code-gauge 5.0.0 → 5.1.0

package/dist/diffCommand.js.map

@@ -1 +1 @@
-{"version":3,"file":"diffCommand.js","names":[],"sources":["../src/diffCommand.ts"],"sourcesContent":["import { lstat, readFile, realpath, stat } from 'node:fs/promises';\nimport os from 'node:os';\nimport path from 'node:path';\nimport { loadConfig, resolveGateOptions, resolveOptions, type ResolvedOptions } from './cliConfig.js';\nimport {\n  measureCrossFileDuplication,\n  type CrossFileDuplicateOccurrence,\n  type CrossFileDuplicationMetrics,\n} from './crossFileDuplication.js';\nimport type { CrossFileDuplicationFileData, Token } from './duplication.js';\nimport { keepPaths, loadRepositoryExclusion, type Exclusion } from './exclusion.js';\nimport {\n  listChangedFiles,\n  listLineHunks,\n  listRepositoryFiles,\n  listSymlinkPathsAtRevision,\n  readFileAtRevision,\n  resolveMergeBase,\n  resolveRepoRoot,\n  type ChangedFile,\n} from './git.js';\nimport { collectFunctionTokenSequences } from './metrics.js';\nimport {\n  findNewlyDuplicatedLines,\n  indexPartnersByLine,\n  type CloneOccurrence,\n  type DuplicationChange,\n  type LineHunk,\n} from './newDuplication.js';\nimport {\n  evaluateRegressionGate,\n  type CheckedFunctionReport,\n  type GateFileInput,\n  type GateFunctionValues,\n  type GateResult,\n} from './regressionGate.js';\nimport {\n  collectDuplicatedLineNumbers,\n  configSearchDirectory,\n  formatError,\n  formatPath,\n  getLanguage,\n  isScannedPath,\n  measureWithCrossFileData,\n  resolveTarget,\n  scanListedFiles,\n  writeStderr,\n  writeStdout,\n  type FileMetrics,\n} from './scan.js';\nimport type { CodeMetrics, LanguageName } from './types.js';\n\n/** Raw options of the `diff` subcommand; every field but base is undefined unless the flag was passed. */\nexport interface DiffCliOptions {\n  base: string;\n  config?: string;\n  duplicationMinTokens?: number;\n  duplicationMaxGapTokens?: number;\n  duplicationMinSimilarityPercent?: number;\n  includeTests?: boolean;\n  json?: boolean;\n  full?: boolean;\n}\n\n/** One changed file measured at both revisions, plus its duplication-universe contribution. */\ninterface PreparedFile {\n  changed: ChangedFile;\n  /** Repository-relative display path: the head path, or the base path for deleted files. */\n  displayFile: string;\n  /** Whether the file is gated (under the target directory); others only feed the base universe. */\n  gated: boolean;\n  headFile?: FileMetrics;\n  headContent?: string;\n  baseMetrics?: CodeMetrics;\n  baseContent?: string;\n  /**\n   * The line diff from the measured base to the measured head; with one side unmeasured, every\n   * line of the other counts as added or deleted.\n   */\n  hunks?: LineHunk[];\n  baseCandidates?: CrossFileDuplicationFileData;\n  baseFunctionTokens?: Int32Array[];\n  headFunctionTokens?: Int32Array[];\n}\n\n/** A scanned file that git considers part of the project, keyed by its repository-relative path. */\ninterface ScannedFile {\n  relativePath: string;\n  file: FileMetrics;\n}\n\n/**\n * Runs the regression gate: measures the files changed relative to the merge-base with the base\n * ref, at both revisions (`git cat-file`; no checkout, no persisted baseline), and reports only\n * violations. Exit codes: 0 all gates passed, 1 violations, 2 changed files could not be measured.\n */\nexport async function runDiffCommand(target: string, cliOptions: DiffCliOptions): Promise<void> {\n  try {\n    await runGate(target, cliOptions);\n  } catch (error) {\n    writeStderr(`Error: ${formatError(error)}\\n`);\n    process.exitCode = 2;\n  }\n}\n\nasync function runGate(target: string, cliOptions: DiffCliOptions): Promise<void> {\n  const resolvedTarget = resolveTarget(target);\n  const config = await loadConfig(cliOptions.config, await configSearchDirectory(resolvedTarget));\n  const options = resolveOptions(cliOptions, config);\n  const gateOptions = resolveGateOptions(config);\n\n  // The target may be a typo'd path whose ancestors don't exist either; repository discovery must\n  // still run so the mistyped target gets its own diagnostic instead of a git spawn failure.\n  const repoRoot = await realpath(\n    await resolveRepoRoot(await firstExistingDirectory(await configSearchDirectory(resolvedTarget)))\n  );\n  const mergeBase = await resolveMergeBase(repoRoot, cliOptions.base);\n  const changedFiles = await listChangedFiles(repoRoot, mergeBase);\n\n  // Every git-visible file (tracked or untracked non-ignored) is measured at head: that provides\n  // the head metrics of changed files and the project-wide duplication universe, so copy-paste\n  // from unchanged code into changed files is caught. Scanning the explicit git list (instead of\n  // walking the tree) keeps ignored artifact directories from ever being parsed: they exist in\n  // neither the base commit nor CI, so they would only cost time and skew duplication counts.\n  // Unchanged files are byte-identical at both revisions, so the base universe is the same scan\n  // with the changed files' contents swapped for their merge-base blobs.\n  const repositoryFiles = await listRepositoryFiles(repoRoot);\n  const baseSymlinkPaths = await listSymlinkPathsAtRevision(repoRoot, mergeBase);\n  const { canonicalTarget, targetExists } = await canonicalizeTarget(resolvedTarget);\n  const targetStat = await stat(canonicalTarget).catch(() => {});\n  const targetFile = targetStat?.isFile() ? canonicalTarget : undefined;\n  // A targeted file git does not list (an ignored one) is still measured, as an addition.\n  const unlistedTarget = targetFile && path.relative(repoRoot, targetFile).split(path.sep).join('/');\n  if (unlistedTarget !== undefined && !unlistedTarget.startsWith('../') && !repositoryFiles.has(unlistedTarget)) {\n    repositoryFiles.add(unlistedTarget);\n    changedFiles.push({ status: 'added', headPath: unlistedTarget });\n  }\n  const explicitFiles = listExplicitlyTargetedPaths(targetFile, repoRoot, changedFiles);\n  // Base blobs are excluded by the attributes of the revision they come from. A changed\n  // .gitattributes can re-include files whose bytes did not change, so then every file's base\n  // attributes are needed to find them.\n  const attributesChanged = changedFiles.some((changed) =>\n    [changed.headPath, changed.basePath].some(\n      (file) => file !== undefined && path.posix.basename(file) === '.gitattributes'\n    )\n  );\n  const basePaths = changedFiles.flatMap((changed) => (changed.basePath === undefined ? [] : [changed.basePath]));\n  const [headAttributesExclusion, baseAttributesExclusion] = await Promise.all([\n    loadRepositoryExclusion(repoRoot, repositoryFiles, options.exclude),\n    loadRepositoryExclusion(\n      repoRoot,\n      attributesChanged ? [...new Set([...basePaths, ...repositoryFiles])] : basePaths,\n      options.exclude,\n      mergeBase\n    ),\n  ]);\n  const headExclusion = keepPaths(headAttributesExclusion, explicitFiles);\n  const baseExclusion = keepPaths(baseAttributesExclusion, explicitFiles);\n  const scan = await scanListedFiles(\n    repoRoot,\n    repositoryFiles,\n    { ...options, loadExclusion: () => Promise.resolve(headExclusion) },\n    explicitFiles\n  );\n  // A run-wide failure (a missing native addon) invalidates the whole gate: surface it once as\n  // the fatal error (exit 2) instead of diagnosing every changed file as unmeasured.\n  if (scan.fatalError) {\n    throw new Error(scan.fatalError);\n  }\n  const scannedFiles: ScannedFile[] = scan.files.map((file) => ({\n    relativePath: formatPath(file.file, scan.displayRoot),\n    file,\n  }));\n  if (attributesChanged) {\n    changedFiles.push(\n      ...(await listReincludedFiles(scannedFiles, changedFiles, repoRoot, headExclusion, baseExclusion))\n    );\n  }\n\n  // A measurement failure on ANY scannable changed file forces exit 2 — deliberately including\n  // files outside a scoped target, because cross-file function matching and the base duplication\n  // universe for the gated files depend on them. Failures elsewhere (unchanged files) and\n  // unsupported changed paths degrade to warnings.\n  const changedPaths = new Set(\n    changedFiles\n      .flatMap((changed) => [changed.headPath, ...(changed.basePath === undefined ? [] : [changed.basePath])])\n      .filter((changedPath) => isScannedPath(changedPath, options, explicitFiles.has(path.join(repoRoot, changedPath))))\n  );\n  const errors: string[] = [];\n  const warnings = [...scan.warnings];\n  for (const error of scan.errors) {\n    if ([...changedPaths].some((changedPath) => error.startsWith(`${changedPath}:`))) {\n      errors.push(error);\n    } else {\n      warnings.push(error);\n    }\n  }\n\n  const prepared = await prepareChangedFiles(\n    changedFiles,\n    {\n      repoRoot,\n      mergeBase,\n      canonicalTarget,\n      options,\n      scannedFiles,\n      baseSymlinkPaths,\n      scanErrors: [...errors],\n      headExclusion,\n      baseExclusion,\n      explicitFiles,\n      generatedHeadFiles: new Set(scan.generatedFiles),\n    },\n    errors,\n    warnings\n  );\n  // A gate must not fail open on a mistyped target: a nonexistent path is only acceptable when it\n  // still matches changed files (e.g. a fully deleted directory).\n  if (!targetExists && !prepared.some((file) => file.gated)) {\n    throw new Error(`target \"${target}\" does not exist and matches no changed file`);\n  }\n\n  // Non-gated files (outside the target, or renamed out of scan scope) still feed function\n  // matching and the duplication universes; the evaluator reports nothing for them.\n  const { baseCross, headCross } = measureDuplicationUniverses(prepared, scannedFiles, options);\n  const renamedPaths = new Map(\n    prepared.flatMap(({ changed }) =>\n      changed.status === 'renamed' && changed.basePath !== undefined\n        ? [[changed.basePath, changed.headPath] as const]\n        : []\n    )\n  );\n  const headPathOf = (basePath: string): string => renamedPaths.get(basePath) ?? basePath;\n  const lineSignaturesOf = createLineSignatureIndex(scannedFiles);\n  const newlyDuplicatedLines = findNewlyDuplicatedLines(\n    prepared.map((file) => toDuplicationChange(file, baseCross, headCross, headPathOf, lineSignaturesOf))\n  );\n  const inputs = prepared.map((file, index) => toGateInput(file, newlyDuplicatedLines[index] ?? [], headCross));\n  const result = evaluateRegressionGate(inputs, gateOptions);\n\n  if (cliOptions.json) {\n    printJsonReport(cliOptions, mergeBase, result, inputs, errors, warnings);\n  } else {\n    printTextReport(cliOptions, mergeBase, result, errors, warnings);\n  }\n\n  if (errors.length > 0) {\n    process.exitCode = 2;\n  } else if (result.violations.length > 0) {\n    process.exitCode = 1;\n  }\n}\n\n/** Content with a generated-code marker, to ask whether an exclusion honors the marker for a path. */\nconst markedCode = '/* @generated */';\n\n/**\n * Unchanged files the head scan measured but the merge-base would not have: excluded by its\n * attributes, or generated by content without the `-linguist-generated` override the working tree\n * adds. Their code is newly measured, so it gates as a modification from an unmeasurable base. The\n * head side comes from the scan, which already applied scope, symbolic-link, and exclusion rules.\n */\nasync function listReincludedFiles(\n  scannedFiles: readonly ScannedFile[],\n  changedFiles: readonly ChangedFile[],\n  repoRoot: string,\n  headExclusion: Exclusion,\n  baseExclusion: Exclusion\n): Promise<ChangedFile[]> {\n  const changedPaths = new Set(changedFiles.flatMap((changed) => [changed.headPath, changed.basePath]));\n  const reincluded: ChangedFile[] = [];\n  for (const { relativePath, file } of scannedFiles) {\n    const absolutePath = path.join(repoRoot, relativePath);\n    if (changedPaths.has(relativePath)) {\n      continue;\n    }\n    // The same content is generated at one revision only when its override differs, so the\n    // content is read just for files the working tree newly marks -linguist-generated.\n    const overrideAdded =\n      baseExclusion.isGeneratedCode(absolutePath, markedCode) &&\n      !headExclusion.isGeneratedCode(absolutePath, markedCode);\n    if (\n      baseExclusion.isExcludedPath(absolutePath) ||\n      (overrideAdded && baseExclusion.isGeneratedCode(absolutePath, await readFile(file.file, 'utf8')))\n    ) {\n      reincluded.push({ status: 'modified', basePath: relativePath, headPath: relativePath });\n    }\n  }\n  return reincluded;\n}\n\n/**\n * An explicitly targeted file is measured even when excluded, like the ranking command's single\n * file target: the file itself, and its base path when it was renamed.\n */\nfunction listExplicitlyTargetedPaths(\n  targetFile: string | undefined,\n  repoRoot: string,\n  changedFiles: ChangedFile[]\n): Set<string> {\n  if (targetFile === undefined) {\n    return new Set();\n  }\n  const paths = new Set([targetFile]);\n  for (const changed of changedFiles) {\n    if (changed.basePath !== undefined && path.join(repoRoot, changed.headPath) === targetFile) {\n      paths.add(path.join(repoRoot, changed.basePath));\n    }\n  }\n  return paths;\n}\n\n/** The target may not exist (e.g. only deleted files under it); fall back to the resolved path. */\nasync function canonicalizeTarget(resolvedTarget: string): Promise<{ canonicalTarget: string; targetExists: boolean }> {\n  try {\n    return { canonicalTarget: await realpath(resolvedTarget), targetExists: true };\n  } catch {\n    return { canonicalTarget: resolvedTarget, targetExists: false };\n  }\n}\n\ninterface GateContext {\n  repoRoot: string;\n  mergeBase: string;\n  canonicalTarget: string;\n  options: ResolvedOptions;\n  scannedFiles: ScannedFile[];\n  /** Paths that are symbolic links at the merge-base; like head symlinks, they are not gated. */\n  baseSymlinkPaths: Set<string>;\n  /** Errors the head scan recorded against changed files. */\n  scanErrors: readonly string[];\n  headExclusion: Exclusion;\n  baseExclusion: Exclusion;\n  /** The explicitly targeted file (and its renamed base path), measured despite every exclusion. */\n  explicitFiles: ReadonlySet<string>;\n  /** Absolute paths the head scan skipped as generated code. */\n  generatedHeadFiles: ReadonlySet<string>;\n}\n\nasync function prepareChangedFiles(\n  changedFiles: ChangedFile[],\n  context: GateContext,\n  errors: string[],\n  warnings: string[]\n): Promise<PreparedFile[]> {\n  const headByPath = new Map(context.scannedFiles.map(({ relativePath, file }) => [relativePath, file]));\n  // Files are prepared concurrently (git reads and base measurements); each records its diagnostics\n  // on its own, and they are appended in changed-file order so the report stays deterministic.\n  const results = await mapConcurrently(changedFiles, os.availableParallelism() * 2, async (changed) => {\n    const fileErrors: string[] = [];\n    const fileWarnings: string[] = [];\n    const file = await prepareChangedFile(changed, context, headByPath, fileErrors, fileWarnings);\n    return { file, fileErrors, fileWarnings };\n  });\n  const prepared: PreparedFile[] = [];\n  for (const { file, fileErrors, fileWarnings } of results) {\n    errors.push(...fileErrors);\n    warnings.push(...fileWarnings);\n    if (file) {\n      prepared.push(file);\n    }\n  }\n  return prepared;\n}\n\n/** `Promise.all(items.map(map))` with at most `limit` calls pending at once, results in input order. */\nasync function mapConcurrently<T, R>(items: readonly T[], limit: number, map: (item: T) => Promise<R>): Promise<R[]> {\n  const results: R[] = [];\n  let nextIndex = 0;\n  const work = async (): Promise<void> => {\n    while (nextIndex < items.length) {\n      const index = nextIndex;\n      nextIndex += 1;\n      results[index] = await map(items[index] as T);\n    }\n  };\n  await Promise.all(Array.from({ length: Math.min(limit, items.length) }, work));\n  return results;\n}\n\nasync function prepareChangedFile(\n  changed: ChangedFile,\n  context: GateContext,\n  headByPath: Map<string, FileMetrics>,\n  errors: string[],\n  warnings: string[]\n): Promise<PreparedFile | undefined> {\n  // Symbolic links are skipped on both sides, mirroring scanListedFiles: git stores only the\n  // target string, so a symlink blob is not measurable source.\n  const absoluteHeadPath = path.join(context.repoRoot, changed.headPath);\n  const headScannable =\n    changed.status !== 'deleted' &&\n    isScannedPath(changed.headPath, context.options, context.explicitFiles.has(absoluteHeadPath)) &&\n    !context.headExclusion.isExcludedPath(absoluteHeadPath) &&\n    !context.generatedHeadFiles.has(absoluteHeadPath) &&\n    !(await isSymbolicLink(absoluteHeadPath));\n  // A base path outside the scan scope (renamed from a test/ignored directory, or an unsupported\n  // extension) was never measurable code: its content gates as new code instead of ratcheting\n  // against a blob the scanner would not have measured.\n  const absoluteBasePath = changed.basePath === undefined ? '' : path.join(context.repoRoot, changed.basePath);\n  const baseScannable =\n    changed.basePath !== undefined &&\n    isScannedPath(changed.basePath, context.options, context.explicitFiles.has(absoluteBasePath)) &&\n    !context.baseExclusion.isExcludedPath(absoluteBasePath) &&\n    !context.baseSymlinkPaths.has(changed.basePath);\n  if (!headScannable && !baseScannable) {\n    return undefined;\n  }\n\n  const displayFile = changed.status === 'deleted' ? (changed.basePath as string) : changed.headPath;\n  const headFile = headScannable ? headByPath.get(changed.headPath) : undefined;\n  if (headScannable && !headFile) {\n    reportUnmeasuredChangedFile(changed.headPath, context.scanErrors, errors);\n    return undefined;\n  }\n\n  const file: PreparedFile = {\n    changed,\n    displayFile,\n    // A file whose head left the scan scope still contributes its base functions to matching and\n    // its base blob to the base universe, but nothing about it is gated or reported.\n    gated:\n      headScannable || changed.status === 'deleted'\n        ? isWithinTarget(path.join(context.repoRoot, displayFile), context.canonicalTarget)\n        : false,\n    headFile,\n  };\n\n  if (baseScannable) {\n    const outcome = await measureBaseRevision(file, changed.basePath as string, context, errors, warnings);\n    // A base that turns out to be generated code gates its head as new code, like any other base\n    // outside the scan scope.\n    if (outcome === 'failed' || (outcome === 'generated' && !headFile)) {\n      return undefined;\n    }\n  }\n\n  try {\n    if (headFile) {\n      // The duplication gate needs the head lines, so an unreadable head fails like an unmeasured one.\n      file.headContent = await readFile(headFile.file, 'utf8');\n      collectHeadFunctionTokens(file, file.headContent, context, warnings);\n    }\n    file.hunks = await diffMeasuredRevisions(file, context);\n  } catch (error) {\n    errors.push(`${displayFile}: ${formatError(error)}`);\n    return undefined;\n  }\n  return file;\n}\n\nasync function diffMeasuredRevisions(file: PreparedFile, context: GateContext): Promise<LineHunk[]> {\n  const { changed, baseContent, headContent } = file;\n  if (baseContent !== undefined && headContent !== undefined) {\n    return await listLineHunks(context.repoRoot, context.mergeBase, changed.basePath as string, changed.headPath);\n  }\n  return [\n    {\n      baseStart: 1,\n      baseCount: baseContent === undefined ? 0 : splitLines(baseContent).length,\n      headStart: 1,\n      headCount: headContent === undefined ? 0 : splitLines(headContent).length,\n    },\n  ];\n}\n\n/** Line numbering of the measured metrics (classifyLines splits on every line terminator). */\nfunction splitLines(content: string): string[] {\n  return content.split(/\\r\\n|\\n|\\r/u);\n}\n\n/**\n * The scan covers exactly the git-visible list, so a scannable changed path can only be missing\n * after a measurement failure (already recorded as an error) or a silent exclusion (an alias of\n * an already-visited file, or absence from the git list). Failing loudly keeps the gate from\n * passing with the file unchecked.\n */\nfunction reportUnmeasuredChangedFile(headPath: string, scanErrors: readonly string[], errors: string[]): void {\n  if (!scanErrors.some((error) => error.startsWith(`${headPath}:`))) {\n    errors.push(`${headPath}: changed file was not measured`);\n  }\n}\n\nfunction collectHeadFunctionTokens(\n  file: PreparedFile,\n  headContent: string,\n  context: GateContext,\n  warnings: string[]\n): void {\n  try {\n    file.headFunctionTokens = collectFunctionTokenSequences(headContent, {\n      language: languageOf(file.changed.headPath, context),\n      duplication: context.options.duplication,\n    });\n  } catch (error) {\n    // Only rename re-matching degrades without token sequences; the head metrics still gate.\n    warnings.push(`${file.displayFile}: function token sequences unavailable: ${formatError(error)}`);\n  }\n}\n\n/**\n * Measures the merge-base blob into `file`; 'failed' (with an error recorded) only when the\n * metrics themselves cannot be measured, and 'generated' (nothing measured) for generated code.\n * The auxiliary collections (duplication candidates, token sequences) may fail independently of\n * the metrics, so their failure only degrades duplication data and rename re-matching — the\n * function-level ratchets still run.\n */\nasync function measureBaseRevision(\n  file: PreparedFile,\n  basePath: string,\n  context: GateContext,\n  errors: string[],\n  warnings: string[]\n): Promise<'measured' | 'generated' | 'failed'> {\n  const measureOptions = {\n    language: languageOf(basePath, context),\n    duplication: context.options.duplication,\n  };\n  let baseContent;\n  try {\n    baseContent = await readFileAtRevision(context.repoRoot, context.mergeBase, basePath);\n    if (context.baseExclusion.isGeneratedCode(path.join(context.repoRoot, basePath), baseContent)) {\n      return 'generated';\n    }\n    const measured = await measureWithCrossFileData(baseContent, measureOptions);\n    file.baseMetrics = measured.metrics;\n    file.baseContent = baseContent;\n    file.baseCandidates = measured.crossFileData;\n    if (measured.crossFileError !== undefined) {\n      warnings.push(`${basePath} (at merge-base): duplication candidates unavailable: ${measured.crossFileError}`);\n    }\n  } catch (error) {\n    errors.push(`${basePath} (at merge-base): ${formatError(error)}`);\n    return 'failed';\n  }\n  try {\n    file.baseFunctionTokens = collectFunctionTokenSequences(baseContent, measureOptions);\n  } catch (error) {\n    warnings.push(`${basePath} (at merge-base): function token sequences unavailable: ${formatError(error)}`);\n  }\n  return 'measured';\n}\n\n/** The language of a scannable path; an explicitly targeted file skips the test-file name rules. */\nfunction languageOf(relativePath: string, context: GateContext): LanguageName {\n  const explicitTarget = context.explicitFiles.has(path.join(context.repoRoot, relativePath));\n  return getLanguage(relativePath, context.options, explicitTarget) as LanguageName;\n}\n\nasync function isSymbolicLink(absolutePath: string): Promise<boolean> {\n  const stats = await lstat(absolutePath).catch(() => {});\n  return stats?.isSymbolicLink() ?? false;\n}\n\n/** Walks up to the nearest existing DIRECTORY, so git commands never spawn in a missing or non-directory cwd. */\nasync function firstExistingDirectory(directory: string): Promise<string> {\n  let current = directory;\n  while (true) {\n    const stats = await stat(current).catch(() => {});\n    if (stats?.isDirectory()) {\n      return current;\n    }\n    const parent = path.dirname(current);\n    if (parent === current) {\n      return current;\n    }\n    current = parent;\n  }\n}\n\nfunction isWithinTarget(candidate: string, targetDirectory: string): boolean {\n  const relative = path.relative(targetDirectory, candidate);\n  return relative === '' || (!relative.startsWith(`..${path.sep}`) && relative !== '..' && !path.isAbsolute(relative));\n}\n\nfunction measureDuplicationUniverses(\n  prepared: PreparedFile[],\n  scannedFiles: ScannedFile[],\n  options: ResolvedOptions\n): { baseCross?: CrossFileDuplicationMetrics; headCross?: CrossFileDuplicationMetrics } {\n  const headSources = scannedFiles.flatMap(({ relativePath, file }) =>\n    file.duplicationCandidates ? [{ file: relativePath, ...file.duplicationCandidates }] : []\n  );\n\n  const changedHeadPaths = new Set(\n    prepared.flatMap((file) => (file.changed.status === 'deleted' ? [] : [file.changed.headPath]))\n  );\n  const baseSources = headSources.filter((source) => !changedHeadPaths.has(source.file));\n  for (const file of prepared) {\n    if (file.baseCandidates && file.changed.basePath !== undefined) {\n      baseSources.push({ file: file.changed.basePath, ...file.baseCandidates });\n    }\n  }\n\n  return {\n    baseCross: baseSources.length >= 2 ? measureCrossFileDuplication(baseSources, options.duplication) : undefined,\n    headCross: headSources.length >= 2 ? measureCrossFileDuplication(headSources, options.duplication) : undefined,\n  };\n}\n\nfunction toDuplicationChange(\n  file: PreparedFile,\n  baseCross: CrossFileDuplicationMetrics | undefined,\n  headCross: CrossFileDuplicationMetrics | undefined,\n  headPathOf: (basePath: string) => string,\n  lineSignaturesOf: (file: string) => LineSignatures | undefined\n): DuplicationChange {\n  const headOccurrences = collectOccurrences(file.headFile?.metrics, headCross, file.changed.headPath);\n  return {\n    baseLines: file.baseContent === undefined ? undefined : splitLines(file.baseContent),\n    headLines: file.headContent === undefined ? undefined : splitLines(file.headContent),\n    baseDuplicatedLines:\n      file.baseMetrics === undefined || file.changed.basePath === undefined\n        ? new Set()\n        : collectDuplicatedLineNumbers(file.baseMetrics, baseCross, file.changed.basePath),\n    headDuplicatedLines:\n      file.headFile === undefined\n        ? new Set()\n        : keepRepeatedLines(\n            collectDuplicatedLineNumbers(file.headFile.metrics, headCross, file.changed.headPath),\n            file.changed.headPath,\n            headOccurrences,\n            lineSignaturesOf\n          ),\n    baseOccurrences: collectOccurrences(file.baseMetrics, baseCross, file.changed.basePath, headPathOf),\n    headOccurrences,\n    hunks: file.hunks ?? [],\n  };\n}\n\n/** A file's per-line normalized token signatures and how many lines carry each. */\ninterface LineSignatures {\n  byLine: Map<number, string>;\n  lineCounts: Map<string, number>;\n}\n\n/** Line signatures of the scanned head files, computed on first use. */\nfunction createLineSignatureIndex(scannedFiles: ScannedFile[]): (file: string) => LineSignatures | undefined {\n  const tokensByFile = new Map(\n    scannedFiles.map(({ relativePath, file }) => [relativePath, file.duplicationCandidates?.tokens])\n  );\n  const signaturesByFile = new Map<string, LineSignatures | undefined>();\n  return (file) => {\n    if (!signaturesByFile.has(file)) {\n      const tokens = tokensByFile.get(file);\n      signaturesByFile.set(file, tokens && signLines(tokens));\n    }\n    return signaturesByFile.get(file);\n  };\n}\n\n/**\n * Each line's tokens with identifiers numbered by first occurrence, as clone detection anonymizes\n * them.\n */\nfunction signLines(tokens: readonly Token[]): LineSignatures {\n  const byLine = new Map<number, string>();\n  const identifierIndexesByLine = new Map<number, Map<string, number>>();\n  for (const token of tokens) {\n    const line = token.startRow + 1;\n    let text = token.text;\n    if (token.kind === 'id') {\n      // Numbered by first occurrence within the line, so `a + a` and `a + b` stay distinct.\n      const identifierIndexes = identifierIndexesByLine.get(line) ?? new Map<string, number>();\n      identifierIndexesByLine.set(line, identifierIndexes);\n      if (!identifierIndexes.has(token.text)) {\n        identifierIndexes.set(token.text, identifierIndexes.size);\n      }\n      text = `\\u0001${identifierIndexes.get(token.text)}`;\n    }\n    byLine.set(line, `${byLine.get(line) ?? ''}${text}\\u0000`);\n  }\n  const lineCounts = new Map<string, number>();\n  for (const signature of byLine.values()) {\n    lineCounts.set(signature, (lineCounts.get(signature) ?? 0) + 1);\n  }\n  return { byLine, lineCounts };\n}\n\n/**\n * The duplicated lines whose normalized tokens a partner file of their clones (the file itself,\n * on another line, for `''`) repeats on some line. Near-miss coverage also marks lines the copies\n * do not share, such as a new line inserted into an old near-miss clone, which copied nothing.\n */\nfunction keepRepeatedLines(\n  lines: ReadonlySet<number>,\n  file: string,\n  occurrences: readonly CloneOccurrence[],\n  lineSignaturesOf: (file: string) => LineSignatures | undefined\n): Set<number> {\n  const own = lineSignaturesOf(file);\n  if (own === undefined) {\n    return new Set(lines);\n  }\n  const repeats = (signature: string, partner: string): boolean => {\n    if (partner === '') {\n      return (own.lineCounts.get(signature) ?? 0) > 1;\n    }\n    const partnerSignatures = lineSignaturesOf(partner);\n    return partnerSignatures === undefined || partnerSignatures.lineCounts.has(signature);\n  };\n  const partnersByLine = indexPartnersByLine(occurrences);\n  return new Set(\n    [...lines].filter((line) => {\n      const signature = own.byLine.get(line);\n      return (\n        signature === undefined || [...(partnersByLine.get(line) ?? [])].some((partner) => repeats(signature, partner))\n      );\n    })\n  );\n}\n\n/**\n * The file's clone occurrences with their partner files (`''` for the file itself), partner paths\n * renamed by `renamePartner` so base partners compare with head ones.\n */\nfunction collectOccurrences(\n  metrics: CodeMetrics | undefined,\n  cross: CrossFileDuplicationMetrics | undefined,\n  file: string | undefined,\n  renamePartner: (partner: string) => string = (partner) => partner\n): CloneOccurrence[] {\n  if (metrics === undefined || file === undefined) {\n    return [];\n  }\n  const withinFile = metrics.duplication.duplicateBlockGroups.flatMap((group) =>\n    group.map(({ startLine, endLine }) => ({ startLine, endLine, partners: [''] }))\n  );\n  const crossFile = (cross?.groups ?? []).flatMap((group) => {\n    const own = group.occurrences.filter((occurrence) => occurrence.file === file);\n    const partners = [\n      ...group.files.filter((partner) => partner !== file).map(renamePartner),\n      ...(own.length > 1 ? [''] : []),\n    ];\n    return own.map(({ startLine, endLine }) => ({ startLine, endLine, partners }));\n  });\n  return [...withinFile, ...crossFile];\n}\n\nfunction toGateInput(\n  file: PreparedFile,\n  newlyDuplicatedLines: number[],\n  headCross: CrossFileDuplicationMetrics | undefined\n): GateFileInput {\n  return {\n    file: file.displayFile,\n    baseMetrics: file.baseMetrics,\n    headMetrics: file.headFile?.metrics,\n    baseFunctionTokens: file.baseFunctionTokens,\n    headFunctionTokens: file.headFunctionTokens,\n    newlyDuplicatedLines,\n    duplicationPartners: collectPartners(file, newlyDuplicatedLines, headCross),\n    gated: file.gated,\n  };\n}\n\n/** Other files sharing a cross-file clone that covers the given lines. */\nfunction collectPartners(\n  file: PreparedFile,\n  lines: number[],\n  headCross: CrossFileDuplicationMetrics | undefined\n): string[] {\n  const headPath = file.changed.headPath;\n  const covers = (occurrence: CrossFileDuplicateOccurrence): boolean =>\n    occurrence.file === headPath && lines.some((line) => line >= occurrence.startLine && line <= occurrence.endLine);\n  const partners = new Set<string>();\n  for (const group of headCross?.groups ?? []) {\n    if (group.occurrences.some(covers)) {\n      for (const partner of group.files) {\n        if (partner !== headPath) {\n          partners.add(partner);\n        }\n      }\n    }\n  }\n  return [...partners].toSorted();\n}\n\nfunction printTextReport(\n  cliOptions: DiffCliOptions,\n  mergeBase: string,\n  result: GateResult,\n  errors: string[],\n  warnings: string[]\n): void {\n  const shortBase = mergeBase.slice(0, 12);\n  if (errors.length > 0) {\n    // Unmeasured files were not gated, so \"0 violations\" would be vacuous; never claim a pass.\n    writeStdout(\n      `Regression gate could not complete: ${errors.length} measurement failures (details on stderr)` +\n        `${result.violations.length > 0 ? `; ${result.violations.length} violations in the measured files` : ''} (base ${cliOptions.base}, merge-base ${shortBase}).\\n`\n    );\n    printViolations(result);\n  } else if (result.violations.length === 0) {\n    writeStdout(\n      `Regression gate passed: ${result.checkedFileCount} changed files, ${result.checkedFunctionCount} functions checked (base ${cliOptions.base}, merge-base ${shortBase}).\\n`\n    );\n  } else {\n    writeStdout(\n      `Regression gate vs ${cliOptions.base} (merge-base ${shortBase}): ${result.violations.length} violations\\n`\n    );\n    printViolations(result);\n  }\n\n  if (cliOptions.full) {\n    printFullDetails(result);\n  }\n\n  for (const warning of warnings) {\n    writeStderr(`Warning: ${warning}\\n`);\n  }\n  for (const error of errors) {\n    writeStderr(`Error: ${error}\\n`);\n  }\n}\n\nfunction printViolations(result: GateResult): void {\n  for (const [index, violation] of result.violations.entries()) {\n    writeStdout(`${index + 1}. ${violation.message}\\n`);\n  }\n}\n\n/** Base -> head values of every checked function; kept behind --full for humans and trending. */\nfunction printFullDetails(result: GateResult): void {\n  if (result.checkedFunctions.length === 0) {\n    return;\n  }\n  writeStdout('\\nChecked functions (base -> head):\\n');\n  for (const report of result.checkedFunctions) {\n    writeStdout(`- ${formatFunctionReport(report)}\\n`);\n  }\n}\n\nfunction formatFunctionReport(report: CheckedFunctionReport): string {\n  const range = (\n    select: (values: GateFunctionValues) => number,\n    format: (value: number) => string = String\n  ): string => {\n    const head = format(select(report.head));\n    return report.base ? `${format(select(report.base))} -> ${head}` : head;\n  };\n  const values = [\n    `cognitive ${range((fn) => fn.cognitiveComplexity)}`,\n    `NCSS ${range((fn) => fn.ncss)}`,\n    `nesting ${range((fn) => fn.nestingDepth)}`,\n    `DepDegree ${range((fn) => fn.depDegree)}`,\n    `volume ${range(\n      (fn) => fn.halsteadVolume,\n      (value) => value.toFixed(1)\n    )}`,\n  ];\n  return `${report.file}:${report.startLine}-${report.endLine} ${report.name}${report.base ? '' : ' (new)'}: ${values.join(', ')}`;\n}\n\nfunction printJsonReport(\n  cliOptions: DiffCliOptions,\n  mergeBase: string,\n  result: GateResult,\n  inputs: GateFileInput[],\n  errors: string[],\n  warnings: string[]\n): void {\n  const report: Record<string, unknown> = {\n    base: cliOptions.base,\n    mergeBase,\n    passed: result.violations.length === 0 && errors.length === 0,\n    violations: result.violations,\n    checkedFileCount: result.checkedFileCount,\n    checkedFunctionCount: result.checkedFunctionCount,\n    newFunctionCount: result.newFunctionCount,\n    errors,\n    warnings,\n  };\n  if (cliOptions.full) {\n    report.files = inputs\n      .filter((input) => input.gated !== false)\n      .map((input) => ({\n        file: input.file,\n        baseFunctionCount: input.baseMetrics?.functions.length ?? 0,\n        headFunctionCount: input.headMetrics?.functions.length ?? 0,\n        baseNcss: input.baseMetrics?.ncssCount ?? 0,\n        headNcss: input.headMetrics?.ncssCount ?? 0,\n        baseMaxCognitiveComplexity: input.baseMetrics?.maxCognitiveComplexity ?? 0,\n        headMaxCognitiveComplexity: input.headMetrics?.maxCognitiveComplexity ?? 0,\n        newlyDuplicatedLines: input.newlyDuplicatedLines,\n        duplicationPartners: input.duplicationPartners,\n        functions: result.checkedFunctions.filter((fn) => fn.file === input.file),\n      }));\n  }\n  writeStdout(JSON.stringify(report, undefined, 2) + '\\n');\n}\n"],"mappings":"w/BAgGA,eAAsB,EAAe,EAAgB,EAA2C,CAC9F,GAAI,CACF,MAAM,EAAQ,EAAQ,CAAU,CAClC,OAAS,EAAO,CACd,EAAY,UAAU,EAAY,CAAK,EAAE,GAAG,EAC5C,QAAQ,SAAW,CACrB,CACF,CAEA,eAAe,EAAQ,EAAgB,EAA2C,CAChF,IAAM,EAAiB,GAAc,CAAM,EACrC,EAAS,MAAM,EAAW,EAAW,OAAQ,MAAM,EAAsB,CAAc,CAAC,EACxF,EAAU,EAAe,EAAY,CAAM,EAC3C,EAAc,EAAmB,CAAM,EAIvC,EAAW,MAAM,EACrB,MAAM,EAAgB,MAAM,GAAuB,MAAM,EAAsB,CAAc,CAAC,CAAC,CACjG,EACM,EAAY,MAAM,EAAiB,EAAU,EAAW,IAAI,EAC5D,EAAe,MAAM,EAAiB,EAAU,CAAS,EASzD,EAAkB,MAAM,EAAoB,CAAQ,EACpD,EAAmB,MAAM,EAA2B,EAAU,CAAS,EACvE,CAAE,kBAAiB,gBAAiB,MAAM,GAAmB,CAAc,EAE3E,GAAa,MADM,EAAK,CAAe,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EAC9B,OAAO,EAAI,EAAkB,IAAA,GAEtD,EAAiB,GAAc,EAAK,SAAS,EAAU,CAAU,CAAC,CAAC,MAAM,EAAK,GAAG,CAAC,CAAC,KAAK,GAAG,EAC7F,IAAmB,IAAA,IAAa,CAAC,EAAe,WAAW,KAAK,GAAK,CAAC,EAAgB,IAAI,CAAc,IAC1G,EAAgB,IAAI,CAAc,EAClC,EAAa,KAAK,CAAE,OAAQ,QAAS,SAAU,CAAe,CAAC,GAEjE,IAAM,EAAgB,GAA4B,EAAY,EAAU,CAAY,EAI9E,EAAoB,EAAa,KAAM,GAC3C,CAAC,EAAQ,SAAU,EAAQ,QAAQ,CAAC,CAAC,KAClC,GAAS,IAAS,IAAA,IAAa,EAAK,MAAM,SAAS,CAAI,IAAM,gBAChE,CACF,EACM,EAAY,EAAa,QAAS,GAAa,EAAQ,WAAa,IAAA,GAAY,CAAC,EAAI,CAAC,EAAQ,QAAQ,CAAE,EACxG,CAAC,EAAyB,GAA2B,MAAM,QAAQ,IAAI,CAC3E,EAAwB,EAAU,EAAiB,EAAQ,OAAO,EAClE,EACE,EACA,EAAoB,CAAC,mBAAG,IAAI,IAAI,CAAC,GAAG,EAAW,GAAG,CAAe,CAAC,CAAC,EAAI,EACvE,EAAQ,QACR,CACF,CACF,CAAC,EACK,EAAgB,EAAU,EAAyB,CAAa,EAChE,EAAgB,EAAU,EAAyB,CAAa,EAChE,EAAO,MAAM,GACjB,EACA,EACA,CAAE,GAAG,EAAS,kBAAqB,QAAQ,QAAQ,CAAa,CAAE,EAClE,CACF,EAGA,GAAI,EAAK,WACP,MAAU,MAAM,EAAK,UAAU,EAEjC,IAAM,EAA8B,EAAK,MAAM,IAAK,IAAU,CAC5D,aAAc,GAAW,EAAK,KAAM,EAAK,WAAW,EACpD,MACF,EAAE,EACE,GACF,EAAa,KACX,GAAI,MAAM,GAAoB,EAAc,EAAc,EAAU,EAAe,CAAa,CAClG,EAOF,IAAM,EAAe,IAAI,IACvB,EACG,QAAS,GAAY,CAAC,EAAQ,SAAU,GAAI,EAAQ,WAAa,IAAA,GAAY,CAAC,EAAI,CAAC,EAAQ,QAAQ,CAAE,CAAC,CAAC,CACvG,OAAQ,GAAgB,EAAc,EAAa,EAAS,EAAc,IAAI,EAAK,KAAK,EAAU,CAAW,CAAC,CAAC,CAAC,CACrH,EACM,EAAmB,CAAC,EACpB,EAAW,CAAC,GAAG,EAAK,QAAQ,EAClC,IAAK,IAAM,KAAS,EAAK,OACnB,CAAC,GAAG,CAAY,CAAC,CAAC,KAAM,GAAgB,EAAM,WAAW,GAAG,EAAY,EAAE,CAAC,EAC7E,EAAO,KAAK,CAAK,EAEjB,EAAS,KAAK,CAAK,EAIvB,IAAM,EAAW,MAAM,EACrB,EACA,CACE,WACA,YACA,kBACA,UACA,eACA,mBACA,WAAY,CAAC,GAAG,CAAM,EACtB,gBACA,gBACA,gBACA,mBAAoB,IAAI,IAAI,EAAK,cAAc,CACjD,EACA,EACA,CACF,EAGA,GAAI,CAAC,GAAgB,CAAC,EAAS,KAAM,GAAS,EAAK,KAAK,EACtD,MAAU,MAAM,WAAW,EAAO,6CAA6C,EAKjF,GAAM,CAAE,YAAW,aAAc,GAA4B,EAAU,EAAc,CAAO,EACtF,EAAe,IAAI,IACvB,EAAS,SAAS,CAAE,aAClB,EAAQ,SAAW,WAAa,EAAQ,WAAa,IAAA,GACjD,CAAC,CAAC,EAAQ,SAAU,EAAQ,QAAQ,CAAU,EAC9C,CAAC,CACP,CACF,EACM,EAAc,GAA6B,EAAa,IAAI,CAAQ,GAAK,EACzE,EAAmB,GAAyB,CAAY,EACxD,EAAuB,EAC3B,EAAS,IAAK,GAAS,GAAoB,EAAM,EAAW,EAAW,EAAY,CAAgB,CAAC,CACtG,EACM,EAAS,EAAS,KAAK,EAAM,IAAU,GAAY,EAAM,EAAqB,IAAU,CAAC,EAAG,CAAS,CAAC,EACtG,EAAS,EAAuB,EAAQ,CAAW,EAErD,EAAW,KACb,GAAgB,EAAY,EAAW,EAAQ,EAAQ,EAAQ,CAAQ,EAEvE,GAAgB,EAAY,EAAW,EAAQ,EAAQ,CAAQ,EAG7D,EAAO,OAAS,EAClB,QAAQ,SAAW,EACV,EAAO,WAAW,OAAS,IACpC,QAAQ,SAAW,EAEvB,CAGA,MAAM,EAAa,mBAQnB,eAAe,GACb,EACA,EACA,EACA,EACA,EACwB,CACxB,IAAM,EAAe,IAAI,IAAI,EAAa,QAAS,GAAY,CAAC,EAAQ,SAAU,EAAQ,QAAQ,CAAC,CAAC,EAC9F,EAA4B,CAAC,EACnC,IAAK,GAAM,CAAE,eAAc,UAAU,EAAc,CACjD,IAAM,EAAe,EAAK,KAAK,EAAU,CAAY,EACrD,GAAI,EAAa,IAAI,CAAY,EAC/B,SAIF,IAAM,EACJ,EAAc,gBAAgB,EAAc,CAAU,GACtD,CAAC,EAAc,gBAAgB,EAAc,CAAU,GAEvD,EAAc,eAAe,CAAY,GACxC,GAAiB,EAAc,gBAAgB,EAAc,MAAM,EAAS,EAAK,KAAM,MAAM,CAAC,IAE/F,EAAW,KAAK,CAAE,OAAQ,WAAY,SAAU,EAAc,SAAU,CAAa,CAAC,CAE1F,CACA,OAAO,CACT,CAMA,SAAS,GACP,EACA,EACA,EACa,CACb,GAAI,IAAe,IAAA,GACjB,sBAAO,IAAI,IAEb,IAAM,kBAAQ,IAAI,IAAI,CAAC,CAAU,CAAC,EAClC,IAAK,IAAM,KAAW,EAChB,EAAQ,WAAa,IAAA,IAAa,EAAK,KAAK,EAAU,EAAQ,QAAQ,IAAM,GAC9E,EAAM,IAAI,EAAK,KAAK,EAAU,EAAQ,QAAQ,CAAC,EAGnD,OAAO,CACT,CAGA,eAAe,GAAmB,EAAqF,CACrH,GAAI,CACF,MAAO,CAAE,gBAAiB,MAAM,EAAS,CAAc,EAAG,aAAc,EAAK,CAC/E,MAAQ,CACN,MAAO,CAAE,gBAAiB,EAAgB,aAAc,EAAM,CAChE,CACF,CAoBA,eAAe,EACb,EACA,EACA,EACA,EACyB,CACzB,IAAM,EAAa,IAAI,IAAI,EAAQ,aAAa,KAAK,CAAE,eAAc,UAAW,CAAC,EAAc,CAAI,CAAC,CAAC,EAG/F,EAAU,MAAM,EAAgB,EAAc,EAAG,qBAAqB,EAAI,EAAG,KAAO,IAAY,CACpG,IAAM,EAAuB,CAAC,EACxB,EAAyB,CAAC,EAEhC,MAAO,CAAE,KAAA,MADU,EAAmB,EAAS,EAAS,EAAY,EAAY,CAAY,EAC7E,aAAY,cAAa,CAC1C,CAAC,EACK,EAA2B,CAAC,EAClC,IAAK,GAAM,CAAE,OAAM,aAAY,kBAAkB,EAC/C,EAAO,KAAK,GAAG,CAAU,EACzB,EAAS,KAAK,GAAG,CAAY,EACzB,GACF,EAAS,KAAK,CAAI,EAGtB,OAAO,CACT,CAGA,eAAe,EAAsB,EAAqB,EAAe,EAA4C,CACnH,IAAM,EAAe,CAAC,EAClB,EAAY,EAShB,OADA,MAAM,QAAQ,IAAI,MAAM,KAAK,CAAE,OAAQ,KAAK,IAAI,EAAO,EAAM,MAAM,CAAE,EAAG,SAPhC,CACtC,KAAO,EAAY,EAAM,QAAQ,CAC/B,IAAM,EAAQ,EACd,GAAa,EACb,EAAQ,GAAS,MAAM,EAAI,EAAM,EAAW,CAC9C,CACF,CAC4E,CAAC,EACtE,CACT,CAEA,eAAe,EACb,EACA,EACA,EACA,EACA,EACmC,CAGnC,IAAM,EAAmB,EAAK,KAAK,EAAQ,SAAU,EAAQ,QAAQ,EAC/D,EACJ,EAAQ,SAAW,WACnB,EAAc,EAAQ,SAAU,EAAQ,QAAS,EAAQ,cAAc,IAAI,CAAgB,CAAC,GAC5F,CAAC,EAAQ,cAAc,eAAe,CAAgB,GACtD,CAAC,EAAQ,mBAAmB,IAAI,CAAgB,GAChD,CAAE,MAAM,EAAe,CAAgB,EAInC,EAAmB,EAAQ,WAAa,IAAA,GAAY,GAAK,EAAK,KAAK,EAAQ,SAAU,EAAQ,QAAQ,EACrG,EACJ,EAAQ,WAAa,IAAA,IACrB,EAAc,EAAQ,SAAU,EAAQ,QAAS,EAAQ,cAAc,IAAI,CAAgB,CAAC,GAC5F,CAAC,EAAQ,cAAc,eAAe,CAAgB,GACtD,CAAC,EAAQ,iBAAiB,IAAI,EAAQ,QAAQ,EAChD,GAAI,CAAC,GAAiB,CAAC,EACrB,OAGF,IAAM,EAAc,EAAQ,SAAW,UAAa,EAAQ,SAAsB,EAAQ,SACpF,EAAW,EAAgB,EAAW,IAAI,EAAQ,QAAQ,EAAI,IAAA,GACpE,GAAI,GAAiB,CAAC,EAAU,CAC9B,EAA4B,EAAQ,SAAU,EAAQ,WAAY,CAAM,EACxE,MACF,CAEA,IAAM,EAAqB,CACzB,UACA,cAGA,MACE,GAAiB,EAAQ,SAAW,UAChC,EAAe,EAAK,KAAK,EAAQ,SAAU,CAAW,EAAG,EAAQ,eAAe,EAChF,GACN,UACF,EAEA,GAAI,EAAe,CACjB,IAAM,EAAU,MAAM,EAAoB,EAAM,EAAQ,SAAoB,EAAS,EAAQ,CAAQ,EAGrG,GAAI,IAAY,UAAa,IAAY,aAAe,CAAC,EACvD,MAEJ,CAEA,GAAI,CACE,IAEF,EAAK,YAAc,MAAM,EAAS,EAAS,KAAM,MAAM,EACvD,EAA0B,EAAM,EAAK,YAAa,EAAS,CAAQ,GAErE,EAAK,MAAQ,MAAM,EAAsB,EAAM,CAAO,CACxD,OAAS,EAAO,CACd,EAAO,KAAK,GAAG,EAAY,IAAI,EAAY,CAAK,GAAG,EACnD,MACF,CACA,OAAO,CACT,CAEA,eAAe,EAAsB,EAAoB,EAA2C,CAClG,GAAM,CAAE,UAAS,cAAa,eAAgB,EAI9C,OAHI,IAAgB,IAAA,IAAa,IAAgB,IAAA,GACxC,MAAM,EAAc,EAAQ,SAAU,EAAQ,UAAW,EAAQ,SAAoB,EAAQ,QAAQ,EAEvG,CACL,CACE,UAAW,EACX,UAAW,IAAgB,IAAA,GAAY,EAAI,EAAW,CAAW,CAAC,CAAC,OACnE,UAAW,EACX,UAAW,IAAgB,IAAA,GAAY,EAAI,EAAW,CAAW,CAAC,CAAC,MACrE,CACF,CACF,CAGA,SAAS,EAAW,EAA2B,CAC7C,OAAO,EAAQ,MAAM,aAAa,CACpC,CAQA,SAAS,EAA4B,EAAkB,EAA+B,EAAwB,CACvG,EAAW,KAAM,GAAU,EAAM,WAAW,GAAG,EAAS,EAAE,CAAC,GAC9D,EAAO,KAAK,GAAG,EAAS,gCAAgC,CAE5D,CAEA,SAAS,EACP,EACA,EACA,EACA,EACM,CACN,GAAI,CACF,EAAK,mBAAqB,EAA8B,EAAa,CACnE,SAAU,EAAW,EAAK,QAAQ,SAAU,CAAO,EACnD,YAAa,EAAQ,QAAQ,WAC/B,CAAC,CACH,OAAS,EAAO,CAEd,EAAS,KAAK,GAAG,EAAK,YAAY,0CAA0C,EAAY,CAAK,GAAG,CAClG,CACF,CASA,eAAe,EACb,EACA,EACA,EACA,EACA,EAC8C,CAC9C,IAAM,EAAiB,CACrB,SAAU,EAAW,EAAU,CAAO,EACtC,YAAa,EAAQ,QAAQ,WAC/B,EACI,EACJ,GAAI,CAEF,GADA,EAAc,MAAM,EAAmB,EAAQ,SAAU,EAAQ,UAAW,CAAQ,EAChF,EAAQ,cAAc,gBAAgB,EAAK,KAAK,EAAQ,SAAU,CAAQ,EAAG,CAAW,EAC1F,MAAO,YAET,IAAM,EAAW,MAAM,EAAyB,EAAa,CAAc,EAC3E,EAAK,YAAc,EAAS,QAC5B,EAAK,YAAc,EACnB,EAAK,eAAiB,EAAS,cAC3B,EAAS,iBAAmB,IAAA,IAC9B,EAAS,KAAK,GAAG,EAAS,wDAAwD,EAAS,gBAAgB,CAE/G,OAAS,EAAO,CAEd,OADA,EAAO,KAAK,GAAG,EAAS,oBAAoB,EAAY,CAAK,GAAG,EACzD,QACT,CACA,GAAI,CACF,EAAK,mBAAqB,EAA8B,EAAa,CAAc,CACrF,OAAS,EAAO,CACd,EAAS,KAAK,GAAG,EAAS,0DAA0D,EAAY,CAAK,GAAG,CAC1G,CACA,MAAO,UACT,CAGA,SAAS,EAAW,EAAsB,EAAoC,CAC5E,IAAM,EAAiB,EAAQ,cAAc,IAAI,EAAK,KAAK,EAAQ,SAAU,CAAY,CAAC,EAC1F,OAAO,EAAY,EAAc,EAAQ,QAAS,CAAc,CAClE,CAEA,eAAe,EAAe,EAAwC,CAEpE,OAAO,MADa,EAAM,CAAY,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EACxC,eAAe,GAAK,EACpC,CAGA,eAAe,GAAuB,EAAoC,CACxE,IAAI,EAAU,EACd,OAAa,CAEX,IAAI,MADgB,EAAK,CAAO,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EACrC,YAAY,EACrB,OAAO,EAET,IAAM,EAAS,EAAK,QAAQ,CAAO,EACnC,GAAI,IAAW,EACb,OAAO,EAET,EAAU,CACZ,CACF,CAEA,SAAS,EAAe,EAAmB,EAAkC,CAC3E,IAAM,EAAW,EAAK,SAAS,EAAiB,CAAS,EACzD,OAAO,IAAa,IAAO,CAAC,EAAS,WAAW,KAAK,EAAK,KAAK,GAAK,IAAa,MAAQ,CAAC,EAAK,WAAW,CAAQ,CACpH,CAEA,SAAS,GACP,EACA,EACA,EACsF,CACtF,IAAM,EAAc,EAAa,SAAS,CAAE,eAAc,UACxD,EAAK,sBAAwB,CAAC,CAAE,KAAM,EAAc,GAAG,EAAK,qBAAsB,CAAC,EAAI,CAAC,CAC1F,EAEM,EAAmB,IAAI,IAC3B,EAAS,QAAS,GAAU,EAAK,QAAQ,SAAW,UAAY,CAAC,EAAI,CAAC,EAAK,QAAQ,QAAQ,CAAE,CAC/F,EACM,EAAc,EAAY,OAAQ,GAAW,CAAC,EAAiB,IAAI,EAAO,IAAI,CAAC,EACrF,IAAK,IAAM,KAAQ,EACb,EAAK,gBAAkB,EAAK,QAAQ,WAAa,IAAA,IACnD,EAAY,KAAK,CAAE,KAAM,EAAK,QAAQ,SAAU,GAAG,EAAK,cAAe,CAAC,EAI5E,MAAO,CACL,UAAW,EAAY,QAAU,EAAI,EAA4B,EAAa,EAAQ,WAAW,EAAI,IAAA,GACrG,UAAW,EAAY,QAAU,EAAI,EAA4B,EAAa,EAAQ,WAAW,EAAI,IAAA,EACvG,CACF,CAEA,SAAS,GACP,EACA,EACA,EACA,EACA,EACmB,CACnB,IAAM,EAAkB,EAAmB,EAAK,UAAU,QAAS,EAAW,EAAK,QAAQ,QAAQ,EACnG,MAAO,CACL,UAAW,EAAK,cAAgB,IAAA,GAAY,IAAA,GAAY,EAAW,EAAK,WAAW,EACnF,UAAW,EAAK,cAAgB,IAAA,GAAY,IAAA,GAAY,EAAW,EAAK,WAAW,EACnF,oBACE,EAAK,cAAgB,IAAA,IAAa,EAAK,QAAQ,WAAa,IAAA,mBACxD,IAAI,IACJ,EAA6B,EAAK,YAAa,EAAW,EAAK,QAAQ,QAAQ,EACrF,oBACE,EAAK,WAAa,IAAA,mBACd,IAAI,IACJ,EACE,EAA6B,EAAK,SAAS,QAAS,EAAW,EAAK,QAAQ,QAAQ,EACpF,EAAK,QAAQ,SACb,EACA,CACF,EACN,gBAAiB,EAAmB,EAAK,YAAa,EAAW,EAAK,QAAQ,SAAU,CAAU,EAClG,kBACA,MAAO,EAAK,OAAS,CAAC,CACxB,CACF,CASA,SAAS,GAAyB,EAA2E,CAC3G,IAAM,EAAe,IAAI,IACvB,EAAa,KAAK,CAAE,eAAc,UAAW,CAAC,EAAc,EAAK,uBAAuB,MAAM,CAAC,CACjG,EACM,kBAAmB,IAAI,IAC7B,MAAQ,IAAS,CACf,GAAI,CAAC,EAAiB,IAAI,CAAI,EAAG,CAC/B,IAAM,EAAS,EAAa,IAAI,CAAI,EACpC,EAAiB,IAAI,EAAM,GAAU,EAAU,CAAM,CAAC,CACxD,CACA,OAAO,EAAiB,IAAI,CAAI,CAClC,CACF,CAMA,SAAS,EAAU,EAA0C,CAC3D,IAAM,kBAAS,IAAI,IACb,kBAA0B,IAAI,IACpC,IAAK,IAAM,KAAS,EAAQ,CAC1B,IAAM,EAAO,EAAM,SAAW,EAC1B,EAAO,EAAM,KACjB,GAAI,EAAM,OAAS,KAAM,CAEvB,IAAM,EAAoB,EAAwB,IAAI,CAAI,mBAAK,IAAI,IACnE,EAAwB,IAAI,EAAM,CAAiB,EAC9C,EAAkB,IAAI,EAAM,IAAI,GACnC,EAAkB,IAAI,EAAM,KAAM,EAAkB,IAAI,EAE1D,EAAO,SAAS,EAAkB,IAAI,EAAM,IAAI,GAClD,CACA,EAAO,IAAI,EAAM,GAAG,EAAO,IAAI,CAAI,GAAK,KAAK,EAAK,OAAO,CAC3D,CACA,IAAM,kBAAa,IAAI,IACvB,IAAK,IAAM,KAAa,EAAO,OAAO,EACpC,EAAW,IAAI,GAAY,EAAW,IAAI,CAAS,GAAK,GAAK,CAAC,EAEhE,MAAO,CAAE,SAAQ,YAAW,CAC9B,CAOA,SAAS,EACP,EACA,EACA,EACA,EACa,CACb,IAAM,EAAM,EAAiB,CAAI,EACjC,GAAI,IAAQ,IAAA,GACV,OAAO,IAAI,IAAI,CAAK,EAEtB,IAAM,GAAW,EAAmB,IAA6B,CAC/D,GAAI,IAAY,GACd,OAAQ,EAAI,WAAW,IAAI,CAAS,GAAK,GAAK,EAEhD,IAAM,EAAoB,EAAiB,CAAO,EAClD,OAAO,IAAsB,IAAA,IAAa,EAAkB,WAAW,IAAI,CAAS,CACtF,EACM,EAAiB,EAAoB,CAAW,EACtD,OAAO,IAAI,IACT,CAAC,GAAG,CAAK,CAAC,CAAC,OAAQ,GAAS,CAC1B,IAAM,EAAY,EAAI,OAAO,IAAI,CAAI,EACrC,OACE,IAAc,IAAA,IAAa,CAAC,GAAI,EAAe,IAAI,CAAI,GAAK,CAAC,CAAE,CAAC,CAAC,KAAM,GAAY,EAAQ,EAAW,CAAO,CAAC,CAElH,CAAC,CACH,CACF,CAMA,SAAS,EACP,EACA,EACA,EACA,EAA8C,GAAY,EACvC,CACnB,GAAI,IAAY,IAAA,IAAa,IAAS,IAAA,GACpC,MAAO,CAAC,EAEV,IAAM,EAAa,EAAQ,YAAY,qBAAqB,QAAS,GACnE,EAAM,KAAK,CAAE,YAAW,cAAe,CAAE,YAAW,UAAS,SAAU,CAAC,EAAE,CAAE,EAAE,CAChF,EACM,GAAa,GAAO,QAAU,CAAC,EAAA,CAAG,QAAS,GAAU,CACzD,IAAM,EAAM,EAAM,YAAY,OAAQ,GAAe,EAAW,OAAS,CAAI,EACvE,EAAW,CACf,GAAG,EAAM,MAAM,OAAQ,GAAY,IAAY,CAAI,CAAC,CAAC,IAAI,CAAa,EACtE,GAAI,EAAI,OAAS,EAAI,CAAC,EAAE,EAAI,CAAC,CAC/B,EACA,OAAO,EAAI,KAAK,CAAE,YAAW,cAAe,CAAE,YAAW,UAAS,UAAS,EAAE,CAC/E,CAAC,EACD,MAAO,CAAC,GAAG,EAAY,GAAG,CAAS,CACrC,CAEA,SAAS,GACP,EACA,EACA,EACe,CACf,MAAO,CACL,KAAM,EAAK,YACX,YAAa,EAAK,YAClB,YAAa,EAAK,UAAU,QAC5B,mBAAoB,EAAK,mBACzB,mBAAoB,EAAK,mBACzB,uBACA,oBAAqB,EAAgB,EAAM,EAAsB,CAAS,EAC1E,MAAO,EAAK,KACd,CACF,CAGA,SAAS,EACP,EACA,EACA,EACU,CACV,IAAM,EAAW,EAAK,QAAQ,SACxB,EAAU,GACd,EAAW,OAAS,GAAY,EAAM,KAAM,GAAS,GAAQ,EAAW,WAAa,GAAQ,EAAW,OAAO,EAC3G,kBAAW,IAAI,IACrB,IAAK,IAAM,KAAS,GAAW,QAAU,CAAC,EACxC,GAAI,EAAM,YAAY,KAAK,CAAM,EAC1B,IAAA,IAAM,KAAW,EAAM,MACtB,IAAY,GACd,EAAS,IAAI,CAAO,EAK5B,MAAO,CAAC,GAAG,CAAQ,CAAC,CAAC,SAAS,CAChC,CAEA,SAAS,GACP,EACA,EACA,EACA,EACA,EACM,CACN,IAAM,EAAY,EAAU,MAAM,EAAG,EAAE,EACnC,EAAO,OAAS,GAElB,EACE,uCAAuC,EAAO,OAAO,2CAChD,EAAO,WAAW,OAAS,EAAI,KAAK,EAAO,WAAW,OAAO,mCAAqC,GAAG,SAAS,EAAW,KAAK,eAAe,EAAU,KAC9J,EACA,EAAgB,CAAM,GACb,EAAO,WAAW,SAAW,EACtC,EACE,2BAA2B,EAAO,iBAAiB,kBAAkB,EAAO,qBAAqB,2BAA2B,EAAW,KAAK,eAAe,EAAU,KACvK,GAEA,EACE,sBAAsB,EAAW,KAAK,eAAe,EAAU,KAAK,EAAO,WAAW,OAAO,cAC/F,EACA,EAAgB,CAAM,GAGpB,EAAW,MACb,EAAiB,CAAM,EAGzB,IAAK,IAAM,KAAW,EACpB,EAAY,YAAY,EAAQ,GAAG,EAErC,IAAK,IAAM,KAAS,EAClB,EAAY,UAAU,EAAM,GAAG,CAEnC,CAEA,SAAS,EAAgB,EAA0B,CACjD,IAAK,GAAM,CAAC,EAAO,KAAc,EAAO,WAAW,QAAQ,EACzD,EAAY,GAAG,EAAQ,EAAE,IAAI,EAAU,QAAQ,GAAG,CAEtD,CAGA,SAAS,EAAiB,EAA0B,CAC9C,KAAO,iBAAiB,SAAW,EAGvC,GAAY;;CAAuC,EACnD,IAAK,IAAM,KAAU,EAAO,iBAC1B,EAAY,KAAK,EAAqB,CAAM,EAAE,GAAG,CAFA,CAIrD,CAEA,SAAS,EAAqB,EAAuC,CACnE,IAAM,GACJ,EACA,EAAoC,SACzB,CACX,IAAM,EAAO,EAAO,EAAO,EAAO,IAAI,CAAC,EACvC,OAAO,EAAO,KAAO,GAAG,EAAO,EAAO,EAAO,IAAI,CAAC,EAAE,MAAM,IAAS,CACrE,EACM,EAAS,CACb,aAAa,EAAO,GAAO,EAAG,mBAAmB,IACjD,QAAQ,EAAO,GAAO,EAAG,IAAI,IAC7B,WAAW,EAAO,GAAO,EAAG,YAAY,IACxC,aAAa,EAAO,GAAO,EAAG,SAAS,IACvC,UAAU,EACP,GAAO,EAAG,eACV,GAAU,EAAM,QAAQ,CAAC,CAC5B,GACF,EACA,MAAO,GAAG,EAAO,KAAK,GAAG,EAAO,UAAU,GAAG,EAAO,QAAQ,GAAG,EAAO,OAAO,EAAO,KAAO,GAAK,SAAS,IAAI,EAAO,KAAK,IAAI,GAC/H,CAEA,SAAS,GACP,EACA,EACA,EACA,EACA,EACA,EACM,CACN,IAAM,EAAkC,CACtC,KAAM,EAAW,KACjB,YACA,OAAQ,EAAO,WAAW,SAAW,GAAK,EAAO,SAAW,EAC5D,WAAY,EAAO,WACnB,iBAAkB,EAAO,iBACzB,qBAAsB,EAAO,qBAC7B,iBAAkB,EAAO,iBACzB,SACA,UACF,EACI,EAAW,OACb,EAAO,MAAQ,EACZ,OAAQ,GAAU,EAAM,QAAU,EAAK,CAAC,CACxC,IAAK,IAAW,CACf,KAAM,EAAM,KACZ,kBAAmB,EAAM,aAAa,UAAU,QAAU,EAC1D,kBAAmB,EAAM,aAAa,UAAU,QAAU,EAC1D,SAAU,EAAM,aAAa,WAAa,EAC1C,SAAU,EAAM,aAAa,WAAa,EAC1C,2BAA4B,EAAM,aAAa,wBAA0B,EACzE,2BAA4B,EAAM,aAAa,wBAA0B,EACzE,qBAAsB,EAAM,qBAC5B,oBAAqB,EAAM,oBAC3B,UAAW,EAAO,iBAAiB,OAAQ,GAAO,EAAG,OAAS,EAAM,IAAI,CAC1E,EAAE,GAEN,EAAY,KAAK,UAAU,EAAQ,IAAA,GAAW,CAAC,EAAI;CAAI,CACzD"}
+{"version":3,"file":"diffCommand.js","names":[],"sources":["../src/diffCommand.ts"],"sourcesContent":["import { lstat, readFile, realpath, stat } from 'node:fs/promises';\nimport os from 'node:os';\nimport path from 'node:path';\nimport { getErrorMessage, mapConcurrently } from '@willbooster/shared-lib';\nimport { loadConfig, resolveGateOptions, resolveOptions, type ResolvedOptions } from './cliConfig.js';\nimport {\n  measureCrossFileDuplication,\n  type CrossFileDuplicateOccurrence,\n  type CrossFileDuplicationMetrics,\n} from './crossFileDuplication.js';\nimport type { CrossFileDuplicationFileData, Token } from './duplication.js';\nimport { keepPaths, loadRepositoryExclusion, type Exclusion } from './exclusion.js';\nimport {\n  listChangedFiles,\n  listLineHunks,\n  listRepositoryFiles,\n  listSymlinkPathsAtRevision,\n  readFileAtRevision,\n  resolveMergeBase,\n  resolveRepoRoot,\n  type ChangedFile,\n} from './git.js';\nimport { collectFunctionTokenSequences } from './metrics.js';\nimport {\n  findNewlyDuplicatedLines,\n  indexPartnersByLine,\n  type CloneOccurrence,\n  type DuplicationChange,\n  type LineHunk,\n} from './newDuplication.js';\nimport {\n  evaluateRegressionGate,\n  type CheckedFunctionReport,\n  type GateFileInput,\n  type GateFunctionValues,\n  type GateResult,\n} from './regressionGate.js';\nimport {\n  collectDuplicatedLineNumbers,\n  configSearchDirectory,\n  formatPath,\n  getLanguage,\n  isScannedPath,\n  measureWithCrossFileData,\n  resolveTarget,\n  scanListedFiles,\n  writeStderr,\n  writeStdout,\n  type FileMetrics,\n} from './scan.js';\nimport type { CodeMetrics, LanguageName } from './types.js';\n\n/** Raw options of the `diff` subcommand; every field but base is undefined unless the flag was passed. */\nexport interface DiffCliOptions {\n  base: string;\n  config?: string;\n  duplicationMinTokens?: number;\n  duplicationMaxGapTokens?: number;\n  duplicationMinSimilarityPercent?: number;\n  includeTests?: boolean;\n  json?: boolean;\n  full?: boolean;\n}\n\n/** One changed file measured at both revisions, plus its duplication-universe contribution. */\ninterface PreparedFile {\n  changed: ChangedFile;\n  /** Repository-relative display path: the head path, or the base path for deleted files. */\n  displayFile: string;\n  /** Whether the file is gated (under the target directory); others only feed the base universe. */\n  gated: boolean;\n  headFile?: FileMetrics;\n  headContent?: string;\n  baseMetrics?: CodeMetrics;\n  baseContent?: string;\n  /**\n   * The line diff from the measured base to the measured head; with one side unmeasured, every\n   * line of the other counts as added or deleted.\n   */\n  hunks?: LineHunk[];\n  baseCandidates?: CrossFileDuplicationFileData;\n  baseFunctionTokens?: Int32Array[];\n  headFunctionTokens?: Int32Array[];\n}\n\n/** A scanned file that git considers part of the project, keyed by its repository-relative path. */\ninterface ScannedFile {\n  relativePath: string;\n  file: FileMetrics;\n}\n\n/**\n * Runs the regression gate: measures the files changed relative to the merge-base with the base\n * ref, at both revisions (`git cat-file`; no checkout, no persisted baseline), and reports only\n * violations. Exit codes: 0 all gates passed, 1 violations, 2 changed files could not be measured.\n */\nexport async function runDiffCommand(target: string, cliOptions: DiffCliOptions): Promise<void> {\n  try {\n    await runGate(target, cliOptions);\n  } catch (error) {\n    writeStderr(`Error: ${getErrorMessage(error)}\\n`);\n    process.exitCode = 2;\n  }\n}\n\nasync function runGate(target: string, cliOptions: DiffCliOptions): Promise<void> {\n  const resolvedTarget = resolveTarget(target);\n  const config = await loadConfig(cliOptions.config, await configSearchDirectory(resolvedTarget));\n  const options = resolveOptions(cliOptions, config);\n  const gateOptions = resolveGateOptions(config);\n\n  // The target may be a typo'd path whose ancestors don't exist either; repository discovery must\n  // still run so the mistyped target gets its own diagnostic instead of a git spawn failure.\n  const repoRoot = await realpath(\n    await resolveRepoRoot(await firstExistingDirectory(await configSearchDirectory(resolvedTarget)))\n  );\n  const mergeBase = await resolveMergeBase(repoRoot, cliOptions.base);\n  const changedFiles = await listChangedFiles(repoRoot, mergeBase);\n\n  // Every git-visible file (tracked or untracked non-ignored) is measured at head: that provides\n  // the head metrics of changed files and the project-wide duplication universe, so copy-paste\n  // from unchanged code into changed files is caught. Scanning the explicit git list (instead of\n  // walking the tree) keeps ignored artifact directories from ever being parsed: they exist in\n  // neither the base commit nor CI, so they would only cost time and skew duplication counts.\n  // Unchanged files are byte-identical at both revisions, so the base universe is the same scan\n  // with the changed files' contents swapped for their merge-base blobs.\n  const repositoryFiles = await listRepositoryFiles(repoRoot);\n  const baseSymlinkPaths = await listSymlinkPathsAtRevision(repoRoot, mergeBase);\n  const { canonicalTarget, targetExists } = await canonicalizeTarget(resolvedTarget);\n  const targetStat = await stat(canonicalTarget).catch(() => {});\n  const targetFile = targetStat?.isFile() ? canonicalTarget : undefined;\n  // A targeted file git does not list (an ignored one) is still measured, as an addition.\n  const unlistedTarget = targetFile && path.relative(repoRoot, targetFile).split(path.sep).join('/');\n  if (unlistedTarget !== undefined && !unlistedTarget.startsWith('../') && !repositoryFiles.has(unlistedTarget)) {\n    repositoryFiles.add(unlistedTarget);\n    changedFiles.push({ status: 'added', headPath: unlistedTarget });\n  }\n  const explicitFiles = listExplicitlyTargetedPaths(targetFile, repoRoot, changedFiles);\n  // Base blobs are excluded by the attributes of the revision they come from. A changed\n  // .gitattributes can re-include files whose bytes did not change, so then every file's base\n  // attributes are needed to find them.\n  const attributesChanged = changedFiles.some((changed) =>\n    [changed.headPath, changed.basePath].some(\n      (file) => file !== undefined && path.posix.basename(file) === '.gitattributes'\n    )\n  );\n  const basePaths = changedFiles.flatMap((changed) => (changed.basePath === undefined ? [] : [changed.basePath]));\n  const [headAttributesExclusion, baseAttributesExclusion] = await Promise.all([\n    loadRepositoryExclusion(repoRoot, repositoryFiles, options.exclude),\n    loadRepositoryExclusion(\n      repoRoot,\n      attributesChanged ? [...new Set([...basePaths, ...repositoryFiles])] : basePaths,\n      options.exclude,\n      mergeBase\n    ),\n  ]);\n  const headExclusion = keepPaths(headAttributesExclusion, explicitFiles);\n  const baseExclusion = keepPaths(baseAttributesExclusion, explicitFiles);\n  const scan = await scanListedFiles(\n    repoRoot,\n    repositoryFiles,\n    { ...options, loadExclusion: () => Promise.resolve(headExclusion) },\n    explicitFiles\n  );\n  // A run-wide failure (a missing native addon) invalidates the whole gate: surface it once as\n  // the fatal error (exit 2) instead of diagnosing every changed file as unmeasured.\n  if (scan.fatalError) {\n    throw new Error(scan.fatalError);\n  }\n  const scannedFiles: ScannedFile[] = scan.files.map((file) => ({\n    relativePath: formatPath(file.file, scan.displayRoot),\n    file,\n  }));\n  if (attributesChanged) {\n    changedFiles.push(\n      ...(await listReincludedFiles(scannedFiles, changedFiles, repoRoot, headExclusion, baseExclusion))\n    );\n  }\n\n  // A measurement failure on ANY scannable changed file forces exit 2 — deliberately including\n  // files outside a scoped target, because cross-file function matching and the base duplication\n  // universe for the gated files depend on them. Failures elsewhere (unchanged files) and\n  // unsupported changed paths degrade to warnings.\n  const changedPaths = new Set(\n    changedFiles\n      .flatMap((changed) => [changed.headPath, ...(changed.basePath === undefined ? [] : [changed.basePath])])\n      .filter((changedPath) => isScannedPath(changedPath, options, explicitFiles.has(path.join(repoRoot, changedPath))))\n  );\n  const errors: string[] = [];\n  const warnings = [...scan.warnings];\n  for (const error of scan.errors) {\n    if ([...changedPaths].some((changedPath) => error.startsWith(`${changedPath}:`))) {\n      errors.push(error);\n    } else {\n      warnings.push(error);\n    }\n  }\n\n  const prepared = await prepareChangedFiles(\n    changedFiles,\n    {\n      repoRoot,\n      mergeBase,\n      canonicalTarget,\n      options,\n      scannedFiles,\n      baseSymlinkPaths,\n      scanErrors: [...errors],\n      headExclusion,\n      baseExclusion,\n      explicitFiles,\n      generatedHeadFiles: new Set(scan.generatedFiles),\n    },\n    errors,\n    warnings\n  );\n  // A gate must not fail open on a mistyped target: a nonexistent path is only acceptable when it\n  // still matches changed files (e.g. a fully deleted directory).\n  if (!targetExists && !prepared.some((file) => file.gated)) {\n    throw new Error(`target \"${target}\" does not exist and matches no changed file`);\n  }\n\n  // Non-gated files (outside the target, or renamed out of scan scope) still feed function\n  // matching and the duplication universes; the evaluator reports nothing for them.\n  const { baseCross, headCross } = measureDuplicationUniverses(prepared, scannedFiles, options);\n  const renamedPaths = new Map(\n    prepared.flatMap(({ changed }) =>\n      changed.status === 'renamed' && changed.basePath !== undefined\n        ? [[changed.basePath, changed.headPath] as const]\n        : []\n    )\n  );\n  const headPathOf = (basePath: string): string => renamedPaths.get(basePath) ?? basePath;\n  const lineSignaturesOf = createLineSignatureIndex(scannedFiles);\n  const newlyDuplicatedLines = findNewlyDuplicatedLines(\n    prepared.map((file) => toDuplicationChange(file, baseCross, headCross, headPathOf, lineSignaturesOf))\n  );\n  const inputs = prepared.map((file, index) => toGateInput(file, newlyDuplicatedLines[index] ?? [], headCross));\n  const result = evaluateRegressionGate(inputs, gateOptions);\n\n  if (cliOptions.json) {\n    printJsonReport(cliOptions, mergeBase, result, inputs, errors, warnings);\n  } else {\n    printTextReport(cliOptions, mergeBase, result, errors, warnings);\n  }\n\n  if (errors.length > 0) {\n    process.exitCode = 2;\n  } else if (result.violations.length > 0) {\n    process.exitCode = 1;\n  }\n}\n\n/** Content with a generated-code marker, to ask whether an exclusion honors the marker for a path. */\nconst markedCode = '/* @generated */';\n\n/**\n * Unchanged files the head scan measured but the merge-base would not have: excluded by its\n * attributes, or generated by content without the `-linguist-generated` override the working tree\n * adds. Their code is newly measured, so it gates as a modification from an unmeasurable base. The\n * head side comes from the scan, which already applied scope, symbolic-link, and exclusion rules.\n */\nasync function listReincludedFiles(\n  scannedFiles: readonly ScannedFile[],\n  changedFiles: readonly ChangedFile[],\n  repoRoot: string,\n  headExclusion: Exclusion,\n  baseExclusion: Exclusion\n): Promise<ChangedFile[]> {\n  const changedPaths = new Set(changedFiles.flatMap((changed) => [changed.headPath, changed.basePath]));\n  const reincluded: ChangedFile[] = [];\n  for (const { relativePath, file } of scannedFiles) {\n    const absolutePath = path.join(repoRoot, relativePath);\n    if (changedPaths.has(relativePath)) {\n      continue;\n    }\n    // The same content is generated at one revision only when its override differs, so the\n    // content is read just for files the working tree newly marks -linguist-generated.\n    const overrideAdded =\n      baseExclusion.isGeneratedCode(absolutePath, markedCode) &&\n      !headExclusion.isGeneratedCode(absolutePath, markedCode);\n    if (\n      baseExclusion.isExcludedPath(absolutePath) ||\n      (overrideAdded && baseExclusion.isGeneratedCode(absolutePath, await readFile(file.file, 'utf8')))\n    ) {\n      reincluded.push({ status: 'modified', basePath: relativePath, headPath: relativePath });\n    }\n  }\n  return reincluded;\n}\n\n/**\n * An explicitly targeted file is measured even when excluded, like the ranking command's single\n * file target: the file itself, and its base path when it was renamed.\n */\nfunction listExplicitlyTargetedPaths(\n  targetFile: string | undefined,\n  repoRoot: string,\n  changedFiles: ChangedFile[]\n): Set<string> {\n  if (targetFile === undefined) {\n    return new Set();\n  }\n  const paths = new Set([targetFile]);\n  for (const changed of changedFiles) {\n    if (changed.basePath !== undefined && path.join(repoRoot, changed.headPath) === targetFile) {\n      paths.add(path.join(repoRoot, changed.basePath));\n    }\n  }\n  return paths;\n}\n\n/** The target may not exist (e.g. only deleted files under it); fall back to the resolved path. */\nasync function canonicalizeTarget(resolvedTarget: string): Promise<{ canonicalTarget: string; targetExists: boolean }> {\n  try {\n    return { canonicalTarget: await realpath(resolvedTarget), targetExists: true };\n  } catch {\n    return { canonicalTarget: resolvedTarget, targetExists: false };\n  }\n}\n\ninterface GateContext {\n  repoRoot: string;\n  mergeBase: string;\n  canonicalTarget: string;\n  options: ResolvedOptions;\n  scannedFiles: ScannedFile[];\n  /** Paths that are symbolic links at the merge-base; like head symlinks, they are not gated. */\n  baseSymlinkPaths: Set<string>;\n  /** Errors the head scan recorded against changed files. */\n  scanErrors: readonly string[];\n  headExclusion: Exclusion;\n  baseExclusion: Exclusion;\n  /** The explicitly targeted file (and its renamed base path), measured despite every exclusion. */\n  explicitFiles: ReadonlySet<string>;\n  /** Absolute paths the head scan skipped as generated code. */\n  generatedHeadFiles: ReadonlySet<string>;\n}\n\nasync function prepareChangedFiles(\n  changedFiles: ChangedFile[],\n  context: GateContext,\n  errors: string[],\n  warnings: string[]\n): Promise<PreparedFile[]> {\n  const headByPath = new Map(context.scannedFiles.map(({ relativePath, file }) => [relativePath, file]));\n  // Files are prepared concurrently (git reads and base measurements); each records its diagnostics\n  // on its own, and they are appended in changed-file order so the report stays deterministic.\n  const results = await mapConcurrently(changedFiles, os.availableParallelism() * 2, async (changed) => {\n    const fileErrors: string[] = [];\n    const fileWarnings: string[] = [];\n    const file = await prepareChangedFile(changed, context, headByPath, fileErrors, fileWarnings);\n    return { file, fileErrors, fileWarnings };\n  });\n  const prepared: PreparedFile[] = [];\n  for (const { file, fileErrors, fileWarnings } of results) {\n    errors.push(...fileErrors);\n    warnings.push(...fileWarnings);\n    if (file) {\n      prepared.push(file);\n    }\n  }\n  return prepared;\n}\n\nasync function prepareChangedFile(\n  changed: ChangedFile,\n  context: GateContext,\n  headByPath: Map<string, FileMetrics>,\n  errors: string[],\n  warnings: string[]\n): Promise<PreparedFile | undefined> {\n  // Symbolic links are skipped on both sides, mirroring scanListedFiles: git stores only the\n  // target string, so a symlink blob is not measurable source.\n  const absoluteHeadPath = path.join(context.repoRoot, changed.headPath);\n  const headScannable =\n    changed.status !== 'deleted' &&\n    isScannedPath(changed.headPath, context.options, context.explicitFiles.has(absoluteHeadPath)) &&\n    !context.headExclusion.isExcludedPath(absoluteHeadPath) &&\n    !context.generatedHeadFiles.has(absoluteHeadPath) &&\n    !(await isSymbolicLink(absoluteHeadPath));\n  // A base path outside the scan scope (renamed from a test/ignored directory, or an unsupported\n  // extension) was never measurable code: its content gates as new code instead of ratcheting\n  // against a blob the scanner would not have measured.\n  const absoluteBasePath = changed.basePath === undefined ? '' : path.join(context.repoRoot, changed.basePath);\n  const baseScannable =\n    changed.basePath !== undefined &&\n    isScannedPath(changed.basePath, context.options, context.explicitFiles.has(absoluteBasePath)) &&\n    !context.baseExclusion.isExcludedPath(absoluteBasePath) &&\n    !context.baseSymlinkPaths.has(changed.basePath);\n  if (!headScannable && !baseScannable) {\n    return undefined;\n  }\n\n  const displayFile = changed.status === 'deleted' ? (changed.basePath as string) : changed.headPath;\n  const headFile = headScannable ? headByPath.get(changed.headPath) : undefined;\n  if (headScannable && !headFile) {\n    reportUnmeasuredChangedFile(changed.headPath, context.scanErrors, errors);\n    return undefined;\n  }\n\n  const file: PreparedFile = {\n    changed,\n    displayFile,\n    // A file whose head left the scan scope still contributes its base functions to matching and\n    // its base blob to the base universe, but nothing about it is gated or reported.\n    gated:\n      headScannable || changed.status === 'deleted'\n        ? isWithinTarget(path.join(context.repoRoot, displayFile), context.canonicalTarget)\n        : false,\n    headFile,\n  };\n\n  if (baseScannable) {\n    const outcome = await measureBaseRevision(file, changed.basePath as string, context, errors, warnings);\n    // A base that turns out to be generated code gates its head as new code, like any other base\n    // outside the scan scope.\n    if (outcome === 'failed' || (outcome === 'generated' && !headFile)) {\n      return undefined;\n    }\n  }\n\n  try {\n    if (headFile) {\n      // The duplication gate needs the head lines, so an unreadable head fails like an unmeasured one.\n      file.headContent = await readFile(headFile.file, 'utf8');\n      collectHeadFunctionTokens(file, file.headContent, context, warnings);\n    }\n    file.hunks = await diffMeasuredRevisions(file, context);\n  } catch (error) {\n    errors.push(`${displayFile}: ${getErrorMessage(error)}`);\n    return undefined;\n  }\n  return file;\n}\n\nasync function diffMeasuredRevisions(file: PreparedFile, context: GateContext): Promise<LineHunk[]> {\n  const { changed, baseContent, headContent } = file;\n  if (baseContent !== undefined && headContent !== undefined) {\n    return await listLineHunks(context.repoRoot, context.mergeBase, changed.basePath as string, changed.headPath);\n  }\n  return [\n    {\n      baseStart: 1,\n      baseCount: baseContent === undefined ? 0 : splitLines(baseContent).length,\n      headStart: 1,\n      headCount: headContent === undefined ? 0 : splitLines(headContent).length,\n    },\n  ];\n}\n\n/** Line numbering of the measured metrics (classifyLines splits on every line terminator). */\nfunction splitLines(content: string): string[] {\n  return content.split(/\\r\\n|\\n|\\r/u);\n}\n\n/**\n * The scan covers exactly the git-visible list, so a scannable changed path can only be missing\n * after a measurement failure (already recorded as an error) or a silent exclusion (an alias of\n * an already-visited file, or absence from the git list). Failing loudly keeps the gate from\n * passing with the file unchecked.\n */\nfunction reportUnmeasuredChangedFile(headPath: string, scanErrors: readonly string[], errors: string[]): void {\n  if (!scanErrors.some((error) => error.startsWith(`${headPath}:`))) {\n    errors.push(`${headPath}: changed file was not measured`);\n  }\n}\n\nfunction collectHeadFunctionTokens(\n  file: PreparedFile,\n  headContent: string,\n  context: GateContext,\n  warnings: string[]\n): void {\n  try {\n    file.headFunctionTokens = collectFunctionTokenSequences(headContent, {\n      language: languageOf(file.changed.headPath, context),\n      duplication: context.options.duplication,\n    });\n  } catch (error) {\n    // Only rename re-matching degrades without token sequences; the head metrics still gate.\n    warnings.push(`${file.displayFile}: function token sequences unavailable: ${getErrorMessage(error)}`);\n  }\n}\n\n/**\n * Measures the merge-base blob into `file`; 'failed' (with an error recorded) only when the\n * metrics themselves cannot be measured, and 'generated' (nothing measured) for generated code.\n * The auxiliary collections (duplication candidates, token sequences) may fail independently of\n * the metrics, so their failure only degrades duplication data and rename re-matching — the\n * function-level ratchets still run.\n */\nasync function measureBaseRevision(\n  file: PreparedFile,\n  basePath: string,\n  context: GateContext,\n  errors: string[],\n  warnings: string[]\n): Promise<'measured' | 'generated' | 'failed'> {\n  const measureOptions = {\n    language: languageOf(basePath, context),\n    duplication: context.options.duplication,\n  };\n  let baseContent;\n  try {\n    baseContent = await readFileAtRevision(context.repoRoot, context.mergeBase, basePath);\n    if (context.baseExclusion.isGeneratedCode(path.join(context.repoRoot, basePath), baseContent)) {\n      return 'generated';\n    }\n    const measured = await measureWithCrossFileData(baseContent, measureOptions);\n    file.baseMetrics = measured.metrics;\n    file.baseContent = baseContent;\n    file.baseCandidates = measured.crossFileData;\n    if (measured.crossFileError !== undefined) {\n      warnings.push(`${basePath} (at merge-base): duplication candidates unavailable: ${measured.crossFileError}`);\n    }\n  } catch (error) {\n    errors.push(`${basePath} (at merge-base): ${getErrorMessage(error)}`);\n    return 'failed';\n  }\n  try {\n    file.baseFunctionTokens = collectFunctionTokenSequences(baseContent, measureOptions);\n  } catch (error) {\n    warnings.push(`${basePath} (at merge-base): function token sequences unavailable: ${getErrorMessage(error)}`);\n  }\n  return 'measured';\n}\n\n/** The language of a scannable path; an explicitly targeted file skips the test-file name rules. */\nfunction languageOf(relativePath: string, context: GateContext): LanguageName {\n  const explicitTarget = context.explicitFiles.has(path.join(context.repoRoot, relativePath));\n  return getLanguage(relativePath, context.options, explicitTarget) as LanguageName;\n}\n\nasync function isSymbolicLink(absolutePath: string): Promise<boolean> {\n  const stats = await lstat(absolutePath).catch(() => {});\n  return stats?.isSymbolicLink() ?? false;\n}\n\n/** Walks up to the nearest existing DIRECTORY, so git commands never spawn in a missing or non-directory cwd. */\nasync function firstExistingDirectory(directory: string): Promise<string> {\n  let current = directory;\n  while (true) {\n    const stats = await stat(current).catch(() => {});\n    if (stats?.isDirectory()) {\n      return current;\n    }\n    const parent = path.dirname(current);\n    if (parent === current) {\n      return current;\n    }\n    current = parent;\n  }\n}\n\nfunction isWithinTarget(candidate: string, targetDirectory: string): boolean {\n  const relative = path.relative(targetDirectory, candidate);\n  return relative === '' || (!relative.startsWith(`..${path.sep}`) && relative !== '..' && !path.isAbsolute(relative));\n}\n\nfunction measureDuplicationUniverses(\n  prepared: PreparedFile[],\n  scannedFiles: ScannedFile[],\n  options: ResolvedOptions\n): { baseCross?: CrossFileDuplicationMetrics; headCross?: CrossFileDuplicationMetrics } {\n  const headSources = scannedFiles.flatMap(({ relativePath, file }) =>\n    file.duplicationCandidates ? [{ file: relativePath, ...file.duplicationCandidates }] : []\n  );\n\n  const changedHeadPaths = new Set(\n    prepared.flatMap((file) => (file.changed.status === 'deleted' ? [] : [file.changed.headPath]))\n  );\n  const baseSources = headSources.filter((source) => !changedHeadPaths.has(source.file));\n  for (const file of prepared) {\n    if (file.baseCandidates && file.changed.basePath !== undefined) {\n      baseSources.push({ file: file.changed.basePath, ...file.baseCandidates });\n    }\n  }\n\n  return {\n    baseCross: baseSources.length >= 2 ? measureCrossFileDuplication(baseSources, options.duplication) : undefined,\n    headCross: headSources.length >= 2 ? measureCrossFileDuplication(headSources, options.duplication) : undefined,\n  };\n}\n\nfunction toDuplicationChange(\n  file: PreparedFile,\n  baseCross: CrossFileDuplicationMetrics | undefined,\n  headCross: CrossFileDuplicationMetrics | undefined,\n  headPathOf: (basePath: string) => string,\n  lineSignaturesOf: (file: string) => LineSignatures | undefined\n): DuplicationChange {\n  const headOccurrences = collectOccurrences(file.headFile?.metrics, headCross, file.changed.headPath);\n  return {\n    baseLines: file.baseContent === undefined ? undefined : splitLines(file.baseContent),\n    headLines: file.headContent === undefined ? undefined : splitLines(file.headContent),\n    baseDuplicatedLines:\n      file.baseMetrics === undefined || file.changed.basePath === undefined\n        ? new Set()\n        : collectDuplicatedLineNumbers(file.baseMetrics, baseCross, file.changed.basePath),\n    headDuplicatedLines:\n      file.headFile === undefined\n        ? new Set()\n        : keepRepeatedLines(\n            collectDuplicatedLineNumbers(file.headFile.metrics, headCross, file.changed.headPath),\n            file.changed.headPath,\n            headOccurrences,\n            lineSignaturesOf\n          ),\n    baseOccurrences: collectOccurrences(file.baseMetrics, baseCross, file.changed.basePath, headPathOf),\n    headOccurrences,\n    hunks: file.hunks ?? [],\n  };\n}\n\n/** A file's per-line normalized token signatures and how many lines carry each. */\ninterface LineSignatures {\n  byLine: Map<number, string>;\n  lineCounts: Map<string, number>;\n}\n\n/** Line signatures of the scanned head files, computed on first use. */\nfunction createLineSignatureIndex(scannedFiles: ScannedFile[]): (file: string) => LineSignatures | undefined {\n  const tokensByFile = new Map(\n    scannedFiles.map(({ relativePath, file }) => [relativePath, file.duplicationCandidates?.tokens])\n  );\n  const signaturesByFile = new Map<string, LineSignatures | undefined>();\n  return (file) => {\n    if (!signaturesByFile.has(file)) {\n      const tokens = tokensByFile.get(file);\n      signaturesByFile.set(file, tokens && signLines(tokens));\n    }\n    return signaturesByFile.get(file);\n  };\n}\n\n/**\n * Each line's tokens with identifiers numbered by first occurrence, as clone detection anonymizes\n * them.\n */\nfunction signLines(tokens: readonly Token[]): LineSignatures {\n  const byLine = new Map<number, string>();\n  const identifierIndexesByLine = new Map<number, Map<string, number>>();\n  for (const token of tokens) {\n    const line = token.startRow + 1;\n    let text = token.text;\n    if (token.kind === 'id') {\n      // Numbered by first occurrence within the line, so `a + a` and `a + b` stay distinct.\n      const identifierIndexes = identifierIndexesByLine.get(line) ?? new Map<string, number>();\n      identifierIndexesByLine.set(line, identifierIndexes);\n      if (!identifierIndexes.has(token.text)) {\n        identifierIndexes.set(token.text, identifierIndexes.size);\n      }\n      text = `\\u0001${identifierIndexes.get(token.text)}`;\n    }\n    byLine.set(line, `${byLine.get(line) ?? ''}${text}\\u0000`);\n  }\n  const lineCounts = new Map<string, number>();\n  for (const signature of byLine.values()) {\n    lineCounts.set(signature, (lineCounts.get(signature) ?? 0) + 1);\n  }\n  return { byLine, lineCounts };\n}\n\n/**\n * The duplicated lines whose normalized tokens a partner file of their clones (the file itself,\n * on another line, for `''`) repeats on some line. Near-miss coverage also marks lines the copies\n * do not share, such as a new line inserted into an old near-miss clone, which copied nothing.\n */\nfunction keepRepeatedLines(\n  lines: ReadonlySet<number>,\n  file: string,\n  occurrences: readonly CloneOccurrence[],\n  lineSignaturesOf: (file: string) => LineSignatures | undefined\n): Set<number> {\n  const own = lineSignaturesOf(file);\n  if (own === undefined) {\n    return new Set(lines);\n  }\n  const repeats = (signature: string, partner: string): boolean => {\n    if (partner === '') {\n      return (own.lineCounts.get(signature) ?? 0) > 1;\n    }\n    const partnerSignatures = lineSignaturesOf(partner);\n    return partnerSignatures === undefined || partnerSignatures.lineCounts.has(signature);\n  };\n  const partnersByLine = indexPartnersByLine(occurrences);\n  return new Set(\n    [...lines].filter((line) => {\n      const signature = own.byLine.get(line);\n      return (\n        signature === undefined || [...(partnersByLine.get(line) ?? [])].some((partner) => repeats(signature, partner))\n      );\n    })\n  );\n}\n\n/**\n * The file's clone occurrences with their partner files (`''` for the file itself), partner paths\n * renamed by `renamePartner` so base partners compare with head ones.\n */\nfunction collectOccurrences(\n  metrics: CodeMetrics | undefined,\n  cross: CrossFileDuplicationMetrics | undefined,\n  file: string | undefined,\n  renamePartner: (partner: string) => string = (partner) => partner\n): CloneOccurrence[] {\n  if (metrics === undefined || file === undefined) {\n    return [];\n  }\n  const withinFile = metrics.duplication.duplicateBlockGroups.flatMap((group) =>\n    group.map(({ startLine, endLine }) => ({ startLine, endLine, partners: [''] }))\n  );\n  const crossFile = (cross?.groups ?? []).flatMap((group) => {\n    const own = group.occurrences.filter((occurrence) => occurrence.file === file);\n    const partners = [\n      ...group.files.filter((partner) => partner !== file).map(renamePartner),\n      ...(own.length > 1 ? [''] : []),\n    ];\n    return own.map(({ startLine, endLine }) => ({ startLine, endLine, partners }));\n  });\n  return [...withinFile, ...crossFile];\n}\n\nfunction toGateInput(\n  file: PreparedFile,\n  newlyDuplicatedLines: number[],\n  headCross: CrossFileDuplicationMetrics | undefined\n): GateFileInput {\n  return {\n    file: file.displayFile,\n    baseMetrics: file.baseMetrics,\n    headMetrics: file.headFile?.metrics,\n    baseFunctionTokens: file.baseFunctionTokens,\n    headFunctionTokens: file.headFunctionTokens,\n    newlyDuplicatedLines,\n    duplicationPartners: collectPartners(file, newlyDuplicatedLines, headCross),\n    gated: file.gated,\n  };\n}\n\n/** Other files sharing a cross-file clone that covers the given lines. */\nfunction collectPartners(\n  file: PreparedFile,\n  lines: number[],\n  headCross: CrossFileDuplicationMetrics | undefined\n): string[] {\n  const headPath = file.changed.headPath;\n  const covers = (occurrence: CrossFileDuplicateOccurrence): boolean =>\n    occurrence.file === headPath && lines.some((line) => line >= occurrence.startLine && line <= occurrence.endLine);\n  const partners = new Set<string>();\n  for (const group of headCross?.groups ?? []) {\n    if (group.occurrences.some(covers)) {\n      for (const partner of group.files) {\n        if (partner !== headPath) {\n          partners.add(partner);\n        }\n      }\n    }\n  }\n  return [...partners].toSorted();\n}\n\nfunction printTextReport(\n  cliOptions: DiffCliOptions,\n  mergeBase: string,\n  result: GateResult,\n  errors: string[],\n  warnings: string[]\n): void {\n  const shortBase = mergeBase.slice(0, 12);\n  if (errors.length > 0) {\n    // Unmeasured files were not gated, so \"0 violations\" would be vacuous; never claim a pass.\n    writeStdout(\n      `Regression gate could not complete: ${errors.length} measurement failures (details on stderr)` +\n        `${result.violations.length > 0 ? `; ${result.violations.length} violations in the measured files` : ''} (base ${cliOptions.base}, merge-base ${shortBase}).\\n`\n    );\n    printViolations(result);\n  } else if (result.violations.length === 0) {\n    writeStdout(\n      `Regression gate passed: ${result.checkedFileCount} changed files, ${result.checkedFunctionCount} functions checked (base ${cliOptions.base}, merge-base ${shortBase}).\\n`\n    );\n  } else {\n    writeStdout(\n      `Regression gate vs ${cliOptions.base} (merge-base ${shortBase}): ${result.violations.length} violations\\n`\n    );\n    printViolations(result);\n  }\n\n  if (cliOptions.full) {\n    printFullDetails(result);\n  }\n\n  for (const warning of warnings) {\n    writeStderr(`Warning: ${warning}\\n`);\n  }\n  for (const error of errors) {\n    writeStderr(`Error: ${error}\\n`);\n  }\n}\n\nfunction printViolations(result: GateResult): void {\n  for (const [index, violation] of result.violations.entries()) {\n    writeStdout(`${index + 1}. ${violation.message}\\n`);\n  }\n}\n\n/** Base -> head values of every checked function; kept behind --full for humans and trending. */\nfunction printFullDetails(result: GateResult): void {\n  if (result.checkedFunctions.length === 0) {\n    return;\n  }\n  writeStdout('\\nChecked functions (base -> head):\\n');\n  for (const report of result.checkedFunctions) {\n    writeStdout(`- ${formatFunctionReport(report)}\\n`);\n  }\n}\n\nfunction formatFunctionReport(report: CheckedFunctionReport): string {\n  const range = (\n    select: (values: GateFunctionValues) => number,\n    format: (value: number) => string = String\n  ): string => {\n    const head = format(select(report.head));\n    return report.base ? `${format(select(report.base))} -> ${head}` : head;\n  };\n  const values = [\n    `cognitive ${range((fn) => fn.cognitiveComplexity)}`,\n    `NCSS ${range((fn) => fn.ncss)}`,\n    `nesting ${range((fn) => fn.nestingDepth)}`,\n    `DepDegree ${range((fn) => fn.depDegree)}`,\n    `volume ${range(\n      (fn) => fn.halsteadVolume,\n      (value) => value.toFixed(1)\n    )}`,\n  ];\n  return `${report.file}:${report.startLine}-${report.endLine} ${report.name}${report.base ? '' : ' (new)'}: ${values.join(', ')}`;\n}\n\nfunction printJsonReport(\n  cliOptions: DiffCliOptions,\n  mergeBase: string,\n  result: GateResult,\n  inputs: GateFileInput[],\n  errors: string[],\n  warnings: string[]\n): void {\n  const report: Record<string, unknown> = {\n    base: cliOptions.base,\n    mergeBase,\n    passed: result.violations.length === 0 && errors.length === 0,\n    violations: result.violations,\n    checkedFileCount: result.checkedFileCount,\n    checkedFunctionCount: result.checkedFunctionCount,\n    newFunctionCount: result.newFunctionCount,\n    errors,\n    warnings,\n  };\n  if (cliOptions.full) {\n    report.files = inputs\n      .filter((input) => input.gated !== false)\n      .map((input) => ({\n        file: input.file,\n        baseFunctionCount: input.baseMetrics?.functions.length ?? 0,\n        headFunctionCount: input.headMetrics?.functions.length ?? 0,\n        baseNcss: input.baseMetrics?.ncssCount ?? 0,\n        headNcss: input.headMetrics?.ncssCount ?? 0,\n        baseMaxCognitiveComplexity: input.baseMetrics?.maxCognitiveComplexity ?? 0,\n        headMaxCognitiveComplexity: input.headMetrics?.maxCognitiveComplexity ?? 0,\n        newlyDuplicatedLines: input.newlyDuplicatedLines,\n        duplicationPartners: input.duplicationPartners,\n        functions: result.checkedFunctions.filter((fn) => fn.file === input.file),\n      }));\n  }\n  writeStdout(JSON.stringify(report, undefined, 2) + '\\n');\n}\n"],"mappings":"wjCAgGA,eAAsB,EAAe,EAAgB,EAA2C,CAC9F,GAAI,CACF,MAAM,EAAQ,EAAQ,CAAU,CAClC,OAAS,EAAO,CACd,EAAY,UAAU,EAAgB,CAAK,EAAE,GAAG,EAChD,QAAQ,SAAW,CACrB,CACF,CAEA,eAAe,EAAQ,EAAgB,EAA2C,CAChF,IAAM,EAAiB,GAAc,CAAM,EACrC,EAAS,MAAM,EAAW,EAAW,OAAQ,MAAM,EAAsB,CAAc,CAAC,EACxF,EAAU,EAAe,EAAY,CAAM,EAC3C,EAAc,EAAmB,CAAM,EAIvC,EAAW,MAAM,EACrB,MAAM,GAAgB,MAAM,GAAuB,MAAM,EAAsB,CAAc,CAAC,CAAC,CACjG,EACM,EAAY,MAAM,GAAiB,EAAU,EAAW,IAAI,EAC5D,EAAe,MAAM,EAAiB,EAAU,CAAS,EASzD,EAAkB,MAAM,EAAoB,CAAQ,EACpD,EAAmB,MAAM,EAA2B,EAAU,CAAS,EACvE,CAAE,kBAAiB,gBAAiB,MAAM,EAAmB,CAAc,EAE3E,GAAa,MADM,EAAK,CAAe,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EAC9B,OAAO,EAAI,EAAkB,IAAA,GAEtD,EAAiB,GAAc,EAAK,SAAS,EAAU,CAAU,CAAC,CAAC,MAAM,EAAK,GAAG,CAAC,CAAC,KAAK,GAAG,EAC7F,IAAmB,IAAA,IAAa,CAAC,EAAe,WAAW,KAAK,GAAK,CAAC,EAAgB,IAAI,CAAc,IAC1G,EAAgB,IAAI,CAAc,EAClC,EAAa,KAAK,CAAE,OAAQ,QAAS,SAAU,CAAe,CAAC,GAEjE,IAAM,EAAgB,EAA4B,EAAY,EAAU,CAAY,EAI9E,EAAoB,EAAa,KAAM,GAC3C,CAAC,EAAQ,SAAU,EAAQ,QAAQ,CAAC,CAAC,KAClC,GAAS,IAAS,IAAA,IAAa,EAAK,MAAM,SAAS,CAAI,IAAM,gBAChE,CACF,EACM,EAAY,EAAa,QAAS,GAAa,EAAQ,WAAa,IAAA,GAAY,CAAC,EAAI,CAAC,EAAQ,QAAQ,CAAE,EACxG,CAAC,EAAyB,GAA2B,MAAM,QAAQ,IAAI,CAC3E,EAAwB,EAAU,EAAiB,EAAQ,OAAO,EAClE,EACE,EACA,EAAoB,CAAC,mBAAG,IAAI,IAAI,CAAC,GAAG,EAAW,GAAG,CAAe,CAAC,CAAC,EAAI,EACvE,EAAQ,QACR,CACF,CACF,CAAC,EACK,EAAgB,EAAU,EAAyB,CAAa,EAChE,EAAgB,EAAU,EAAyB,CAAa,EAChE,EAAO,MAAM,GACjB,EACA,EACA,CAAE,GAAG,EAAS,kBAAqB,QAAQ,QAAQ,CAAa,CAAE,EAClE,CACF,EAGA,GAAI,EAAK,WACP,MAAU,MAAM,EAAK,UAAU,EAEjC,IAAM,EAA8B,EAAK,MAAM,IAAK,IAAU,CAC5D,aAAc,GAAW,EAAK,KAAM,EAAK,WAAW,EACpD,MACF,EAAE,EACE,GACF,EAAa,KACX,GAAI,MAAM,GAAoB,EAAc,EAAc,EAAU,EAAe,CAAa,CAClG,EAOF,IAAM,EAAe,IAAI,IACvB,EACG,QAAS,GAAY,CAAC,EAAQ,SAAU,GAAI,EAAQ,WAAa,IAAA,GAAY,CAAC,EAAI,CAAC,EAAQ,QAAQ,CAAE,CAAC,CAAC,CACvG,OAAQ,GAAgB,EAAc,EAAa,EAAS,EAAc,IAAI,EAAK,KAAK,EAAU,CAAW,CAAC,CAAC,CAAC,CACrH,EACM,EAAmB,CAAC,EACpB,EAAW,CAAC,GAAG,EAAK,QAAQ,EAClC,IAAK,IAAM,KAAS,EAAK,OACnB,CAAC,GAAG,CAAY,CAAC,CAAC,KAAM,GAAgB,EAAM,WAAW,GAAG,EAAY,EAAE,CAAC,EAC7E,EAAO,KAAK,CAAK,EAEjB,EAAS,KAAK,CAAK,EAIvB,IAAM,EAAW,MAAM,EACrB,EACA,CACE,WACA,YACA,kBACA,UACA,eACA,mBACA,WAAY,CAAC,GAAG,CAAM,EACtB,gBACA,gBACA,gBACA,mBAAoB,IAAI,IAAI,EAAK,cAAc,CACjD,EACA,EACA,CACF,EAGA,GAAI,CAAC,GAAgB,CAAC,EAAS,KAAM,GAAS,EAAK,KAAK,EACtD,MAAU,MAAM,WAAW,EAAO,6CAA6C,EAKjF,GAAM,CAAE,YAAW,aAAc,GAA4B,EAAU,EAAc,CAAO,EACtF,EAAe,IAAI,IACvB,EAAS,SAAS,CAAE,aAClB,EAAQ,SAAW,WAAa,EAAQ,WAAa,IAAA,GACjD,CAAC,CAAC,EAAQ,SAAU,EAAQ,QAAQ,CAAU,EAC9C,CAAC,CACP,CACF,EACM,EAAc,GAA6B,EAAa,IAAI,CAAQ,GAAK,EACzE,EAAmB,GAAyB,CAAY,EACxD,EAAuB,EAC3B,EAAS,IAAK,GAAS,GAAoB,EAAM,EAAW,EAAW,EAAY,CAAgB,CAAC,CACtG,EACM,EAAS,EAAS,KAAK,EAAM,IAAU,GAAY,EAAM,EAAqB,IAAU,CAAC,EAAG,CAAS,CAAC,EACtG,EAAS,EAAuB,EAAQ,CAAW,EAErD,EAAW,KACb,GAAgB,EAAY,EAAW,EAAQ,EAAQ,EAAQ,CAAQ,EAEvE,GAAgB,EAAY,EAAW,EAAQ,EAAQ,CAAQ,EAG7D,EAAO,OAAS,EAClB,QAAQ,SAAW,EACV,EAAO,WAAW,OAAS,IACpC,QAAQ,SAAW,EAEvB,CAGA,MAAM,EAAa,mBAQnB,eAAe,GACb,EACA,EACA,EACA,EACA,EACwB,CACxB,IAAM,EAAe,IAAI,IAAI,EAAa,QAAS,GAAY,CAAC,EAAQ,SAAU,EAAQ,QAAQ,CAAC,CAAC,EAC9F,EAA4B,CAAC,EACnC,IAAK,GAAM,CAAE,eAAc,UAAU,EAAc,CACjD,IAAM,EAAe,EAAK,KAAK,EAAU,CAAY,EACrD,GAAI,EAAa,IAAI,CAAY,EAC/B,SAIF,IAAM,EACJ,EAAc,gBAAgB,EAAc,CAAU,GACtD,CAAC,EAAc,gBAAgB,EAAc,CAAU,GAEvD,EAAc,eAAe,CAAY,GACxC,GAAiB,EAAc,gBAAgB,EAAc,MAAM,EAAS,EAAK,KAAM,MAAM,CAAC,IAE/F,EAAW,KAAK,CAAE,OAAQ,WAAY,SAAU,EAAc,SAAU,CAAa,CAAC,CAE1F,CACA,OAAO,CACT,CAMA,SAAS,EACP,EACA,EACA,EACa,CACb,GAAI,IAAe,IAAA,GACjB,sBAAO,IAAI,IAEb,IAAM,kBAAQ,IAAI,IAAI,CAAC,CAAU,CAAC,EAClC,IAAK,IAAM,KAAW,EAChB,EAAQ,WAAa,IAAA,IAAa,EAAK,KAAK,EAAU,EAAQ,QAAQ,IAAM,GAC9E,EAAM,IAAI,EAAK,KAAK,EAAU,EAAQ,QAAQ,CAAC,EAGnD,OAAO,CACT,CAGA,eAAe,EAAmB,EAAqF,CACrH,GAAI,CACF,MAAO,CAAE,gBAAiB,MAAM,EAAS,CAAc,EAAG,aAAc,EAAK,CAC/E,MAAQ,CACN,MAAO,CAAE,gBAAiB,EAAgB,aAAc,EAAM,CAChE,CACF,CAoBA,eAAe,EACb,EACA,EACA,EACA,EACyB,CACzB,IAAM,EAAa,IAAI,IAAI,EAAQ,aAAa,KAAK,CAAE,eAAc,UAAW,CAAC,EAAc,CAAI,CAAC,CAAC,EAG/F,EAAU,MAAM,EAAgB,EAAc,EAAG,qBAAqB,EAAI,EAAG,KAAO,IAAY,CACpG,IAAM,EAAuB,CAAC,EACxB,EAAyB,CAAC,EAEhC,MAAO,CAAE,KAAA,MADU,EAAmB,EAAS,EAAS,EAAY,EAAY,CAAY,EAC7E,aAAY,cAAa,CAC1C,CAAC,EACK,EAA2B,CAAC,EAClC,IAAK,GAAM,CAAE,OAAM,aAAY,kBAAkB,EAC/C,EAAO,KAAK,GAAG,CAAU,EACzB,EAAS,KAAK,GAAG,CAAY,EACzB,GACF,EAAS,KAAK,CAAI,EAGtB,OAAO,CACT,CAEA,eAAe,EACb,EACA,EACA,EACA,EACA,EACmC,CAGnC,IAAM,EAAmB,EAAK,KAAK,EAAQ,SAAU,EAAQ,QAAQ,EAC/D,EACJ,EAAQ,SAAW,WACnB,EAAc,EAAQ,SAAU,EAAQ,QAAS,EAAQ,cAAc,IAAI,CAAgB,CAAC,GAC5F,CAAC,EAAQ,cAAc,eAAe,CAAgB,GACtD,CAAC,EAAQ,mBAAmB,IAAI,CAAgB,GAChD,CAAE,MAAM,EAAe,CAAgB,EAInC,EAAmB,EAAQ,WAAa,IAAA,GAAY,GAAK,EAAK,KAAK,EAAQ,SAAU,EAAQ,QAAQ,EACrG,EACJ,EAAQ,WAAa,IAAA,IACrB,EAAc,EAAQ,SAAU,EAAQ,QAAS,EAAQ,cAAc,IAAI,CAAgB,CAAC,GAC5F,CAAC,EAAQ,cAAc,eAAe,CAAgB,GACtD,CAAC,EAAQ,iBAAiB,IAAI,EAAQ,QAAQ,EAChD,GAAI,CAAC,GAAiB,CAAC,EACrB,OAGF,IAAM,EAAc,EAAQ,SAAW,UAAa,EAAQ,SAAsB,EAAQ,SACpF,EAAW,EAAgB,EAAW,IAAI,EAAQ,QAAQ,EAAI,IAAA,GACpE,GAAI,GAAiB,CAAC,EAAU,CAC9B,EAA4B,EAAQ,SAAU,EAAQ,WAAY,CAAM,EACxE,MACF,CAEA,IAAM,EAAqB,CACzB,UACA,cAGA,MACE,GAAiB,EAAQ,SAAW,UAChC,EAAe,EAAK,KAAK,EAAQ,SAAU,CAAW,EAAG,EAAQ,eAAe,EAChF,GACN,UACF,EAEA,GAAI,EAAe,CACjB,IAAM,EAAU,MAAM,EAAoB,EAAM,EAAQ,SAAoB,EAAS,EAAQ,CAAQ,EAGrG,GAAI,IAAY,UAAa,IAAY,aAAe,CAAC,EACvD,MAEJ,CAEA,GAAI,CACE,IAEF,EAAK,YAAc,MAAM,EAAS,EAAS,KAAM,MAAM,EACvD,EAA0B,EAAM,EAAK,YAAa,EAAS,CAAQ,GAErE,EAAK,MAAQ,MAAM,EAAsB,EAAM,CAAO,CACxD,OAAS,EAAO,CACd,EAAO,KAAK,GAAG,EAAY,IAAI,EAAgB,CAAK,GAAG,EACvD,MACF,CACA,OAAO,CACT,CAEA,eAAe,EAAsB,EAAoB,EAA2C,CAClG,GAAM,CAAE,UAAS,cAAa,eAAgB,EAI9C,OAHI,IAAgB,IAAA,IAAa,IAAgB,IAAA,GACxC,MAAM,EAAc,EAAQ,SAAU,EAAQ,UAAW,EAAQ,SAAoB,EAAQ,QAAQ,EAEvG,CACL,CACE,UAAW,EACX,UAAW,IAAgB,IAAA,GAAY,EAAI,EAAW,CAAW,CAAC,CAAC,OACnE,UAAW,EACX,UAAW,IAAgB,IAAA,GAAY,EAAI,EAAW,CAAW,CAAC,CAAC,MACrE,CACF,CACF,CAGA,SAAS,EAAW,EAA2B,CAC7C,OAAO,EAAQ,MAAM,aAAa,CACpC,CAQA,SAAS,EAA4B,EAAkB,EAA+B,EAAwB,CACvG,EAAW,KAAM,GAAU,EAAM,WAAW,GAAG,EAAS,EAAE,CAAC,GAC9D,EAAO,KAAK,GAAG,EAAS,gCAAgC,CAE5D,CAEA,SAAS,EACP,EACA,EACA,EACA,EACM,CACN,GAAI,CACF,EAAK,mBAAqB,EAA8B,EAAa,CACnE,SAAU,EAAW,EAAK,QAAQ,SAAU,CAAO,EACnD,YAAa,EAAQ,QAAQ,WAC/B,CAAC,CACH,OAAS,EAAO,CAEd,EAAS,KAAK,GAAG,EAAK,YAAY,0CAA0C,EAAgB,CAAK,GAAG,CACtG,CACF,CASA,eAAe,EACb,EACA,EACA,EACA,EACA,EAC8C,CAC9C,IAAM,EAAiB,CACrB,SAAU,EAAW,EAAU,CAAO,EACtC,YAAa,EAAQ,QAAQ,WAC/B,EACI,EACJ,GAAI,CAEF,GADA,EAAc,MAAM,EAAmB,EAAQ,SAAU,EAAQ,UAAW,CAAQ,EAChF,EAAQ,cAAc,gBAAgB,EAAK,KAAK,EAAQ,SAAU,CAAQ,EAAG,CAAW,EAC1F,MAAO,YAET,IAAM,EAAW,MAAM,EAAyB,EAAa,CAAc,EAC3E,EAAK,YAAc,EAAS,QAC5B,EAAK,YAAc,EACnB,EAAK,eAAiB,EAAS,cAC3B,EAAS,iBAAmB,IAAA,IAC9B,EAAS,KAAK,GAAG,EAAS,wDAAwD,EAAS,gBAAgB,CAE/G,OAAS,EAAO,CAEd,OADA,EAAO,KAAK,GAAG,EAAS,oBAAoB,EAAgB,CAAK,GAAG,EAC7D,QACT,CACA,GAAI,CACF,EAAK,mBAAqB,EAA8B,EAAa,CAAc,CACrF,OAAS,EAAO,CACd,EAAS,KAAK,GAAG,EAAS,0DAA0D,EAAgB,CAAK,GAAG,CAC9G,CACA,MAAO,UACT,CAGA,SAAS,EAAW,EAAsB,EAAoC,CAC5E,IAAM,EAAiB,EAAQ,cAAc,IAAI,EAAK,KAAK,EAAQ,SAAU,CAAY,CAAC,EAC1F,OAAO,EAAY,EAAc,EAAQ,QAAS,CAAc,CAClE,CAEA,eAAe,EAAe,EAAwC,CAEpE,OAAO,MADa,EAAM,CAAY,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EACxC,eAAe,GAAK,EACpC,CAGA,eAAe,GAAuB,EAAoC,CACxE,IAAI,EAAU,EACd,OAAa,CAEX,IAAI,MADgB,EAAK,CAAO,CAAC,CAAC,UAAY,CAAC,CAAC,EAAA,EACrC,YAAY,EACrB,OAAO,EAET,IAAM,EAAS,EAAK,QAAQ,CAAO,EACnC,GAAI,IAAW,EACb,OAAO,EAET,EAAU,CACZ,CACF,CAEA,SAAS,EAAe,EAAmB,EAAkC,CAC3E,IAAM,EAAW,EAAK,SAAS,EAAiB,CAAS,EACzD,OAAO,IAAa,IAAO,CAAC,EAAS,WAAW,KAAK,EAAK,KAAK,GAAK,IAAa,MAAQ,CAAC,EAAK,WAAW,CAAQ,CACpH,CAEA,SAAS,GACP,EACA,EACA,EACsF,CACtF,IAAM,EAAc,EAAa,SAAS,CAAE,eAAc,UACxD,EAAK,sBAAwB,CAAC,CAAE,KAAM,EAAc,GAAG,EAAK,qBAAsB,CAAC,EAAI,CAAC,CAC1F,EAEM,EAAmB,IAAI,IAC3B,EAAS,QAAS,GAAU,EAAK,QAAQ,SAAW,UAAY,CAAC,EAAI,CAAC,EAAK,QAAQ,QAAQ,CAAE,CAC/F,EACM,EAAc,EAAY,OAAQ,GAAW,CAAC,EAAiB,IAAI,EAAO,IAAI,CAAC,EACrF,IAAK,IAAM,KAAQ,EACb,EAAK,gBAAkB,EAAK,QAAQ,WAAa,IAAA,IACnD,EAAY,KAAK,CAAE,KAAM,EAAK,QAAQ,SAAU,GAAG,EAAK,cAAe,CAAC,EAI5E,MAAO,CACL,UAAW,EAAY,QAAU,EAAI,EAA4B,EAAa,EAAQ,WAAW,EAAI,IAAA,GACrG,UAAW,EAAY,QAAU,EAAI,EAA4B,EAAa,EAAQ,WAAW,EAAI,IAAA,EACvG,CACF,CAEA,SAAS,GACP,EACA,EACA,EACA,EACA,EACmB,CACnB,IAAM,EAAkB,EAAmB,EAAK,UAAU,QAAS,EAAW,EAAK,QAAQ,QAAQ,EACnG,MAAO,CACL,UAAW,EAAK,cAAgB,IAAA,GAAY,IAAA,GAAY,EAAW,EAAK,WAAW,EACnF,UAAW,EAAK,cAAgB,IAAA,GAAY,IAAA,GAAY,EAAW,EAAK,WAAW,EACnF,oBACE,EAAK,cAAgB,IAAA,IAAa,EAAK,QAAQ,WAAa,IAAA,mBACxD,IAAI,IACJ,EAA6B,EAAK,YAAa,EAAW,EAAK,QAAQ,QAAQ,EACrF,oBACE,EAAK,WAAa,IAAA,mBACd,IAAI,IACJ,EACE,EAA6B,EAAK,SAAS,QAAS,EAAW,EAAK,QAAQ,QAAQ,EACpF,EAAK,QAAQ,SACb,EACA,CACF,EACN,gBAAiB,EAAmB,EAAK,YAAa,EAAW,EAAK,QAAQ,SAAU,CAAU,EAClG,kBACA,MAAO,EAAK,OAAS,CAAC,CACxB,CACF,CASA,SAAS,GAAyB,EAA2E,CAC3G,IAAM,EAAe,IAAI,IACvB,EAAa,KAAK,CAAE,eAAc,UAAW,CAAC,EAAc,EAAK,uBAAuB,MAAM,CAAC,CACjG,EACM,kBAAmB,IAAI,IAC7B,MAAQ,IAAS,CACf,GAAI,CAAC,EAAiB,IAAI,CAAI,EAAG,CAC/B,IAAM,EAAS,EAAa,IAAI,CAAI,EACpC,EAAiB,IAAI,EAAM,GAAU,EAAU,CAAM,CAAC,CACxD,CACA,OAAO,EAAiB,IAAI,CAAI,CAClC,CACF,CAMA,SAAS,EAAU,EAA0C,CAC3D,IAAM,kBAAS,IAAI,IACb,kBAA0B,IAAI,IACpC,IAAK,IAAM,KAAS,EAAQ,CAC1B,IAAM,EAAO,EAAM,SAAW,EAC1B,EAAO,EAAM,KACjB,GAAI,EAAM,OAAS,KAAM,CAEvB,IAAM,EAAoB,EAAwB,IAAI,CAAI,mBAAK,IAAI,IACnE,EAAwB,IAAI,EAAM,CAAiB,EAC9C,EAAkB,IAAI,EAAM,IAAI,GACnC,EAAkB,IAAI,EAAM,KAAM,EAAkB,IAAI,EAE1D,EAAO,SAAS,EAAkB,IAAI,EAAM,IAAI,GAClD,CACA,EAAO,IAAI,EAAM,GAAG,EAAO,IAAI,CAAI,GAAK,KAAK,EAAK,OAAO,CAC3D,CACA,IAAM,kBAAa,IAAI,IACvB,IAAK,IAAM,KAAa,EAAO,OAAO,EACpC,EAAW,IAAI,GAAY,EAAW,IAAI,CAAS,GAAK,GAAK,CAAC,EAEhE,MAAO,CAAE,SAAQ,YAAW,CAC9B,CAOA,SAAS,EACP,EACA,EACA,EACA,EACa,CACb,IAAM,EAAM,EAAiB,CAAI,EACjC,GAAI,IAAQ,IAAA,GACV,OAAO,IAAI,IAAI,CAAK,EAEtB,IAAM,GAAW,EAAmB,IAA6B,CAC/D,GAAI,IAAY,GACd,OAAQ,EAAI,WAAW,IAAI,CAAS,GAAK,GAAK,EAEhD,IAAM,EAAoB,EAAiB,CAAO,EAClD,OAAO,IAAsB,IAAA,IAAa,EAAkB,WAAW,IAAI,CAAS,CACtF,EACM,EAAiB,EAAoB,CAAW,EACtD,OAAO,IAAI,IACT,CAAC,GAAG,CAAK,CAAC,CAAC,OAAQ,GAAS,CAC1B,IAAM,EAAY,EAAI,OAAO,IAAI,CAAI,EACrC,OACE,IAAc,IAAA,IAAa,CAAC,GAAI,EAAe,IAAI,CAAI,GAAK,CAAC,CAAE,CAAC,CAAC,KAAM,GAAY,EAAQ,EAAW,CAAO,CAAC,CAElH,CAAC,CACH,CACF,CAMA,SAAS,EACP,EACA,EACA,EACA,EAA8C,GAAY,EACvC,CACnB,GAAI,IAAY,IAAA,IAAa,IAAS,IAAA,GACpC,MAAO,CAAC,EAEV,IAAM,EAAa,EAAQ,YAAY,qBAAqB,QAAS,GACnE,EAAM,KAAK,CAAE,YAAW,cAAe,CAAE,YAAW,UAAS,SAAU,CAAC,EAAE,CAAE,EAAE,CAChF,EACM,GAAa,GAAO,QAAU,CAAC,EAAA,CAAG,QAAS,GAAU,CACzD,IAAM,EAAM,EAAM,YAAY,OAAQ,GAAe,EAAW,OAAS,CAAI,EACvE,EAAW,CACf,GAAG,EAAM,MAAM,OAAQ,GAAY,IAAY,CAAI,CAAC,CAAC,IAAI,CAAa,EACtE,GAAI,EAAI,OAAS,EAAI,CAAC,EAAE,EAAI,CAAC,CAC/B,EACA,OAAO,EAAI,KAAK,CAAE,YAAW,cAAe,CAAE,YAAW,UAAS,UAAS,EAAE,CAC/E,CAAC,EACD,MAAO,CAAC,GAAG,EAAY,GAAG,CAAS,CACrC,CAEA,SAAS,GACP,EACA,EACA,EACe,CACf,MAAO,CACL,KAAM,EAAK,YACX,YAAa,EAAK,YAClB,YAAa,EAAK,UAAU,QAC5B,mBAAoB,EAAK,mBACzB,mBAAoB,EAAK,mBACzB,uBACA,oBAAqB,EAAgB,EAAM,EAAsB,CAAS,EAC1E,MAAO,EAAK,KACd,CACF,CAGA,SAAS,EACP,EACA,EACA,EACU,CACV,IAAM,EAAW,EAAK,QAAQ,SACxB,EAAU,GACd,EAAW,OAAS,GAAY,EAAM,KAAM,GAAS,GAAQ,EAAW,WAAa,GAAQ,EAAW,OAAO,EAC3G,kBAAW,IAAI,IACrB,IAAK,IAAM,KAAS,GAAW,QAAU,CAAC,EACxC,GAAI,EAAM,YAAY,KAAK,CAAM,EAC1B,IAAA,IAAM,KAAW,EAAM,MACtB,IAAY,GACd,EAAS,IAAI,CAAO,EAK5B,MAAO,CAAC,GAAG,CAAQ,CAAC,CAAC,SAAS,CAChC,CAEA,SAAS,GACP,EACA,EACA,EACA,EACA,EACM,CACN,IAAM,EAAY,EAAU,MAAM,EAAG,EAAE,EACnC,EAAO,OAAS,GAElB,EACE,uCAAuC,EAAO,OAAO,2CAChD,EAAO,WAAW,OAAS,EAAI,KAAK,EAAO,WAAW,OAAO,mCAAqC,GAAG,SAAS,EAAW,KAAK,eAAe,EAAU,KAC9J,EACA,EAAgB,CAAM,GACb,EAAO,WAAW,SAAW,EACtC,EACE,2BAA2B,EAAO,iBAAiB,kBAAkB,EAAO,qBAAqB,2BAA2B,EAAW,KAAK,eAAe,EAAU,KACvK,GAEA,EACE,sBAAsB,EAAW,KAAK,eAAe,EAAU,KAAK,EAAO,WAAW,OAAO,cAC/F,EACA,EAAgB,CAAM,GAGpB,EAAW,MACb,EAAiB,CAAM,EAGzB,IAAK,IAAM,KAAW,EACpB,EAAY,YAAY,EAAQ,GAAG,EAErC,IAAK,IAAM,KAAS,EAClB,EAAY,UAAU,EAAM,GAAG,CAEnC,CAEA,SAAS,EAAgB,EAA0B,CACjD,IAAK,GAAM,CAAC,EAAO,KAAc,EAAO,WAAW,QAAQ,EACzD,EAAY,GAAG,EAAQ,EAAE,IAAI,EAAU,QAAQ,GAAG,CAEtD,CAGA,SAAS,EAAiB,EAA0B,CAC9C,KAAO,iBAAiB,SAAW,EAGvC,GAAY;;CAAuC,EACnD,IAAK,IAAM,KAAU,EAAO,iBAC1B,EAAY,KAAK,EAAqB,CAAM,EAAE,GAAG,CAFA,CAIrD,CAEA,SAAS,EAAqB,EAAuC,CACnE,IAAM,GACJ,EACA,EAAoC,SACzB,CACX,IAAM,EAAO,EAAO,EAAO,EAAO,IAAI,CAAC,EACvC,OAAO,EAAO,KAAO,GAAG,EAAO,EAAO,EAAO,IAAI,CAAC,EAAE,MAAM,IAAS,CACrE,EACM,EAAS,CACb,aAAa,EAAO,GAAO,EAAG,mBAAmB,IACjD,QAAQ,EAAO,GAAO,EAAG,IAAI,IAC7B,WAAW,EAAO,GAAO,EAAG,YAAY,IACxC,aAAa,EAAO,GAAO,EAAG,SAAS,IACvC,UAAU,EACP,GAAO,EAAG,eACV,GAAU,EAAM,QAAQ,CAAC,CAC5B,GACF,EACA,MAAO,GAAG,EAAO,KAAK,GAAG,EAAO,UAAU,GAAG,EAAO,QAAQ,GAAG,EAAO,OAAO,EAAO,KAAO,GAAK,SAAS,IAAI,EAAO,KAAK,IAAI,GAC/H,CAEA,SAAS,GACP,EACA,EACA,EACA,EACA,EACA,EACM,CACN,IAAM,EAAkC,CACtC,KAAM,EAAW,KACjB,YACA,OAAQ,EAAO,WAAW,SAAW,GAAK,EAAO,SAAW,EAC5D,WAAY,EAAO,WACnB,iBAAkB,EAAO,iBACzB,qBAAsB,EAAO,qBAC7B,iBAAkB,EAAO,iBACzB,SACA,UACF,EACI,EAAW,OACb,EAAO,MAAQ,EACZ,OAAQ,GAAU,EAAM,QAAU,EAAK,CAAC,CACxC,IAAK,IAAW,CACf,KAAM,EAAM,KACZ,kBAAmB,EAAM,aAAa,UAAU,QAAU,EAC1D,kBAAmB,EAAM,aAAa,UAAU,QAAU,EAC1D,SAAU,EAAM,aAAa,WAAa,EAC1C,SAAU,EAAM,aAAa,WAAa,EAC1C,2BAA4B,EAAM,aAAa,wBAA0B,EACzE,2BAA4B,EAAM,aAAa,wBAA0B,EACzE,qBAAsB,EAAM,qBAC5B,oBAAqB,EAAM,oBAC3B,UAAW,EAAO,iBAAiB,OAAQ,GAAO,EAAG,OAAS,EAAM,IAAI,CAC1E,EAAE,GAEN,EAAY,KAAK,UAAU,EAAQ,IAAA,GAAW,CAAC,EAAI;CAAI,CACzD"}

package/dist/duplication.cjs

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

package/dist/duplication.cjs.map

@@ -1 +1 @@
-{"version":3,"file":"duplication.cjs","names":[],"sources":["../src/duplication.ts"],"sourcesContent":["import type { DuplicationOptions } from './types.js';\n\n/**\n * Project-level duplication machinery operating on normalized token streams. Tokenization itself\n * (parsing, identifier anonymization, literal normalization) happens in the Rust addon, which\n * serializes each file's Token stream and statement structure; the helpers here match statement\n * windows across files, merge gap-adjacent groups, and count duplicated lines over that data.\n */\n\nexport const defaultDuplicationOptions: Required<DuplicationOptions> = {\n  minTokens: 40,\n  maxGapTokens: 30,\n  minSimilarityPercent: 70,\n};\n\n/**\n * Fills defaults for absent settings, applying the same normalization as the native boundary's\n * clampToU32 — NaN (e.g. `Number(unsetEnvVariable)`) counts as absent, and other values truncate\n * and clamp to [0, u32::MAX] — so the TypeScript half of cross-file matching cannot diverge from\n * the natively collected candidates on such input.\n */\nexport function resolveDuplicationOptions(options?: DuplicationOptions): Required<DuplicationOptions> {\n  return {\n    minTokens: resolveOption(options?.minTokens, defaultDuplicationOptions.minTokens),\n    maxGapTokens: resolveOption(options?.maxGapTokens, defaultDuplicationOptions.maxGapTokens),\n    minSimilarityPercent: resolveOption(options?.minSimilarityPercent, defaultDuplicationOptions.minSimilarityPercent),\n  };\n}\n\nfunction resolveOption(value: number | undefined, fallback: number): number {\n  return value === undefined || Number.isNaN(value)\n    ? fallback\n    : Math.min(Math.max(Math.trunc(value), 0), 0xFF_FF_FF_FF);\n}\n\n/** Minimum consecutive statements for a statement-sequence duplicate candidate. */\nconst minSequenceStatementCount = 2;\n/**\n * Caps the window length so statement-sequence enumeration stays linear in the statement count.\n * Heterogeneous clones longer than the cap are reported as capped windows (a deliberate\n * conservative undercount trading completeness for bounded discovery cost).\n */\nconst maxSequenceStatementCount = 100;\n\n/**\n * A region whose normalized tokens are at least 20% literal values is data-like (a lookup table, a\n * constant list, a value-mapping switch), not logic: literal values re-enter its fingerprint so\n * tables that merely share their shape stop counting as copy-paste. Compared in integer math\n * (5 * literals >= total) so the TypeScript and native sides cannot disagree on the boundary.\n */\nfunction isLiteralDense(literalCount: number, tokenCount: number): boolean {\n  return literalCount * 5 >= tokenCount;\n}\n\nexport interface Token {\n  /** Normalization target: identifiers to anonymize, literal kind tags, or the raw token text. */\n  kind: 'id' | 'text';\n  text: string;\n  /**\n   * Two INDEPENDENT hashes of `text` (djb2 and FNV-1a), precomputed so fingerprinting nested\n   * regions never re-hashes a token. Feeding the same per-token hash to both fingerprint\n   * accumulators would collapse the key to 32 effective bits: one djb2 collision between two\n   * token texts would then equate whole regions.\n   */\n  textHash: number;\n  textHash2: number;\n  /** Hash pair of a value-carrying literal's value, folded into data-like region fingerprints. */\n  literalHash?: number;\n  literalHash2?: number;\n  /**\n   * True for verbatim-kept NAMES (member/callee/type names, named grammar leaves): together with\n   * value-carrying literals these are the content-bearing tokens the near-miss content gate\n   * counts. Keywords, operators, and punctuation come from unnamed nodes and stay false.\n   */\n  isName?: boolean;\n  /** 0-based source rows the token occupies, so line coverage counts only matched-token lines. */\n  startRow: number;\n  endRow: number;\n}\n\n/** A token span with its source position; plain data so cross-file matching can retain it. */\nexport interface TokenRange {\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A contiguous run of matched tokens; gapped (merged) duplicates carry several per occurrence. */\ninterface TokenSegment {\n  startTokenIndex: number;\n  endTokenIndex: number;\n}\n\ninterface DuplicateCandidate {\n  fingerprint: string;\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\nexport interface CountedOccurrence {\n  /** Matched token runs; more than one once gapped groups are merged. */\n  segments: TokenSegment[];\n  /**\n   * Set on occurrences whose span another reported group already counts: a retained group's\n   * occurrences that a partial gapped merge also paired into a merged group, and cross-file copies\n   * nested inside a larger group's region. Block counting must not count them again.\n   */\n  spanCountedElsewhere?: boolean;\n  /**\n   * Set on cross-file copies nested inside a larger group's region (they also set\n   * `spanCountedElsewhere`). They never pair in gapped merging, and they do not keep their group's\n   * standalone copies from merging, which they are not copies of. They do keep the merged group\n   * from taking their group's place, since only the original group reports the nesting.\n   */\n  nestedInLargerGroup?: boolean;\n  /** Sum of segment token counts (the gap tokens are not matched content). */\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A duplicate region found in one file, exported for cross-file matching by fingerprint. */\nexport interface CrossFileDuplicateCandidate {\n  /** Content key: equal fingerprints mean equal normalized token sequences (up to hash collision). */\n  fingerprint: string;\n  tokenCount: number;\n  /** Token positions within the owning file, for cross-file gapped (Type-3) merging. */\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/**\n * One file's contribution to cross-file clone detection: catalogued candidates plus the normalized\n * token stream and statement structure, so the project-level pass can match partial statement runs\n * (windows that a single file cannot know repeat elsewhere) and merge gap-adjacent groups.\n */\nexport interface CrossFileDuplicationFileData {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  /**\n   * The mutually disjoint blocks (at least `minTokens` long, not literal-dense) compared for\n   * cross-file near-miss (Type-3) clones. Optional for backward compatibility; without it, the\n   * file takes part in exact and gapped cross-file matching only.\n   */\n  nearMissBlocks?: TokenRange[];\n  /**\n   * 1-based lines that are neither blank nor comment-only, so cross-file line coverage counts only\n   * code lines (blank rows inside multi-row tokens such as template literals carry no content).\n   * Optional for backward compatibility; without it, every matched-token row counts.\n   */\n  codeLineNumbers?: Set<number>;\n}\n\n/** literalCountPrefix[i] = value-carrying literal tokens in tokens[0..i), for O(1) density checks. */\nexport function buildLiteralCountPrefix(tokens: Token[]): Int32Array {\n  const prefix = new Int32Array(tokens.length + 1);\n  for (const [index, token] of tokens.entries()) {\n    prefix[index + 1] = (prefix[index] ?? 0) + (token.literalHash === undefined ? 0 : 1);\n  }\n  return prefix;\n}\n\ninterface WindowOccurrences {\n  count: number;\n  /** -1 once occurrences span more than one container. */\n  containerIndex: number;\n  /** -1 once occurrences span more than one context (file). */\n  contextIndex: number;\n  minStart: number;\n  maxStart: number;\n}\n\ninterface SequenceWindow {\n  containerIndex: number;\n  start: number;\n  length: number;\n}\n\n/** One file's token stream and statement containers, as a window-matching context. */\nexport interface SequenceWindowContext {\n  tokens: Token[];\n  literalCountPrefix: Int32Array;\n  containers: TokenRange[][];\n}\n\nexport interface ContextualSequenceCandidate {\n  candidate: DuplicateCandidate;\n  contextIndex: number;\n}\n\n/**\n * Enumerates runs of consecutive sibling statements over one or more contexts (files). Every\n * container statement participates; only the window length is capped, so enumeration stays linear\n * in the statement count. Windows are grouped by a cheap rolling hash of per-statement\n * fingerprints, and only locally maximal repeated windows — those whose one-statement extensions\n * stop repeating — become candidates with an exact (window-consistent) fingerprint. Without the\n * maximality filter a degenerate file of near-identical statements would fingerprint every\n * sub-window of every repeated region. With `requireMultipleContexts` a window only counts as\n * repeated when its occurrences span at least two contexts (CPD-style cross-file matching): a\n * repeat confined to one file is that file's own concern, and emitting it here would flood the\n * project-level selection with unusable single-file groups.\n */\nexport function collectSequenceWindowCandidates(\n  contexts: SequenceWindowContext[],\n  minTokens: number,\n  requireMultipleContexts: boolean\n): ContextualSequenceCandidate[] {\n  const candidates: ContextualSequenceCandidate[] = [];\n  const contextIndexByContainer: number[] = [];\n  const containers: TokenRange[][] = [];\n  for (const [contextIndex, context] of contexts.entries()) {\n    for (const statements of context.containers) {\n      contextIndexByContainer.push(contextIndex);\n      containers.push(statements);\n    }\n  }\n  const contextAt = (containerIndex: number): SequenceWindowContext | undefined =>\n    contexts[contextIndexByContainer[containerIndex] ?? 0];\n  const occurrencesByWindowKey = new Map<number, WindowOccurrences>();\n  const containerWindows = containers.map((statements, containerIndex) =>\n    enumerateContainerWindows(contextAt(containerIndex)?.tokens ?? [], statements, minTokens)\n  );\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    const contextIndex = contextIndexByContainer[containerIndex] ?? 0;\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const windowKey of row) {\n        if (windowKey === undefined) {\n          continue;\n        }\n        const occurrences = occurrencesByWindowKey.get(windowKey);\n        if (occurrences) {\n          occurrences.count += 1;\n          if (occurrences.containerIndex !== containerIndex) {\n            occurrences.containerIndex = -1;\n          }\n          if (occurrences.contextIndex !== contextIndex) {\n            occurrences.contextIndex = -1;\n          }\n          occurrences.minStart = Math.min(occurrences.minStart, start);\n          occurrences.maxStart = Math.max(occurrences.maxStart, start);\n        } else {\n          occurrencesByWindowKey.set(windowKey, {\n            count: 1,\n            containerIndex,\n            contextIndex,\n            minStart: start,\n            maxStart: start,\n          });\n        }\n      }\n    }\n  }\n\n  // A window only \"repeats\" when two of its occurrences can coexist without overlapping: sliding\n  // matches inside a homogeneous run (start spread smaller than the window length) can never both\n  // be counted and must neither qualify a window nor dominate its sub-windows. Cross-context\n  // matching instead requires occurrences in two contexts, which coexist by construction.\n  const repeats = (windowKey: number | undefined, length: number): boolean => {\n    if (windowKey === undefined) {\n      return false;\n    }\n    const occurrences = occurrencesByWindowKey.get(windowKey);\n    if (occurrences === undefined || occurrences.count < 2) {\n      return false;\n    }\n    if (requireMultipleContexts) {\n      return occurrences.contextIndex === -1;\n    }\n    return occurrences.containerIndex === -1 || occurrences.maxStart - occurrences.minStart >= length;\n  };\n\n  // A window whose statements all share one normalized shape (sixteen `let x = 0;` declarations,\n  // a constant table) is a homogeneous preamble, not a copy-paste: requiring two distinct\n  // per-statement shapes keeps such runs out of duplicate groups and the duplication ratio.\n  const hasDistinctStatements = (window: SequenceWindow): boolean => {\n    const hashes = containerWindows[window.containerIndex]?.statementHashes ?? [];\n    const firstHash = hashes[window.start];\n    for (let index = window.start + 1; index < window.start + window.length; index += 1) {\n      if (hashes[index] !== firstHash) {\n        return true;\n      }\n    }\n    return false;\n  };\n\n  const maximalWindows: SequenceWindow[] = [];\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const [length, windowKey] of row.entries()) {\n        if (!repeats(windowKey, length) || !hasDistinctStatements({ containerIndex, start, length })) {\n          continue;\n        }\n        // Dominated windows are skipped: the one-statement extension also repeats, so a larger\n        // candidate covering this window exists.\n        const extendedRight = windows.windowKeysByStart[start]?.[length + 1];\n        const extendedLeft = windows.windowKeysByStart[start - 1]?.[length + 1];\n        if (repeats(extendedRight, length + 1) || repeats(extendedLeft, length + 1)) {\n          continue;\n        }\n        maximalWindows.push({ containerIndex, start, length });\n      }\n    }\n  }\n\n  // The rolling hash anonymizes identifiers per statement, so a window can look repeated coarsely\n  // while its exact (window-consistent) fingerprints differ, and a longer window's match can\n  // dominate sub-windows that other copies still need (three copies where only two extend one\n  // statement further). Every emitted window therefore exposes its repeating, unvisited\n  // sub-windows; `visited` bounds the worklist and lengths strictly decrease, so it terminates.\n  const visited = new Set(maximalWindows.map(windowId));\n  let frontier = maximalWindows;\n  while (frontier.length > 0) {\n    const emitted: SequenceWindow[] = [];\n    for (const window of frontier) {\n      const statements = containers[window.containerIndex];\n      const first = statements?.[window.start];\n      const last = statements?.[window.start + window.length - 1];\n      const context = contextAt(window.containerIndex);\n      if (!first || !last || !context) {\n        continue;\n      }\n      const fingerprint = `s:${fingerprintKey(context.tokens, context.literalCountPrefix, first.startTokenIndex, last.endTokenIndex)}`;\n      candidates.push({\n        candidate: toCandidate(fingerprint, first.startTokenIndex, last.endTokenIndex, first, last),\n        contextIndex: contextIndexByContainer[window.containerIndex] ?? 0,\n      });\n      emitted.push(window);\n    }\n    frontier = [];\n    for (const window of emitted) {\n      for (const start of [window.start, window.start + 1]) {\n        const subWindow = { containerIndex: window.containerIndex, start, length: window.length - 1 };\n        const subWindowKey = containerWindows[window.containerIndex]?.windowKeysByStart[start]?.[subWindow.length];\n        if (\n          visited.has(windowId(subWindow)) ||\n          !repeats(subWindowKey, subWindow.length) ||\n          !hasDistinctStatements(subWindow)\n        ) {\n          continue;\n        }\n        visited.add(windowId(subWindow));\n        frontier.push(subWindow);\n      }\n    }\n  }\n  return candidates;\n}\n\nfunction windowId(window: SequenceWindow): string {\n  return `${window.containerIndex}:${window.start}:${window.length}`;\n}\n\ninterface ContainerWindows {\n  /** windowKeysByStart[start][length] is the rolling-hash key of the window, or undefined if it is below the size thresholds. */\n  windowKeysByStart: (number | undefined)[][];\n  /** Per-statement fingerprint hashes, for the distinct-shape requirement on windows. */\n  statementHashes: number[];\n}\n\nfunction enumerateContainerWindows(tokens: Token[], statements: TokenRange[], minTokens: number): ContainerWindows {\n  const statementHashes = statements.map((statement) =>\n    fingerprintHash(tokens, statement.startTokenIndex, statement.endTokenIndex)\n  );\n  const windowKeysByStart: (number | undefined)[][] = [];\n  for (let start = 0; start < statements.length; start += 1) {\n    const row: (number | undefined)[] = [];\n    let hash = 5381;\n    let tokenCount = 0;\n    const maxEnd = Math.min(statements.length, start + maxSequenceStatementCount);\n    for (let end = start; end < maxEnd; end += 1) {\n      const statement = statements[end];\n      const statementHash = statementHashes[end];\n      if (!statement || statementHash === undefined) {\n        break;\n      }\n      hash = combineHashes(hash, statementHash);\n      tokenCount += statement.endTokenIndex - statement.startTokenIndex;\n      const statementCount = end - start + 1;\n      row[statementCount] =\n        statementCount >= minSequenceStatementCount && tokenCount >= minTokens\n          ? combineHashes(hash, statementCount)\n          : undefined;\n    }\n    windowKeysByStart.push(row);\n  }\n  return { windowKeysByStart, statementHashes };\n}\n\n/**\n * Longest-common-subsequence LENGTH of two symbol sequences via the Allison–Dix bit-parallel\n * recurrence (O(|a|/32 · |b|) words): per symbol of `b`, `x = match | v` and\n * `v = x & ~(x - ((v << 1) | 1))` over multi-word bit vectors; the set bits of `v` count the LCS.\n */\nexport function lcsLength(a: Int32Array, b: Int32Array): number {\n  return createLcsLengthCounter(a)(b);\n}\n\n/**\n * lcsLength against a fixed `a`, whose per-symbol position masks are built once and reused for\n * every `b` compared with it.\n */\nexport function createLcsLengthCounter(a: Int32Array): (b: Int32Array) => number {\n  const wordCount = (a.length + 31) >>> 5;\n  const positionMasks = new Map<number, Uint32Array>();\n  for (const [index, symbol] of a.entries()) {\n    let mask = positionMasks.get(symbol);\n    if (!mask) {\n      mask = new Uint32Array(wordCount);\n      positionMasks.set(symbol, mask);\n    }\n    const word = index >>> 5;\n    // The Uint32Array store wraps the signed int32 bit pattern to unsigned. The `?? 0` guards in\n    // this function are required by noUncheckedIndexedAccess (typed-array reads type as\n    // `number | undefined`), not redundancy: every index is in bounds.\n    mask[word] = (mask[word] ?? 0) | (1 << (index & 31));\n  }\n\n  const v = new Uint32Array(wordCount);\n  return (b) => {\n    v.fill(0);\n    for (const symbol of b) {\n      const matchMask = positionMasks.get(symbol);\n      // `(v << 1) | 1` shifts a carry bit across words; subtraction borrows across words.\n      let shiftCarry = 1;\n      let borrow = 0;\n      for (let word = 0; word < wordCount; word += 1) {\n        const previous = v[word] ?? 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- `>>> 0` reinterprets the signed int32 bit pattern as unsigned so the borrow subtraction below compares magnitudes; Math.trunc would keep it negative.\n        const x = ((matchMask?.[word] ?? 0) | previous) >>> 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- same unsigned reinterpretation as `x`.\n        const shifted = ((previous << 1) | shiftCarry) >>> 0;\n        shiftCarry = previous >>> 31;\n        const difference = x - shifted - borrow;\n        borrow = difference < 0 ? 1 : 0;\n        // The Uint32Array store wraps the signed int32 bit pattern to unsigned.\n        v[word] = x & ~difference;\n      }\n    }\n\n    let length = 0;\n    for (const word of v) {\n      length += popCount(word);\n    }\n    return length;\n  };\n}\n\nfunction popCount(value: number): number {\n  let count = value - ((value >>> 1) & 0x55_55_55_55);\n  count = (count & 0x33_33_33_33) + ((count >>> 2) & 0x33_33_33_33);\n  return (Math.imul((count + (count >>> 4)) & 0x0F_0F_0F_0F, 0x01_01_01_01) >>> 24) & 0xFF;\n}\n\nfunction toCandidate(\n  fingerprint: string,\n  startTokenIndex: number,\n  endTokenIndex: number,\n  first: TokenRange,\n  last: TokenRange\n): DuplicateCandidate {\n  return {\n    fingerprint,\n    tokenCount: endTokenIndex - startTokenIndex,\n    startTokenIndex,\n    endTokenIndex,\n    startIndex: first.startIndex,\n    endIndex: last.endIndex,\n    startLine: first.startLine,\n    endLine: last.endLine,\n  };\n}\n\n/** Caches of hashText/hashText2 over '$0', '$1', ... so anonymized identifiers hash without allocating. */\nconst anonymizedIndexHashes: number[] = [];\nconst anonymizedIndexHashes2: number[] = [];\n\nfunction anonymizedIndexHash(index: number): number {\n  let hash = anonymizedIndexHashes[index];\n  if (hash === undefined) {\n    hash = hashText(`$${index}`);\n    anonymizedIndexHashes[index] = hash;\n  }\n  return hash;\n}\n\nfunction anonymizedIndexHash2(index: number): number {\n  let hash = anonymizedIndexHashes2[index];\n  if (hash === undefined) {\n    hash = hashText2(`$${index}`);\n    anonymizedIndexHashes2[index] = hash;\n  }\n  return hash;\n}\n\n/**\n * Content key of a token range: two independent 32-bit hashes over the normalized token sequence\n * (identifiers anonymized consistently by first-occurrence order) plus the token count. Regions\n * with equal keys are treated as equal content; a collision would need both 32-bit hashes and the\n * length to coincide, which is negligible for a metrics report. The format and arithmetic match\n * fingerprint_key in native/src/duplication.rs exactly, so window candidates fingerprinted here\n * group together with the per-file candidates the addon catalogues.\n */\nfunction fingerprintKey(\n  tokens: Token[],\n  literalCountPrefix: Int32Array,\n  startTokenIndex: number,\n  endTokenIndex: number\n): string {\n  const literalCount = (literalCountPrefix[endTokenIndex] ?? 0) - (literalCountPrefix[startTokenIndex] ?? 0);\n  const literalDense = isLiteralDense(literalCount, endTokenIndex - startTokenIndex);\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, literalDense);\n  return `${primary}:${secondary}:${endTokenIndex - startTokenIndex}`;\n}\n\n/**\n * A single 32-bit summary of a range for the coarse rolling-hash phase. Deliberately\n * density-agnostic: density is a property of the final candidate REGION, and folding literal\n * values into per-statement hashes would make a dense statement inside a logic-heavy window\n * (`const weights = [1, 2, 3];`) block the window from ever being enumerated. The coarse phase\n * over-approximates on shape alone; the exact region fingerprint still applies the density rule.\n */\nfunction fingerprintHash(tokens: Token[], startTokenIndex: number, endTokenIndex: number): number {\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, false);\n  // XOR already coerces to int32, matching the native side's i32 arithmetic.\n  return primary ^ Math.imul(secondary, 31);\n}\n\nfunction fingerprintHashPair(\n  tokens: Token[],\n  startTokenIndex: number,\n  endTokenIndex: number,\n  foldLiteralValues: boolean\n): [number, number] {\n  const indexByIdentifier = new Map<string, number>();\n  let primary = 5381;\n  let secondary = 52_711;\n  for (let index = startTokenIndex; index < endTokenIndex; index += 1) {\n    const token = tokens[index];\n    if (!token) {\n      continue;\n    }\n    // Each accumulator consumes its own independent per-token hash: sharing one would collapse\n    // the key to 32 effective bits (a single djb2 collision would equate whole regions).\n    let part: number;\n    let part2: number;\n    if (token.kind === 'id') {\n      let identifierIndex = indexByIdentifier.get(token.text);\n      if (identifierIndex === undefined) {\n        identifierIndex = indexByIdentifier.size;\n        indexByIdentifier.set(token.text, identifierIndex);\n      }\n      part = anonymizedIndexHash(identifierIndex);\n      part2 = anonymizedIndexHash2(identifierIndex);\n    } else {\n      part = token.textHash;\n      part2 = token.textHash2;\n    }\n    // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n    primary = (Math.imul(primary, 31) + part) | 0;\n    secondary = Math.imul(secondary, 37) ^ part2;\n    if (foldLiteralValues && token.literalHash !== undefined && token.literalHash2 !== undefined) {\n      // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n      primary = (Math.imul(primary, 31) + token.literalHash) | 0;\n      secondary = Math.imul(secondary, 37) ^ token.literalHash2;\n    }\n  }\n  return [primary, secondary];\n}\n\n/** djb2-style hash; XOR keeps the value in signed 32-bit range, which is fine for a grouping key. */\nfunction hashText(text: string): number {\n  let hash = 5381;\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- djb2 hashes UTF-16 code units; codePointAt would hash surrogate pairs twice (full code point, then the lone low surrogate).\n    hash = Math.imul(hash, 33) ^ text.charCodeAt(index);\n  }\n  return hash;\n}\n\n/** FNV-1a over UTF-16 code units: independent of hashText so the two accumulators never share input. */\nfunction hashText2(text: string): number {\n  let hash = -2_128_831_035; // 2166136261 as int32 (the FNV-1a offset basis)\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- hashes UTF-16 code units like hashText.\n    hash = Math.imul(hash ^ text.charCodeAt(index), 16_777_619);\n  }\n  return hash;\n}\n\nfunction combineHashes(hash: number, value: number): number {\n  return Math.imul(hash, 31) + value;\n}\n\n/**\n * Merges duplicate groups separated by a small token gap into one gapped (Type-3) clone group: a\n * copy edited in one spot splits into two exact groups whose occurrences sit side by side in the\n * same order. Occurrences are paired greedily in source order; a merge happens when the pairing\n * fully pairs at least one group with at least two pairs. Equal-cardinality groups whose\n * occurrences all pair merge into one group as before. When cardinalities differ (a fragment also\n * occurs standalone: prefix ×3, suffix ×2), the fully-paired group is subsumed into the merged\n * gapped group while the other group is RETAINED with ALL its occurrences: dropping the leftover\n * would lose duplicated-line coverage, and reporting it alone would make a single-occurrence group\n * (contradicting duplicateBlockGroupCount's \"appears more than once\" meaning). Line coverage\n * unions ranges, so the overlap between the retained exact group and the merged group is harmless.\n * Merging repeats to a fixpoint so a clone edited in several spots still reassembles; it\n * terminates because a merge either removes the input groups it replaces or marks their paired\n * occurrences as counted elsewhere, and only unmarked occurrences of remaining groups pair, so the\n * number of pairable occurrences strictly decreases with every merge. Gap tokens are not matched\n * content: line coverage and sizes count only the matched segments. Generic so cross-file merging\n * can thread file identity through occurrences.\n */\nexport function mergeAdjacentGroups<T extends CountedOccurrence>(\n  groups: T[][],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean = () => true\n): T[][] {\n  if (maxGapTokens <= 0 || groups.length < 2) {\n    return groups;\n  }\n  // Deterministic processing order (mirrored by the native side): by first occurrence position.\n  groups.sort(compareGroups);\n  for (let restart = true; restart;) {\n    restart = false;\n    const partnersByGroup = collectGapAdjacentPartners(groups, maxGapTokens);\n    for (let leftIndex = 0; leftIndex < groups.length && !restart; leftIndex += 1) {\n      for (const rightIndex of partnersByGroup[leftIndex] ?? []) {\n        const left = groups[leftIndex];\n        const right = groups[rightIndex];\n        if (!left || !right) {\n          continue;\n        }\n        const forward = mergeGroups(left, right, maxGapTokens, isReportableGroup);\n        const result = forward ?? mergeGroups(right, left, maxGapTokens, isReportableGroup);\n        if (!result) {\n          continue;\n        }\n        const leftReplaced = forward ? result.firstReplaced : result.secondReplaced;\n        const rightReplaced = forward ? result.secondReplaced : result.firstReplaced;\n        if (leftReplaced && rightReplaced) {\n          groups[leftIndex] = result.merged;\n          groups.splice(rightIndex, 1);\n        } else if (rightReplaced) {\n          groups[rightIndex] = result.merged;\n        } else if (leftReplaced) {\n          groups[leftIndex] = result.merged;\n        } else {\n          // Both groups stay (each is only partly paired, or reports nested copies of its own), so\n          // the merged group joins them instead of taking a place.\n          groups.push(result.merged);\n        }\n        // A group that stays keeps ALL its occurrences (line coverage must not shrink, and a\n        // reported group must keep >= 2 occurrences), so its paired occurrences now also live\n        // inside the merged group's occurrences: mark them so duplicateBlockCount counts each\n        // token span once, and so the same pair cannot merge again.\n        for (const occurrence of result.pairedRetained) {\n          occurrence.spanCountedElsewhere = true;\n        }\n        groups.sort(compareGroups);\n        restart = true;\n        break;\n      }\n    }\n  }\n  return groups;\n}\n\n/**\n * Per group index, the ascending indexes of later groups that mergeGroups can pair with it in\n * either direction: an occurrence of one starts within `maxGapTokens` after an occurrence of the\n * other ends. Every other pair fails mergeGroups, so skipping it leaves the fixpoint unchanged\n * while each restart stops costing a comparison per pair of groups.\n */\nfunction collectGapAdjacentPartners(groups: CountedOccurrence[][], maxGapTokens: number): number[][] {\n  const starts: { startTokenIndex: number; groupIndex: number }[] = [];\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      starts.push({ startTokenIndex: occurrence.startTokenIndex, groupIndex });\n    }\n  }\n  starts.sort((left, right) => left.startTokenIndex - right.startTokenIndex);\n  const partners = groups.map(() => new Set<number>());\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      for (\n        let index = lowerBoundByStart(starts, occurrence.endTokenIndex);\n        index < starts.length && (starts[index]?.startTokenIndex ?? 0) <= occurrence.endTokenIndex + maxGapTokens;\n        index += 1\n      ) {\n        const other = starts[index]?.groupIndex ?? groupIndex;\n        if (other !== groupIndex) {\n          partners[Math.min(other, groupIndex)]?.add(Math.max(other, groupIndex));\n        }\n      }\n    }\n  }\n  return partners.map((set) => [...set].toSorted((left, right) => left - right));\n}\n\nfunction lowerBoundByStart(sorted: { startTokenIndex: number }[], target: number): number {\n  let low = 0;\n  let high = sorted.length;\n  while (low < high) {\n    const middle = (low + high) >>> 1;\n    if ((sorted[middle]?.startTokenIndex ?? 0) < target) {\n      low = middle + 1;\n    } else {\n      high = middle;\n    }\n  }\n  return low;\n}\n\nfunction compareGroups(left: CountedOccurrence[], right: CountedOccurrence[]): number {\n  const leftFirst = left[0];\n  const rightFirst = right[0];\n  return (\n    (leftFirst?.startTokenIndex ?? 0) - (rightFirst?.startTokenIndex ?? 0) ||\n    (leftFirst?.endTokenIndex ?? 0) - (rightFirst?.endTokenIndex ?? 0)\n  );\n}\n\ninterface MergeResult<T> {\n  merged: T[];\n  /**\n   * Whether the merged group takes the respective input group's place: its standalone occurrences\n   * were all paired and it holds no nested copies that only it can report.\n   */\n  firstReplaced: boolean;\n  secondReplaced: boolean;\n  /** The occurrences of groups that stay, which the merged group's spans now also cover. */\n  pairedRetained: T[];\n}\n\n/**\n * Pairs `second` occurrences with gap-preceding `first` occurrences, greedily in source order:\n * each trailing occurrence takes the earliest unused leading occurrence within the gap, and a\n * pair's leading must start at or after the previous pair's trailing end so merged spans never\n * overlap. A merge needs at least two pairs (a merged group must still mean \"appears more than\n * once\") and must fully consume at least one group; for equal cardinalities this reduces to the\n * strict all-pairs merge, so pre-partial-merge behavior is unchanged there.\n */\nfunction mergeGroups<T extends CountedOccurrence>(\n  first: T[],\n  second: T[],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean\n): MergeResult<T> | undefined {\n  // Occurrences a previous partial merge already paired into a merged group must not pair again:\n  // their spans already live inside that merged group, so re-pairing them would assemble a second,\n  // competing merged group instead of letting the existing merged group extend (and would count\n  // the same span twice). Consumption is still judged against the FULL group, so a group holding\n  // shared occurrences is never subsumed away; nested copies, which the merged span would not\n  // cover anyway, are left out of that judgment so they cannot veto a merge of the standalone\n  // copies.\n  const [leadings, firstLength] = pairableOccurrences(first);\n  const [trailings, secondLength] = pairableOccurrences(second);\n  const pairs: [T, T][] = [];\n  let leadingIndex = 0;\n  let previousTrailingEnd = -1;\n  for (const trailing of trailings) {\n    // Leadings ending too far before this trailing can never pair a later (even farther) one.\n    while (leadingIndex < leadings.length) {\n      const leading = leadings[leadingIndex];\n      if (leading && leading.endTokenIndex + maxGapTokens < trailing.startTokenIndex) {\n        leadingIndex += 1;\n      } else {\n        break;\n      }\n    }\n    const leading = leadings[leadingIndex];\n    if (\n      leading &&\n      leading.endTokenIndex <= trailing.startTokenIndex &&\n      leading.startTokenIndex >= previousTrailingEnd\n    ) {\n      pairs.push([leading, trailing]);\n      previousTrailingEnd = trailing.endTokenIndex;\n      leadingIndex += 1;\n    }\n  }\n  const firstFullyPaired = pairs.length === firstLength;\n  const secondFullyPaired = pairs.length === secondLength;\n  if (pairs.length < 2 || (!firstFullyPaired && !secondFullyPaired)) {\n    return undefined;\n  }\n  // A group holding nested copies stays even when all its standalone copies pair: the merged\n  // group's content is larger than what those copies matched, so only the original group can\n  // report which files share the matched fragment.\n  const firstReplaced = firstFullyPaired && !first.some((occurrence) => occurrence.nestedInLargerGroup);\n  const secondReplaced = secondFullyPaired && !second.some((occurrence) => occurrence.nestedInLargerGroup);\n  const merged = pairs.map(([leading, trailing]) => ({\n    ...leading,\n    // A merged occurrence is a fresh span combination; it never inherits shared-span marks.\n    spanCountedElsewhere: undefined,\n    nestedInLargerGroup: undefined,\n    segments: [...leading.segments, ...trailing.segments],\n    tokenCount: leading.tokenCount + trailing.tokenCount,\n    endTokenIndex: trailing.endTokenIndex,\n    endIndex: trailing.endIndex,\n    endLine: trailing.endLine,\n  }));\n  // Only occurrences of one file pair (file offsets exceed the gap), so a merged group spans the\n  // files its pairs sit in: with nested copies left out of pairing, that can be fewer files than\n  // the input groups covered, and a merged group that is no longer reportable must not form.\n  if (!isReportableGroup(merged)) {\n    return undefined;\n  }\n  const pairedRetained = [\n    ...(firstReplaced ? [] : pairs.map(([leading]) => leading)),\n    ...(secondReplaced ? [] : pairs.map(([, trailing]) => trailing)),\n  ];\n  return { merged, firstReplaced, secondReplaced, pairedRetained };\n}\n\n/** One pass over a group: its pairable occurrences and its non-nested occurrence count. */\nfunction pairableOccurrences<T extends CountedOccurrence>(group: T[]): [T[], number] {\n  const pairable: T[] = [];\n  let length = 0;\n  for (const occurrence of group) {\n    if (!occurrence.nestedInLargerGroup) {\n      length += 1;\n    }\n    // Either flag keeps an occurrence out of pairing: its span is already counted elsewhere, or it\n    // is a nested copy of content the merged span would not cover.\n    if (!occurrence.spanCountedElsewhere && !occurrence.nestedInLargerGroup) {\n      pairable.push(occurrence);\n    }\n  }\n  return [pairable, length];\n}\n\n/**\n * Redundant copies one group adds to duplicateBlockCount. Each redundant occurrence contributes\n * one count per matched fragment, so merging a gapped clone's fragments into one group does not\n * halve duplicateBlockCount: an edited two-fragment pair still counts 2, exactly as its unmerged\n * fragments did. Occurrence shapes can differ within one group (a\n * gap-merged exact pair plus an appended whole-block near-miss copy), so every occurrence's\n * fragments are summed and one representative — the largest — is deducted, keeping the count\n * independent of source order. Occurrences a partial gapped merge also paired into a merged group\n * are skipped: their spans are already counted there, and such a retained group deducts no\n * representative of its own — the merged group's representative already stands for the shared\n * content — so no token span contributes to the count twice.\n */\nexport function countRedundantFragments(group: CountedOccurrence[]): number {\n  let fragmentCount = 0;\n  let maxFragmentCount = 0;\n  let hasSharedOccurrence = false;\n  for (const occurrence of group) {\n    if (occurrence.spanCountedElsewhere) {\n      hasSharedOccurrence = true;\n      continue;\n    }\n    fragmentCount += occurrence.segments.length;\n    maxFragmentCount = Math.max(maxFragmentCount, occurrence.segments.length);\n  }\n  return hasSharedOccurrence ? fragmentCount : fragmentCount - maxFragmentCount;\n}\n\n/** Adds the 1-based code lines the segment's matched tokens cover; shared with cross-file coverage. */\nexport function collectSegmentLines(\n  segment: { startTokenIndex: number; endTokenIndex: number },\n  tokens: Token[],\n  codeLineNumbers: Set<number> | undefined,\n  duplicatedLines: Set<number>\n): void {\n  for (let index = segment.startTokenIndex; index < segment.endTokenIndex; index += 1) {\n    const token = tokens[index];\n    for (let row = token?.startRow ?? 0; row <= (token?.endRow ?? -1); row += 1) {\n      if (!codeLineNumbers || codeLineNumbers.has(row + 1)) {\n        duplicatedLines.add(row + 1);\n      }\n    }\n  }\n}\n"],"mappings":"aASA,MAAa,EAA0D,CACrE,UAAW,GACX,aAAc,GACd,qBAAsB,EACxB,EAQA,SAAgB,EAA0B,EAA4D,CACpG,MAAO,CACL,UAAW,EAAc,GAAS,UAAW,EAA0B,SAAS,EAChF,aAAc,EAAc,GAAS,aAAc,EAA0B,YAAY,EACzF,qBAAsB,EAAc,GAAS,qBAAsB,EAA0B,oBAAoB,CACnH,CACF,CAEA,SAAS,EAAc,EAA2B,EAA0B,CAC1E,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAC5C,EACA,KAAK,IAAI,KAAK,IAAI,KAAK,MAAM,CAAK,EAAG,CAAC,EAAG,UAAa,CAC5D,CAiBA,SAAS,EAAe,EAAsB,EAA6B,CACzE,OAAO,EAAe,GAAK,CAC7B,CAuHA,SAAgB,EAAwB,EAA6B,CACnE,IAAM,EAAS,IAAI,WAAW,EAAO,OAAS,CAAC,EAC/C,IAAK,GAAM,CAAC,EAAO,KAAU,EAAO,QAAQ,EAC1C,EAAO,EAAQ,IAAM,EAAO,IAAU,IAAM,EAAM,cAAgB,IAAA,GAAY,EAAI,GAEpF,OAAO,CACT,CA0CA,SAAgB,EACd,EACA,EACA,EAC+B,CAC/B,IAAM,EAA4C,CAAC,EAC7C,EAAoC,CAAC,EACrC,EAA6B,CAAC,EACpC,IAAK,GAAM,CAAC,EAAc,KAAY,EAAS,QAAQ,EACrD,IAAK,IAAM,KAAc,EAAQ,WAC/B,EAAwB,KAAK,CAAY,EACzC,EAAW,KAAK,CAAU,EAG9B,IAAM,EAAa,GACjB,EAAS,EAAwB,IAAmB,GAChD,kBAAyB,IAAI,IAC7B,EAAmB,EAAW,KAAK,EAAY,IACnD,EAA0B,EAAU,CAAc,CAAC,EAAE,QAAU,CAAC,EAAG,EAAY,CAAS,CAC1F,EACA,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAAG,CAClE,IAAM,EAAe,EAAwB,IAAmB,EAChE,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,IAAM,KAAa,EAAK,CAC3B,GAAI,IAAc,IAAA,GAChB,SAEF,IAAM,EAAc,EAAuB,IAAI,CAAS,EACpD,GACF,EAAY,OAAS,EACjB,EAAY,iBAAmB,IACjC,EAAY,eAAiB,IAE3B,EAAY,eAAiB,IAC/B,EAAY,aAAe,IAE7B,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,EAC3D,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,GAE3D,EAAuB,IAAI,EAAW,CACpC,MAAO,EACP,iBACA,eACA,SAAU,EACV,SAAU,CACZ,CAAC,CAEL,CAEJ,CAMA,IAAM,GAAW,EAA+B,IAA4B,CAC1E,GAAI,IAAc,IAAA,GAChB,MAAO,GAET,IAAM,EAAc,EAAuB,IAAI,CAAS,EAOxD,OANI,IAAgB,IAAA,IAAa,EAAY,MAAQ,EAC5C,GAEL,EACK,EAAY,eAAiB,GAE/B,EAAY,iBAAmB,IAAM,EAAY,SAAW,EAAY,UAAY,CAC7F,EAKM,EAAyB,GAAoC,CACjE,IAAM,EAAS,EAAiB,EAAO,eAAe,EAAE,iBAAmB,CAAC,EACtE,EAAY,EAAO,EAAO,OAChC,IAAK,IAAI,EAAQ,EAAO,MAAQ,EAAG,EAAQ,EAAO,MAAQ,EAAO,OAAQ,GAAS,EAChF,GAAI,EAAO,KAAW,EACpB,MAAO,GAGX,MAAO,EACT,EAEM,EAAmC,CAAC,EAC1C,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAC/D,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,GAAM,CAAC,EAAQ,KAAc,EAAI,QAAQ,EAAG,CAC/C,GAAI,CAAC,EAAQ,EAAW,CAAM,GAAK,CAAC,EAAsB,CAAE,iBAAgB,QAAO,QAAO,CAAC,EACzF,SAIF,IAAM,EAAgB,EAAQ,kBAAkB,EAAM,GAAG,EAAS,GAC5D,EAAe,EAAQ,kBAAkB,EAAQ,EAAE,GAAG,EAAS,GACjE,EAAQ,EAAe,EAAS,CAAC,GAAK,EAAQ,EAAc,EAAS,CAAC,GAG1E,EAAe,KAAK,CAAE,iBAAgB,QAAO,QAAO,CAAC,CACvD,CASJ,IAAM,EAAU,IAAI,IAAI,EAAe,IAAI,CAAQ,CAAC,EAChD,EAAW,EACf,KAAO,EAAS,OAAS,GAAG,CAC1B,IAAM,EAA4B,CAAC,EACnC,IAAK,IAAM,KAAU,EAAU,CAC7B,IAAM,EAAa,EAAW,EAAO,gBAC/B,EAAQ,IAAa,EAAO,OAC5B,EAAO,IAAa,EAAO,MAAQ,EAAO,OAAS,GACnD,EAAU,EAAU,EAAO,cAAc,EAC/C,GAAI,CAAC,GAAS,CAAC,GAAQ,CAAC,EACtB,SAEF,IAAM,EAAc,KAAK,EAAe,EAAQ,OAAQ,EAAQ,mBAAoB,EAAM,gBAAiB,EAAK,aAAa,IAC7H,EAAW,KAAK,CACd,UAAW,EAAY,EAAa,EAAM,gBAAiB,EAAK,cAAe,EAAO,CAAI,EAC1F,aAAc,EAAwB,EAAO,iBAAmB,CAClE,CAAC,EACD,EAAQ,KAAK,CAAM,CACrB,CACA,EAAW,CAAC,EACZ,IAAK,IAAM,KAAU,EACnB,IAAK,IAAM,IAAS,CAAC,EAAO,MAAO,EAAO,MAAQ,CAAC,EAAG,CACpD,IAAM,EAAY,CAAE,eAAgB,EAAO,eAAgB,QAAO,OAAQ,EAAO,OAAS,CAAE,EACtF,EAAe,EAAiB,EAAO,eAAe,EAAE,kBAAkB,EAAM,GAAG,EAAU,QAEjG,GAAQ,IAAI,EAAS,CAAS,CAAC,GAC9B,EAAQ,EAAc,EAAU,MAAM,GACtC,EAAsB,CAAS,IAIlC,EAAQ,IAAI,EAAS,CAAS,CAAC,EAC/B,EAAS,KAAK,CAAS,EACzB,CAEJ,CACA,OAAO,CACT,CAEA,SAAS,EAAS,EAAgC,CAChD,MAAO,GAAG,EAAO,eAAe,GAAG,EAAO,MAAM,GAAG,EAAO,QAC5D,CASA,SAAS,EAA0B,EAAiB,EAA0B,EAAqC,CACjH,IAAM,EAAkB,EAAW,IAAK,GACtC,EAAgB,EAAQ,EAAU,gBAAiB,EAAU,aAAa,CAC5E,EACM,EAA8C,CAAC,EACrD,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAW,OAAQ,GAAS,EAAG,CACzD,IAAM,EAA8B,CAAC,EACjC,EAAO,KACP,EAAa,EACX,EAAS,KAAK,IAAI,EAAW,OAAQ,EAAQ,GAAyB,EAC5E,IAAK,IAAI,EAAM,EAAO,EAAM,EAAQ,GAAO,EAAG,CAC5C,IAAM,EAAY,EAAW,GACvB,EAAgB,EAAgB,GACtC,GAAI,CAAC,GAAa,IAAkB,IAAA,GAClC,MAEF,EAAO,EAAc,EAAM,CAAa,EACxC,GAAc,EAAU,cAAgB,EAAU,gBAClD,IAAM,EAAiB,EAAM,EAAQ,EACrC,EAAI,GACF,GAAkB,GAA6B,GAAc,EACzD,EAAc,EAAM,CAAc,EAClC,IAAA,EACR,CACA,EAAkB,KAAK,CAAG,CAC5B,CACA,MAAO,CAAE,oBAAmB,iBAAgB,CAC9C,CAOA,SAAgB,EAAU,EAAe,EAAuB,CAC9D,OAAO,EAAuB,CAAC,CAAC,CAAC,CAAC,CACpC,CAMA,SAAgB,EAAuB,EAA0C,CAC/E,IAAM,EAAa,EAAE,OAAS,KAAQ,EAChC,kBAAgB,IAAI,IAC1B,IAAK,GAAM,CAAC,EAAO,KAAW,EAAE,QAAQ,EAAG,CACzC,IAAI,EAAO,EAAc,IAAI,CAAM,EAC9B,IACH,EAAO,IAAI,YAAY,CAAS,EAChC,EAAc,IAAI,EAAQ,CAAI,GAEhC,IAAM,EAAO,IAAU,EAIvB,EAAK,IAAS,EAAK,IAAS,GAAM,IAAM,EAAQ,GAClD,CAEA,IAAM,EAAI,IAAI,YAAY,CAAS,EACnC,MAAQ,IAAM,CACZ,EAAE,KAAK,CAAC,EACR,IAAK,IAAM,KAAU,EAAG,CACtB,IAAM,EAAY,EAAc,IAAI,CAAM,EAEtC,EAAa,EACb,EAAS,EACb,IAAK,IAAI,EAAO,EAAG,EAAO,EAAW,GAAQ,EAAG,CAC9C,IAAM,EAAW,EAAE,IAAS,EAEtB,IAAM,IAAY,IAAS,GAAK,KAAc,EAE9C,GAAY,GAAY,EAAK,KAAgB,EACnD,EAAa,IAAa,GAC1B,IAAM,EAAa,EAAI,EAAU,EACjC,EAAS,IAAa,GAEtB,EAAE,GAAQ,EAAI,CAAC,CACjB,CACF,CAEA,IAAI,EAAS,EACb,IAAK,IAAM,KAAQ,EACjB,GAAU,EAAS,CAAI,EAEzB,OAAO,CACT,CACF,CAEA,SAAS,EAAS,EAAuB,CACvC,IAAI,EAAQ,GAAU,IAAU,EAAK,YAErC,MADA,IAAS,EAAQ,YAAmB,IAAU,EAAK,WAC3C,KAAK,KAAM,GAAS,IAAU,GAAM,UAAe,QAAa,IAAM,GAAM,GACtF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACoB,CACpB,MAAO,CACL,cACA,WAAY,EAAgB,EAC5B,kBACA,gBACA,WAAY,EAAM,WAClB,SAAU,EAAK,SACf,UAAW,EAAM,UACjB,QAAS,EAAK,OAChB,CACF,CAGA,MAAM,EAAkC,CAAC,EACnC,EAAmC,CAAC,EAE1C,SAAS,EAAoB,EAAuB,CAClD,IAAI,EAAO,EAAsB,GAKjC,OAJI,IAAS,IAAA,KACX,EAAO,EAAS,IAAI,GAAO,EAC3B,EAAsB,GAAS,GAE1B,CACT,CAEA,SAAS,EAAqB,EAAuB,CACnD,IAAI,EAAO,EAAuB,GAKlC,OAJI,IAAS,IAAA,KACX,EAAO,EAAU,IAAI,GAAO,EAC5B,EAAuB,GAAS,GAE3B,CACT,CAUA,SAAS,EACP,EACA,EACA,EACA,EACQ,CAGR,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EADrD,GADC,EAAmB,IAAkB,IAAM,EAAmB,IAAoB,GACtD,EAAgB,CACkC,CAAC,EACrG,MAAO,GAAG,EAAQ,GAAG,EAAU,GAAG,EAAgB,GACpD,CASA,SAAS,EAAgB,EAAiB,EAAyB,EAA+B,CAChG,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EAAe,EAAK,EAE9F,OAAO,EAAU,KAAK,KAAK,EAAW,EAAE,CAC1C,CAEA,SAAS,EACP,EACA,EACA,EACA,EACkB,CAClB,IAAM,kBAAoB,IAAI,IAC1B,EAAU,KACV,EAAY,MAChB,IAAK,IAAI,EAAQ,EAAiB,EAAQ,EAAe,GAAS,EAAG,CACnE,IAAM,EAAQ,EAAO,GACrB,GAAI,CAAC,EACH,SAIF,IAAI,EACA,EACJ,GAAI,EAAM,OAAS,KAAM,CACvB,IAAI,EAAkB,EAAkB,IAAI,EAAM,IAAI,EAClD,IAAoB,IAAA,KACtB,EAAkB,EAAkB,KACpC,EAAkB,IAAI,EAAM,KAAM,CAAe,GAEnD,EAAO,EAAoB,CAAe,EAC1C,EAAQ,EAAqB,CAAe,CAC9C,KACE,GAAO,EAAM,SACb,EAAQ,EAAM,UAGhB,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAQ,EAC5C,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EACnC,GAAqB,EAAM,cAAgB,IAAA,IAAa,EAAM,eAAiB,IAAA,KAEjF,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAM,YAAe,EACzD,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EAAM,aAEjD,CACA,MAAO,CAAC,EAAS,CAAS,CAC5B,CAGA,SAAS,EAAS,EAAsB,CACtC,IAAI,EAAO,KACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAM,EAAE,EAAI,EAAK,WAAW,CAAK,EAEpD,OAAO,CACT,CAGA,SAAS,EAAU,EAAsB,CACvC,IAAI,EAAO,YACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAO,EAAK,WAAW,CAAK,EAAG,QAAU,EAE5D,OAAO,CACT,CAEA,SAAS,EAAc,EAAc,EAAuB,CAC1D,OAAO,KAAK,KAAK,EAAM,EAAE,EAAI,CAC/B,CAoBA,SAAgB,EACd,EACA,EACA,MAAmD,GAC5C,CACP,GAAI,GAAgB,GAAK,EAAO,OAAS,EACvC,OAAO,EAGT,EAAO,KAAK,CAAa,EACzB,IAAK,IAAI,EAAU,GAAM,GAAU,CACjC,EAAU,GACV,IAAM,EAAkB,EAA2B,EAAQ,CAAY,EACvE,IAAK,IAAI,EAAY,EAAG,EAAY,EAAO,QAAU,CAAC,EAAS,GAAa,EAC1E,IAAK,IAAM,KAAc,EAAgB,IAAc,CAAC,EAAG,CACzD,IAAM,EAAO,EAAO,GACd,EAAQ,EAAO,GACrB,GAAI,CAAC,GAAQ,CAAC,EACZ,SAEF,IAAM,EAAU,EAAY,EAAM,EAAO,EAAc,CAAiB,EAClE,EAAS,GAAW,EAAY,EAAO,EAAM,EAAc,CAAiB,EAClF,GAAI,CAAC,EACH,SAEF,IAAM,EAAe,EAAU,EAAO,cAAgB,EAAO,eACvD,EAAgB,EAAU,EAAO,eAAiB,EAAO,cAC3D,GAAgB,GAClB,EAAO,GAAa,EAAO,OAC3B,EAAO,OAAO,EAAY,CAAC,GAClB,EACT,EAAO,GAAc,EAAO,OACnB,EACT,EAAO,GAAa,EAAO,OAI3B,EAAO,KAAK,EAAO,MAAM,EAM3B,IAAK,IAAM,KAAc,EAAO,eAC9B,EAAW,qBAAuB,GAEpC,EAAO,KAAK,CAAa,EACzB,EAAU,GACV,KACF,CAEJ,CACA,OAAO,CACT,CAQA,SAAS,EAA2B,EAA+B,EAAkC,CACnG,IAAM,EAA4D,CAAC,EACnE,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,EAAO,KAAK,CAAE,gBAAiB,EAAW,gBAAiB,YAAW,CAAC,EAG3E,EAAO,MAAM,EAAM,IAAU,EAAK,gBAAkB,EAAM,eAAe,EACzE,IAAM,EAAW,EAAO,wBAAU,IAAI,GAAa,EACnD,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,IACE,IAAI,EAAQ,EAAkB,EAAQ,EAAW,aAAa,EAC9D,EAAQ,EAAO,SAAW,EAAO,EAAM,EAAE,iBAAmB,IAAM,EAAW,cAAgB,EAC7F,GAAS,EACT,CACA,IAAM,EAAQ,EAAO,EAAM,EAAE,YAAc,EACvC,IAAU,GACZ,EAAS,KAAK,IAAI,EAAO,CAAU,EAAE,EAAE,IAAI,KAAK,IAAI,EAAO,CAAU,CAAC,CAE1E,CAGJ,OAAO,EAAS,IAAK,GAAQ,CAAC,GAAG,CAAG,CAAC,CAAC,UAAU,EAAM,IAAU,EAAO,CAAK,CAAC,CAC/E,CAEA,SAAS,EAAkB,EAAuC,EAAwB,CACxF,IAAI,EAAM,EACN,EAAO,EAAO,OAClB,KAAO,EAAM,GAAM,CACjB,IAAM,EAAU,EAAM,IAAU,GAC3B,EAAO,EAAO,EAAE,iBAAmB,GAAK,EAC3C,EAAM,EAAS,EAEf,EAAO,CAEX,CACA,OAAO,CACT,CAEA,SAAS,EAAc,EAA2B,EAAoC,CACpF,IAAM,EAAY,EAAK,GACjB,EAAa,EAAM,GACzB,OACG,GAAW,iBAAmB,IAAM,GAAY,iBAAmB,KACnE,GAAW,eAAiB,IAAM,GAAY,eAAiB,EAEpE,CAsBA,SAAS,EACP,EACA,EACA,EACA,EAC4B,CAQ5B,GAAM,CAAC,EAAU,GAAe,EAAoB,CAAK,EACnD,CAAC,EAAW,GAAgB,EAAoB,CAAM,EACtD,EAAkB,CAAC,EACrB,EAAe,EACf,EAAsB,GAC1B,IAAK,IAAM,KAAY,EAAW,CAEhC,KAAO,EAAe,EAAS,QAAQ,CACrC,IAAM,EAAU,EAAS,GACzB,GAAI,GAAW,EAAQ,cAAgB,EAAe,EAAS,gBAC7D,GAAgB,OAEhB,KAEJ,CACA,IAAM,EAAU,EAAS,GAEvB,GACA,EAAQ,eAAiB,EAAS,iBAClC,EAAQ,iBAAmB,IAE3B,EAAM,KAAK,CAAC,EAAS,CAAQ,CAAC,EAC9B,EAAsB,EAAS,cAC/B,GAAgB,EAEpB,CACA,IAAM,EAAmB,EAAM,SAAW,EACpC,EAAoB,EAAM,SAAW,EAC3C,GAAI,EAAM,OAAS,GAAM,CAAC,GAAoB,CAAC,EAC7C,OAKF,IAAM,EAAgB,GAAoB,CAAC,EAAM,KAAM,GAAe,EAAW,mBAAmB,EAC9F,EAAiB,GAAqB,CAAC,EAAO,KAAM,GAAe,EAAW,mBAAmB,EACjG,EAAS,EAAM,KAAK,CAAC,EAAS,MAAe,CACjD,GAAG,EAEH,qBAAsB,IAAA,GACtB,oBAAqB,IAAA,GACrB,SAAU,CAAC,GAAG,EAAQ,SAAU,GAAG,EAAS,QAAQ,EACpD,WAAY,EAAQ,WAAa,EAAS,WAC1C,cAAe,EAAS,cACxB,SAAU,EAAS,SACnB,QAAS,EAAS,OACpB,EAAE,EAIG,KAAkB,CAAM,EAO7B,MAAO,CAAE,SAAQ,gBAAe,iBAAgB,eAAA,CAH9C,GAAI,EAAgB,CAAC,EAAI,EAAM,KAAK,CAAC,KAAa,CAAO,EACzD,GAAI,EAAiB,CAAC,EAAI,EAAM,KAAK,EAAG,KAAc,CAAQ,CAEH,CAAE,CACjE,CAGA,SAAS,EAAiD,EAA2B,CACnF,IAAM,EAAgB,CAAC,EACnB,EAAS,EACb,IAAK,IAAM,KAAc,EAClB,EAAW,sBACd,GAAU,GAIR,CAAC,EAAW,sBAAwB,CAAC,EAAW,qBAClD,EAAS,KAAK,CAAU,EAG5B,MAAO,CAAC,EAAU,CAAM,CAC1B,CAcA,SAAgB,EAAwB,EAAoC,CAC1E,IAAI,EAAgB,EAChB,EAAmB,EACnB,EAAsB,GAC1B,IAAK,IAAM,KAAc,EAAO,CAC9B,GAAI,EAAW,qBAAsB,CACnC,EAAsB,GACtB,QACF,CACA,GAAiB,EAAW,SAAS,OACrC,EAAmB,KAAK,IAAI,EAAkB,EAAW,SAAS,MAAM,CAC1E,CACA,OAAO,EAAsB,EAAgB,EAAgB,CAC/D,CAGA,SAAgB,EACd,EACA,EACA,EACA,EACM,CACN,IAAK,IAAI,EAAQ,EAAQ,gBAAiB,EAAQ,EAAQ,cAAe,GAAS,EAAG,CACnF,IAAM,EAAQ,EAAO,GACrB,IAAK,IAAI,EAAM,GAAO,UAAY,EAAG,IAAQ,GAAO,QAAU,IAAK,GAAO,GACpE,CAAC,GAAmB,EAAgB,IAAI,EAAM,CAAC,IACjD,EAAgB,IAAI,EAAM,CAAC,CAGjC,CACF"}
+{"version":3,"file":"duplication.cjs","names":["clamp"],"sources":["../src/duplication.ts"],"sourcesContent":["import { clamp } from '@willbooster/shared-lib';\nimport type { DuplicationOptions } from './types.js';\n\n/**\n * Project-level duplication machinery operating on normalized token streams. Tokenization itself\n * (parsing, identifier anonymization, literal normalization) happens in the Rust addon, which\n * serializes each file's Token stream and statement structure; the helpers here match statement\n * windows across files, merge gap-adjacent groups, and count duplicated lines over that data.\n */\n\nexport const defaultDuplicationOptions: Required<DuplicationOptions> = {\n  minTokens: 40,\n  maxGapTokens: 30,\n  minSimilarityPercent: 70,\n};\n\n/**\n * Fills defaults for absent settings, applying the same normalization as the native boundary's\n * clampToU32 — NaN (e.g. `Number(unsetEnvVariable)`) counts as absent, and other values truncate\n * and clamp to [0, u32::MAX] — so the TypeScript half of cross-file matching cannot diverge from\n * the natively collected candidates on such input.\n */\nexport function resolveDuplicationOptions(options?: DuplicationOptions): Required<DuplicationOptions> {\n  return {\n    minTokens: resolveOption(options?.minTokens, defaultDuplicationOptions.minTokens),\n    maxGapTokens: resolveOption(options?.maxGapTokens, defaultDuplicationOptions.maxGapTokens),\n    minSimilarityPercent: resolveOption(options?.minSimilarityPercent, defaultDuplicationOptions.minSimilarityPercent),\n  };\n}\n\nfunction resolveOption(value: number | undefined, fallback: number): number {\n  return value === undefined || Number.isNaN(value) ? fallback : clamp(Math.trunc(value), 0, 0xFF_FF_FF_FF);\n}\n\n/** Minimum consecutive statements for a statement-sequence duplicate candidate. */\nconst minSequenceStatementCount = 2;\n/**\n * Caps the window length so statement-sequence enumeration stays linear in the statement count.\n * Heterogeneous clones longer than the cap are reported as capped windows (a deliberate\n * conservative undercount trading completeness for bounded discovery cost).\n */\nconst maxSequenceStatementCount = 100;\n\n/**\n * A region whose normalized tokens are at least 20% literal values is data-like (a lookup table, a\n * constant list, a value-mapping switch), not logic: literal values re-enter its fingerprint so\n * tables that merely share their shape stop counting as copy-paste. Compared in integer math\n * (5 * literals >= total) so the TypeScript and native sides cannot disagree on the boundary.\n */\nfunction isLiteralDense(literalCount: number, tokenCount: number): boolean {\n  return literalCount * 5 >= tokenCount;\n}\n\nexport interface Token {\n  /** Normalization target: identifiers to anonymize, literal kind tags, or the raw token text. */\n  kind: 'id' | 'text';\n  text: string;\n  /**\n   * Two INDEPENDENT hashes of `text` (djb2 and FNV-1a), precomputed so fingerprinting nested\n   * regions never re-hashes a token. Feeding the same per-token hash to both fingerprint\n   * accumulators would collapse the key to 32 effective bits: one djb2 collision between two\n   * token texts would then equate whole regions.\n   */\n  textHash: number;\n  textHash2: number;\n  /** Hash pair of a value-carrying literal's value, folded into data-like region fingerprints. */\n  literalHash?: number;\n  literalHash2?: number;\n  /**\n   * True for verbatim-kept NAMES (member/callee/type names, named grammar leaves): together with\n   * value-carrying literals these are the content-bearing tokens the near-miss content gate\n   * counts. Keywords, operators, and punctuation come from unnamed nodes and stay false.\n   */\n  isName?: boolean;\n  /** 0-based source rows the token occupies, so line coverage counts only matched-token lines. */\n  startRow: number;\n  endRow: number;\n}\n\n/** A token span with its source position; plain data so cross-file matching can retain it. */\nexport interface TokenRange {\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A contiguous run of matched tokens; gapped (merged) duplicates carry several per occurrence. */\ninterface TokenSegment {\n  startTokenIndex: number;\n  endTokenIndex: number;\n}\n\ninterface DuplicateCandidate {\n  fingerprint: string;\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\nexport interface CountedOccurrence {\n  /** Matched token runs; more than one once gapped groups are merged. */\n  segments: TokenSegment[];\n  /**\n   * Set on occurrences whose span another reported group already counts: a retained group's\n   * occurrences that a partial gapped merge also paired into a merged group, and cross-file copies\n   * nested inside a larger group's region. Block counting must not count them again.\n   */\n  spanCountedElsewhere?: boolean;\n  /**\n   * Set on cross-file copies nested inside a larger group's region (they also set\n   * `spanCountedElsewhere`). They never pair in gapped merging, and they do not keep their group's\n   * standalone copies from merging, which they are not copies of. They do keep the merged group\n   * from taking their group's place, since only the original group reports the nesting.\n   */\n  nestedInLargerGroup?: boolean;\n  /** Sum of segment token counts (the gap tokens are not matched content). */\n  tokenCount: number;\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/** A duplicate region found in one file, exported for cross-file matching by fingerprint. */\nexport interface CrossFileDuplicateCandidate {\n  /** Content key: equal fingerprints mean equal normalized token sequences (up to hash collision). */\n  fingerprint: string;\n  tokenCount: number;\n  /** Token positions within the owning file, for cross-file gapped (Type-3) merging. */\n  startTokenIndex: number;\n  endTokenIndex: number;\n  startIndex: number;\n  endIndex: number;\n  startLine: number;\n  endLine: number;\n}\n\n/**\n * One file's contribution to cross-file clone detection: catalogued candidates plus the normalized\n * token stream and statement structure, so the project-level pass can match partial statement runs\n * (windows that a single file cannot know repeat elsewhere) and merge gap-adjacent groups.\n */\nexport interface CrossFileDuplicationFileData {\n  candidates: CrossFileDuplicateCandidate[];\n  tokens: Token[];\n  containerStatements: TokenRange[][];\n  /**\n   * The mutually disjoint blocks (at least `minTokens` long, not literal-dense) compared for\n   * cross-file near-miss (Type-3) clones. Optional for backward compatibility; without it, the\n   * file takes part in exact and gapped cross-file matching only.\n   */\n  nearMissBlocks?: TokenRange[];\n  /**\n   * 1-based lines that are neither blank nor comment-only, so cross-file line coverage counts only\n   * code lines (blank rows inside multi-row tokens such as template literals carry no content).\n   * Optional for backward compatibility; without it, every matched-token row counts.\n   */\n  codeLineNumbers?: Set<number>;\n}\n\n/** literalCountPrefix[i] = value-carrying literal tokens in tokens[0..i), for O(1) density checks. */\nexport function buildLiteralCountPrefix(tokens: Token[]): Int32Array {\n  const prefix = new Int32Array(tokens.length + 1);\n  for (const [index, token] of tokens.entries()) {\n    prefix[index + 1] = (prefix[index] ?? 0) + (token.literalHash === undefined ? 0 : 1);\n  }\n  return prefix;\n}\n\ninterface WindowOccurrences {\n  count: number;\n  /** -1 once occurrences span more than one container. */\n  containerIndex: number;\n  /** -1 once occurrences span more than one context (file). */\n  contextIndex: number;\n  minStart: number;\n  maxStart: number;\n}\n\ninterface SequenceWindow {\n  containerIndex: number;\n  start: number;\n  length: number;\n}\n\n/** One file's token stream and statement containers, as a window-matching context. */\nexport interface SequenceWindowContext {\n  tokens: Token[];\n  literalCountPrefix: Int32Array;\n  containers: TokenRange[][];\n}\n\nexport interface ContextualSequenceCandidate {\n  candidate: DuplicateCandidate;\n  contextIndex: number;\n}\n\n/**\n * Enumerates runs of consecutive sibling statements over one or more contexts (files). Every\n * container statement participates; only the window length is capped, so enumeration stays linear\n * in the statement count. Windows are grouped by a cheap rolling hash of per-statement\n * fingerprints, and only locally maximal repeated windows — those whose one-statement extensions\n * stop repeating — become candidates with an exact (window-consistent) fingerprint. Without the\n * maximality filter a degenerate file of near-identical statements would fingerprint every\n * sub-window of every repeated region. With `requireMultipleContexts` a window only counts as\n * repeated when its occurrences span at least two contexts (CPD-style cross-file matching): a\n * repeat confined to one file is that file's own concern, and emitting it here would flood the\n * project-level selection with unusable single-file groups.\n */\nexport function collectSequenceWindowCandidates(\n  contexts: SequenceWindowContext[],\n  minTokens: number,\n  requireMultipleContexts: boolean\n): ContextualSequenceCandidate[] {\n  const candidates: ContextualSequenceCandidate[] = [];\n  const contextIndexByContainer: number[] = [];\n  const containers: TokenRange[][] = [];\n  for (const [contextIndex, context] of contexts.entries()) {\n    for (const statements of context.containers) {\n      contextIndexByContainer.push(contextIndex);\n      containers.push(statements);\n    }\n  }\n  const contextAt = (containerIndex: number): SequenceWindowContext | undefined =>\n    contexts[contextIndexByContainer[containerIndex] ?? 0];\n  const occurrencesByWindowKey = new Map<number, WindowOccurrences>();\n  const containerWindows = containers.map((statements, containerIndex) =>\n    enumerateContainerWindows(contextAt(containerIndex)?.tokens ?? [], statements, minTokens)\n  );\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    const contextIndex = contextIndexByContainer[containerIndex] ?? 0;\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const windowKey of row) {\n        if (windowKey === undefined) {\n          continue;\n        }\n        const occurrences = occurrencesByWindowKey.get(windowKey);\n        if (occurrences) {\n          occurrences.count += 1;\n          if (occurrences.containerIndex !== containerIndex) {\n            occurrences.containerIndex = -1;\n          }\n          if (occurrences.contextIndex !== contextIndex) {\n            occurrences.contextIndex = -1;\n          }\n          occurrences.minStart = Math.min(occurrences.minStart, start);\n          occurrences.maxStart = Math.max(occurrences.maxStart, start);\n        } else {\n          occurrencesByWindowKey.set(windowKey, {\n            count: 1,\n            containerIndex,\n            contextIndex,\n            minStart: start,\n            maxStart: start,\n          });\n        }\n      }\n    }\n  }\n\n  // A window only \"repeats\" when two of its occurrences can coexist without overlapping: sliding\n  // matches inside a homogeneous run (start spread smaller than the window length) can never both\n  // be counted and must neither qualify a window nor dominate its sub-windows. Cross-context\n  // matching instead requires occurrences in two contexts, which coexist by construction.\n  const repeats = (windowKey: number | undefined, length: number): boolean => {\n    if (windowKey === undefined) {\n      return false;\n    }\n    const occurrences = occurrencesByWindowKey.get(windowKey);\n    if (occurrences === undefined || occurrences.count < 2) {\n      return false;\n    }\n    if (requireMultipleContexts) {\n      return occurrences.contextIndex === -1;\n    }\n    return occurrences.containerIndex === -1 || occurrences.maxStart - occurrences.minStart >= length;\n  };\n\n  // A window whose statements all share one normalized shape (sixteen `let x = 0;` declarations,\n  // a constant table) is a homogeneous preamble, not a copy-paste: requiring two distinct\n  // per-statement shapes keeps such runs out of duplicate groups and the duplication ratio.\n  const hasDistinctStatements = (window: SequenceWindow): boolean => {\n    const hashes = containerWindows[window.containerIndex]?.statementHashes ?? [];\n    const firstHash = hashes[window.start];\n    for (let index = window.start + 1; index < window.start + window.length; index += 1) {\n      if (hashes[index] !== firstHash) {\n        return true;\n      }\n    }\n    return false;\n  };\n\n  const maximalWindows: SequenceWindow[] = [];\n  for (const [containerIndex, windows] of containerWindows.entries()) {\n    for (const [start, row] of windows.windowKeysByStart.entries()) {\n      for (const [length, windowKey] of row.entries()) {\n        if (!repeats(windowKey, length) || !hasDistinctStatements({ containerIndex, start, length })) {\n          continue;\n        }\n        // Dominated windows are skipped: the one-statement extension also repeats, so a larger\n        // candidate covering this window exists.\n        const extendedRight = windows.windowKeysByStart[start]?.[length + 1];\n        const extendedLeft = windows.windowKeysByStart[start - 1]?.[length + 1];\n        if (repeats(extendedRight, length + 1) || repeats(extendedLeft, length + 1)) {\n          continue;\n        }\n        maximalWindows.push({ containerIndex, start, length });\n      }\n    }\n  }\n\n  // The rolling hash anonymizes identifiers per statement, so a window can look repeated coarsely\n  // while its exact (window-consistent) fingerprints differ, and a longer window's match can\n  // dominate sub-windows that other copies still need (three copies where only two extend one\n  // statement further). Every emitted window therefore exposes its repeating, unvisited\n  // sub-windows; `visited` bounds the worklist and lengths strictly decrease, so it terminates.\n  const visited = new Set(maximalWindows.map(windowId));\n  let frontier = maximalWindows;\n  while (frontier.length > 0) {\n    const emitted: SequenceWindow[] = [];\n    for (const window of frontier) {\n      const statements = containers[window.containerIndex];\n      const first = statements?.[window.start];\n      const last = statements?.[window.start + window.length - 1];\n      const context = contextAt(window.containerIndex);\n      if (!first || !last || !context) {\n        continue;\n      }\n      const fingerprint = `s:${fingerprintKey(context.tokens, context.literalCountPrefix, first.startTokenIndex, last.endTokenIndex)}`;\n      candidates.push({\n        candidate: toCandidate(fingerprint, first.startTokenIndex, last.endTokenIndex, first, last),\n        contextIndex: contextIndexByContainer[window.containerIndex] ?? 0,\n      });\n      emitted.push(window);\n    }\n    frontier = [];\n    for (const window of emitted) {\n      for (const start of [window.start, window.start + 1]) {\n        const subWindow = { containerIndex: window.containerIndex, start, length: window.length - 1 };\n        const subWindowKey = containerWindows[window.containerIndex]?.windowKeysByStart[start]?.[subWindow.length];\n        if (\n          visited.has(windowId(subWindow)) ||\n          !repeats(subWindowKey, subWindow.length) ||\n          !hasDistinctStatements(subWindow)\n        ) {\n          continue;\n        }\n        visited.add(windowId(subWindow));\n        frontier.push(subWindow);\n      }\n    }\n  }\n  return candidates;\n}\n\nfunction windowId(window: SequenceWindow): string {\n  return `${window.containerIndex}:${window.start}:${window.length}`;\n}\n\ninterface ContainerWindows {\n  /** windowKeysByStart[start][length] is the rolling-hash key of the window, or undefined if it is below the size thresholds. */\n  windowKeysByStart: (number | undefined)[][];\n  /** Per-statement fingerprint hashes, for the distinct-shape requirement on windows. */\n  statementHashes: number[];\n}\n\nfunction enumerateContainerWindows(tokens: Token[], statements: TokenRange[], minTokens: number): ContainerWindows {\n  const statementHashes = statements.map((statement) =>\n    fingerprintHash(tokens, statement.startTokenIndex, statement.endTokenIndex)\n  );\n  const windowKeysByStart: (number | undefined)[][] = [];\n  for (let start = 0; start < statements.length; start += 1) {\n    const row: (number | undefined)[] = [];\n    let hash = 5381;\n    let tokenCount = 0;\n    const maxEnd = Math.min(statements.length, start + maxSequenceStatementCount);\n    for (let end = start; end < maxEnd; end += 1) {\n      const statement = statements[end];\n      const statementHash = statementHashes[end];\n      if (!statement || statementHash === undefined) {\n        break;\n      }\n      hash = combineHashes(hash, statementHash);\n      tokenCount += statement.endTokenIndex - statement.startTokenIndex;\n      const statementCount = end - start + 1;\n      row[statementCount] =\n        statementCount >= minSequenceStatementCount && tokenCount >= minTokens\n          ? combineHashes(hash, statementCount)\n          : undefined;\n    }\n    windowKeysByStart.push(row);\n  }\n  return { windowKeysByStart, statementHashes };\n}\n\n/**\n * Longest-common-subsequence LENGTH of two symbol sequences via the Allison–Dix bit-parallel\n * recurrence (O(|a|/32 · |b|) words): per symbol of `b`, `x = match | v` and\n * `v = x & ~(x - ((v << 1) | 1))` over multi-word bit vectors; the set bits of `v` count the LCS.\n */\nexport function lcsLength(a: Int32Array, b: Int32Array): number {\n  return createLcsLengthCounter(a)(b);\n}\n\n/**\n * lcsLength against a fixed `a`, whose per-symbol position masks are built once and reused for\n * every `b` compared with it.\n */\nexport function createLcsLengthCounter(a: Int32Array): (b: Int32Array) => number {\n  const wordCount = (a.length + 31) >>> 5;\n  const positionMasks = new Map<number, Uint32Array>();\n  for (const [index, symbol] of a.entries()) {\n    let mask = positionMasks.get(symbol);\n    if (!mask) {\n      mask = new Uint32Array(wordCount);\n      positionMasks.set(symbol, mask);\n    }\n    const word = index >>> 5;\n    // The Uint32Array store wraps the signed int32 bit pattern to unsigned. The `?? 0` guards in\n    // this function are required by noUncheckedIndexedAccess (typed-array reads type as\n    // `number | undefined`), not redundancy: every index is in bounds.\n    mask[word] = (mask[word] ?? 0) | (1 << (index & 31));\n  }\n\n  const v = new Uint32Array(wordCount);\n  return (b) => {\n    v.fill(0);\n    for (const symbol of b) {\n      const matchMask = positionMasks.get(symbol);\n      // `(v << 1) | 1` shifts a carry bit across words; subtraction borrows across words.\n      let shiftCarry = 1;\n      let borrow = 0;\n      for (let word = 0; word < wordCount; word += 1) {\n        const previous = v[word] ?? 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- `>>> 0` reinterprets the signed int32 bit pattern as unsigned so the borrow subtraction below compares magnitudes; Math.trunc would keep it negative.\n        const x = ((matchMask?.[word] ?? 0) | previous) >>> 0;\n        // oxlint-disable-next-line unicorn/prefer-math-trunc -- same unsigned reinterpretation as `x`.\n        const shifted = ((previous << 1) | shiftCarry) >>> 0;\n        shiftCarry = previous >>> 31;\n        const difference = x - shifted - borrow;\n        borrow = difference < 0 ? 1 : 0;\n        // The Uint32Array store wraps the signed int32 bit pattern to unsigned.\n        v[word] = x & ~difference;\n      }\n    }\n\n    let length = 0;\n    for (const word of v) {\n      length += popCount(word);\n    }\n    return length;\n  };\n}\n\nfunction popCount(value: number): number {\n  let count = value - ((value >>> 1) & 0x55_55_55_55);\n  count = (count & 0x33_33_33_33) + ((count >>> 2) & 0x33_33_33_33);\n  return (Math.imul((count + (count >>> 4)) & 0x0F_0F_0F_0F, 0x01_01_01_01) >>> 24) & 0xFF;\n}\n\nfunction toCandidate(\n  fingerprint: string,\n  startTokenIndex: number,\n  endTokenIndex: number,\n  first: TokenRange,\n  last: TokenRange\n): DuplicateCandidate {\n  return {\n    fingerprint,\n    tokenCount: endTokenIndex - startTokenIndex,\n    startTokenIndex,\n    endTokenIndex,\n    startIndex: first.startIndex,\n    endIndex: last.endIndex,\n    startLine: first.startLine,\n    endLine: last.endLine,\n  };\n}\n\n/** Caches of hashText/hashText2 over '$0', '$1', ... so anonymized identifiers hash without allocating. */\nconst anonymizedIndexHashes: number[] = [];\nconst anonymizedIndexHashes2: number[] = [];\n\nfunction anonymizedIndexHash(index: number): number {\n  let hash = anonymizedIndexHashes[index];\n  if (hash === undefined) {\n    hash = hashText(`$${index}`);\n    anonymizedIndexHashes[index] = hash;\n  }\n  return hash;\n}\n\nfunction anonymizedIndexHash2(index: number): number {\n  let hash = anonymizedIndexHashes2[index];\n  if (hash === undefined) {\n    hash = hashText2(`$${index}`);\n    anonymizedIndexHashes2[index] = hash;\n  }\n  return hash;\n}\n\n/**\n * Content key of a token range: two independent 32-bit hashes over the normalized token sequence\n * (identifiers anonymized consistently by first-occurrence order) plus the token count. Regions\n * with equal keys are treated as equal content; a collision would need both 32-bit hashes and the\n * length to coincide, which is negligible for a metrics report. The format and arithmetic match\n * fingerprint_key in native/src/duplication.rs exactly, so window candidates fingerprinted here\n * group together with the per-file candidates the addon catalogues.\n */\nfunction fingerprintKey(\n  tokens: Token[],\n  literalCountPrefix: Int32Array,\n  startTokenIndex: number,\n  endTokenIndex: number\n): string {\n  const literalCount = (literalCountPrefix[endTokenIndex] ?? 0) - (literalCountPrefix[startTokenIndex] ?? 0);\n  const literalDense = isLiteralDense(literalCount, endTokenIndex - startTokenIndex);\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, literalDense);\n  return `${primary}:${secondary}:${endTokenIndex - startTokenIndex}`;\n}\n\n/**\n * A single 32-bit summary of a range for the coarse rolling-hash phase. Deliberately\n * density-agnostic: density is a property of the final candidate REGION, and folding literal\n * values into per-statement hashes would make a dense statement inside a logic-heavy window\n * (`const weights = [1, 2, 3];`) block the window from ever being enumerated. The coarse phase\n * over-approximates on shape alone; the exact region fingerprint still applies the density rule.\n */\nfunction fingerprintHash(tokens: Token[], startTokenIndex: number, endTokenIndex: number): number {\n  const [primary, secondary] = fingerprintHashPair(tokens, startTokenIndex, endTokenIndex, false);\n  // XOR already coerces to int32, matching the native side's i32 arithmetic.\n  return primary ^ Math.imul(secondary, 31);\n}\n\nfunction fingerprintHashPair(\n  tokens: Token[],\n  startTokenIndex: number,\n  endTokenIndex: number,\n  foldLiteralValues: boolean\n): [number, number] {\n  const indexByIdentifier = new Map<string, number>();\n  let primary = 5381;\n  let secondary = 52_711;\n  for (let index = startTokenIndex; index < endTokenIndex; index += 1) {\n    const token = tokens[index];\n    if (!token) {\n      continue;\n    }\n    // Each accumulator consumes its own independent per-token hash: sharing one would collapse\n    // the key to 32 effective bits (a single djb2 collision would equate whole regions).\n    let part: number;\n    let part2: number;\n    if (token.kind === 'id') {\n      let identifierIndex = indexByIdentifier.get(token.text);\n      if (identifierIndex === undefined) {\n        identifierIndex = indexByIdentifier.size;\n        indexByIdentifier.set(token.text, identifierIndex);\n      }\n      part = anonymizedIndexHash(identifierIndex);\n      part2 = anonymizedIndexHash2(identifierIndex);\n    } else {\n      part = token.textHash;\n      part2 = token.textHash2;\n    }\n    // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n    primary = (Math.imul(primary, 31) + part) | 0;\n    secondary = Math.imul(secondary, 37) ^ part2;\n    if (foldLiteralValues && token.literalHash !== undefined && token.literalHash2 !== undefined) {\n      // oxlint-disable-next-line unicorn/prefer-math-trunc -- `| 0` wraps the sum to int32 (Math.trunc does not), which must match the native side's wrapping i32 arithmetic.\n      primary = (Math.imul(primary, 31) + token.literalHash) | 0;\n      secondary = Math.imul(secondary, 37) ^ token.literalHash2;\n    }\n  }\n  return [primary, secondary];\n}\n\n/** djb2-style hash; XOR keeps the value in signed 32-bit range, which is fine for a grouping key. */\nfunction hashText(text: string): number {\n  let hash = 5381;\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- djb2 hashes UTF-16 code units; codePointAt would hash surrogate pairs twice (full code point, then the lone low surrogate).\n    hash = Math.imul(hash, 33) ^ text.charCodeAt(index);\n  }\n  return hash;\n}\n\n/** FNV-1a over UTF-16 code units: independent of hashText so the two accumulators never share input. */\nfunction hashText2(text: string): number {\n  let hash = -2_128_831_035; // 2166136261 as int32 (the FNV-1a offset basis)\n  for (let index = 0; index < text.length; index += 1) {\n    // oxlint-disable-next-line unicorn/prefer-code-point -- hashes UTF-16 code units like hashText.\n    hash = Math.imul(hash ^ text.charCodeAt(index), 16_777_619);\n  }\n  return hash;\n}\n\nfunction combineHashes(hash: number, value: number): number {\n  return Math.imul(hash, 31) + value;\n}\n\n/**\n * Merges duplicate groups separated by a small token gap into one gapped (Type-3) clone group: a\n * copy edited in one spot splits into two exact groups whose occurrences sit side by side in the\n * same order. Occurrences are paired greedily in source order; a merge happens when the pairing\n * fully pairs at least one group with at least two pairs. Equal-cardinality groups whose\n * occurrences all pair merge into one group as before. When cardinalities differ (a fragment also\n * occurs standalone: prefix ×3, suffix ×2), the fully-paired group is subsumed into the merged\n * gapped group while the other group is RETAINED with ALL its occurrences: dropping the leftover\n * would lose duplicated-line coverage, and reporting it alone would make a single-occurrence group\n * (contradicting duplicateBlockGroupCount's \"appears more than once\" meaning). Line coverage\n * unions ranges, so the overlap between the retained exact group and the merged group is harmless.\n * Merging repeats to a fixpoint so a clone edited in several spots still reassembles; it\n * terminates because a merge either removes the input groups it replaces or marks their paired\n * occurrences as counted elsewhere, and only unmarked occurrences of remaining groups pair, so the\n * number of pairable occurrences strictly decreases with every merge. Gap tokens are not matched\n * content: line coverage and sizes count only the matched segments. Generic so cross-file merging\n * can thread file identity through occurrences.\n */\nexport function mergeAdjacentGroups<T extends CountedOccurrence>(\n  groups: T[][],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean = () => true\n): T[][] {\n  if (maxGapTokens <= 0 || groups.length < 2) {\n    return groups;\n  }\n  // Deterministic processing order (mirrored by the native side): by first occurrence position.\n  groups.sort(compareGroups);\n  for (let restart = true; restart;) {\n    restart = false;\n    const partnersByGroup = collectGapAdjacentPartners(groups, maxGapTokens);\n    for (let leftIndex = 0; leftIndex < groups.length && !restart; leftIndex += 1) {\n      for (const rightIndex of partnersByGroup[leftIndex] ?? []) {\n        const left = groups[leftIndex];\n        const right = groups[rightIndex];\n        if (!left || !right) {\n          continue;\n        }\n        const forward = mergeGroups(left, right, maxGapTokens, isReportableGroup);\n        const result = forward ?? mergeGroups(right, left, maxGapTokens, isReportableGroup);\n        if (!result) {\n          continue;\n        }\n        const leftReplaced = forward ? result.firstReplaced : result.secondReplaced;\n        const rightReplaced = forward ? result.secondReplaced : result.firstReplaced;\n        if (leftReplaced && rightReplaced) {\n          groups[leftIndex] = result.merged;\n          groups.splice(rightIndex, 1);\n        } else if (rightReplaced) {\n          groups[rightIndex] = result.merged;\n        } else if (leftReplaced) {\n          groups[leftIndex] = result.merged;\n        } else {\n          // Both groups stay (each is only partly paired, or reports nested copies of its own), so\n          // the merged group joins them instead of taking a place.\n          groups.push(result.merged);\n        }\n        // A group that stays keeps ALL its occurrences (line coverage must not shrink, and a\n        // reported group must keep >= 2 occurrences), so its paired occurrences now also live\n        // inside the merged group's occurrences: mark them so duplicateBlockCount counts each\n        // token span once, and so the same pair cannot merge again.\n        for (const occurrence of result.pairedRetained) {\n          occurrence.spanCountedElsewhere = true;\n        }\n        groups.sort(compareGroups);\n        restart = true;\n        break;\n      }\n    }\n  }\n  return groups;\n}\n\n/**\n * Per group index, the ascending indexes of later groups that mergeGroups can pair with it in\n * either direction: an occurrence of one starts within `maxGapTokens` after an occurrence of the\n * other ends. Every other pair fails mergeGroups, so skipping it leaves the fixpoint unchanged\n * while each restart stops costing a comparison per pair of groups.\n */\nfunction collectGapAdjacentPartners(groups: CountedOccurrence[][], maxGapTokens: number): number[][] {\n  const starts: { startTokenIndex: number; groupIndex: number }[] = [];\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      starts.push({ startTokenIndex: occurrence.startTokenIndex, groupIndex });\n    }\n  }\n  starts.sort((left, right) => left.startTokenIndex - right.startTokenIndex);\n  const partners = groups.map(() => new Set<number>());\n  for (const [groupIndex, group] of groups.entries()) {\n    for (const occurrence of group) {\n      for (\n        let index = lowerBoundByStart(starts, occurrence.endTokenIndex);\n        index < starts.length && (starts[index]?.startTokenIndex ?? 0) <= occurrence.endTokenIndex + maxGapTokens;\n        index += 1\n      ) {\n        const other = starts[index]?.groupIndex ?? groupIndex;\n        if (other !== groupIndex) {\n          partners[Math.min(other, groupIndex)]?.add(Math.max(other, groupIndex));\n        }\n      }\n    }\n  }\n  return partners.map((set) => [...set].toSorted((left, right) => left - right));\n}\n\nfunction lowerBoundByStart(sorted: { startTokenIndex: number }[], target: number): number {\n  let low = 0;\n  let high = sorted.length;\n  while (low < high) {\n    const middle = (low + high) >>> 1;\n    if ((sorted[middle]?.startTokenIndex ?? 0) < target) {\n      low = middle + 1;\n    } else {\n      high = middle;\n    }\n  }\n  return low;\n}\n\nfunction compareGroups(left: CountedOccurrence[], right: CountedOccurrence[]): number {\n  const leftFirst = left[0];\n  const rightFirst = right[0];\n  return (\n    (leftFirst?.startTokenIndex ?? 0) - (rightFirst?.startTokenIndex ?? 0) ||\n    (leftFirst?.endTokenIndex ?? 0) - (rightFirst?.endTokenIndex ?? 0)\n  );\n}\n\ninterface MergeResult<T> {\n  merged: T[];\n  /**\n   * Whether the merged group takes the respective input group's place: its standalone occurrences\n   * were all paired and it holds no nested copies that only it can report.\n   */\n  firstReplaced: boolean;\n  secondReplaced: boolean;\n  /** The occurrences of groups that stay, which the merged group's spans now also cover. */\n  pairedRetained: T[];\n}\n\n/**\n * Pairs `second` occurrences with gap-preceding `first` occurrences, greedily in source order:\n * each trailing occurrence takes the earliest unused leading occurrence within the gap, and a\n * pair's leading must start at or after the previous pair's trailing end so merged spans never\n * overlap. A merge needs at least two pairs (a merged group must still mean \"appears more than\n * once\") and must fully consume at least one group; for equal cardinalities this reduces to the\n * strict all-pairs merge, so pre-partial-merge behavior is unchanged there.\n */\nfunction mergeGroups<T extends CountedOccurrence>(\n  first: T[],\n  second: T[],\n  maxGapTokens: number,\n  isReportableGroup: (group: T[]) => boolean\n): MergeResult<T> | undefined {\n  // Occurrences a previous partial merge already paired into a merged group must not pair again:\n  // their spans already live inside that merged group, so re-pairing them would assemble a second,\n  // competing merged group instead of letting the existing merged group extend (and would count\n  // the same span twice). Consumption is still judged against the FULL group, so a group holding\n  // shared occurrences is never subsumed away; nested copies, which the merged span would not\n  // cover anyway, are left out of that judgment so they cannot veto a merge of the standalone\n  // copies.\n  const [leadings, firstLength] = pairableOccurrences(first);\n  const [trailings, secondLength] = pairableOccurrences(second);\n  const pairs: [T, T][] = [];\n  let leadingIndex = 0;\n  let previousTrailingEnd = -1;\n  for (const trailing of trailings) {\n    // Leadings ending too far before this trailing can never pair a later (even farther) one.\n    while (leadingIndex < leadings.length) {\n      const leading = leadings[leadingIndex];\n      if (leading && leading.endTokenIndex + maxGapTokens < trailing.startTokenIndex) {\n        leadingIndex += 1;\n      } else {\n        break;\n      }\n    }\n    const leading = leadings[leadingIndex];\n    if (\n      leading &&\n      leading.endTokenIndex <= trailing.startTokenIndex &&\n      leading.startTokenIndex >= previousTrailingEnd\n    ) {\n      pairs.push([leading, trailing]);\n      previousTrailingEnd = trailing.endTokenIndex;\n      leadingIndex += 1;\n    }\n  }\n  const firstFullyPaired = pairs.length === firstLength;\n  const secondFullyPaired = pairs.length === secondLength;\n  if (pairs.length < 2 || (!firstFullyPaired && !secondFullyPaired)) {\n    return undefined;\n  }\n  // A group holding nested copies stays even when all its standalone copies pair: the merged\n  // group's content is larger than what those copies matched, so only the original group can\n  // report which files share the matched fragment.\n  const firstReplaced = firstFullyPaired && !first.some((occurrence) => occurrence.nestedInLargerGroup);\n  const secondReplaced = secondFullyPaired && !second.some((occurrence) => occurrence.nestedInLargerGroup);\n  const merged = pairs.map(([leading, trailing]) => ({\n    ...leading,\n    // A merged occurrence is a fresh span combination; it never inherits shared-span marks.\n    spanCountedElsewhere: undefined,\n    nestedInLargerGroup: undefined,\n    segments: [...leading.segments, ...trailing.segments],\n    tokenCount: leading.tokenCount + trailing.tokenCount,\n    endTokenIndex: trailing.endTokenIndex,\n    endIndex: trailing.endIndex,\n    endLine: trailing.endLine,\n  }));\n  // Only occurrences of one file pair (file offsets exceed the gap), so a merged group spans the\n  // files its pairs sit in: with nested copies left out of pairing, that can be fewer files than\n  // the input groups covered, and a merged group that is no longer reportable must not form.\n  if (!isReportableGroup(merged)) {\n    return undefined;\n  }\n  const pairedRetained = [\n    ...(firstReplaced ? [] : pairs.map(([leading]) => leading)),\n    ...(secondReplaced ? [] : pairs.map(([, trailing]) => trailing)),\n  ];\n  return { merged, firstReplaced, secondReplaced, pairedRetained };\n}\n\n/** One pass over a group: its pairable occurrences and its non-nested occurrence count. */\nfunction pairableOccurrences<T extends CountedOccurrence>(group: T[]): [T[], number] {\n  const pairable: T[] = [];\n  let length = 0;\n  for (const occurrence of group) {\n    if (!occurrence.nestedInLargerGroup) {\n      length += 1;\n    }\n    // Either flag keeps an occurrence out of pairing: its span is already counted elsewhere, or it\n    // is a nested copy of content the merged span would not cover.\n    if (!occurrence.spanCountedElsewhere && !occurrence.nestedInLargerGroup) {\n      pairable.push(occurrence);\n    }\n  }\n  return [pairable, length];\n}\n\n/**\n * Redundant copies one group adds to duplicateBlockCount. Each redundant occurrence contributes\n * one count per matched fragment, so merging a gapped clone's fragments into one group does not\n * halve duplicateBlockCount: an edited two-fragment pair still counts 2, exactly as its unmerged\n * fragments did. Occurrence shapes can differ within one group (a\n * gap-merged exact pair plus an appended whole-block near-miss copy), so every occurrence's\n * fragments are summed and one representative — the largest — is deducted, keeping the count\n * independent of source order. Occurrences a partial gapped merge also paired into a merged group\n * are skipped: their spans are already counted there, and such a retained group deducts no\n * representative of its own — the merged group's representative already stands for the shared\n * content — so no token span contributes to the count twice.\n */\nexport function countRedundantFragments(group: CountedOccurrence[]): number {\n  let fragmentCount = 0;\n  let maxFragmentCount = 0;\n  let hasSharedOccurrence = false;\n  for (const occurrence of group) {\n    if (occurrence.spanCountedElsewhere) {\n      hasSharedOccurrence = true;\n      continue;\n    }\n    fragmentCount += occurrence.segments.length;\n    maxFragmentCount = Math.max(maxFragmentCount, occurrence.segments.length);\n  }\n  return hasSharedOccurrence ? fragmentCount : fragmentCount - maxFragmentCount;\n}\n\n/** Adds the 1-based code lines the segment's matched tokens cover; shared with cross-file coverage. */\nexport function collectSegmentLines(\n  segment: { startTokenIndex: number; endTokenIndex: number },\n  tokens: Token[],\n  codeLineNumbers: Set<number> | undefined,\n  duplicatedLines: Set<number>\n): void {\n  for (let index = segment.startTokenIndex; index < segment.endTokenIndex; index += 1) {\n    const token = tokens[index];\n    for (let row = token?.startRow ?? 0; row <= (token?.endRow ?? -1); row += 1) {\n      if (!codeLineNumbers || codeLineNumbers.has(row + 1)) {\n        duplicatedLines.add(row + 1);\n      }\n    }\n  }\n}\n"],"mappings":"sDAUA,MAAa,EAA0D,CACrE,UAAW,GACX,aAAc,GACd,qBAAsB,EACxB,EAQA,SAAgB,EAA0B,EAA4D,CACpG,MAAO,CACL,UAAW,EAAc,GAAS,UAAW,EAA0B,SAAS,EAChF,aAAc,EAAc,GAAS,aAAc,EAA0B,YAAY,EACzF,qBAAsB,EAAc,GAAS,qBAAsB,EAA0B,oBAAoB,CACnH,CACF,CAEA,SAAS,EAAc,EAA2B,EAA0B,CAC1E,OAAO,IAAU,IAAA,IAAa,OAAO,MAAM,CAAK,EAAI,GAAA,EAAWA,EAAAA,MAAAA,CAAM,KAAK,MAAM,CAAK,EAAG,EAAG,UAAa,CAC1G,CAiBA,SAAS,EAAe,EAAsB,EAA6B,CACzE,OAAO,EAAe,GAAK,CAC7B,CAuHA,SAAgB,EAAwB,EAA6B,CACnE,IAAM,EAAS,IAAI,WAAW,EAAO,OAAS,CAAC,EAC/C,IAAK,GAAM,CAAC,EAAO,KAAU,EAAO,QAAQ,EAC1C,EAAO,EAAQ,IAAM,EAAO,IAAU,IAAM,EAAM,cAAgB,IAAA,GAAY,EAAI,GAEpF,OAAO,CACT,CA0CA,SAAgB,EACd,EACA,EACA,EAC+B,CAC/B,IAAM,EAA4C,CAAC,EAC7C,EAAoC,CAAC,EACrC,EAA6B,CAAC,EACpC,IAAK,GAAM,CAAC,EAAc,KAAY,EAAS,QAAQ,EACrD,IAAK,IAAM,KAAc,EAAQ,WAC/B,EAAwB,KAAK,CAAY,EACzC,EAAW,KAAK,CAAU,EAG9B,IAAM,EAAa,GACjB,EAAS,EAAwB,IAAmB,GAChD,kBAAyB,IAAI,IAC7B,EAAmB,EAAW,KAAK,EAAY,IACnD,EAA0B,EAAU,CAAc,CAAC,EAAE,QAAU,CAAC,EAAG,EAAY,CAAS,CAC1F,EACA,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAAG,CAClE,IAAM,EAAe,EAAwB,IAAmB,EAChE,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,IAAM,KAAa,EAAK,CAC3B,GAAI,IAAc,IAAA,GAChB,SAEF,IAAM,EAAc,EAAuB,IAAI,CAAS,EACpD,GACF,EAAY,OAAS,EACjB,EAAY,iBAAmB,IACjC,EAAY,eAAiB,IAE3B,EAAY,eAAiB,IAC/B,EAAY,aAAe,IAE7B,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,EAC3D,EAAY,SAAW,KAAK,IAAI,EAAY,SAAU,CAAK,GAE3D,EAAuB,IAAI,EAAW,CACpC,MAAO,EACP,iBACA,eACA,SAAU,EACV,SAAU,CACZ,CAAC,CAEL,CAEJ,CAMA,IAAM,GAAW,EAA+B,IAA4B,CAC1E,GAAI,IAAc,IAAA,GAChB,MAAO,GAET,IAAM,EAAc,EAAuB,IAAI,CAAS,EAOxD,OANI,IAAgB,IAAA,IAAa,EAAY,MAAQ,EAC5C,GAEL,EACK,EAAY,eAAiB,GAE/B,EAAY,iBAAmB,IAAM,EAAY,SAAW,EAAY,UAAY,CAC7F,EAKM,EAAyB,GAAoC,CACjE,IAAM,EAAS,EAAiB,EAAO,eAAe,EAAE,iBAAmB,CAAC,EACtE,EAAY,EAAO,EAAO,OAChC,IAAK,IAAI,EAAQ,EAAO,MAAQ,EAAG,EAAQ,EAAO,MAAQ,EAAO,OAAQ,GAAS,EAChF,GAAI,EAAO,KAAW,EACpB,MAAO,GAGX,MAAO,EACT,EAEM,EAAmC,CAAC,EAC1C,IAAK,GAAM,CAAC,EAAgB,KAAY,EAAiB,QAAQ,EAC/D,IAAK,GAAM,CAAC,EAAO,KAAQ,EAAQ,kBAAkB,QAAQ,EAC3D,IAAK,GAAM,CAAC,EAAQ,KAAc,EAAI,QAAQ,EAAG,CAC/C,GAAI,CAAC,EAAQ,EAAW,CAAM,GAAK,CAAC,EAAsB,CAAE,iBAAgB,QAAO,QAAO,CAAC,EACzF,SAIF,IAAM,EAAgB,EAAQ,kBAAkB,EAAM,GAAG,EAAS,GAC5D,EAAe,EAAQ,kBAAkB,EAAQ,EAAE,GAAG,EAAS,GACjE,EAAQ,EAAe,EAAS,CAAC,GAAK,EAAQ,EAAc,EAAS,CAAC,GAG1E,EAAe,KAAK,CAAE,iBAAgB,QAAO,QAAO,CAAC,CACvD,CASJ,IAAM,EAAU,IAAI,IAAI,EAAe,IAAI,CAAQ,CAAC,EAChD,EAAW,EACf,KAAO,EAAS,OAAS,GAAG,CAC1B,IAAM,EAA4B,CAAC,EACnC,IAAK,IAAM,KAAU,EAAU,CAC7B,IAAM,EAAa,EAAW,EAAO,gBAC/B,EAAQ,IAAa,EAAO,OAC5B,EAAO,IAAa,EAAO,MAAQ,EAAO,OAAS,GACnD,EAAU,EAAU,EAAO,cAAc,EAC/C,GAAI,CAAC,GAAS,CAAC,GAAQ,CAAC,EACtB,SAEF,IAAM,EAAc,KAAK,EAAe,EAAQ,OAAQ,EAAQ,mBAAoB,EAAM,gBAAiB,EAAK,aAAa,IAC7H,EAAW,KAAK,CACd,UAAW,EAAY,EAAa,EAAM,gBAAiB,EAAK,cAAe,EAAO,CAAI,EAC1F,aAAc,EAAwB,EAAO,iBAAmB,CAClE,CAAC,EACD,EAAQ,KAAK,CAAM,CACrB,CACA,EAAW,CAAC,EACZ,IAAK,IAAM,KAAU,EACnB,IAAK,IAAM,IAAS,CAAC,EAAO,MAAO,EAAO,MAAQ,CAAC,EAAG,CACpD,IAAM,EAAY,CAAE,eAAgB,EAAO,eAAgB,QAAO,OAAQ,EAAO,OAAS,CAAE,EACtF,EAAe,EAAiB,EAAO,eAAe,EAAE,kBAAkB,EAAM,GAAG,EAAU,QAEjG,GAAQ,IAAI,EAAS,CAAS,CAAC,GAC9B,EAAQ,EAAc,EAAU,MAAM,GACtC,EAAsB,CAAS,IAIlC,EAAQ,IAAI,EAAS,CAAS,CAAC,EAC/B,EAAS,KAAK,CAAS,EACzB,CAEJ,CACA,OAAO,CACT,CAEA,SAAS,EAAS,EAAgC,CAChD,MAAO,GAAG,EAAO,eAAe,GAAG,EAAO,MAAM,GAAG,EAAO,QAC5D,CASA,SAAS,EAA0B,EAAiB,EAA0B,EAAqC,CACjH,IAAM,EAAkB,EAAW,IAAK,GACtC,EAAgB,EAAQ,EAAU,gBAAiB,EAAU,aAAa,CAC5E,EACM,EAA8C,CAAC,EACrD,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAW,OAAQ,GAAS,EAAG,CACzD,IAAM,EAA8B,CAAC,EACjC,EAAO,KACP,EAAa,EACX,EAAS,KAAK,IAAI,EAAW,OAAQ,EAAQ,GAAyB,EAC5E,IAAK,IAAI,EAAM,EAAO,EAAM,EAAQ,GAAO,EAAG,CAC5C,IAAM,EAAY,EAAW,GACvB,EAAgB,EAAgB,GACtC,GAAI,CAAC,GAAa,IAAkB,IAAA,GAClC,MAEF,EAAO,EAAc,EAAM,CAAa,EACxC,GAAc,EAAU,cAAgB,EAAU,gBAClD,IAAM,EAAiB,EAAM,EAAQ,EACrC,EAAI,GACF,GAAkB,GAA6B,GAAc,EACzD,EAAc,EAAM,CAAc,EAClC,IAAA,EACR,CACA,EAAkB,KAAK,CAAG,CAC5B,CACA,MAAO,CAAE,oBAAmB,iBAAgB,CAC9C,CAOA,SAAgB,EAAU,EAAe,EAAuB,CAC9D,OAAO,EAAuB,CAAC,CAAC,CAAC,CAAC,CACpC,CAMA,SAAgB,EAAuB,EAA0C,CAC/E,IAAM,EAAa,EAAE,OAAS,KAAQ,EAChC,kBAAgB,IAAI,IAC1B,IAAK,GAAM,CAAC,EAAO,KAAW,EAAE,QAAQ,EAAG,CACzC,IAAI,EAAO,EAAc,IAAI,CAAM,EAC9B,IACH,EAAO,IAAI,YAAY,CAAS,EAChC,EAAc,IAAI,EAAQ,CAAI,GAEhC,IAAM,EAAO,IAAU,EAIvB,EAAK,IAAS,EAAK,IAAS,GAAM,IAAM,EAAQ,GAClD,CAEA,IAAM,EAAI,IAAI,YAAY,CAAS,EACnC,MAAQ,IAAM,CACZ,EAAE,KAAK,CAAC,EACR,IAAK,IAAM,KAAU,EAAG,CACtB,IAAM,EAAY,EAAc,IAAI,CAAM,EAEtC,EAAa,EACb,EAAS,EACb,IAAK,IAAI,EAAO,EAAG,EAAO,EAAW,GAAQ,EAAG,CAC9C,IAAM,EAAW,EAAE,IAAS,EAEtB,IAAM,IAAY,IAAS,GAAK,KAAc,EAE9C,GAAY,GAAY,EAAK,KAAgB,EACnD,EAAa,IAAa,GAC1B,IAAM,EAAa,EAAI,EAAU,EACjC,EAAS,IAAa,GAEtB,EAAE,GAAQ,EAAI,CAAC,CACjB,CACF,CAEA,IAAI,EAAS,EACb,IAAK,IAAM,KAAQ,EACjB,GAAU,EAAS,CAAI,EAEzB,OAAO,CACT,CACF,CAEA,SAAS,EAAS,EAAuB,CACvC,IAAI,EAAQ,GAAU,IAAU,EAAK,YAErC,MADA,IAAS,EAAQ,YAAmB,IAAU,EAAK,WAC3C,KAAK,KAAM,GAAS,IAAU,GAAM,UAAe,QAAa,IAAM,GAAM,GACtF,CAEA,SAAS,EACP,EACA,EACA,EACA,EACA,EACoB,CACpB,MAAO,CACL,cACA,WAAY,EAAgB,EAC5B,kBACA,gBACA,WAAY,EAAM,WAClB,SAAU,EAAK,SACf,UAAW,EAAM,UACjB,QAAS,EAAK,OAChB,CACF,CAGA,MAAM,EAAkC,CAAC,EACnC,EAAmC,CAAC,EAE1C,SAAS,EAAoB,EAAuB,CAClD,IAAI,EAAO,EAAsB,GAKjC,OAJI,IAAS,IAAA,KACX,EAAO,EAAS,IAAI,GAAO,EAC3B,EAAsB,GAAS,GAE1B,CACT,CAEA,SAAS,EAAqB,EAAuB,CACnD,IAAI,EAAO,EAAuB,GAKlC,OAJI,IAAS,IAAA,KACX,EAAO,EAAU,IAAI,GAAO,EAC5B,EAAuB,GAAS,GAE3B,CACT,CAUA,SAAS,EACP,EACA,EACA,EACA,EACQ,CAGR,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EADrD,GADC,EAAmB,IAAkB,IAAM,EAAmB,IAAoB,GACtD,EAAgB,CACkC,CAAC,EACrG,MAAO,GAAG,EAAQ,GAAG,EAAU,GAAG,EAAgB,GACpD,CASA,SAAS,EAAgB,EAAiB,EAAyB,EAA+B,CAChG,GAAM,CAAC,EAAS,GAAa,EAAoB,EAAQ,EAAiB,EAAe,EAAK,EAE9F,OAAO,EAAU,KAAK,KAAK,EAAW,EAAE,CAC1C,CAEA,SAAS,EACP,EACA,EACA,EACA,EACkB,CAClB,IAAM,kBAAoB,IAAI,IAC1B,EAAU,KACV,EAAY,MAChB,IAAK,IAAI,EAAQ,EAAiB,EAAQ,EAAe,GAAS,EAAG,CACnE,IAAM,EAAQ,EAAO,GACrB,GAAI,CAAC,EACH,SAIF,IAAI,EACA,EACJ,GAAI,EAAM,OAAS,KAAM,CACvB,IAAI,EAAkB,EAAkB,IAAI,EAAM,IAAI,EAClD,IAAoB,IAAA,KACtB,EAAkB,EAAkB,KACpC,EAAkB,IAAI,EAAM,KAAM,CAAe,GAEnD,EAAO,EAAoB,CAAe,EAC1C,EAAQ,EAAqB,CAAe,CAC9C,KACE,GAAO,EAAM,SACb,EAAQ,EAAM,UAGhB,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAQ,EAC5C,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EACnC,GAAqB,EAAM,cAAgB,IAAA,IAAa,EAAM,eAAiB,IAAA,KAEjF,EAAW,KAAK,KAAK,EAAS,EAAE,EAAI,EAAM,YAAe,EACzD,EAAY,KAAK,KAAK,EAAW,EAAE,EAAI,EAAM,aAEjD,CACA,MAAO,CAAC,EAAS,CAAS,CAC5B,CAGA,SAAS,EAAS,EAAsB,CACtC,IAAI,EAAO,KACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAM,EAAE,EAAI,EAAK,WAAW,CAAK,EAEpD,OAAO,CACT,CAGA,SAAS,EAAU,EAAsB,CACvC,IAAI,EAAO,YACX,IAAK,IAAI,EAAQ,EAAG,EAAQ,EAAK,OAAQ,GAAS,EAEhD,EAAO,KAAK,KAAK,EAAO,EAAK,WAAW,CAAK,EAAG,QAAU,EAE5D,OAAO,CACT,CAEA,SAAS,EAAc,EAAc,EAAuB,CAC1D,OAAO,KAAK,KAAK,EAAM,EAAE,EAAI,CAC/B,CAoBA,SAAgB,EACd,EACA,EACA,MAAmD,GAC5C,CACP,GAAI,GAAgB,GAAK,EAAO,OAAS,EACvC,OAAO,EAGT,EAAO,KAAK,CAAa,EACzB,IAAK,IAAI,EAAU,GAAM,GAAU,CACjC,EAAU,GACV,IAAM,EAAkB,EAA2B,EAAQ,CAAY,EACvE,IAAK,IAAI,EAAY,EAAG,EAAY,EAAO,QAAU,CAAC,EAAS,GAAa,EAC1E,IAAK,IAAM,KAAc,EAAgB,IAAc,CAAC,EAAG,CACzD,IAAM,EAAO,EAAO,GACd,EAAQ,EAAO,GACrB,GAAI,CAAC,GAAQ,CAAC,EACZ,SAEF,IAAM,EAAU,EAAY,EAAM,EAAO,EAAc,CAAiB,EAClE,EAAS,GAAW,EAAY,EAAO,EAAM,EAAc,CAAiB,EAClF,GAAI,CAAC,EACH,SAEF,IAAM,EAAe,EAAU,EAAO,cAAgB,EAAO,eACvD,EAAgB,EAAU,EAAO,eAAiB,EAAO,cAC3D,GAAgB,GAClB,EAAO,GAAa,EAAO,OAC3B,EAAO,OAAO,EAAY,CAAC,GAClB,EACT,EAAO,GAAc,EAAO,OACnB,EACT,EAAO,GAAa,EAAO,OAI3B,EAAO,KAAK,EAAO,MAAM,EAM3B,IAAK,IAAM,KAAc,EAAO,eAC9B,EAAW,qBAAuB,GAEpC,EAAO,KAAK,CAAa,EACzB,EAAU,GACV,KACF,CAEJ,CACA,OAAO,CACT,CAQA,SAAS,EAA2B,EAA+B,EAAkC,CACnG,IAAM,EAA4D,CAAC,EACnE,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,EAAO,KAAK,CAAE,gBAAiB,EAAW,gBAAiB,YAAW,CAAC,EAG3E,EAAO,MAAM,EAAM,IAAU,EAAK,gBAAkB,EAAM,eAAe,EACzE,IAAM,EAAW,EAAO,wBAAU,IAAI,GAAa,EACnD,IAAK,GAAM,CAAC,EAAY,KAAU,EAAO,QAAQ,EAC/C,IAAK,IAAM,KAAc,EACvB,IACE,IAAI,EAAQ,EAAkB,EAAQ,EAAW,aAAa,EAC9D,EAAQ,EAAO,SAAW,EAAO,EAAM,EAAE,iBAAmB,IAAM,EAAW,cAAgB,EAC7F,GAAS,EACT,CACA,IAAM,EAAQ,EAAO,EAAM,EAAE,YAAc,EACvC,IAAU,GACZ,EAAS,KAAK,IAAI,EAAO,CAAU,EAAE,EAAE,IAAI,KAAK,IAAI,EAAO,CAAU,CAAC,CAE1E,CAGJ,OAAO,EAAS,IAAK,GAAQ,CAAC,GAAG,CAAG,CAAC,CAAC,UAAU,EAAM,IAAU,EAAO,CAAK,CAAC,CAC/E,CAEA,SAAS,EAAkB,EAAuC,EAAwB,CACxF,IAAI,EAAM,EACN,EAAO,EAAO,OAClB,KAAO,EAAM,GAAM,CACjB,IAAM,EAAU,EAAM,IAAU,GAC3B,EAAO,EAAO,EAAE,iBAAmB,GAAK,EAC3C,EAAM,EAAS,EAEf,EAAO,CAEX,CACA,OAAO,CACT,CAEA,SAAS,EAAc,EAA2B,EAAoC,CACpF,IAAM,EAAY,EAAK,GACjB,EAAa,EAAM,GACzB,OACG,GAAW,iBAAmB,IAAM,GAAY,iBAAmB,KACnE,GAAW,eAAiB,IAAM,GAAY,eAAiB,EAEpE,CAsBA,SAAS,EACP,EACA,EACA,EACA,EAC4B,CAQ5B,GAAM,CAAC,EAAU,GAAe,EAAoB,CAAK,EACnD,CAAC,EAAW,GAAgB,EAAoB,CAAM,EACtD,EAAkB,CAAC,EACrB,EAAe,EACf,EAAsB,GAC1B,IAAK,IAAM,KAAY,EAAW,CAEhC,KAAO,EAAe,EAAS,QAAQ,CACrC,IAAM,EAAU,EAAS,GACzB,GAAI,GAAW,EAAQ,cAAgB,EAAe,EAAS,gBAC7D,GAAgB,OAEhB,KAEJ,CACA,IAAM,EAAU,EAAS,GAEvB,GACA,EAAQ,eAAiB,EAAS,iBAClC,EAAQ,iBAAmB,IAE3B,EAAM,KAAK,CAAC,EAAS,CAAQ,CAAC,EAC9B,EAAsB,EAAS,cAC/B,GAAgB,EAEpB,CACA,IAAM,EAAmB,EAAM,SAAW,EACpC,EAAoB,EAAM,SAAW,EAC3C,GAAI,EAAM,OAAS,GAAM,CAAC,GAAoB,CAAC,EAC7C,OAKF,IAAM,EAAgB,GAAoB,CAAC,EAAM,KAAM,GAAe,EAAW,mBAAmB,EAC9F,EAAiB,GAAqB,CAAC,EAAO,KAAM,GAAe,EAAW,mBAAmB,EACjG,EAAS,EAAM,KAAK,CAAC,EAAS,MAAe,CACjD,GAAG,EAEH,qBAAsB,IAAA,GACtB,oBAAqB,IAAA,GACrB,SAAU,CAAC,GAAG,EAAQ,SAAU,GAAG,EAAS,QAAQ,EACpD,WAAY,EAAQ,WAAa,EAAS,WAC1C,cAAe,EAAS,cACxB,SAAU,EAAS,SACnB,QAAS,EAAS,OACpB,EAAE,EAIG,KAAkB,CAAM,EAO7B,MAAO,CAAE,SAAQ,gBAAe,iBAAgB,eAAA,CAH9C,GAAI,EAAgB,CAAC,EAAI,EAAM,KAAK,CAAC,KAAa,CAAO,EACzD,GAAI,EAAiB,CAAC,EAAI,EAAM,KAAK,EAAG,KAAc,CAAQ,CAEH,CAAE,CACjE,CAGA,SAAS,EAAiD,EAA2B,CACnF,IAAM,EAAgB,CAAC,EACnB,EAAS,EACb,IAAK,IAAM,KAAc,EAClB,EAAW,sBACd,GAAU,GAIR,CAAC,EAAW,sBAAwB,CAAC,EAAW,qBAClD,EAAS,KAAK,CAAU,EAG5B,MAAO,CAAC,EAAU,CAAM,CAC1B,CAcA,SAAgB,EAAwB,EAAoC,CAC1E,IAAI,EAAgB,EAChB,EAAmB,EACnB,EAAsB,GAC1B,IAAK,IAAM,KAAc,EAAO,CAC9B,GAAI,EAAW,qBAAsB,CACnC,EAAsB,GACtB,QACF,CACA,GAAiB,EAAW,SAAS,OACrC,EAAmB,KAAK,IAAI,EAAkB,EAAW,SAAS,MAAM,CAC1E,CACA,OAAO,EAAsB,EAAgB,EAAgB,CAC/D,CAGA,SAAgB,EACd,EACA,EACA,EACA,EACM,CACN,IAAK,IAAI,EAAQ,EAAQ,gBAAiB,EAAQ,EAAQ,cAAe,GAAS,EAAG,CACnF,IAAM,EAAQ,EAAO,GACrB,IAAK,IAAI,EAAM,GAAO,UAAY,EAAG,IAAQ,GAAO,QAAU,IAAK,GAAO,GACpE,CAAC,GAAmB,EAAgB,IAAI,EAAM,CAAC,IACjD,EAAgB,IAAI,EAAM,CAAC,CAGjC,CACF"}

package/dist/duplication.js

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