\n/** The scan settings shared by every command (a structural subset of each command's options). */\nexport interface ScanOptions {\n  duplication: Required<DuplicationOptions>;\n  includeTests: boolean;\n}\n\nexport interface FileMetrics {\n  file: string;\n  metrics: CodeMetrics;\n  /** Cross-file duplicate candidates and token/statement data, collected only for directory scans. */\n  duplicationCandidates?: CrossFileDuplicationFileData;\n}\n\nexport interface ScanResult {\n  crossFileDuplication?: CrossFileDuplicationMetrics;\n  displayRoot: string;\n  errors: string[];\n  /** Non-fatal degradations (e.g. cross-file candidates unavailable); the file is still measured. */\n  warnings: string[];\n  fatalError?: string;\n  files: FileMetrics[];\n}\n\nconst ignoredDirectoryNames = new Set([\n  '.agents',\n  '.claude',\n  '.cursor',\n  '.git',\n  '.next',\n  '.playwright-cli',\n  '.tox',\n  '.tmp',\n  '.turbo',\n  '.venv',\n  '.yarn',\n  '__fixtures__',\n  '__generated__',\n  '__pycache__',\n  'coverage',\n  'dist',\n  'fixtures',\n  'generated',\n  'node_modules',\n  // .NET SDK intermediate output (generated sources such as `*.GlobalUsings.g.cs`).\n  'obj',\n  'target',\n  'test-fixtures',\n  'vendor',\n  'venv',\n]);\n\nconst testDirectoryNames = new Set(['__tests__', 'test', 'tests', 'spec']);\nconst testFilePattern = /(?:^test(?:[_-].*)?|\\.(?:spec|test)|[_-](?:test|spec))\\.[^.]+$/iu;\n// JUnit (Java/Kotlin) and xUnit/NUnit (C#) tests use case-sensitive `Test`/`Tests` class-name\n// suffixes; case-insensitive matching would catch production files like `contest.java`.\nconst suffixTestFilePattern = /Tests?\\.(?:java|kt|cs)$/u;\n\nexport function resolveTarget(target: string): string {\n  if (target === '~') {\n    return os.homedir();\n  }\n\n  if (target.startsWith('~/')) {\n    return path.join(os.homedir(), target.slice(2));\n  }\n\n  return path.resolve(target);\n}\n\n/** Returns the directory from which the config file search should start (the target itself if it is a directory). */\nexport async function configSearchDirectory(target: string): Promise<string> {\n  try {\n    const targetStat = await stat(target);\n    return targetStat.isDirectory() ? target : path.dirname(target);\n  } catch {\n    return path.dirname(target);\n  }\n}\n\n/** Shared state of one scan, threaded through the directory walk instead of positional plumbing. */\ninterface ScanContext {\n  options: ScanOptions;\n  files: FileMetrics[];\n  errors: string[];\n  warnings: string[];\n  visitedDirectories: Set<string>;\n  visitedFiles: Set<string>;\n  /** Scan root: paths are displayed relative to it, and symbolic links may not escape it. */\n  rootDirectory: string;\n}\n\nexport async function scanTarget(target: string, options: ScanOptions): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  let canonicalTarget = target;\n  try {\n    canonicalTarget = await realpath(target);\n  } catch {\n    // stat below reports missing targets with the original path.\n  }\n\n  const fallbackDisplayRoot = path.dirname(canonicalTarget);\n  let targetStat;\n\n  try {\n    targetStat = await stat(canonicalTarget);\n  } catch (error) {\n    const fatalError = `${formatPath(canonicalTarget, fallbackDisplayRoot)}: ${formatError(error)}`;\n    return { displayRoot: fallbackDisplayRoot, files, errors: [fatalError], warnings, fatalError };\n  }\n\n  if (targetStat.isFile()) {\n    const displayRoot = path.dirname(canonicalTarget);\n    const language = getLanguage(canonicalTarget, options, true);\n    if (!language) {\n      const fatalError = `${formatPath(canonicalTarget, displayRoot)}: unsupported file type`;\n      return { displayRoot, files, errors: [fatalError], warnings, fatalError };\n    }\n\n    const context = makeScanContext(options, files, errors, warnings, displayRoot);\n    try {\n      await measureFile(canonicalTarget, language, 'single-file', context, canonicalTarget);\n    } catch (error) {\n      return toFatalResult(error, displayRoot, files, errors, warnings);\n    }\n    return { displayRoot, files, errors, warnings };\n  }\n\n  try {\n    await scanDirectory(canonicalTarget, makeScanContext(options, files, errors, warnings, canonicalTarget));\n  } catch (error) {\n    return toFatalResult(error, canonicalTarget, files, errors, warnings);\n  }\n  return { displayRoot: canonicalTarget, files, errors, warnings };\n}\n\n/**\n * Measures an explicit list of repository-relative files (the diff gate's git-visible allowlist)\n * instead of walking the directory tree, so ignored artifact directories are never parsed. Paths\n * outside the scan scope (ignored/test directories, unsupported or test file names) are skipped\n * with the same rules as the walk.\n */\nexport async function scanListedFiles(\n  rootDirectory: string,\n  relativePaths: Iterable<string>,\n  options: ScanOptions\n): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  const context = makeScanContext(options, files, errors, warnings, rootDirectory);\n  for (const relativePath of relativePaths) {\n    const language = isScannedPath(relativePath, options) ? getLanguage(relativePath, options) : undefined;\n    if (!language) {\n      continue;\n    }\n    const absolutePath = path.join(rootDirectory, relativePath);\n    // Symbolic links are not source files: git stores only their target string, so measuring\n    // through them would diverge from what any revision of the repository actually contains.\n    const stats = await lstat(absolutePath).catch(() => {});\n    if (stats?.isSymbolicLink()) {\n      continue;\n    }\n    try {\n      await measureFile(absolutePath, language, 'directory', context);\n    } catch (error) {\n      return toFatalResult(error, rootDirectory, files, errors, warnings);\n    }\n  }\n  return { displayRoot: rootDirectory, files, errors, warnings };\n}\n\n/** A run-wide failure (a missing native addon) as a fatal result; anything else keeps throwing. */\nfunction toFatalResult(\n  error: unknown,\n  displayRoot: string,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[]\n): ScanResult {\n  if (!(error instanceof NativeAddonError)) {\n    throw error;\n  }\n  const fatalError = formatError(error);\n  // Errors the walk accumulated before the fatal failure stay reported alongside it.\n  return { displayRoot, files, errors: [...errors, fatalError], warnings, fatalError };\n}\n\nfunction makeScanContext(\n  options: ScanOptions,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[],\n  rootDirectory: string\n): ScanContext {\n  return { options, files, errors, warnings, visitedDirectories: new Set(), visitedFiles: new Set(), rootDirectory };\n}\n\n/** Runs a filesystem operation, recording a scan error and returning undefined when it fails. */\nasync function tryFileSystem<T>(\n  operation: () => Promise<T>,\n  target: string,\n  context: ScanContext\n): Promise<T | undefined> {\n  try {\n    return await operation();\n  } catch (error) {\n    context.errors.push(`${formatPath(target, context.rootDirectory)}: ${formatError(error)}`);\n    return undefined;\n  }\n}\n\n/** Resolves the path (recording errors); undefined when that fails or the result escapes the root. */\nasync function resolveWithinRoot(target: string, context: ScanContext): Promise<string | undefined> {\n  const resolved = await tryFileSystem(() => realpath(target), target, context);\n  return resolved !== undefined && isWithinDirectory(resolved, context.rootDirectory) ? resolved : undefined;\n}\n\nasync function scanDirectory(directory: string, context: ScanContext): Promise<void> {\n  const resolvedDirectory = await resolveWithinRoot(directory, context);\n  if (resolvedDirectory === undefined || context.visitedDirectories.has(resolvedDirectory)) {\n    return;\n  }\n  context.visitedDirectories.add(resolvedDirectory);\n\n  const entries = await tryFileSystem(() => readdir(directory, { withFileTypes: true }), directory, context);\n  if (entries === undefined) {\n    return;\n  }\n\n  for (const entry of entries) {\n    const entryPath = path.join(directory, entry.name);\n    if (entry.isSymbolicLink()) {\n      await scanSymbolicLink(entry.name, entryPath, context);\n      continue;\n    }\n\n    if (entry.isDirectory()) {\n      if (shouldSkipDirectory(entry.name, context.options)) {\n        continue;\n      }\n      await scanDirectory(entryPath, context);\n      continue;\n    }\n\n    if (entry.isFile()) {\n      await measureScannableFile(entryPath, context);\n    }\n  }\n}\n\nasync function scanSymbolicLink(name: string, entryPath: string, context: ScanContext): Promise<void> {\n  const resolvedPath = await resolveWithinRoot(entryPath, context);\n  if (resolvedPath === undefined) {\n    return;\n  }\n\n  const entryStat = await tryFileSystem(() => stat(entryPath), entryPath, context);\n  if (entryStat === undefined) {\n    return;\n  }\n\n  if (entryStat.isDirectory()) {\n    if (\n      shouldSkipDirectory(name, context.options) ||\n      shouldSkipDirectory(path.basename(resolvedPath), context.options)\n    ) {\n      return;\n    }\n    await scanDirectory(entryPath, context);\n    return;\n  }\n\n  if (entryStat.isFile()) {\n    await measureScannableFile(entryPath, context, resolvedPath, resolvedPath);\n  }\n}\n\nasync function measureScannableFile(\n  file: string,\n  context: ScanContext,\n  languageFile = file,\n  realFile?: string\n): Promise<void> {\n  const language = getLanguage(languageFile, context.options);\n  if (language) {\n    await measureFile(file, language, 'directory', context, realFile);\n  }\n}\n\nasync function measureFile(\n  file: string,\n  language: LanguageName,\n  mode: 'single-file' | 'directory',\n  context: ScanContext,\n  realFile?: string\n): Promise<void> {\n  try {\n    const resolvedFile = realFile ?? (await realpath(file));\n    if (context.visitedFiles.has(resolvedFile)) {\n      return;\n    }\n    context.visitedFiles.add(resolvedFile);\n\n    const code = await readFile(file, 'utf8');\n    const measureOptions = { language, duplication: context.options.duplication };\n    const fileMetrics: FileMetrics = { file, metrics: measureCode(code, measureOptions) };\n    // Only directory scans compare files against each other; a single-file target has no peers.\n    // Candidate collection is an auxiliary pass: if it fails where measureCode succeeded (an\n    // addon error specific to this pass), that must not discard the measured metrics.\n    if (mode === 'directory') {\n      try {\n        fileMetrics.duplicationCandidates = collectCrossFileDuplicationFileData(code, measureOptions);\n      } catch (error) {\n        // A warning, not an error: the file's metrics are complete, only its participation in\n        // cross-file matching is lost, so it is not \"skipped\" and must not fail --fail-on-error.\n        context.warnings.push(\n          `${formatPath(file, context.rootDirectory)}: cross-file duplication candidates unavailable: ${formatError(error)}`\n        );\n      }\n    }\n    context.files.push(fileMetrics);\n  } catch (error) {\n    // A missing native addon fails every file identically: propagate it once as a fatal scan\n    // error instead of recording one \"skipped\" entry per file behind a successful exit code.\n    if (error instanceof NativeAddonError) {\n      throw error;\n    }\n    context.errors.push(`${formatPath(file, context.rootDirectory)}: ${formatError(error)}`);\n  }\n}\n\n/** Runs after the scan so every measured file's candidates participate. */\nexport function addCrossFileDuplication(result: ScanResult, options: ScanOptions): void {\n  if (result.fatalError || result.files.length < 2) {\n    return;\n  }\n  const sourceFiles = result.files.flatMap(({ file, duplicationCandidates }) =>\n    duplicationCandidates ? [{ file: formatPath(file, result.displayRoot), ...duplicationCandidates }] : []\n  );\n  if (sourceFiles.length < 2) {\n    return;\n  }\n  result.crossFileDuplication = measureCrossFileDuplication(sourceFiles, options.duplication);\n}\n\n/**\n * Distinct 1-based code lines covered by within-file or cross-file duplicated content. Both\n * sources expose the exact lines carrying matched tokens (never block bounding ranges, which\n * would over-count comment/blank lines and the unmatched gap of a merged clone), so the union is\n * a subset of the file's code lines and a ratio derived over code lines can never exceed 1.\n */\nexport function collectDuplicatedLineNumbers(\n  metrics: CodeMetrics | undefined,\n  crossFileDuplication: CrossFileDuplicationMetrics | undefined,\n  formattedFile: string\n): Set<number> {\n  const lines = new Set(metrics?.duplication.duplicateLineNumbers);\n  // Object.hasOwn: a file named like an Object.prototype member must not read an inherited value.\n  const crossFileLines =\n    crossFileDuplication && Object.hasOwn(crossFileDuplication.duplicateLineNumbersByFile, formattedFile)\n      ? (crossFileDuplication.duplicateLineNumbersByFile[formattedFile] ?? [])\n      : [];\n  for (const line of crossFileLines) {\n    lines.add(line);\n  }\n  return lines;\n}\n\nfunction shouldSkipDirectory(name: string, options: ScanOptions): boolean {\n  if (ignoredDirectoryNames.has(name)) {\n    return true;\n  }\n\n  if (options.includeTests) {\n    return false;\n  }\n\n  return testDirectoryNames.has(name);\n}\n\nfunction isWithinDirectory(candidate: string, directory: string): boolean {\n  const relative = path.relative(directory, candidate);\n  return relative === '' || (relative !== '..' && !relative.startsWith(`..${path.sep}`) && !path.isAbsolute(relative));\n}\n\n/**\n * Whether a repository-relative path would be scanned: no ignored or excluded-test directory\n * segment and a supported, non-test file name. The diff gate uses this for base-revision\n * eligibility, so code renamed into scan scope gates as new code instead of ratcheting against\n * a blob the scanner would never have measured.\n */\nexport function isScannedPath(relativePath: string, options: ScanOptions): boolean {\n  const segments = relativePath.split('/');\n  for (const segment of segments.slice(0, -1)) {\n    if (ignoredDirectoryNames.has(segment) || (!options.includeTests && testDirectoryNames.has(segment))) {\n      return false;\n    }\n  }\n  return getLanguage(relativePath, options) !== undefined;\n}\n\nexport function getLanguage(file: string, options: ScanOptions, explicitTarget = false): LanguageName | undefined {\n  const lowerFile = file.toLowerCase();\n  if (\n    !explicitTarget &&\n    (lowerFile.endsWith('.d.ts') ||\n      lowerFile.endsWith('.d.mts') ||\n      lowerFile.endsWith('.d.cts') ||\n      lowerFile.endsWith('.min.js') ||\n      lowerFile.endsWith('.pnp.cjs'))\n  ) {\n    return undefined;\n  }\n\n  if (\n    !explicitTarget &&\n    !options.includeTests &&\n    (testFilePattern.test(path.basename(file)) || suffixTestFilePattern.test(path.basename(file)))\n  ) {\n    return undefined;\n  }\n\n  return detectLanguage(file);\n}\n\nexport function formatPath(file: string, base: string): string {\n  return path.relative(base, file) || path.basename(file);\n}\n\nexport function writeStdout(message: string): void {\n  process.stdout.write(message);\n}\n\nexport function writeStderr(message: string): void {\n  process.stderr.write(message);\n}\n\nexport function formatError(error: unknown): string {\n  return error instanceof Error ? error.message : String(error);\n}\n"],"mappings":"sTAiCA,MAAM,EAAwB,IAAI,IAAI,CACpC,UACA,UACA,UACA,OACA,QACA,kBACA,OACA,OACA,SACA,QACA,QACA,eACA,gBACA,cACA,WACA,OACA,WACA,YACA,eAEA,MACA,SACA,gBACA,SACA,MACF,CAAC,EAEK,EAAqB,IAAI,IAAI,CAAC,YAAa,OAAQ,QAAS,MAAM,CAAC,EACnE,EAAkB,mEAGlB,EAAwB,2BAE9B,SAAgB,EAAc,EAAwB,CASpD,OARI,IAAW,IACNA,EAAAA,QAAG,QAAQ,EAGhB,EAAO,WAAW,IAAI,EACjBC,EAAAA,QAAK,KAAKD,EAAAA,QAAG,QAAQ,EAAG,EAAO,MAAM,CAAC,CAAC,EAGzCC,EAAAA,QAAK,QAAQ,CAAM,CAC5B,CAGA,eAAsB,EAAsB,EAAiC,CAC3E,GAAI,CAEF,OAAO,MAAA,EADkBC,EAAAA,KAAAA,CAAK,CAAM,EAAA,CAClB,YAAY,EAAI,EAASD,EAAAA,QAAK,QAAQ,CAAM,CAChE,MAAQ,CACN,OAAOA,EAAAA,QAAK,QAAQ,CAAM,CAC5B,CACF,CAcA,eAAsB,EAAW,EAAgB,EAA2C,CAC1F,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACxB,EAAkB,EACtB,GAAI,CACF,EAAkB,MAAA,EAAME,EAAAA,SAAAA,CAAS,CAAM,CACzC,MAAQ,CAER,CAEA,IAAM,EAAsBF,EAAAA,QAAK,QAAQ,CAAe,EACpD,EAEJ,GAAI,CACF,EAAa,MAAA,EAAMC,EAAAA,KAAAA,CAAK,CAAe,CACzC,OAAS,EAAO,CACd,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAmB,EAAE,IAAI,EAAY,CAAK,IAC5F,MAAO,CAAE,YAAa,EAAqB,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC/F,CAEA,GAAI,EAAW,OAAO,EAAG,CACvB,IAAM,EAAcD,EAAAA,QAAK,QAAQ,CAAe,EAC1C,EAAW,EAAY,EAAiB,EAAS,EAAI,EAC3D,GAAI,CAAC,EAAU,CACb,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAW,EAAE,yBAC/D,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC1E,CAEA,IAAM,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAW,EAC7E,GAAI,CACF,MAAM,EAAY,EAAiB,EAAU,cAAe,EAAS,CAAe,CACtF,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAa,EAAO,EAAQ,CAAQ,CAClE,CACA,MAAO,CAAE,cAAa,QAAO,SAAQ,UAAS,CAChD,CAEA,GAAI,CACF,MAAM,EAAc,EAAiB,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAe,CAAC,CACzG,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAiB,EAAO,EAAQ,CAAQ,CACtE,CACA,MAAO,CAAE,YAAa,EAAiB,QAAO,SAAQ,UAAS,CACjE,CAQA,eAAsB,EACpB,EACA,EACA,EACqB,CACrB,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACtB,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAa,EAC/E,IAAK,IAAM,KAAgB,EAAe,CACxC,IAAM,EAAW,EAAc,EAAc,CAAO,EAAI,EAAY,EAAc,CAAO,EAAI,IAAA,GAC7F,GAAI,CAAC,EACH,SAEF,IAAM,EAAeA,EAAAA,QAAK,KAAK,EAAe,CAAY,EAI1D,KAAI,MAAA,EADgBG,EAAAA,MAAAA,CAAM,CAAY,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EAC3C,eAAe,EAG1B,GAAI,CACF,MAAM,EAAY,EAAc,EAAU,YAAa,CAAO,CAChE,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAe,EAAO,EAAQ,CAAQ,CACpE,CACF,CACA,MAAO,CAAE,YAAa,EAAe,QAAO,SAAQ,UAAS,CAC/D,CAGA,SAAS,EACP,EACA,EACA,EACA,EACA,EACY,CACZ,GAAI,EAAE,aAAiBC,EAAAA,kBACrB,MAAM,EAER,IAAM,EAAa,EAAY,CAAK,EAEpC,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,GAAG,EAAQ,CAAU,EAAG,WAAU,YAAW,CACrF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACa,CACb,MAAO,CAAE,UAAS,QAAO,SAAQ,WAAU,mBAAoB,IAAI,IAAO,aAAc,IAAI,IAAO,eAAc,CACnH,CAGA,eAAe,EACb,EACA,EACA,EACwB,CACxB,GAAI,CACF,OAAO,MAAM,EAAU,CACzB,OAAS,EAAO,CACd,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAQ,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,EACzF,MACF,CACF,CAGA,eAAe,EAAkB,EAAgB,EAAmD,CAClG,IAAM,EAAW,MAAM,OAAA,EAAoBF,EAAAA,SAAAA,CAAS,CAAM,EAAG,EAAQ,CAAO,EAC5E,OAAO,IAAa,IAAA,IAAa,EAAkB,EAAU,EAAQ,aAAa,EAAI,EAAW,IAAA,EACnG,CAEA,eAAe,EAAc,EAAmB,EAAqC,CACnF,IAAM,EAAoB,MAAM,EAAkB,EAAW,CAAO,EACpE,GAAI,IAAsB,IAAA,IAAa,EAAQ,mBAAmB,IAAI,CAAiB,EACrF,OAEF,EAAQ,mBAAmB,IAAI,CAAiB,EAEhD,IAAM,EAAU,MAAM,OAAA,EAAoBG,EAAAA,QAAAA,CAAQ,EAAW,CAAE,cAAe,EAAK,CAAC,EAAG,EAAW,CAAO,EACrG,OAAY,IAAA,GAIhB,IAAK,IAAM,KAAS,EAAS,CAC3B,IAAM,EAAYL,EAAAA,QAAK,KAAK,EAAW,EAAM,IAAI,EACjD,GAAI,EAAM,eAAe,EAAG,CAC1B,MAAM,EAAiB,EAAM,KAAM,EAAW,CAAO,EACrD,QACF,CAEA,GAAI,EAAM,YAAY,EAAG,CACvB,GAAI,EAAoB,EAAM,KAAM,EAAQ,OAAO,EACjD,SAEF,MAAM,EAAc,EAAW,CAAO,EACtC,QACF,CAEI,EAAM,OAAO,GACf,MAAM,EAAqB,EAAW,CAAO,CAEjD,CACF,CAEA,eAAe,EAAiB,EAAc,EAAmB,EAAqC,CACpG,IAAM,EAAe,MAAM,EAAkB,EAAW,CAAO,EAC/D,GAAI,IAAiB,IAAA,GACnB,OAGF,IAAM,EAAY,MAAM,OAAA,EAAoBC,EAAAA,KAAAA,CAAK,CAAS,EAAG,EAAW,CAAO,EAC3E,OAAc,IAAA,GAIlB,IAAI,EAAU,YAAY,EAAG,CAC3B,GACE,EAAoB,EAAM,EAAQ,OAAO,GACzC,EAAoBD,EAAAA,QAAK,SAAS,CAAY,EAAG,EAAQ,OAAO,EAEhE,OAEF,MAAM,EAAc,EAAW,CAAO,EACtC,MACF,CAEI,EAAU,OAAO,GACnB,MAAM,EAAqB,EAAW,EAAS,EAAc,CAAY,CAH3E,CAKF,CAEA,eAAe,EACb,EACA,EACA,EAAe,EACf,EACe,CACf,IAAM,EAAW,EAAY,EAAc,EAAQ,OAAO,EACtD,GACF,MAAM,EAAY,EAAM,EAAU,YAAa,EAAS,CAAQ,CAEpE,CAEA,eAAe,EACb,EACA,EACA,EACA,EACA,EACe,CACf,GAAI,CACF,IAAM,EAAe,GAAa,MAAA,EAAME,EAAAA,SAAAA,CAAS,CAAI,EACrD,GAAI,EAAQ,aAAa,IAAI,CAAY,EACvC,OAEF,EAAQ,aAAa,IAAI,CAAY,EAErC,IAAM,EAAO,MAAA,EAAMI,EAAAA,SAAAA,CAAS,EAAM,MAAM,EAClC,EAAiB,CAAE,WAAU,YAAa,EAAQ,QAAQ,WAAY,EACtE,EAA2B,CAAE,OAAM,QAASC,EAAAA,YAAY,EAAM,CAAc,CAAE,EAIpF,GAAI,IAAS,YACX,GAAI,CACF,EAAY,sBAAwBC,EAAAA,oCAAoC,EAAM,CAAc,CAC9F,OAAS,EAAO,CAGd,EAAQ,SAAS,KACf,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,mDAAmD,EAAY,CAAK,GACjH,CACF,CAEF,EAAQ,MAAM,KAAK,CAAW,CAChC,OAAS,EAAO,CAGd,GAAI,aAAiBJ,EAAAA,iBACnB,MAAM,EAER,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,CACzF,CACF,CAGA,SAAgB,EAAwB,EAAoB,EAA4B,CACtF,GAAI,EAAO,YAAc,EAAO,MAAM,OAAS,EAC7C,OAEF,IAAM,EAAc,EAAO,MAAM,SAAS,CAAE,OAAM,2BAChD,EAAwB,CAAC,CAAE,KAAM,EAAW,EAAM,EAAO,WAAW,EAAG,GAAG,CAAsB,CAAC,EAAI,CAAC,CACxG,EACI,EAAY,OAAS,IAGzB,EAAO,qBAAuBK,EAAAA,4BAA4B,EAAa,EAAQ,WAAW,EAC5F,CAQA,SAAgB,EACd,EACA,EACA,EACa,CACb,IAAM,EAAQ,IAAI,IAAI,GAAS,YAAY,oBAAoB,EAEzD,EACJ,GAAwB,OAAO,OAAO,EAAqB,2BAA4B,CAAa,EAC/F,EAAqB,2BAA2B,IAAkB,CAAC,EACpE,CAAC,EACP,IAAK,IAAM,KAAQ,EACjB,EAAM,IAAI,CAAI,EAEhB,OAAO,CACT,CAEA,SAAS,EAAoB,EAAc,EAA+B,CASxE,OARI,EAAsB,IAAI,CAAI,EACzB,GAGT,CAAI,EAAQ,cAIL,EAAmB,IAAI,CAAI,CACpC,CAEA,SAAS,EAAkB,EAAmB,EAA4B,CACxE,IAAM,EAAWT,EAAAA,QAAK,SAAS,EAAW,CAAS,EACnD,OAAO,IAAa,IAAO,IAAa,MAAQ,CAAC,EAAS,WAAW,KAAKA,EAAAA,QAAK,KAAK,GAAK,CAACA,EAAAA,QAAK,WAAW,CAAQ,CACpH,CAQA,SAAgB,EAAc,EAAsB,EAA+B,CACjF,IAAM,EAAW,EAAa,MAAM,GAAG,EACvC,IAAK,IAAM,KAAW,EAAS,MAAM,EAAG,EAAE,EACxC,GAAI,EAAsB,IAAI,CAAO,GAAM,CAAC,EAAQ,cAAgB,EAAmB,IAAI,CAAO,EAChG,MAAO,GAGX,OAAO,EAAY,EAAc,CAAO,IAAM,IAAA,EAChD,CAEA,SAAgB,EAAY,EAAc,EAAsB,EAAiB,GAAiC,CAChH,IAAM,EAAY,EAAK,YAAY,EAEjC,MAAC,IACA,EAAU,SAAS,OAAO,GACzB,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,SAAS,GAC5B,EAAU,SAAS,UAAU,MAM9B,GACA,EAAQ,cACR,GAAgB,KAAKA,EAAAA,QAAK,SAAS,CAAI,CAAC,GAAK,GAAsB,KAAKA,EAAAA,QAAK,SAAS,CAAI,CAAC,GAK9F,OAAOU,EAAAA,eAAe,CAAI,CAC5B,CAEA,SAAgB,EAAW,EAAc,EAAsB,CAC7D,OAAOV,EAAAA,QAAK,SAAS,EAAM,CAAI,GAAKA,EAAAA,QAAK,SAAS,CAAI,CACxD,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAwB,CAClD,OAAO,aAAiB,MAAQ,EAAM,QAAU,OAAO,CAAK,CAC9D"}
+{"version":3,"file":"scan.cjs","names":["os","path","stat","realpath","lstat","NativeAddonError","readdir","readFile","measureCode","measureCodeWithCrossFileData","measureCrossFileDuplication","detectLanguage"],"sources":["../src/scan.ts"],"sourcesContent":["import { lstat, readdir, readFile, realpath, stat } from 'node:fs/promises';\nimport os from 'node:os';\nimport path from 'node:path';\nimport { measureCrossFileDuplication, type CrossFileDuplicationMetrics } from './crossFileDuplication.js';\nimport type { CrossFileDuplicationFileData } from './duplication.js';\nimport { detectLanguage } from './languages.js';\nimport { measureCode, measureCodeWithCrossFileData } from './metrics.js';\nimport { NativeAddonError } from './nativeMetrics.js';\nimport type { CodeMetrics, DuplicationOptions, LanguageName, MeasureOptions } from './types.js';\n\n/** The scan settings shared by every command (a structural subset of each command's options). */\nexport interface ScanOptions {\n  duplication: Required<DuplicationOptions>;\n  includeTests: boolean;\n}\n\nexport interface FileMetrics {\n  file: string;\n  metrics: CodeMetrics;\n  /** Cross-file duplicate candidates and token/statement data, collected only for directory scans. */\n  duplicationCandidates?: CrossFileDuplicationFileData;\n}\n\nexport interface ScanResult {\n  crossFileDuplication?: CrossFileDuplicationMetrics;\n  displayRoot: string;\n  errors: string[];\n  /** Non-fatal degradations (e.g. cross-file candidates unavailable); the file is still measured. */\n  warnings: string[];\n  fatalError?: string;\n  files: FileMetrics[];\n}\n\nconst ignoredDirectoryNames = new Set([\n  '.agents',\n  '.claude',\n  '.cursor',\n  '.git',\n  '.next',\n  '.playwright-cli',\n  '.tox',\n  '.tmp',\n  '.turbo',\n  '.venv',\n  '.yarn',\n  '__fixtures__',\n  '__generated__',\n  '__pycache__',\n  'coverage',\n  'dist',\n  'fixtures',\n  'generated',\n  'node_modules',\n  // .NET SDK intermediate output (generated sources such as `*.GlobalUsings.g.cs`).\n  'obj',\n  'target',\n  'test-fixtures',\n  'vendor',\n  'venv',\n]);\n\nconst testDirectoryNames = new Set(['__tests__', 'test', 'tests', 'spec']);\nconst testFilePattern = /(?:^test(?:[_-].*)?|\\.(?:spec|test)|[_-](?:test|spec))\\.[^.]+$/iu;\n// JUnit (Java/Kotlin) and xUnit/NUnit (C#) tests use case-sensitive `Test`/`Tests` class-name\n// suffixes; case-insensitive matching would catch production files like `contest.java`.\nconst suffixTestFilePattern = /Tests?\\.(?:java|kt|cs)$/u;\n\nexport function resolveTarget(target: string): string {\n  if (target === '~') {\n    return os.homedir();\n  }\n\n  if (target.startsWith('~/')) {\n    return path.join(os.homedir(), target.slice(2));\n  }\n\n  return path.resolve(target);\n}\n\n/** Returns the directory from which the config file search should start (the target itself if it is a directory). */\nexport async function configSearchDirectory(target: string): Promise<string> {\n  try {\n    const targetStat = await stat(target);\n    return targetStat.isDirectory() ? target : path.dirname(target);\n  } catch {\n    return path.dirname(target);\n  }\n}\n\n/** Shared state of one scan, threaded through the directory walk instead of positional plumbing. */\ninterface ScanContext {\n  options: ScanOptions;\n  files: FileMetrics[];\n  errors: string[];\n  warnings: string[];\n  visitedDirectories: Set<string>;\n  visitedFiles: Set<string>;\n  /** Scan root: paths are displayed relative to it, and symbolic links may not escape it. */\n  rootDirectory: string;\n}\n\nexport async function scanTarget(target: string, options: ScanOptions): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  let canonicalTarget = target;\n  try {\n    canonicalTarget = await realpath(target);\n  } catch {\n    // stat below reports missing targets with the original path.\n  }\n\n  const fallbackDisplayRoot = path.dirname(canonicalTarget);\n  let targetStat;\n\n  try {\n    targetStat = await stat(canonicalTarget);\n  } catch (error) {\n    const fatalError = `${formatPath(canonicalTarget, fallbackDisplayRoot)}: ${formatError(error)}`;\n    return { displayRoot: fallbackDisplayRoot, files, errors: [fatalError], warnings, fatalError };\n  }\n\n  if (targetStat.isFile()) {\n    const displayRoot = path.dirname(canonicalTarget);\n    const language = getLanguage(canonicalTarget, options, true);\n    if (!language) {\n      const fatalError = `${formatPath(canonicalTarget, displayRoot)}: unsupported file type`;\n      return { displayRoot, files, errors: [fatalError], warnings, fatalError };\n    }\n\n    const context = makeScanContext(options, files, errors, warnings, displayRoot);\n    try {\n      await measureFile(canonicalTarget, language, 'single-file', context, canonicalTarget);\n    } catch (error) {\n      return toFatalResult(error, displayRoot, files, errors, warnings);\n    }\n    return { displayRoot, files, errors, warnings };\n  }\n\n  try {\n    await scanDirectory(canonicalTarget, makeScanContext(options, files, errors, warnings, canonicalTarget));\n  } catch (error) {\n    return toFatalResult(error, canonicalTarget, files, errors, warnings);\n  }\n  return { displayRoot: canonicalTarget, files, errors, warnings };\n}\n\n/**\n * Measures an explicit list of repository-relative files (the diff gate's git-visible allowlist)\n * instead of walking the directory tree, so ignored artifact directories are never parsed. Paths\n * outside the scan scope (ignored/test directories, unsupported or test file names) are skipped\n * with the same rules as the walk.\n */\nexport async function scanListedFiles(\n  rootDirectory: string,\n  relativePaths: Iterable<string>,\n  options: ScanOptions\n): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  const context = makeScanContext(options, files, errors, warnings, rootDirectory);\n  for (const relativePath of relativePaths) {\n    const language = isScannedPath(relativePath, options) ? getLanguage(relativePath, options) : undefined;\n    if (!language) {\n      continue;\n    }\n    const absolutePath = path.join(rootDirectory, relativePath);\n    // Symbolic links are not source files: git stores only their target string, so measuring\n    // through them would diverge from what any revision of the repository actually contains.\n    const stats = await lstat(absolutePath).catch(() => {});\n    if (stats?.isSymbolicLink()) {\n      continue;\n    }\n    try {\n      await measureFile(absolutePath, language, 'directory', context);\n    } catch (error) {\n      return toFatalResult(error, rootDirectory, files, errors, warnings);\n    }\n  }\n  return { displayRoot: rootDirectory, files, errors, warnings };\n}\n\n/** A run-wide failure (a missing native addon) as a fatal result; anything else keeps throwing. */\nfunction toFatalResult(\n  error: unknown,\n  displayRoot: string,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[]\n): ScanResult {\n  if (!(error instanceof NativeAddonError)) {\n    throw error;\n  }\n  const fatalError = formatError(error);\n  // Errors the walk accumulated before the fatal failure stay reported alongside it.\n  return { displayRoot, files, errors: [...errors, fatalError], warnings, fatalError };\n}\n\nfunction makeScanContext(\n  options: ScanOptions,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[],\n  rootDirectory: string\n): ScanContext {\n  return { options, files, errors, warnings, visitedDirectories: new Set(), visitedFiles: new Set(), rootDirectory };\n}\n\n/** Runs a filesystem operation, recording a scan error and returning undefined when it fails. */\nasync function tryFileSystem<T>(\n  operation: () => Promise<T>,\n  target: string,\n  context: ScanContext\n): Promise<T | undefined> {\n  try {\n    return await operation();\n  } catch (error) {\n    context.errors.push(`${formatPath(target, context.rootDirectory)}: ${formatError(error)}`);\n    return undefined;\n  }\n}\n\n/** Resolves the path (recording errors); undefined when that fails or the result escapes the root. */\nasync function resolveWithinRoot(target: string, context: ScanContext): Promise<string | undefined> {\n  const resolved = await tryFileSystem(() => realpath(target), target, context);\n  return resolved !== undefined && isWithinDirectory(resolved, context.rootDirectory) ? resolved : undefined;\n}\n\nasync function scanDirectory(directory: string, context: ScanContext): Promise<void> {\n  const resolvedDirectory = await resolveWithinRoot(directory, context);\n  if (resolvedDirectory === undefined || context.visitedDirectories.has(resolvedDirectory)) {\n    return;\n  }\n  context.visitedDirectories.add(resolvedDirectory);\n\n  const entries = await tryFileSystem(() => readdir(directory, { withFileTypes: true }), directory, context);\n  if (entries === undefined) {\n    return;\n  }\n\n  for (const entry of entries) {\n    const entryPath = path.join(directory, entry.name);\n    if (entry.isSymbolicLink()) {\n      await scanSymbolicLink(entry.name, entryPath, context);\n      continue;\n    }\n\n    if (entry.isDirectory()) {\n      if (shouldSkipDirectory(entry.name, context.options)) {\n        continue;\n      }\n      await scanDirectory(entryPath, context);\n      continue;\n    }\n\n    if (entry.isFile()) {\n      await measureScannableFile(entryPath, context);\n    }\n  }\n}\n\nasync function scanSymbolicLink(name: string, entryPath: string, context: ScanContext): Promise<void> {\n  const resolvedPath = await resolveWithinRoot(entryPath, context);\n  if (resolvedPath === undefined) {\n    return;\n  }\n\n  const entryStat = await tryFileSystem(() => stat(entryPath), entryPath, context);\n  if (entryStat === undefined) {\n    return;\n  }\n\n  if (entryStat.isDirectory()) {\n    if (\n      shouldSkipDirectory(name, context.options) ||\n      shouldSkipDirectory(path.basename(resolvedPath), context.options)\n    ) {\n      return;\n    }\n    await scanDirectory(entryPath, context);\n    return;\n  }\n\n  if (entryStat.isFile()) {\n    await measureScannableFile(entryPath, context, resolvedPath, resolvedPath);\n  }\n}\n\nasync function measureScannableFile(\n  file: string,\n  context: ScanContext,\n  languageFile = file,\n  realFile?: string\n): Promise<void> {\n  const language = getLanguage(languageFile, context.options);\n  if (language) {\n    await measureFile(file, language, 'directory', context, realFile);\n  }\n}\n\nasync function measureFile(\n  file: string,\n  language: LanguageName,\n  mode: 'single-file' | 'directory',\n  context: ScanContext,\n  realFile?: string\n): Promise<void> {\n  try {\n    const resolvedFile = realFile ?? (await realpath(file));\n    if (context.visitedFiles.has(resolvedFile)) {\n      return;\n    }\n    context.visitedFiles.add(resolvedFile);\n\n    const code = await readFile(file, 'utf8');\n    const measureOptions = { language, duplication: context.options.duplication };\n    // Only directory scans compare files against each other; a single-file target has no peers.\n    if (mode === 'single-file') {\n      context.files.push({ file, metrics: measureCode(code, measureOptions) });\n      return;\n    }\n    const { metrics, crossFileData, crossFileError } = measureWithCrossFileData(code, measureOptions);\n    if (crossFileError !== undefined) {\n      // A warning, not an error: the file's metrics are complete, only its participation in\n      // cross-file matching is lost, so it is not \"skipped\" and must not fail --fail-on-error.\n      context.warnings.push(\n        `${formatPath(file, context.rootDirectory)}: cross-file duplication candidates unavailable: ${crossFileError}`\n      );\n    }\n    context.files.push({ file, metrics, duplicationCandidates: crossFileData });\n  } catch (error) {\n    // A missing native addon fails every file identically: propagate it once as a fatal scan\n    // error instead of recording one \"skipped\" entry per file behind a successful exit code.\n    if (error instanceof NativeAddonError) {\n      throw error;\n    }\n    context.errors.push(`${formatPath(file, context.rootDirectory)}: ${formatError(error)}`);\n  }\n}\n\n/**\n * Measures a file together with its cross-file contribution from one parse. The contribution is\n * auxiliary: if collecting it fails where plain measurement succeeds (e.g. a payload too large to\n * cross the addon boundary), the metrics are still returned with the failure message, which\n * callers report as a warning rather than an error.\n */\nexport function measureWithCrossFileData(\n  code: string,\n  measureOptions: MeasureOptions\n): { metrics: CodeMetrics; crossFileData?: CrossFileDuplicationFileData; crossFileError?: string } {\n  try {\n    return measureCodeWithCrossFileData(code, measureOptions);\n  } catch (error) {\n    if (error instanceof NativeAddonError) {\n      throw error;\n    }\n    return { metrics: measureCode(code, measureOptions), crossFileError: formatError(error) };\n  }\n}\n\n/** Runs after the scan so every measured file's candidates participate. */\nexport function addCrossFileDuplication(result: ScanResult, options: ScanOptions): void {\n  if (result.fatalError || result.files.length < 2) {\n    return;\n  }\n  const sourceFiles = result.files.flatMap(({ file, duplicationCandidates }) =>\n    duplicationCandidates ? [{ file: formatPath(file, result.displayRoot), ...duplicationCandidates }] : []\n  );\n  if (sourceFiles.length < 2) {\n    return;\n  }\n  result.crossFileDuplication = measureCrossFileDuplication(sourceFiles, options.duplication);\n}\n\n/**\n * Distinct 1-based code lines covered by within-file or cross-file duplicated content. Both\n * sources expose the exact lines carrying matched tokens (never block bounding ranges, which\n * would over-count comment/blank lines and the unmatched gap of a merged clone), so the union is\n * a subset of the file's code lines and a ratio derived over code lines can never exceed 1.\n */\nexport function collectDuplicatedLineNumbers(\n  metrics: CodeMetrics | undefined,\n  crossFileDuplication: CrossFileDuplicationMetrics | undefined,\n  formattedFile: string\n): Set<number> {\n  const lines = new Set(metrics?.duplication.duplicateLineNumbers);\n  // Object.hasOwn: a file named like an Object.prototype member must not read an inherited value.\n  const crossFileLines =\n    crossFileDuplication && Object.hasOwn(crossFileDuplication.duplicateLineNumbersByFile, formattedFile)\n      ? (crossFileDuplication.duplicateLineNumbersByFile[formattedFile] ?? [])\n      : [];\n  for (const line of crossFileLines) {\n    lines.add(line);\n  }\n  return lines;\n}\n\nfunction shouldSkipDirectory(name: string, options: ScanOptions): boolean {\n  if (ignoredDirectoryNames.has(name)) {\n    return true;\n  }\n\n  if (options.includeTests) {\n    return false;\n  }\n\n  return testDirectoryNames.has(name);\n}\n\nfunction isWithinDirectory(candidate: string, directory: string): boolean {\n  const relative = path.relative(directory, candidate);\n  return relative === '' || (relative !== '..' && !relative.startsWith(`..${path.sep}`) && !path.isAbsolute(relative));\n}\n\n/**\n * Whether a repository-relative path would be scanned: no ignored or excluded-test directory\n * segment and a supported, non-test file name. The diff gate uses this for base-revision\n * eligibility, so code renamed into scan scope gates as new code instead of ratcheting against\n * a blob the scanner would never have measured.\n */\nexport function isScannedPath(relativePath: string, options: ScanOptions): boolean {\n  const segments = relativePath.split('/');\n  for (const segment of segments.slice(0, -1)) {\n    if (ignoredDirectoryNames.has(segment) || (!options.includeTests && testDirectoryNames.has(segment))) {\n      return false;\n    }\n  }\n  return getLanguage(relativePath, options) !== undefined;\n}\n\nexport function getLanguage(file: string, options: ScanOptions, explicitTarget = false): LanguageName | undefined {\n  const lowerFile = file.toLowerCase();\n  if (\n    !explicitTarget &&\n    (lowerFile.endsWith('.d.ts') ||\n      lowerFile.endsWith('.d.mts') ||\n      lowerFile.endsWith('.d.cts') ||\n      lowerFile.endsWith('.min.js') ||\n      lowerFile.endsWith('.pnp.cjs'))\n  ) {\n    return undefined;\n  }\n\n  if (\n    !explicitTarget &&\n    !options.includeTests &&\n    (testFilePattern.test(path.basename(file)) || suffixTestFilePattern.test(path.basename(file)))\n  ) {\n    return undefined;\n  }\n\n  return detectLanguage(file);\n}\n\nexport function formatPath(file: string, base: string): string {\n  return path.relative(base, file) || path.basename(file);\n}\n\nexport function writeStdout(message: string): void {\n  process.stdout.write(message);\n}\n\nexport function writeStderr(message: string): void {\n  process.stderr.write(message);\n}\n\nexport function formatError(error: unknown): string {\n  return error instanceof Error ? error.message : String(error);\n}\n"],"mappings":"sTAiCA,MAAM,EAAwB,IAAI,IAAI,CACpC,UACA,UACA,UACA,OACA,QACA,kBACA,OACA,OACA,SACA,QACA,QACA,eACA,gBACA,cACA,WACA,OACA,WACA,YACA,eAEA,MACA,SACA,gBACA,SACA,MACF,CAAC,EAEK,EAAqB,IAAI,IAAI,CAAC,YAAa,OAAQ,QAAS,MAAM,CAAC,EACnE,EAAkB,mEAGlB,EAAwB,2BAE9B,SAAgB,EAAc,EAAwB,CASpD,OARI,IAAW,IACNA,EAAAA,QAAG,QAAQ,EAGhB,EAAO,WAAW,IAAI,EACjBC,EAAAA,QAAK,KAAKD,EAAAA,QAAG,QAAQ,EAAG,EAAO,MAAM,CAAC,CAAC,EAGzCC,EAAAA,QAAK,QAAQ,CAAM,CAC5B,CAGA,eAAsB,EAAsB,EAAiC,CAC3E,GAAI,CAEF,OAAO,MAAA,EADkBC,EAAAA,KAAAA,CAAK,CAAM,EAAA,CAClB,YAAY,EAAI,EAASD,EAAAA,QAAK,QAAQ,CAAM,CAChE,MAAQ,CACN,OAAOA,EAAAA,QAAK,QAAQ,CAAM,CAC5B,CACF,CAcA,eAAsB,EAAW,EAAgB,EAA2C,CAC1F,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACxB,EAAkB,EACtB,GAAI,CACF,EAAkB,MAAA,EAAME,EAAAA,SAAAA,CAAS,CAAM,CACzC,MAAQ,CAER,CAEA,IAAM,EAAsBF,EAAAA,QAAK,QAAQ,CAAe,EACpD,EAEJ,GAAI,CACF,EAAa,MAAA,EAAMC,EAAAA,KAAAA,CAAK,CAAe,CACzC,OAAS,EAAO,CACd,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAmB,EAAE,IAAI,EAAY,CAAK,IAC5F,MAAO,CAAE,YAAa,EAAqB,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC/F,CAEA,GAAI,EAAW,OAAO,EAAG,CACvB,IAAM,EAAcD,EAAAA,QAAK,QAAQ,CAAe,EAC1C,EAAW,EAAY,EAAiB,EAAS,EAAI,EAC3D,GAAI,CAAC,EAAU,CACb,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAW,EAAE,yBAC/D,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC1E,CAEA,IAAM,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAW,EAC7E,GAAI,CACF,MAAM,EAAY,EAAiB,EAAU,cAAe,EAAS,CAAe,CACtF,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAa,EAAO,EAAQ,CAAQ,CAClE,CACA,MAAO,CAAE,cAAa,QAAO,SAAQ,UAAS,CAChD,CAEA,GAAI,CACF,MAAM,EAAc,EAAiB,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAe,CAAC,CACzG,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAiB,EAAO,EAAQ,CAAQ,CACtE,CACA,MAAO,CAAE,YAAa,EAAiB,QAAO,SAAQ,UAAS,CACjE,CAQA,eAAsB,EACpB,EACA,EACA,EACqB,CACrB,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACtB,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAa,EAC/E,IAAK,IAAM,KAAgB,EAAe,CACxC,IAAM,EAAW,EAAc,EAAc,CAAO,EAAI,EAAY,EAAc,CAAO,EAAI,IAAA,GAC7F,GAAI,CAAC,EACH,SAEF,IAAM,EAAeA,EAAAA,QAAK,KAAK,EAAe,CAAY,EAI1D,KAAI,MAAA,EADgBG,EAAAA,MAAAA,CAAM,CAAY,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EAC3C,eAAe,EAG1B,GAAI,CACF,MAAM,EAAY,EAAc,EAAU,YAAa,CAAO,CAChE,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAe,EAAO,EAAQ,CAAQ,CACpE,CACF,CACA,MAAO,CAAE,YAAa,EAAe,QAAO,SAAQ,UAAS,CAC/D,CAGA,SAAS,EACP,EACA,EACA,EACA,EACA,EACY,CACZ,GAAI,EAAE,aAAiBC,EAAAA,kBACrB,MAAM,EAER,IAAM,EAAa,EAAY,CAAK,EAEpC,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,GAAG,EAAQ,CAAU,EAAG,WAAU,YAAW,CACrF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACa,CACb,MAAO,CAAE,UAAS,QAAO,SAAQ,WAAU,mBAAoB,IAAI,IAAO,aAAc,IAAI,IAAO,eAAc,CACnH,CAGA,eAAe,EACb,EACA,EACA,EACwB,CACxB,GAAI,CACF,OAAO,MAAM,EAAU,CACzB,OAAS,EAAO,CACd,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAQ,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,EACzF,MACF,CACF,CAGA,eAAe,EAAkB,EAAgB,EAAmD,CAClG,IAAM,EAAW,MAAM,OAAA,EAAoBF,EAAAA,SAAAA,CAAS,CAAM,EAAG,EAAQ,CAAO,EAC5E,OAAO,IAAa,IAAA,IAAa,EAAkB,EAAU,EAAQ,aAAa,EAAI,EAAW,IAAA,EACnG,CAEA,eAAe,EAAc,EAAmB,EAAqC,CACnF,IAAM,EAAoB,MAAM,EAAkB,EAAW,CAAO,EACpE,GAAI,IAAsB,IAAA,IAAa,EAAQ,mBAAmB,IAAI,CAAiB,EACrF,OAEF,EAAQ,mBAAmB,IAAI,CAAiB,EAEhD,IAAM,EAAU,MAAM,OAAA,EAAoBG,EAAAA,QAAAA,CAAQ,EAAW,CAAE,cAAe,EAAK,CAAC,EAAG,EAAW,CAAO,EACrG,OAAY,IAAA,GAIhB,IAAK,IAAM,KAAS,EAAS,CAC3B,IAAM,EAAYL,EAAAA,QAAK,KAAK,EAAW,EAAM,IAAI,EACjD,GAAI,EAAM,eAAe,EAAG,CAC1B,MAAM,EAAiB,EAAM,KAAM,EAAW,CAAO,EACrD,QACF,CAEA,GAAI,EAAM,YAAY,EAAG,CACvB,GAAI,EAAoB,EAAM,KAAM,EAAQ,OAAO,EACjD,SAEF,MAAM,EAAc,EAAW,CAAO,EACtC,QACF,CAEI,EAAM,OAAO,GACf,MAAM,EAAqB,EAAW,CAAO,CAEjD,CACF,CAEA,eAAe,EAAiB,EAAc,EAAmB,EAAqC,CACpG,IAAM,EAAe,MAAM,EAAkB,EAAW,CAAO,EAC/D,GAAI,IAAiB,IAAA,GACnB,OAGF,IAAM,EAAY,MAAM,OAAA,EAAoBC,EAAAA,KAAAA,CAAK,CAAS,EAAG,EAAW,CAAO,EAC3E,OAAc,IAAA,GAIlB,IAAI,EAAU,YAAY,EAAG,CAC3B,GACE,EAAoB,EAAM,EAAQ,OAAO,GACzC,EAAoBD,EAAAA,QAAK,SAAS,CAAY,EAAG,EAAQ,OAAO,EAEhE,OAEF,MAAM,EAAc,EAAW,CAAO,EACtC,MACF,CAEI,EAAU,OAAO,GACnB,MAAM,EAAqB,EAAW,EAAS,EAAc,CAAY,CAH3E,CAKF,CAEA,eAAe,EACb,EACA,EACA,EAAe,EACf,EACe,CACf,IAAM,EAAW,EAAY,EAAc,EAAQ,OAAO,EACtD,GACF,MAAM,EAAY,EAAM,EAAU,YAAa,EAAS,CAAQ,CAEpE,CAEA,eAAe,EACb,EACA,EACA,EACA,EACA,EACe,CACf,GAAI,CACF,IAAM,EAAe,GAAa,MAAA,EAAME,EAAAA,SAAAA,CAAS,CAAI,EACrD,GAAI,EAAQ,aAAa,IAAI,CAAY,EACvC,OAEF,EAAQ,aAAa,IAAI,CAAY,EAErC,IAAM,EAAO,MAAA,EAAMI,EAAAA,SAAAA,CAAS,EAAM,MAAM,EAClC,EAAiB,CAAE,WAAU,YAAa,EAAQ,QAAQ,WAAY,EAE5E,GAAI,IAAS,cAAe,CAC1B,EAAQ,MAAM,KAAK,CAAE,OAAM,QAASC,EAAAA,YAAY,EAAM,CAAc,CAAE,CAAC,EACvE,MACF,CACA,GAAM,CAAE,UAAS,gBAAe,kBAAmB,EAAyB,EAAM,CAAc,EAC5F,IAAmB,IAAA,IAGrB,EAAQ,SAAS,KACf,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,mDAAmD,GAChG,EAEF,EAAQ,MAAM,KAAK,CAAE,OAAM,UAAS,sBAAuB,CAAc,CAAC,CAC5E,OAAS,EAAO,CAGd,GAAI,aAAiBH,EAAAA,iBACnB,MAAM,EAER,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,CACzF,CACF,CAQA,SAAgB,EACd,EACA,EACiG,CACjG,GAAI,CACF,OAAOI,EAAAA,6BAA6B,EAAM,CAAc,CAC1D,OAAS,EAAO,CACd,GAAI,aAAiBJ,EAAAA,iBACnB,MAAM,EAER,MAAO,CAAE,QAASG,EAAAA,YAAY,EAAM,CAAc,EAAG,eAAgB,EAAY,CAAK,CAAE,CAC1F,CACF,CAGA,SAAgB,EAAwB,EAAoB,EAA4B,CACtF,GAAI,EAAO,YAAc,EAAO,MAAM,OAAS,EAC7C,OAEF,IAAM,EAAc,EAAO,MAAM,SAAS,CAAE,OAAM,2BAChD,EAAwB,CAAC,CAAE,KAAM,EAAW,EAAM,EAAO,WAAW,EAAG,GAAG,CAAsB,CAAC,EAAI,CAAC,CACxG,EACI,EAAY,OAAS,IAGzB,EAAO,qBAAuBE,EAAAA,4BAA4B,EAAa,EAAQ,WAAW,EAC5F,CAQA,SAAgB,EACd,EACA,EACA,EACa,CACb,IAAM,EAAQ,IAAI,IAAI,GAAS,YAAY,oBAAoB,EAEzD,EACJ,GAAwB,OAAO,OAAO,EAAqB,2BAA4B,CAAa,EAC/F,EAAqB,2BAA2B,IAAkB,CAAC,EACpE,CAAC,EACP,IAAK,IAAM,KAAQ,EACjB,EAAM,IAAI,CAAI,EAEhB,OAAO,CACT,CAEA,SAAS,EAAoB,EAAc,EAA+B,CASxE,OARI,EAAsB,IAAI,CAAI,EACzB,GAGT,CAAI,EAAQ,cAIL,EAAmB,IAAI,CAAI,CACpC,CAEA,SAAS,EAAkB,EAAmB,EAA4B,CACxE,IAAM,EAAWT,EAAAA,QAAK,SAAS,EAAW,CAAS,EACnD,OAAO,IAAa,IAAO,IAAa,MAAQ,CAAC,EAAS,WAAW,KAAKA,EAAAA,QAAK,KAAK,GAAK,CAACA,EAAAA,QAAK,WAAW,CAAQ,CACpH,CAQA,SAAgB,EAAc,EAAsB,EAA+B,CACjF,IAAM,EAAW,EAAa,MAAM,GAAG,EACvC,IAAK,IAAM,KAAW,EAAS,MAAM,EAAG,EAAE,EACxC,GAAI,EAAsB,IAAI,CAAO,GAAM,CAAC,EAAQ,cAAgB,EAAmB,IAAI,CAAO,EAChG,MAAO,GAGX,OAAO,EAAY,EAAc,CAAO,IAAM,IAAA,EAChD,CAEA,SAAgB,EAAY,EAAc,EAAsB,EAAiB,GAAiC,CAChH,IAAM,EAAY,EAAK,YAAY,EAEjC,MAAC,IACA,EAAU,SAAS,OAAO,GACzB,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,SAAS,GAC5B,EAAU,SAAS,UAAU,MAM9B,GACA,EAAQ,cACR,GAAgB,KAAKA,EAAAA,QAAK,SAAS,CAAI,CAAC,GAAK,GAAsB,KAAKA,EAAAA,QAAK,SAAS,CAAI,CAAC,GAK9F,OAAOU,EAAAA,eAAe,CAAI,CAC5B,CAEA,SAAgB,EAAW,EAAc,EAAsB,CAC7D,OAAOV,EAAAA,QAAK,SAAS,EAAM,CAAI,GAAKA,EAAAA,QAAK,SAAS,CAAI,CACxD,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAwB,CAClD,OAAO,aAAiB,MAAQ,EAAM,QAAU,OAAO,CAAK,CAC9D"}

package/dist/scan.d.ts

@@ -1,6 +1,6 @@
 import { type CrossFileDuplicationMetrics } from './crossFileDuplication.js';
 import type { CrossFileDuplicationFileData } from './duplication.js';
-import type { CodeMetrics, DuplicationOptions, LanguageName } from './types.js';
+import type { CodeMetrics, DuplicationOptions, LanguageName, MeasureOptions } from './types.js';
 /** The scan settings shared by every command (a structural subset of each command's options). */
 export interface ScanOptions {
     duplication: Required<DuplicationOptions>;
@@ -32,6 +32,17 @@ export declare function scanTarget(target: string, options: ScanOptions): Promis
  * with the same rules as the walk.
  */
 export declare function scanListedFiles(rootDirectory: string, relativePaths: Iterable<string>, options: ScanOptions): Promise<ScanResult>;
+/**
+ * Measures a file together with its cross-file contribution from one parse. The contribution is
+ * auxiliary: if collecting it fails where plain measurement succeeds (e.g. a payload too large to
+ * cross the addon boundary), the metrics are still returned with the failure message, which
+ * callers report as a warning rather than an error.
+ */
+export declare function measureWithCrossFileData(code: string, measureOptions: MeasureOptions): {
+    metrics: CodeMetrics;
+    crossFileData?: CrossFileDuplicationFileData;
+    crossFileError?: string;
+};
 /** Runs after the scan so every measured file's candidates participate. */
 export declare function addCrossFileDuplication(result: ScanResult, options: ScanOptions): void;
 /**

package/dist/scan.js

@@ -1,2 +1,2 @@
-import{measureCrossFileDuplication as e}from"./crossFileDuplication.js";import{detectLanguage as t}from"./languages.js";import{NativeAddonError as n}from"./nativeMetrics.js";import{collectCrossFileDuplicationFileData as r,measureCode as i}from"./metrics.js";import a from"node:path";import{lstat as o,readFile as s,readdir as c,realpath as l,stat as u}from"node:fs/promises";import d from"node:os";const f=new Set([`.agents`,`.claude`,`.cursor`,`.git`,`.next`,`.playwright-cli`,`.tox`,`.tmp`,`.turbo`,`.venv`,`.yarn`,`__fixtures__`,`__generated__`,`__pycache__`,`coverage`,`dist`,`fixtures`,`generated`,`node_modules`,`obj`,`target`,`test-fixtures`,`vendor`,`venv`]),p=new Set([`__tests__`,`test`,`tests`,`spec`]),m=/(?:^test(?:[_-].*)?|\.(?:spec|test)|[_-](?:test|spec))\.[^.]+$/iu,h=/Tests?\.(?:java|kt|cs)$/u;function g(e){return e===`~`?d.homedir():e.startsWith(`~/`)?a.join(d.homedir(),e.slice(2)):a.resolve(e)}async function _(e){try{return(await u(e)).isDirectory()?e:a.dirname(e)}catch{return a.dirname(e)}}async function v(e,t){let n=[],r=[],i=[],o=e;try{o=await l(e)}catch{}let s=a.dirname(o),c;try{c=await u(o)}catch(e){let t=`${P(o,s)}: ${L(e)}`;return{displayRoot:s,files:n,errors:[t],warnings:i,fatalError:t}}if(c.isFile()){let e=a.dirname(o),s=N(o,t,!0);if(!s){let t=`${P(o,e)}: unsupported file type`;return{displayRoot:e,files:n,errors:[t],warnings:i,fatalError:t}}let c=x(t,n,r,i,e);try{await D(o,s,`single-file`,c,o)}catch(t){return b(t,e,n,r,i)}return{displayRoot:e,files:n,errors:r,warnings:i}}try{await w(o,x(t,n,r,i,o))}catch(e){return b(e,o,n,r,i)}return{displayRoot:o,files:n,errors:r,warnings:i}}async function y(e,t,n){let r=[],i=[],s=[],c=x(n,r,i,s,e);for(let l of t){let t=M(l,n)?N(l,n):void 0;if(!t)continue;let u=a.join(e,l);if(!(await o(u).catch(()=>{}))?.isSymbolicLink())try{await D(u,t,`directory`,c)}catch(t){return b(t,e,r,i,s)}}return{displayRoot:e,files:r,errors:i,warnings:s}}function b(e,t,r,i,a){if(!(e instanceof n))throw e;let o=L(e);return{displayRoot:t,files:r,errors:[...i,o],warnings:a,fatalError:o}}function x(e,t,n,r,i){return{options:e,files:t,errors:n,warnings:r,visitedDirectories:new Set,visitedFiles:new Set,rootDirectory:i}}async function S(e,t,n){try{return await e()}catch(e){n.errors.push(`${P(t,n.rootDirectory)}: ${L(e)}`);return}}async function C(e,t){let n=await S(()=>l(e),e,t);return n!==void 0&&j(n,t.rootDirectory)?n:void 0}async function w(e,t){let n=await C(e,t);if(n===void 0||t.visitedDirectories.has(n))return;t.visitedDirectories.add(n);let r=await S(()=>c(e,{withFileTypes:!0}),e,t);if(r!==void 0)for(let n of r){let r=a.join(e,n.name);if(n.isSymbolicLink()){await T(n.name,r,t);continue}if(n.isDirectory()){if(A(n.name,t.options))continue;await w(r,t);continue}n.isFile()&&await E(r,t)}}async function T(e,t,n){let r=await C(t,n);if(r===void 0)return;let i=await S(()=>u(t),t,n);if(i!==void 0){if(i.isDirectory()){if(A(e,n.options)||A(a.basename(r),n.options))return;await w(t,n);return}i.isFile()&&await E(t,n,r,r)}}async function E(e,t,n=e,r){let i=N(n,t.options);i&&await D(e,i,`directory`,t,r)}async function D(e,t,a,o,c){try{let n=c??await l(e);if(o.visitedFiles.has(n))return;o.visitedFiles.add(n);let u=await s(e,`utf8`),d={language:t,duplication:o.options.duplication},f={file:e,metrics:i(u,d)};if(a===`directory`)try{f.duplicationCandidates=r(u,d)}catch(t){o.warnings.push(`${P(e,o.rootDirectory)}: cross-file duplication candidates unavailable: ${L(t)}`)}o.files.push(f)}catch(t){if(t instanceof n)throw t;o.errors.push(`${P(e,o.rootDirectory)}: ${L(t)}`)}}function O(t,n){if(t.fatalError||t.files.length<2)return;let r=t.files.flatMap(({file:e,duplicationCandidates:n})=>n?[{file:P(e,t.displayRoot),...n}]:[]);r.length<2||(t.crossFileDuplication=e(r,n.duplication))}function k(e,t,n){let r=new Set(e?.duplication.duplicateLineNumbers),i=t&&Object.hasOwn(t.duplicateLineNumbersByFile,n)?t.duplicateLineNumbersByFile[n]??[]:[];for(let e of i)r.add(e);return r}function A(e,t){return f.has(e)?!0:!t.includeTests&&p.has(e)}function j(e,t){let n=a.relative(t,e);return n===``||n!==`..`&&!n.startsWith(`..${a.sep}`)&&!a.isAbsolute(n)}function M(e,t){let n=e.split(`/`);for(let e of n.slice(0,-1))if(f.has(e)||!t.includeTests&&p.has(e))return!1;return N(e,t)!==void 0}function N(e,n,r=!1){let i=e.toLowerCase();if(!(!r&&(i.endsWith(`.d.ts`)||i.endsWith(`.d.mts`)||i.endsWith(`.d.cts`)||i.endsWith(`.min.js`)||i.endsWith(`.pnp.cjs`)))&&(r||n.includeTests||!m.test(a.basename(e))&&!h.test(a.basename(e))))return t(e)}function P(e,t){return a.relative(t,e)||a.basename(e)}function F(e){process.stdout.write(e)}function I(e){process.stderr.write(e)}function L(e){return e instanceof Error?e.message:String(e)}export{O as addCrossFileDuplication,k as collectDuplicatedLineNumbers,_ as configSearchDirectory,L as formatError,P as formatPath,N as getLanguage,M as isScannedPath,g as resolveTarget,y as scanListedFiles,v as scanTarget,I as writeStderr,F as writeStdout};
+import{measureCrossFileDuplication as e}from"./crossFileDuplication.js";import{detectLanguage as t}from"./languages.js";import{NativeAddonError as n}from"./nativeMetrics.js";import{measureCode as r,measureCodeWithCrossFileData as i}from"./metrics.js";import a from"node:path";import{lstat as o,readFile as s,readdir as c,realpath as l,stat as u}from"node:fs/promises";import d from"node:os";const f=new Set([`.agents`,`.claude`,`.cursor`,`.git`,`.next`,`.playwright-cli`,`.tox`,`.tmp`,`.turbo`,`.venv`,`.yarn`,`__fixtures__`,`__generated__`,`__pycache__`,`coverage`,`dist`,`fixtures`,`generated`,`node_modules`,`obj`,`target`,`test-fixtures`,`vendor`,`venv`]),p=new Set([`__tests__`,`test`,`tests`,`spec`]),m=/(?:^test(?:[_-].*)?|\.(?:spec|test)|[_-](?:test|spec))\.[^.]+$/iu,h=/Tests?\.(?:java|kt|cs)$/u;function g(e){return e===`~`?d.homedir():e.startsWith(`~/`)?a.join(d.homedir(),e.slice(2)):a.resolve(e)}async function _(e){try{return(await u(e)).isDirectory()?e:a.dirname(e)}catch{return a.dirname(e)}}async function v(e,t){let n=[],r=[],i=[],o=e;try{o=await l(e)}catch{}let s=a.dirname(o),c;try{c=await u(o)}catch(e){let t=`${F(o,s)}: ${R(e)}`;return{displayRoot:s,files:n,errors:[t],warnings:i,fatalError:t}}if(c.isFile()){let e=a.dirname(o),s=P(o,t,!0);if(!s){let t=`${F(o,e)}: unsupported file type`;return{displayRoot:e,files:n,errors:[t],warnings:i,fatalError:t}}let c=x(t,n,r,i,e);try{await D(o,s,`single-file`,c,o)}catch(t){return b(t,e,n,r,i)}return{displayRoot:e,files:n,errors:r,warnings:i}}try{await w(o,x(t,n,r,i,o))}catch(e){return b(e,o,n,r,i)}return{displayRoot:o,files:n,errors:r,warnings:i}}async function y(e,t,n){let r=[],i=[],s=[],c=x(n,r,i,s,e);for(let l of t){let t=N(l,n)?P(l,n):void 0;if(!t)continue;let u=a.join(e,l);if(!(await o(u).catch(()=>{}))?.isSymbolicLink())try{await D(u,t,`directory`,c)}catch(t){return b(t,e,r,i,s)}}return{displayRoot:e,files:r,errors:i,warnings:s}}function b(e,t,r,i,a){if(!(e instanceof n))throw e;let o=R(e);return{displayRoot:t,files:r,errors:[...i,o],warnings:a,fatalError:o}}function x(e,t,n,r,i){return{options:e,files:t,errors:n,warnings:r,visitedDirectories:new Set,visitedFiles:new Set,rootDirectory:i}}async function S(e,t,n){try{return await e()}catch(e){n.errors.push(`${F(t,n.rootDirectory)}: ${R(e)}`);return}}async function C(e,t){let n=await S(()=>l(e),e,t);return n!==void 0&&M(n,t.rootDirectory)?n:void 0}async function w(e,t){let n=await C(e,t);if(n===void 0||t.visitedDirectories.has(n))return;t.visitedDirectories.add(n);let r=await S(()=>c(e,{withFileTypes:!0}),e,t);if(r!==void 0)for(let n of r){let r=a.join(e,n.name);if(n.isSymbolicLink()){await T(n.name,r,t);continue}if(n.isDirectory()){if(j(n.name,t.options))continue;await w(r,t);continue}n.isFile()&&await E(r,t)}}async function T(e,t,n){let r=await C(t,n);if(r===void 0)return;let i=await S(()=>u(t),t,n);if(i!==void 0){if(i.isDirectory()){if(j(e,n.options)||j(a.basename(r),n.options))return;await w(t,n);return}i.isFile()&&await E(t,n,r,r)}}async function E(e,t,n=e,r){let i=P(n,t.options);i&&await D(e,i,`directory`,t,r)}async function D(e,t,i,a,o){try{let n=o??await l(e);if(a.visitedFiles.has(n))return;a.visitedFiles.add(n);let c=await s(e,`utf8`),u={language:t,duplication:a.options.duplication};if(i===`single-file`){a.files.push({file:e,metrics:r(c,u)});return}let{metrics:d,crossFileData:f,crossFileError:p}=O(c,u);p!==void 0&&a.warnings.push(`${F(e,a.rootDirectory)}: cross-file duplication candidates unavailable: ${p}`),a.files.push({file:e,metrics:d,duplicationCandidates:f})}catch(t){if(t instanceof n)throw t;a.errors.push(`${F(e,a.rootDirectory)}: ${R(t)}`)}}function O(e,t){try{return i(e,t)}catch(i){if(i instanceof n)throw i;return{metrics:r(e,t),crossFileError:R(i)}}}function k(t,n){if(t.fatalError||t.files.length<2)return;let r=t.files.flatMap(({file:e,duplicationCandidates:n})=>n?[{file:F(e,t.displayRoot),...n}]:[]);r.length<2||(t.crossFileDuplication=e(r,n.duplication))}function A(e,t,n){let r=new Set(e?.duplication.duplicateLineNumbers),i=t&&Object.hasOwn(t.duplicateLineNumbersByFile,n)?t.duplicateLineNumbersByFile[n]??[]:[];for(let e of i)r.add(e);return r}function j(e,t){return f.has(e)?!0:!t.includeTests&&p.has(e)}function M(e,t){let n=a.relative(t,e);return n===``||n!==`..`&&!n.startsWith(`..${a.sep}`)&&!a.isAbsolute(n)}function N(e,t){let n=e.split(`/`);for(let e of n.slice(0,-1))if(f.has(e)||!t.includeTests&&p.has(e))return!1;return P(e,t)!==void 0}function P(e,n,r=!1){let i=e.toLowerCase();if(!(!r&&(i.endsWith(`.d.ts`)||i.endsWith(`.d.mts`)||i.endsWith(`.d.cts`)||i.endsWith(`.min.js`)||i.endsWith(`.pnp.cjs`)))&&(r||n.includeTests||!m.test(a.basename(e))&&!h.test(a.basename(e))))return t(e)}function F(e,t){return a.relative(t,e)||a.basename(e)}function I(e){process.stdout.write(e)}function L(e){process.stderr.write(e)}function R(e){return e instanceof Error?e.message:String(e)}export{k as addCrossFileDuplication,A as collectDuplicatedLineNumbers,_ as configSearchDirectory,R as formatError,F as formatPath,P as getLanguage,N as isScannedPath,O as measureWithCrossFileData,g as resolveTarget,y as scanListedFiles,v as scanTarget,L as writeStderr,I as writeStdout};
 //# sourceMappingURL=scan.js.map

package/dist/scan.js.map

@@ -1 +1 @@
-{"version":3,"file":"scan.js","names":[],"sources":["../src/scan.ts"],"sourcesContent":["import { lstat, readdir, readFile, realpath, stat } from 'node:fs/promises';\nimport os from 'node:os';\nimport path from 'node:path';\nimport { measureCrossFileDuplication, type CrossFileDuplicationMetrics } from './crossFileDuplication.js';\nimport type { CrossFileDuplicationFileData } from './duplication.js';\nimport { detectLanguage } from './languages.js';\nimport { collectCrossFileDuplicationFileData, measureCode } from './metrics.js';\nimport { NativeAddonError } from './nativeMetrics.js';\nimport type { CodeMetrics, DuplicationOptions, LanguageName } from './types.js';\n\n/** The scan settings shared by every command (a structural subset of each command's options). */\nexport interface ScanOptions {\n  duplication: Required<DuplicationOptions>;\n  includeTests: boolean;\n}\n\nexport interface FileMetrics {\n  file: string;\n  metrics: CodeMetrics;\n  /** Cross-file duplicate candidates and token/statement data, collected only for directory scans. */\n  duplicationCandidates?: CrossFileDuplicationFileData;\n}\n\nexport interface ScanResult {\n  crossFileDuplication?: CrossFileDuplicationMetrics;\n  displayRoot: string;\n  errors: string[];\n  /** Non-fatal degradations (e.g. cross-file candidates unavailable); the file is still measured. */\n  warnings: string[];\n  fatalError?: string;\n  files: FileMetrics[];\n}\n\nconst ignoredDirectoryNames = new Set([\n  '.agents',\n  '.claude',\n  '.cursor',\n  '.git',\n  '.next',\n  '.playwright-cli',\n  '.tox',\n  '.tmp',\n  '.turbo',\n  '.venv',\n  '.yarn',\n  '__fixtures__',\n  '__generated__',\n  '__pycache__',\n  'coverage',\n  'dist',\n  'fixtures',\n  'generated',\n  'node_modules',\n  // .NET SDK intermediate output (generated sources such as `*.GlobalUsings.g.cs`).\n  'obj',\n  'target',\n  'test-fixtures',\n  'vendor',\n  'venv',\n]);\n\nconst testDirectoryNames = new Set(['__tests__', 'test', 'tests', 'spec']);\nconst testFilePattern = /(?:^test(?:[_-].*)?|\\.(?:spec|test)|[_-](?:test|spec))\\.[^.]+$/iu;\n// JUnit (Java/Kotlin) and xUnit/NUnit (C#) tests use case-sensitive `Test`/`Tests` class-name\n// suffixes; case-insensitive matching would catch production files like `contest.java`.\nconst suffixTestFilePattern = /Tests?\\.(?:java|kt|cs)$/u;\n\nexport function resolveTarget(target: string): string {\n  if (target === '~') {\n    return os.homedir();\n  }\n\n  if (target.startsWith('~/')) {\n    return path.join(os.homedir(), target.slice(2));\n  }\n\n  return path.resolve(target);\n}\n\n/** Returns the directory from which the config file search should start (the target itself if it is a directory). */\nexport async function configSearchDirectory(target: string): Promise<string> {\n  try {\n    const targetStat = await stat(target);\n    return targetStat.isDirectory() ? target : path.dirname(target);\n  } catch {\n    return path.dirname(target);\n  }\n}\n\n/** Shared state of one scan, threaded through the directory walk instead of positional plumbing. */\ninterface ScanContext {\n  options: ScanOptions;\n  files: FileMetrics[];\n  errors: string[];\n  warnings: string[];\n  visitedDirectories: Set<string>;\n  visitedFiles: Set<string>;\n  /** Scan root: paths are displayed relative to it, and symbolic links may not escape it. */\n  rootDirectory: string;\n}\n\nexport async function scanTarget(target: string, options: ScanOptions): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  let canonicalTarget = target;\n  try {\n    canonicalTarget = await realpath(target);\n  } catch {\n    // stat below reports missing targets with the original path.\n  }\n\n  const fallbackDisplayRoot = path.dirname(canonicalTarget);\n  let targetStat;\n\n  try {\n    targetStat = await stat(canonicalTarget);\n  } catch (error) {\n    const fatalError = `${formatPath(canonicalTarget, fallbackDisplayRoot)}: ${formatError(error)}`;\n    return { displayRoot: fallbackDisplayRoot, files, errors: [fatalError], warnings, fatalError };\n  }\n\n  if (targetStat.isFile()) {\n    const displayRoot = path.dirname(canonicalTarget);\n    const language = getLanguage(canonicalTarget, options, true);\n    if (!language) {\n      const fatalError = `${formatPath(canonicalTarget, displayRoot)}: unsupported file type`;\n      return { displayRoot, files, errors: [fatalError], warnings, fatalError };\n    }\n\n    const context = makeScanContext(options, files, errors, warnings, displayRoot);\n    try {\n      await measureFile(canonicalTarget, language, 'single-file', context, canonicalTarget);\n    } catch (error) {\n      return toFatalResult(error, displayRoot, files, errors, warnings);\n    }\n    return { displayRoot, files, errors, warnings };\n  }\n\n  try {\n    await scanDirectory(canonicalTarget, makeScanContext(options, files, errors, warnings, canonicalTarget));\n  } catch (error) {\n    return toFatalResult(error, canonicalTarget, files, errors, warnings);\n  }\n  return { displayRoot: canonicalTarget, files, errors, warnings };\n}\n\n/**\n * Measures an explicit list of repository-relative files (the diff gate's git-visible allowlist)\n * instead of walking the directory tree, so ignored artifact directories are never parsed. Paths\n * outside the scan scope (ignored/test directories, unsupported or test file names) are skipped\n * with the same rules as the walk.\n */\nexport async function scanListedFiles(\n  rootDirectory: string,\n  relativePaths: Iterable<string>,\n  options: ScanOptions\n): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  const context = makeScanContext(options, files, errors, warnings, rootDirectory);\n  for (const relativePath of relativePaths) {\n    const language = isScannedPath(relativePath, options) ? getLanguage(relativePath, options) : undefined;\n    if (!language) {\n      continue;\n    }\n    const absolutePath = path.join(rootDirectory, relativePath);\n    // Symbolic links are not source files: git stores only their target string, so measuring\n    // through them would diverge from what any revision of the repository actually contains.\n    const stats = await lstat(absolutePath).catch(() => {});\n    if (stats?.isSymbolicLink()) {\n      continue;\n    }\n    try {\n      await measureFile(absolutePath, language, 'directory', context);\n    } catch (error) {\n      return toFatalResult(error, rootDirectory, files, errors, warnings);\n    }\n  }\n  return { displayRoot: rootDirectory, files, errors, warnings };\n}\n\n/** A run-wide failure (a missing native addon) as a fatal result; anything else keeps throwing. */\nfunction toFatalResult(\n  error: unknown,\n  displayRoot: string,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[]\n): ScanResult {\n  if (!(error instanceof NativeAddonError)) {\n    throw error;\n  }\n  const fatalError = formatError(error);\n  // Errors the walk accumulated before the fatal failure stay reported alongside it.\n  return { displayRoot, files, errors: [...errors, fatalError], warnings, fatalError };\n}\n\nfunction makeScanContext(\n  options: ScanOptions,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[],\n  rootDirectory: string\n): ScanContext {\n  return { options, files, errors, warnings, visitedDirectories: new Set(), visitedFiles: new Set(), rootDirectory };\n}\n\n/** Runs a filesystem operation, recording a scan error and returning undefined when it fails. */\nasync function tryFileSystem<T>(\n  operation: () => Promise<T>,\n  target: string,\n  context: ScanContext\n): Promise<T | undefined> {\n  try {\n    return await operation();\n  } catch (error) {\n    context.errors.push(`${formatPath(target, context.rootDirectory)}: ${formatError(error)}`);\n    return undefined;\n  }\n}\n\n/** Resolves the path (recording errors); undefined when that fails or the result escapes the root. */\nasync function resolveWithinRoot(target: string, context: ScanContext): Promise<string | undefined> {\n  const resolved = await tryFileSystem(() => realpath(target), target, context);\n  return resolved !== undefined && isWithinDirectory(resolved, context.rootDirectory) ? resolved : undefined;\n}\n\nasync function scanDirectory(directory: string, context: ScanContext): Promise<void> {\n  const resolvedDirectory = await resolveWithinRoot(directory, context);\n  if (resolvedDirectory === undefined || context.visitedDirectories.has(resolvedDirectory)) {\n    return;\n  }\n  context.visitedDirectories.add(resolvedDirectory);\n\n  const entries = await tryFileSystem(() => readdir(directory, { withFileTypes: true }), directory, context);\n  if (entries === undefined) {\n    return;\n  }\n\n  for (const entry of entries) {\n    const entryPath = path.join(directory, entry.name);\n    if (entry.isSymbolicLink()) {\n      await scanSymbolicLink(entry.name, entryPath, context);\n      continue;\n    }\n\n    if (entry.isDirectory()) {\n      if (shouldSkipDirectory(entry.name, context.options)) {\n        continue;\n      }\n      await scanDirectory(entryPath, context);\n      continue;\n    }\n\n    if (entry.isFile()) {\n      await measureScannableFile(entryPath, context);\n    }\n  }\n}\n\nasync function scanSymbolicLink(name: string, entryPath: string, context: ScanContext): Promise<void> {\n  const resolvedPath = await resolveWithinRoot(entryPath, context);\n  if (resolvedPath === undefined) {\n    return;\n  }\n\n  const entryStat = await tryFileSystem(() => stat(entryPath), entryPath, context);\n  if (entryStat === undefined) {\n    return;\n  }\n\n  if (entryStat.isDirectory()) {\n    if (\n      shouldSkipDirectory(name, context.options) ||\n      shouldSkipDirectory(path.basename(resolvedPath), context.options)\n    ) {\n      return;\n    }\n    await scanDirectory(entryPath, context);\n    return;\n  }\n\n  if (entryStat.isFile()) {\n    await measureScannableFile(entryPath, context, resolvedPath, resolvedPath);\n  }\n}\n\nasync function measureScannableFile(\n  file: string,\n  context: ScanContext,\n  languageFile = file,\n  realFile?: string\n): Promise<void> {\n  const language = getLanguage(languageFile, context.options);\n  if (language) {\n    await measureFile(file, language, 'directory', context, realFile);\n  }\n}\n\nasync function measureFile(\n  file: string,\n  language: LanguageName,\n  mode: 'single-file' | 'directory',\n  context: ScanContext,\n  realFile?: string\n): Promise<void> {\n  try {\n    const resolvedFile = realFile ?? (await realpath(file));\n    if (context.visitedFiles.has(resolvedFile)) {\n      return;\n    }\n    context.visitedFiles.add(resolvedFile);\n\n    const code = await readFile(file, 'utf8');\n    const measureOptions = { language, duplication: context.options.duplication };\n    const fileMetrics: FileMetrics = { file, metrics: measureCode(code, measureOptions) };\n    // Only directory scans compare files against each other; a single-file target has no peers.\n    // Candidate collection is an auxiliary pass: if it fails where measureCode succeeded (an\n    // addon error specific to this pass), that must not discard the measured metrics.\n    if (mode === 'directory') {\n      try {\n        fileMetrics.duplicationCandidates = collectCrossFileDuplicationFileData(code, measureOptions);\n      } catch (error) {\n        // A warning, not an error: the file's metrics are complete, only its participation in\n        // cross-file matching is lost, so it is not \"skipped\" and must not fail --fail-on-error.\n        context.warnings.push(\n          `${formatPath(file, context.rootDirectory)}: cross-file duplication candidates unavailable: ${formatError(error)}`\n        );\n      }\n    }\n    context.files.push(fileMetrics);\n  } catch (error) {\n    // A missing native addon fails every file identically: propagate it once as a fatal scan\n    // error instead of recording one \"skipped\" entry per file behind a successful exit code.\n    if (error instanceof NativeAddonError) {\n      throw error;\n    }\n    context.errors.push(`${formatPath(file, context.rootDirectory)}: ${formatError(error)}`);\n  }\n}\n\n/** Runs after the scan so every measured file's candidates participate. */\nexport function addCrossFileDuplication(result: ScanResult, options: ScanOptions): void {\n  if (result.fatalError || result.files.length < 2) {\n    return;\n  }\n  const sourceFiles = result.files.flatMap(({ file, duplicationCandidates }) =>\n    duplicationCandidates ? [{ file: formatPath(file, result.displayRoot), ...duplicationCandidates }] : []\n  );\n  if (sourceFiles.length < 2) {\n    return;\n  }\n  result.crossFileDuplication = measureCrossFileDuplication(sourceFiles, options.duplication);\n}\n\n/**\n * Distinct 1-based code lines covered by within-file or cross-file duplicated content. Both\n * sources expose the exact lines carrying matched tokens (never block bounding ranges, which\n * would over-count comment/blank lines and the unmatched gap of a merged clone), so the union is\n * a subset of the file's code lines and a ratio derived over code lines can never exceed 1.\n */\nexport function collectDuplicatedLineNumbers(\n  metrics: CodeMetrics | undefined,\n  crossFileDuplication: CrossFileDuplicationMetrics | undefined,\n  formattedFile: string\n): Set<number> {\n  const lines = new Set(metrics?.duplication.duplicateLineNumbers);\n  // Object.hasOwn: a file named like an Object.prototype member must not read an inherited value.\n  const crossFileLines =\n    crossFileDuplication && Object.hasOwn(crossFileDuplication.duplicateLineNumbersByFile, formattedFile)\n      ? (crossFileDuplication.duplicateLineNumbersByFile[formattedFile] ?? [])\n      : [];\n  for (const line of crossFileLines) {\n    lines.add(line);\n  }\n  return lines;\n}\n\nfunction shouldSkipDirectory(name: string, options: ScanOptions): boolean {\n  if (ignoredDirectoryNames.has(name)) {\n    return true;\n  }\n\n  if (options.includeTests) {\n    return false;\n  }\n\n  return testDirectoryNames.has(name);\n}\n\nfunction isWithinDirectory(candidate: string, directory: string): boolean {\n  const relative = path.relative(directory, candidate);\n  return relative === '' || (relative !== '..' && !relative.startsWith(`..${path.sep}`) && !path.isAbsolute(relative));\n}\n\n/**\n * Whether a repository-relative path would be scanned: no ignored or excluded-test directory\n * segment and a supported, non-test file name. The diff gate uses this for base-revision\n * eligibility, so code renamed into scan scope gates as new code instead of ratcheting against\n * a blob the scanner would never have measured.\n */\nexport function isScannedPath(relativePath: string, options: ScanOptions): boolean {\n  const segments = relativePath.split('/');\n  for (const segment of segments.slice(0, -1)) {\n    if (ignoredDirectoryNames.has(segment) || (!options.includeTests && testDirectoryNames.has(segment))) {\n      return false;\n    }\n  }\n  return getLanguage(relativePath, options) !== undefined;\n}\n\nexport function getLanguage(file: string, options: ScanOptions, explicitTarget = false): LanguageName | undefined {\n  const lowerFile = file.toLowerCase();\n  if (\n    !explicitTarget &&\n    (lowerFile.endsWith('.d.ts') ||\n      lowerFile.endsWith('.d.mts') ||\n      lowerFile.endsWith('.d.cts') ||\n      lowerFile.endsWith('.min.js') ||\n      lowerFile.endsWith('.pnp.cjs'))\n  ) {\n    return undefined;\n  }\n\n  if (\n    !explicitTarget &&\n    !options.includeTests &&\n    (testFilePattern.test(path.basename(file)) || suffixTestFilePattern.test(path.basename(file)))\n  ) {\n    return undefined;\n  }\n\n  return detectLanguage(file);\n}\n\nexport function formatPath(file: string, base: string): string {\n  return path.relative(base, file) || path.basename(file);\n}\n\nexport function writeStdout(message: string): void {\n  process.stdout.write(message);\n}\n\nexport function writeStderr(message: string): void {\n  process.stderr.write(message);\n}\n\nexport function formatError(error: unknown): string {\n  return error instanceof Error ? error.message : String(error);\n}\n"],"mappings":"8YAiCA,MAAM,EAAwB,IAAI,IAAI,CACpC,UACA,UACA,UACA,OACA,QACA,kBACA,OACA,OACA,SACA,QACA,QACA,eACA,gBACA,cACA,WACA,OACA,WACA,YACA,eAEA,MACA,SACA,gBACA,SACA,MACF,CAAC,EAEK,EAAqB,IAAI,IAAI,CAAC,YAAa,OAAQ,QAAS,MAAM,CAAC,EACnE,EAAkB,mEAGlB,EAAwB,2BAE9B,SAAgB,EAAc,EAAwB,CASpD,OARI,IAAW,IACN,EAAG,QAAQ,EAGhB,EAAO,WAAW,IAAI,EACjB,EAAK,KAAK,EAAG,QAAQ,EAAG,EAAO,MAAM,CAAC,CAAC,EAGzC,EAAK,QAAQ,CAAM,CAC5B,CAGA,eAAsB,EAAsB,EAAiC,CAC3E,GAAI,CAEF,OAAO,MADkB,EAAK,CAAM,EAAA,CAClB,YAAY,EAAI,EAAS,EAAK,QAAQ,CAAM,CAChE,MAAQ,CACN,OAAO,EAAK,QAAQ,CAAM,CAC5B,CACF,CAcA,eAAsB,EAAW,EAAgB,EAA2C,CAC1F,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACxB,EAAkB,EACtB,GAAI,CACF,EAAkB,MAAM,EAAS,CAAM,CACzC,MAAQ,CAER,CAEA,IAAM,EAAsB,EAAK,QAAQ,CAAe,EACpD,EAEJ,GAAI,CACF,EAAa,MAAM,EAAK,CAAe,CACzC,OAAS,EAAO,CACd,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAmB,EAAE,IAAI,EAAY,CAAK,IAC5F,MAAO,CAAE,YAAa,EAAqB,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC/F,CAEA,GAAI,EAAW,OAAO,EAAG,CACvB,IAAM,EAAc,EAAK,QAAQ,CAAe,EAC1C,EAAW,EAAY,EAAiB,EAAS,EAAI,EAC3D,GAAI,CAAC,EAAU,CACb,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAW,EAAE,yBAC/D,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC1E,CAEA,IAAM,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAW,EAC7E,GAAI,CACF,MAAM,EAAY,EAAiB,EAAU,cAAe,EAAS,CAAe,CACtF,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAa,EAAO,EAAQ,CAAQ,CAClE,CACA,MAAO,CAAE,cAAa,QAAO,SAAQ,UAAS,CAChD,CAEA,GAAI,CACF,MAAM,EAAc,EAAiB,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAe,CAAC,CACzG,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAiB,EAAO,EAAQ,CAAQ,CACtE,CACA,MAAO,CAAE,YAAa,EAAiB,QAAO,SAAQ,UAAS,CACjE,CAQA,eAAsB,EACpB,EACA,EACA,EACqB,CACrB,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACtB,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAa,EAC/E,IAAK,IAAM,KAAgB,EAAe,CACxC,IAAM,EAAW,EAAc,EAAc,CAAO,EAAI,EAAY,EAAc,CAAO,EAAI,IAAA,GAC7F,GAAI,CAAC,EACH,SAEF,IAAM,EAAe,EAAK,KAAK,EAAe,CAAY,EAI1D,KAAI,MADgB,EAAM,CAAY,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EAC3C,eAAe,EAG1B,GAAI,CACF,MAAM,EAAY,EAAc,EAAU,YAAa,CAAO,CAChE,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAe,EAAO,EAAQ,CAAQ,CACpE,CACF,CACA,MAAO,CAAE,YAAa,EAAe,QAAO,SAAQ,UAAS,CAC/D,CAGA,SAAS,EACP,EACA,EACA,EACA,EACA,EACY,CACZ,GAAI,EAAE,aAAiB,GACrB,MAAM,EAER,IAAM,EAAa,EAAY,CAAK,EAEpC,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,GAAG,EAAQ,CAAU,EAAG,WAAU,YAAW,CACrF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACa,CACb,MAAO,CAAE,UAAS,QAAO,SAAQ,WAAU,mBAAoB,IAAI,IAAO,aAAc,IAAI,IAAO,eAAc,CACnH,CAGA,eAAe,EACb,EACA,EACA,EACwB,CACxB,GAAI,CACF,OAAO,MAAM,EAAU,CACzB,OAAS,EAAO,CACd,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAQ,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,EACzF,MACF,CACF,CAGA,eAAe,EAAkB,EAAgB,EAAmD,CAClG,IAAM,EAAW,MAAM,MAAoB,EAAS,CAAM,EAAG,EAAQ,CAAO,EAC5E,OAAO,IAAa,IAAA,IAAa,EAAkB,EAAU,EAAQ,aAAa,EAAI,EAAW,IAAA,EACnG,CAEA,eAAe,EAAc,EAAmB,EAAqC,CACnF,IAAM,EAAoB,MAAM,EAAkB,EAAW,CAAO,EACpE,GAAI,IAAsB,IAAA,IAAa,EAAQ,mBAAmB,IAAI,CAAiB,EACrF,OAEF,EAAQ,mBAAmB,IAAI,CAAiB,EAEhD,IAAM,EAAU,MAAM,MAAoB,EAAQ,EAAW,CAAE,cAAe,EAAK,CAAC,EAAG,EAAW,CAAO,EACrG,OAAY,IAAA,GAIhB,IAAK,IAAM,KAAS,EAAS,CAC3B,IAAM,EAAY,EAAK,KAAK,EAAW,EAAM,IAAI,EACjD,GAAI,EAAM,eAAe,EAAG,CAC1B,MAAM,EAAiB,EAAM,KAAM,EAAW,CAAO,EACrD,QACF,CAEA,GAAI,EAAM,YAAY,EAAG,CACvB,GAAI,EAAoB,EAAM,KAAM,EAAQ,OAAO,EACjD,SAEF,MAAM,EAAc,EAAW,CAAO,EACtC,QACF,CAEI,EAAM,OAAO,GACf,MAAM,EAAqB,EAAW,CAAO,CAEjD,CACF,CAEA,eAAe,EAAiB,EAAc,EAAmB,EAAqC,CACpG,IAAM,EAAe,MAAM,EAAkB,EAAW,CAAO,EAC/D,GAAI,IAAiB,IAAA,GACnB,OAGF,IAAM,EAAY,MAAM,MAAoB,EAAK,CAAS,EAAG,EAAW,CAAO,EAC3E,OAAc,IAAA,GAIlB,IAAI,EAAU,YAAY,EAAG,CAC3B,GACE,EAAoB,EAAM,EAAQ,OAAO,GACzC,EAAoB,EAAK,SAAS,CAAY,EAAG,EAAQ,OAAO,EAEhE,OAEF,MAAM,EAAc,EAAW,CAAO,EACtC,MACF,CAEI,EAAU,OAAO,GACnB,MAAM,EAAqB,EAAW,EAAS,EAAc,CAAY,CAH3E,CAKF,CAEA,eAAe,EACb,EACA,EACA,EAAe,EACf,EACe,CACf,IAAM,EAAW,EAAY,EAAc,EAAQ,OAAO,EACtD,GACF,MAAM,EAAY,EAAM,EAAU,YAAa,EAAS,CAAQ,CAEpE,CAEA,eAAe,EACb,EACA,EACA,EACA,EACA,EACe,CACf,GAAI,CACF,IAAM,EAAe,GAAa,MAAM,EAAS,CAAI,EACrD,GAAI,EAAQ,aAAa,IAAI,CAAY,EACvC,OAEF,EAAQ,aAAa,IAAI,CAAY,EAErC,IAAM,EAAO,MAAM,EAAS,EAAM,MAAM,EAClC,EAAiB,CAAE,WAAU,YAAa,EAAQ,QAAQ,WAAY,EACtE,EAA2B,CAAE,OAAM,QAAS,EAAY,EAAM,CAAc,CAAE,EAIpF,GAAI,IAAS,YACX,GAAI,CACF,EAAY,sBAAwB,EAAoC,EAAM,CAAc,CAC9F,OAAS,EAAO,CAGd,EAAQ,SAAS,KACf,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,mDAAmD,EAAY,CAAK,GACjH,CACF,CAEF,EAAQ,MAAM,KAAK,CAAW,CAChC,OAAS,EAAO,CAGd,GAAI,aAAiB,EACnB,MAAM,EAER,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,CACzF,CACF,CAGA,SAAgB,EAAwB,EAAoB,EAA4B,CACtF,GAAI,EAAO,YAAc,EAAO,MAAM,OAAS,EAC7C,OAEF,IAAM,EAAc,EAAO,MAAM,SAAS,CAAE,OAAM,2BAChD,EAAwB,CAAC,CAAE,KAAM,EAAW,EAAM,EAAO,WAAW,EAAG,GAAG,CAAsB,CAAC,EAAI,CAAC,CACxG,EACI,EAAY,OAAS,IAGzB,EAAO,qBAAuB,EAA4B,EAAa,EAAQ,WAAW,EAC5F,CAQA,SAAgB,EACd,EACA,EACA,EACa,CACb,IAAM,EAAQ,IAAI,IAAI,GAAS,YAAY,oBAAoB,EAEzD,EACJ,GAAwB,OAAO,OAAO,EAAqB,2BAA4B,CAAa,EAC/F,EAAqB,2BAA2B,IAAkB,CAAC,EACpE,CAAC,EACP,IAAK,IAAM,KAAQ,EACjB,EAAM,IAAI,CAAI,EAEhB,OAAO,CACT,CAEA,SAAS,EAAoB,EAAc,EAA+B,CASxE,OARI,EAAsB,IAAI,CAAI,EACzB,GAGT,CAAI,EAAQ,cAIL,EAAmB,IAAI,CAAI,CACpC,CAEA,SAAS,EAAkB,EAAmB,EAA4B,CACxE,IAAM,EAAW,EAAK,SAAS,EAAW,CAAS,EACnD,OAAO,IAAa,IAAO,IAAa,MAAQ,CAAC,EAAS,WAAW,KAAK,EAAK,KAAK,GAAK,CAAC,EAAK,WAAW,CAAQ,CACpH,CAQA,SAAgB,EAAc,EAAsB,EAA+B,CACjF,IAAM,EAAW,EAAa,MAAM,GAAG,EACvC,IAAK,IAAM,KAAW,EAAS,MAAM,EAAG,EAAE,EACxC,GAAI,EAAsB,IAAI,CAAO,GAAM,CAAC,EAAQ,cAAgB,EAAmB,IAAI,CAAO,EAChG,MAAO,GAGX,OAAO,EAAY,EAAc,CAAO,IAAM,IAAA,EAChD,CAEA,SAAgB,EAAY,EAAc,EAAsB,EAAiB,GAAiC,CAChH,IAAM,EAAY,EAAK,YAAY,EAEjC,MAAC,IACA,EAAU,SAAS,OAAO,GACzB,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,SAAS,GAC5B,EAAU,SAAS,UAAU,MAM9B,GACA,EAAQ,cACR,GAAgB,KAAK,EAAK,SAAS,CAAI,CAAC,GAAK,GAAsB,KAAK,EAAK,SAAS,CAAI,CAAC,GAK9F,OAAO,EAAe,CAAI,CAC5B,CAEA,SAAgB,EAAW,EAAc,EAAsB,CAC7D,OAAO,EAAK,SAAS,EAAM,CAAI,GAAK,EAAK,SAAS,CAAI,CACxD,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAwB,CAClD,OAAO,aAAiB,MAAQ,EAAM,QAAU,OAAO,CAAK,CAC9D"}
+{"version":3,"file":"scan.js","names":[],"sources":["../src/scan.ts"],"sourcesContent":["import { lstat, readdir, readFile, realpath, stat } from 'node:fs/promises';\nimport os from 'node:os';\nimport path from 'node:path';\nimport { measureCrossFileDuplication, type CrossFileDuplicationMetrics } from './crossFileDuplication.js';\nimport type { CrossFileDuplicationFileData } from './duplication.js';\nimport { detectLanguage } from './languages.js';\nimport { measureCode, measureCodeWithCrossFileData } from './metrics.js';\nimport { NativeAddonError } from './nativeMetrics.js';\nimport type { CodeMetrics, DuplicationOptions, LanguageName, MeasureOptions } from './types.js';\n\n/** The scan settings shared by every command (a structural subset of each command's options). */\nexport interface ScanOptions {\n  duplication: Required<DuplicationOptions>;\n  includeTests: boolean;\n}\n\nexport interface FileMetrics {\n  file: string;\n  metrics: CodeMetrics;\n  /** Cross-file duplicate candidates and token/statement data, collected only for directory scans. */\n  duplicationCandidates?: CrossFileDuplicationFileData;\n}\n\nexport interface ScanResult {\n  crossFileDuplication?: CrossFileDuplicationMetrics;\n  displayRoot: string;\n  errors: string[];\n  /** Non-fatal degradations (e.g. cross-file candidates unavailable); the file is still measured. */\n  warnings: string[];\n  fatalError?: string;\n  files: FileMetrics[];\n}\n\nconst ignoredDirectoryNames = new Set([\n  '.agents',\n  '.claude',\n  '.cursor',\n  '.git',\n  '.next',\n  '.playwright-cli',\n  '.tox',\n  '.tmp',\n  '.turbo',\n  '.venv',\n  '.yarn',\n  '__fixtures__',\n  '__generated__',\n  '__pycache__',\n  'coverage',\n  'dist',\n  'fixtures',\n  'generated',\n  'node_modules',\n  // .NET SDK intermediate output (generated sources such as `*.GlobalUsings.g.cs`).\n  'obj',\n  'target',\n  'test-fixtures',\n  'vendor',\n  'venv',\n]);\n\nconst testDirectoryNames = new Set(['__tests__', 'test', 'tests', 'spec']);\nconst testFilePattern = /(?:^test(?:[_-].*)?|\\.(?:spec|test)|[_-](?:test|spec))\\.[^.]+$/iu;\n// JUnit (Java/Kotlin) and xUnit/NUnit (C#) tests use case-sensitive `Test`/`Tests` class-name\n// suffixes; case-insensitive matching would catch production files like `contest.java`.\nconst suffixTestFilePattern = /Tests?\\.(?:java|kt|cs)$/u;\n\nexport function resolveTarget(target: string): string {\n  if (target === '~') {\n    return os.homedir();\n  }\n\n  if (target.startsWith('~/')) {\n    return path.join(os.homedir(), target.slice(2));\n  }\n\n  return path.resolve(target);\n}\n\n/** Returns the directory from which the config file search should start (the target itself if it is a directory). */\nexport async function configSearchDirectory(target: string): Promise<string> {\n  try {\n    const targetStat = await stat(target);\n    return targetStat.isDirectory() ? target : path.dirname(target);\n  } catch {\n    return path.dirname(target);\n  }\n}\n\n/** Shared state of one scan, threaded through the directory walk instead of positional plumbing. */\ninterface ScanContext {\n  options: ScanOptions;\n  files: FileMetrics[];\n  errors: string[];\n  warnings: string[];\n  visitedDirectories: Set<string>;\n  visitedFiles: Set<string>;\n  /** Scan root: paths are displayed relative to it, and symbolic links may not escape it. */\n  rootDirectory: string;\n}\n\nexport async function scanTarget(target: string, options: ScanOptions): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  let canonicalTarget = target;\n  try {\n    canonicalTarget = await realpath(target);\n  } catch {\n    // stat below reports missing targets with the original path.\n  }\n\n  const fallbackDisplayRoot = path.dirname(canonicalTarget);\n  let targetStat;\n\n  try {\n    targetStat = await stat(canonicalTarget);\n  } catch (error) {\n    const fatalError = `${formatPath(canonicalTarget, fallbackDisplayRoot)}: ${formatError(error)}`;\n    return { displayRoot: fallbackDisplayRoot, files, errors: [fatalError], warnings, fatalError };\n  }\n\n  if (targetStat.isFile()) {\n    const displayRoot = path.dirname(canonicalTarget);\n    const language = getLanguage(canonicalTarget, options, true);\n    if (!language) {\n      const fatalError = `${formatPath(canonicalTarget, displayRoot)}: unsupported file type`;\n      return { displayRoot, files, errors: [fatalError], warnings, fatalError };\n    }\n\n    const context = makeScanContext(options, files, errors, warnings, displayRoot);\n    try {\n      await measureFile(canonicalTarget, language, 'single-file', context, canonicalTarget);\n    } catch (error) {\n      return toFatalResult(error, displayRoot, files, errors, warnings);\n    }\n    return { displayRoot, files, errors, warnings };\n  }\n\n  try {\n    await scanDirectory(canonicalTarget, makeScanContext(options, files, errors, warnings, canonicalTarget));\n  } catch (error) {\n    return toFatalResult(error, canonicalTarget, files, errors, warnings);\n  }\n  return { displayRoot: canonicalTarget, files, errors, warnings };\n}\n\n/**\n * Measures an explicit list of repository-relative files (the diff gate's git-visible allowlist)\n * instead of walking the directory tree, so ignored artifact directories are never parsed. Paths\n * outside the scan scope (ignored/test directories, unsupported or test file names) are skipped\n * with the same rules as the walk.\n */\nexport async function scanListedFiles(\n  rootDirectory: string,\n  relativePaths: Iterable<string>,\n  options: ScanOptions\n): Promise<ScanResult> {\n  const files: FileMetrics[] = [];\n  const errors: string[] = [];\n  const warnings: string[] = [];\n  const context = makeScanContext(options, files, errors, warnings, rootDirectory);\n  for (const relativePath of relativePaths) {\n    const language = isScannedPath(relativePath, options) ? getLanguage(relativePath, options) : undefined;\n    if (!language) {\n      continue;\n    }\n    const absolutePath = path.join(rootDirectory, relativePath);\n    // Symbolic links are not source files: git stores only their target string, so measuring\n    // through them would diverge from what any revision of the repository actually contains.\n    const stats = await lstat(absolutePath).catch(() => {});\n    if (stats?.isSymbolicLink()) {\n      continue;\n    }\n    try {\n      await measureFile(absolutePath, language, 'directory', context);\n    } catch (error) {\n      return toFatalResult(error, rootDirectory, files, errors, warnings);\n    }\n  }\n  return { displayRoot: rootDirectory, files, errors, warnings };\n}\n\n/** A run-wide failure (a missing native addon) as a fatal result; anything else keeps throwing. */\nfunction toFatalResult(\n  error: unknown,\n  displayRoot: string,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[]\n): ScanResult {\n  if (!(error instanceof NativeAddonError)) {\n    throw error;\n  }\n  const fatalError = formatError(error);\n  // Errors the walk accumulated before the fatal failure stay reported alongside it.\n  return { displayRoot, files, errors: [...errors, fatalError], warnings, fatalError };\n}\n\nfunction makeScanContext(\n  options: ScanOptions,\n  files: FileMetrics[],\n  errors: string[],\n  warnings: string[],\n  rootDirectory: string\n): ScanContext {\n  return { options, files, errors, warnings, visitedDirectories: new Set(), visitedFiles: new Set(), rootDirectory };\n}\n\n/** Runs a filesystem operation, recording a scan error and returning undefined when it fails. */\nasync function tryFileSystem<T>(\n  operation: () => Promise<T>,\n  target: string,\n  context: ScanContext\n): Promise<T | undefined> {\n  try {\n    return await operation();\n  } catch (error) {\n    context.errors.push(`${formatPath(target, context.rootDirectory)}: ${formatError(error)}`);\n    return undefined;\n  }\n}\n\n/** Resolves the path (recording errors); undefined when that fails or the result escapes the root. */\nasync function resolveWithinRoot(target: string, context: ScanContext): Promise<string | undefined> {\n  const resolved = await tryFileSystem(() => realpath(target), target, context);\n  return resolved !== undefined && isWithinDirectory(resolved, context.rootDirectory) ? resolved : undefined;\n}\n\nasync function scanDirectory(directory: string, context: ScanContext): Promise<void> {\n  const resolvedDirectory = await resolveWithinRoot(directory, context);\n  if (resolvedDirectory === undefined || context.visitedDirectories.has(resolvedDirectory)) {\n    return;\n  }\n  context.visitedDirectories.add(resolvedDirectory);\n\n  const entries = await tryFileSystem(() => readdir(directory, { withFileTypes: true }), directory, context);\n  if (entries === undefined) {\n    return;\n  }\n\n  for (const entry of entries) {\n    const entryPath = path.join(directory, entry.name);\n    if (entry.isSymbolicLink()) {\n      await scanSymbolicLink(entry.name, entryPath, context);\n      continue;\n    }\n\n    if (entry.isDirectory()) {\n      if (shouldSkipDirectory(entry.name, context.options)) {\n        continue;\n      }\n      await scanDirectory(entryPath, context);\n      continue;\n    }\n\n    if (entry.isFile()) {\n      await measureScannableFile(entryPath, context);\n    }\n  }\n}\n\nasync function scanSymbolicLink(name: string, entryPath: string, context: ScanContext): Promise<void> {\n  const resolvedPath = await resolveWithinRoot(entryPath, context);\n  if (resolvedPath === undefined) {\n    return;\n  }\n\n  const entryStat = await tryFileSystem(() => stat(entryPath), entryPath, context);\n  if (entryStat === undefined) {\n    return;\n  }\n\n  if (entryStat.isDirectory()) {\n    if (\n      shouldSkipDirectory(name, context.options) ||\n      shouldSkipDirectory(path.basename(resolvedPath), context.options)\n    ) {\n      return;\n    }\n    await scanDirectory(entryPath, context);\n    return;\n  }\n\n  if (entryStat.isFile()) {\n    await measureScannableFile(entryPath, context, resolvedPath, resolvedPath);\n  }\n}\n\nasync function measureScannableFile(\n  file: string,\n  context: ScanContext,\n  languageFile = file,\n  realFile?: string\n): Promise<void> {\n  const language = getLanguage(languageFile, context.options);\n  if (language) {\n    await measureFile(file, language, 'directory', context, realFile);\n  }\n}\n\nasync function measureFile(\n  file: string,\n  language: LanguageName,\n  mode: 'single-file' | 'directory',\n  context: ScanContext,\n  realFile?: string\n): Promise<void> {\n  try {\n    const resolvedFile = realFile ?? (await realpath(file));\n    if (context.visitedFiles.has(resolvedFile)) {\n      return;\n    }\n    context.visitedFiles.add(resolvedFile);\n\n    const code = await readFile(file, 'utf8');\n    const measureOptions = { language, duplication: context.options.duplication };\n    // Only directory scans compare files against each other; a single-file target has no peers.\n    if (mode === 'single-file') {\n      context.files.push({ file, metrics: measureCode(code, measureOptions) });\n      return;\n    }\n    const { metrics, crossFileData, crossFileError } = measureWithCrossFileData(code, measureOptions);\n    if (crossFileError !== undefined) {\n      // A warning, not an error: the file's metrics are complete, only its participation in\n      // cross-file matching is lost, so it is not \"skipped\" and must not fail --fail-on-error.\n      context.warnings.push(\n        `${formatPath(file, context.rootDirectory)}: cross-file duplication candidates unavailable: ${crossFileError}`\n      );\n    }\n    context.files.push({ file, metrics, duplicationCandidates: crossFileData });\n  } catch (error) {\n    // A missing native addon fails every file identically: propagate it once as a fatal scan\n    // error instead of recording one \"skipped\" entry per file behind a successful exit code.\n    if (error instanceof NativeAddonError) {\n      throw error;\n    }\n    context.errors.push(`${formatPath(file, context.rootDirectory)}: ${formatError(error)}`);\n  }\n}\n\n/**\n * Measures a file together with its cross-file contribution from one parse. The contribution is\n * auxiliary: if collecting it fails where plain measurement succeeds (e.g. a payload too large to\n * cross the addon boundary), the metrics are still returned with the failure message, which\n * callers report as a warning rather than an error.\n */\nexport function measureWithCrossFileData(\n  code: string,\n  measureOptions: MeasureOptions\n): { metrics: CodeMetrics; crossFileData?: CrossFileDuplicationFileData; crossFileError?: string } {\n  try {\n    return measureCodeWithCrossFileData(code, measureOptions);\n  } catch (error) {\n    if (error instanceof NativeAddonError) {\n      throw error;\n    }\n    return { metrics: measureCode(code, measureOptions), crossFileError: formatError(error) };\n  }\n}\n\n/** Runs after the scan so every measured file's candidates participate. */\nexport function addCrossFileDuplication(result: ScanResult, options: ScanOptions): void {\n  if (result.fatalError || result.files.length < 2) {\n    return;\n  }\n  const sourceFiles = result.files.flatMap(({ file, duplicationCandidates }) =>\n    duplicationCandidates ? [{ file: formatPath(file, result.displayRoot), ...duplicationCandidates }] : []\n  );\n  if (sourceFiles.length < 2) {\n    return;\n  }\n  result.crossFileDuplication = measureCrossFileDuplication(sourceFiles, options.duplication);\n}\n\n/**\n * Distinct 1-based code lines covered by within-file or cross-file duplicated content. Both\n * sources expose the exact lines carrying matched tokens (never block bounding ranges, which\n * would over-count comment/blank lines and the unmatched gap of a merged clone), so the union is\n * a subset of the file's code lines and a ratio derived over code lines can never exceed 1.\n */\nexport function collectDuplicatedLineNumbers(\n  metrics: CodeMetrics | undefined,\n  crossFileDuplication: CrossFileDuplicationMetrics | undefined,\n  formattedFile: string\n): Set<number> {\n  const lines = new Set(metrics?.duplication.duplicateLineNumbers);\n  // Object.hasOwn: a file named like an Object.prototype member must not read an inherited value.\n  const crossFileLines =\n    crossFileDuplication && Object.hasOwn(crossFileDuplication.duplicateLineNumbersByFile, formattedFile)\n      ? (crossFileDuplication.duplicateLineNumbersByFile[formattedFile] ?? [])\n      : [];\n  for (const line of crossFileLines) {\n    lines.add(line);\n  }\n  return lines;\n}\n\nfunction shouldSkipDirectory(name: string, options: ScanOptions): boolean {\n  if (ignoredDirectoryNames.has(name)) {\n    return true;\n  }\n\n  if (options.includeTests) {\n    return false;\n  }\n\n  return testDirectoryNames.has(name);\n}\n\nfunction isWithinDirectory(candidate: string, directory: string): boolean {\n  const relative = path.relative(directory, candidate);\n  return relative === '' || (relative !== '..' && !relative.startsWith(`..${path.sep}`) && !path.isAbsolute(relative));\n}\n\n/**\n * Whether a repository-relative path would be scanned: no ignored or excluded-test directory\n * segment and a supported, non-test file name. The diff gate uses this for base-revision\n * eligibility, so code renamed into scan scope gates as new code instead of ratcheting against\n * a blob the scanner would never have measured.\n */\nexport function isScannedPath(relativePath: string, options: ScanOptions): boolean {\n  const segments = relativePath.split('/');\n  for (const segment of segments.slice(0, -1)) {\n    if (ignoredDirectoryNames.has(segment) || (!options.includeTests && testDirectoryNames.has(segment))) {\n      return false;\n    }\n  }\n  return getLanguage(relativePath, options) !== undefined;\n}\n\nexport function getLanguage(file: string, options: ScanOptions, explicitTarget = false): LanguageName | undefined {\n  const lowerFile = file.toLowerCase();\n  if (\n    !explicitTarget &&\n    (lowerFile.endsWith('.d.ts') ||\n      lowerFile.endsWith('.d.mts') ||\n      lowerFile.endsWith('.d.cts') ||\n      lowerFile.endsWith('.min.js') ||\n      lowerFile.endsWith('.pnp.cjs'))\n  ) {\n    return undefined;\n  }\n\n  if (\n    !explicitTarget &&\n    !options.includeTests &&\n    (testFilePattern.test(path.basename(file)) || suffixTestFilePattern.test(path.basename(file)))\n  ) {\n    return undefined;\n  }\n\n  return detectLanguage(file);\n}\n\nexport function formatPath(file: string, base: string): string {\n  return path.relative(base, file) || path.basename(file);\n}\n\nexport function writeStdout(message: string): void {\n  process.stdout.write(message);\n}\n\nexport function writeStderr(message: string): void {\n  process.stderr.write(message);\n}\n\nexport function formatError(error: unknown): string {\n  return error instanceof Error ? error.message : String(error);\n}\n"],"mappings":"uYAiCA,MAAM,EAAwB,IAAI,IAAI,CACpC,UACA,UACA,UACA,OACA,QACA,kBACA,OACA,OACA,SACA,QACA,QACA,eACA,gBACA,cACA,WACA,OACA,WACA,YACA,eAEA,MACA,SACA,gBACA,SACA,MACF,CAAC,EAEK,EAAqB,IAAI,IAAI,CAAC,YAAa,OAAQ,QAAS,MAAM,CAAC,EACnE,EAAkB,mEAGlB,EAAwB,2BAE9B,SAAgB,EAAc,EAAwB,CASpD,OARI,IAAW,IACN,EAAG,QAAQ,EAGhB,EAAO,WAAW,IAAI,EACjB,EAAK,KAAK,EAAG,QAAQ,EAAG,EAAO,MAAM,CAAC,CAAC,EAGzC,EAAK,QAAQ,CAAM,CAC5B,CAGA,eAAsB,EAAsB,EAAiC,CAC3E,GAAI,CAEF,OAAO,MADkB,EAAK,CAAM,EAAA,CAClB,YAAY,EAAI,EAAS,EAAK,QAAQ,CAAM,CAChE,MAAQ,CACN,OAAO,EAAK,QAAQ,CAAM,CAC5B,CACF,CAcA,eAAsB,EAAW,EAAgB,EAA2C,CAC1F,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACxB,EAAkB,EACtB,GAAI,CACF,EAAkB,MAAM,EAAS,CAAM,CACzC,MAAQ,CAER,CAEA,IAAM,EAAsB,EAAK,QAAQ,CAAe,EACpD,EAEJ,GAAI,CACF,EAAa,MAAM,EAAK,CAAe,CACzC,OAAS,EAAO,CACd,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAmB,EAAE,IAAI,EAAY,CAAK,IAC5F,MAAO,CAAE,YAAa,EAAqB,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC/F,CAEA,GAAI,EAAW,OAAO,EAAG,CACvB,IAAM,EAAc,EAAK,QAAQ,CAAe,EAC1C,EAAW,EAAY,EAAiB,EAAS,EAAI,EAC3D,GAAI,CAAC,EAAU,CACb,IAAM,EAAa,GAAG,EAAW,EAAiB,CAAW,EAAE,yBAC/D,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,CAAU,EAAG,WAAU,YAAW,CAC1E,CAEA,IAAM,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAW,EAC7E,GAAI,CACF,MAAM,EAAY,EAAiB,EAAU,cAAe,EAAS,CAAe,CACtF,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAa,EAAO,EAAQ,CAAQ,CAClE,CACA,MAAO,CAAE,cAAa,QAAO,SAAQ,UAAS,CAChD,CAEA,GAAI,CACF,MAAM,EAAc,EAAiB,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAe,CAAC,CACzG,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAiB,EAAO,EAAQ,CAAQ,CACtE,CACA,MAAO,CAAE,YAAa,EAAiB,QAAO,SAAQ,UAAS,CACjE,CAQA,eAAsB,EACpB,EACA,EACA,EACqB,CACrB,IAAM,EAAuB,CAAC,EACxB,EAAmB,CAAC,EACpB,EAAqB,CAAC,EACtB,EAAU,EAAgB,EAAS,EAAO,EAAQ,EAAU,CAAa,EAC/E,IAAK,IAAM,KAAgB,EAAe,CACxC,IAAM,EAAW,EAAc,EAAc,CAAO,EAAI,EAAY,EAAc,CAAO,EAAI,IAAA,GAC7F,GAAI,CAAC,EACH,SAEF,IAAM,EAAe,EAAK,KAAK,EAAe,CAAY,EAI1D,KAAI,MADgB,EAAM,CAAY,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EAC3C,eAAe,EAG1B,GAAI,CACF,MAAM,EAAY,EAAc,EAAU,YAAa,CAAO,CAChE,OAAS,EAAO,CACd,OAAO,EAAc,EAAO,EAAe,EAAO,EAAQ,CAAQ,CACpE,CACF,CACA,MAAO,CAAE,YAAa,EAAe,QAAO,SAAQ,UAAS,CAC/D,CAGA,SAAS,EACP,EACA,EACA,EACA,EACA,EACY,CACZ,GAAI,EAAE,aAAiB,GACrB,MAAM,EAER,IAAM,EAAa,EAAY,CAAK,EAEpC,MAAO,CAAE,cAAa,QAAO,OAAQ,CAAC,GAAG,EAAQ,CAAU,EAAG,WAAU,YAAW,CACrF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACa,CACb,MAAO,CAAE,UAAS,QAAO,SAAQ,WAAU,mBAAoB,IAAI,IAAO,aAAc,IAAI,IAAO,eAAc,CACnH,CAGA,eAAe,EACb,EACA,EACA,EACwB,CACxB,GAAI,CACF,OAAO,MAAM,EAAU,CACzB,OAAS,EAAO,CACd,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAQ,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,EACzF,MACF,CACF,CAGA,eAAe,EAAkB,EAAgB,EAAmD,CAClG,IAAM,EAAW,MAAM,MAAoB,EAAS,CAAM,EAAG,EAAQ,CAAO,EAC5E,OAAO,IAAa,IAAA,IAAa,EAAkB,EAAU,EAAQ,aAAa,EAAI,EAAW,IAAA,EACnG,CAEA,eAAe,EAAc,EAAmB,EAAqC,CACnF,IAAM,EAAoB,MAAM,EAAkB,EAAW,CAAO,EACpE,GAAI,IAAsB,IAAA,IAAa,EAAQ,mBAAmB,IAAI,CAAiB,EACrF,OAEF,EAAQ,mBAAmB,IAAI,CAAiB,EAEhD,IAAM,EAAU,MAAM,MAAoB,EAAQ,EAAW,CAAE,cAAe,EAAK,CAAC,EAAG,EAAW,CAAO,EACrG,OAAY,IAAA,GAIhB,IAAK,IAAM,KAAS,EAAS,CAC3B,IAAM,EAAY,EAAK,KAAK,EAAW,EAAM,IAAI,EACjD,GAAI,EAAM,eAAe,EAAG,CAC1B,MAAM,EAAiB,EAAM,KAAM,EAAW,CAAO,EACrD,QACF,CAEA,GAAI,EAAM,YAAY,EAAG,CACvB,GAAI,EAAoB,EAAM,KAAM,EAAQ,OAAO,EACjD,SAEF,MAAM,EAAc,EAAW,CAAO,EACtC,QACF,CAEI,EAAM,OAAO,GACf,MAAM,EAAqB,EAAW,CAAO,CAEjD,CACF,CAEA,eAAe,EAAiB,EAAc,EAAmB,EAAqC,CACpG,IAAM,EAAe,MAAM,EAAkB,EAAW,CAAO,EAC/D,GAAI,IAAiB,IAAA,GACnB,OAGF,IAAM,EAAY,MAAM,MAAoB,EAAK,CAAS,EAAG,EAAW,CAAO,EAC3E,OAAc,IAAA,GAIlB,IAAI,EAAU,YAAY,EAAG,CAC3B,GACE,EAAoB,EAAM,EAAQ,OAAO,GACzC,EAAoB,EAAK,SAAS,CAAY,EAAG,EAAQ,OAAO,EAEhE,OAEF,MAAM,EAAc,EAAW,CAAO,EACtC,MACF,CAEI,EAAU,OAAO,GACnB,MAAM,EAAqB,EAAW,EAAS,EAAc,CAAY,CAH3E,CAKF,CAEA,eAAe,EACb,EACA,EACA,EAAe,EACf,EACe,CACf,IAAM,EAAW,EAAY,EAAc,EAAQ,OAAO,EACtD,GACF,MAAM,EAAY,EAAM,EAAU,YAAa,EAAS,CAAQ,CAEpE,CAEA,eAAe,EACb,EACA,EACA,EACA,EACA,EACe,CACf,GAAI,CACF,IAAM,EAAe,GAAa,MAAM,EAAS,CAAI,EACrD,GAAI,EAAQ,aAAa,IAAI,CAAY,EACvC,OAEF,EAAQ,aAAa,IAAI,CAAY,EAErC,IAAM,EAAO,MAAM,EAAS,EAAM,MAAM,EAClC,EAAiB,CAAE,WAAU,YAAa,EAAQ,QAAQ,WAAY,EAE5E,GAAI,IAAS,cAAe,CAC1B,EAAQ,MAAM,KAAK,CAAE,OAAM,QAAS,EAAY,EAAM,CAAc,CAAE,CAAC,EACvE,MACF,CACA,GAAM,CAAE,UAAS,gBAAe,kBAAmB,EAAyB,EAAM,CAAc,EAC5F,IAAmB,IAAA,IAGrB,EAAQ,SAAS,KACf,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,mDAAmD,GAChG,EAEF,EAAQ,MAAM,KAAK,CAAE,OAAM,UAAS,sBAAuB,CAAc,CAAC,CAC5E,OAAS,EAAO,CAGd,GAAI,aAAiB,EACnB,MAAM,EAER,EAAQ,OAAO,KAAK,GAAG,EAAW,EAAM,EAAQ,aAAa,EAAE,IAAI,EAAY,CAAK,GAAG,CACzF,CACF,CAQA,SAAgB,EACd,EACA,EACiG,CACjG,GAAI,CACF,OAAO,EAA6B,EAAM,CAAc,CAC1D,OAAS,EAAO,CACd,GAAI,aAAiB,EACnB,MAAM,EAER,MAAO,CAAE,QAAS,EAAY,EAAM,CAAc,EAAG,eAAgB,EAAY,CAAK,CAAE,CAC1F,CACF,CAGA,SAAgB,EAAwB,EAAoB,EAA4B,CACtF,GAAI,EAAO,YAAc,EAAO,MAAM,OAAS,EAC7C,OAEF,IAAM,EAAc,EAAO,MAAM,SAAS,CAAE,OAAM,2BAChD,EAAwB,CAAC,CAAE,KAAM,EAAW,EAAM,EAAO,WAAW,EAAG,GAAG,CAAsB,CAAC,EAAI,CAAC,CACxG,EACI,EAAY,OAAS,IAGzB,EAAO,qBAAuB,EAA4B,EAAa,EAAQ,WAAW,EAC5F,CAQA,SAAgB,EACd,EACA,EACA,EACa,CACb,IAAM,EAAQ,IAAI,IAAI,GAAS,YAAY,oBAAoB,EAEzD,EACJ,GAAwB,OAAO,OAAO,EAAqB,2BAA4B,CAAa,EAC/F,EAAqB,2BAA2B,IAAkB,CAAC,EACpE,CAAC,EACP,IAAK,IAAM,KAAQ,EACjB,EAAM,IAAI,CAAI,EAEhB,OAAO,CACT,CAEA,SAAS,EAAoB,EAAc,EAA+B,CASxE,OARI,EAAsB,IAAI,CAAI,EACzB,GAGT,CAAI,EAAQ,cAIL,EAAmB,IAAI,CAAI,CACpC,CAEA,SAAS,EAAkB,EAAmB,EAA4B,CACxE,IAAM,EAAW,EAAK,SAAS,EAAW,CAAS,EACnD,OAAO,IAAa,IAAO,IAAa,MAAQ,CAAC,EAAS,WAAW,KAAK,EAAK,KAAK,GAAK,CAAC,EAAK,WAAW,CAAQ,CACpH,CAQA,SAAgB,EAAc,EAAsB,EAA+B,CACjF,IAAM,EAAW,EAAa,MAAM,GAAG,EACvC,IAAK,IAAM,KAAW,EAAS,MAAM,EAAG,EAAE,EACxC,GAAI,EAAsB,IAAI,CAAO,GAAM,CAAC,EAAQ,cAAgB,EAAmB,IAAI,CAAO,EAChG,MAAO,GAGX,OAAO,EAAY,EAAc,CAAO,IAAM,IAAA,EAChD,CAEA,SAAgB,EAAY,EAAc,EAAsB,EAAiB,GAAiC,CAChH,IAAM,EAAY,EAAK,YAAY,EAEjC,MAAC,IACA,EAAU,SAAS,OAAO,GACzB,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,QAAQ,GAC3B,EAAU,SAAS,SAAS,GAC5B,EAAU,SAAS,UAAU,MAM9B,GACA,EAAQ,cACR,GAAgB,KAAK,EAAK,SAAS,CAAI,CAAC,GAAK,GAAsB,KAAK,EAAK,SAAS,CAAI,CAAC,GAK9F,OAAO,EAAe,CAAI,CAC5B,CAEA,SAAgB,EAAW,EAAc,EAAsB,CAC7D,OAAO,EAAK,SAAS,EAAM,CAAI,GAAK,EAAK,SAAS,CAAI,CACxD,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAuB,CACjD,QAAQ,OAAO,MAAM,CAAO,CAC9B,CAEA,SAAgB,EAAY,EAAwB,CAClD,OAAO,aAAiB,MAAQ,EAAM,QAAU,OAAO,CAAK,CAC9D"}

package/dist/types.d.ts

@@ -23,7 +23,7 @@ export interface DuplicationOptions {
      * Minimum similarity percent (1-100) for near-miss (Type-3) clone blocks, measured as the
      * token-level longest common subsequence relative to the larger block (NiCad-style per-fragment
      * similarity). 100 disables near-miss detection and reports exact (Type-1/2) matches plus gapped
-     * merges only (default 70). Applies to within-file detection only.
+     * merges only (default 70). Applies to within-file and cross-file detection.
      */
     minSimilarityPercent?: number;
 }
@@ -140,6 +140,16 @@ export interface CodeMetrics {
     bytes: number;
     lines: LineMetrics;
     functions: FunctionMetrics[];
+    /**
+     * The file's cyclomatic complexity as McCabe's v = e - n + 2p over its components: the sum of
+     * every function's value, plus 1 per initializer block (Java static/instance initializers, Kotlin
+     * `init`, JavaScript/TypeScript class `static` blocks), plus the decisions outside functions,
+     * plus 1 for the module body when
+     * the file runs top-level code (always for JavaScript, TypeScript, Python, and Ruby; for C# and
+     * Kotlin when the file has top-level statements, except a Kotlin file with parse errors, whose top
+     * level the grammar cannot be trusted to have read correctly).
+     */
+    cyclomaticComplexity: number;
     cognitiveComplexity: number;
     maxCognitiveComplexity: number;
     nestingDepth: number;