@lota-sdk/core 0.4.8 → 0.4.9

package/src/workers/utils/repo-structure-extractor.ts

@@ -1,6 +1,3 @@
-import { readdir } from 'node:fs/promises'
-import path from 'node:path'
-
 import { RepositoryStructureArtifactSchema, RepositoryStructureSummarySchema } from '@lota-sdk/shared'
 import type {
   RepositoryStructureArtifact,
@@ -9,6 +6,11 @@ import type {
   RepositoryStructureSignal,
   RepositoryStructureSummary,
 } from '@lota-sdk/shared'
+import type { Cause } from 'effect'
+import { Effect } from 'effect'
+
+import { effectTryPromise } from '../../effect/helpers'
+import { nowIsoDateTimeString } from '../../utils/date-time'
 
 const EXTRACTOR_VERSION = 'repository-structure-extractor.v1'
 const IGNORED_DIR_NAMES = new Set([
@@ -54,6 +56,23 @@ type PackageJson = {
   devDependencies?: Record<string, string>
 }
 
+function isPlainRecord(value: unknown): value is Record<string, unknown> {
+  return typeof value === 'object' && value !== null && !Array.isArray(value)
+}
+
+function pickStringRecord(value: unknown): Record<string, string> | undefined {
+  if (!isPlainRecord(value)) return undefined
+
+  const result: Record<string, string> = {}
+  for (const [key, entry] of Object.entries(value)) {
+    if (typeof entry === 'string') {
+      result[key] = entry
+    }
+  }
+
+  return Object.keys(result).length > 0 ? result : undefined
+}
+
 type ExtractParams = {
   rootDir: string
   repositoryKey: string
@@ -70,7 +89,35 @@ type ExtractResult = {
 }
 
 function normalizePath(value: string): string {
-  return value.replaceAll(path.sep, '/')
+  return value.replaceAll('\\', '/').replaceAll(/\/+/g, '/')
+}
+
+function joinPath(...segments: string[]): string {
+  return normalizePath(
+    segments
+      .map((segment) => segment.trim())
+      .filter((segment) => segment.length > 0)
+      .join('/'),
+  )
+}
+
+function extname(filePath: string): string {
+  const base = basename(filePath)
+  const dotIndex = base.lastIndexOf('.')
+  return dotIndex > 0 ? base.slice(dotIndex) : ''
+}
+
+function dirname(filePath: string): string {
+  const normalized = normalizePath(filePath).replace(/\/+$/, '')
+  const slashIndex = normalized.lastIndexOf('/')
+  if (slashIndex <= 0) return '.'
+  return normalized.slice(0, slashIndex)
+}
+
+function basename(filePath: string): string {
+  const normalized = normalizePath(filePath).replace(/\/+$/, '')
+  const slashIndex = normalized.lastIndexOf('/')
+  return slashIndex >= 0 ? normalized.slice(slashIndex + 1) : normalized
 }
 
 function unique<T>(values: T[]): T[] {
@@ -99,31 +146,62 @@ function createSignal(params: {
   }
 }
 
-async function collectRelativeFilePaths(rootDir: string, currentDir = ''): Promise<string[]> {
-  const absoluteDir = path.join(rootDir, currentDir)
-  const entries = await readdir(absoluteDir, { withFileTypes: true })
-  const filePaths: string[] = []
-
-  for (const entry of entries) {
-    const relativePath = normalizePath(path.join(currentDir, entry.name))
-    if (entry.isDirectory()) {
-      if (IGNORED_DIR_NAMES.has(entry.name)) continue
-      filePaths.push(...(await collectRelativeFilePaths(rootDir, relativePath)))
-      continue
+function collectRelativeFilePathsEffect(
+  rootDir: string,
+  currentDir = '',
+): Effect.Effect<string[], Cause.UnknownError, never> {
+  return Effect.gen(function* () {
+    const absoluteDir = currentDir ? joinPath(rootDir, currentDir) : rootDir
+    const entries = yield* effectTryPromise(() =>
+      Array.fromAsync(new Bun.Glob('*').scan({ cwd: absoluteDir, onlyFiles: false, absolute: false })),
+    )
+
+    const filePaths: string[] = []
+    for (const entry of entries) {
+      const relativePath = currentDir ? joinPath(currentDir, entry) : normalizePath(entry)
+      const stats = yield* effectTryPromise(() => Bun.file(joinPath(rootDir, relativePath)).stat())
+
+      if (stats.isDirectory()) {
+        if (IGNORED_DIR_NAMES.has(entry)) continue
+        const nested = yield* collectRelativeFilePathsEffect(rootDir, relativePath)
+        filePaths.push(...nested)
+        continue
+      }
+
+      filePaths.push(relativePath)
     }
-    filePaths.push(relativePath)
-  }
 
-  return filePaths
+    return filePaths
+  })
 }
 
-async function readPackageJson(rootDir: string, relativePath: string): Promise<PackageJson | null> {
-  try {
-    const raw = await Bun.file(path.join(rootDir, relativePath)).text()
-    return JSON.parse(raw) as PackageJson
-  } catch {
-    return null
-  }
+function readPackageJsonEffect(
+  rootDir: string,
+  relativePath: string,
+): Effect.Effect<PackageJson | null, Cause.UnknownError, never> {
+  return effectTryPromise(() => Bun.file(joinPath(rootDir, relativePath)).text()).pipe(
+    Effect.map((raw) => {
+      try {
+        const parsed: unknown = JSON.parse(raw)
+        if (!isPlainRecord(parsed)) return null
+
+        const packageJson: PackageJson = {
+          ...(typeof parsed.name === 'string' ? { name: parsed.name } : {}),
+          ...(typeof parsed.private === 'boolean' ? { private: parsed.private } : {}),
+        }
+        const scripts = pickStringRecord(parsed.scripts)
+        if (scripts) packageJson.scripts = scripts
+        const dependencies = pickStringRecord(parsed.dependencies)
+        if (dependencies) packageJson.dependencies = dependencies
+        const devDependencies = pickStringRecord(parsed.devDependencies)
+        if (devDependencies) packageJson.devDependencies = devDependencies
+        return packageJson
+      } catch {
+        return null
+      }
+    }),
+    Effect.catch(() => Effect.succeed(null)),
+  )
 }
 
 function gatherTopLevelEntries(filePaths: string[]): string[] {
@@ -137,7 +215,7 @@ function gatherLanguageSignals(filePaths: string[]): RepositoryStructureSignal[]
   const languageEvidence = new Map<string, string[]>()
 
   for (const filePath of filePaths) {
-    const language = LANGUAGE_BY_EXTENSION[path.extname(filePath).toLowerCase()]
+    const language = LANGUAGE_BY_EXTENSION[extname(filePath).toLowerCase()]
     if (!language) continue
     const entries = languageEvidence.get(language) ?? []
     entries.push(filePath)
@@ -354,7 +432,7 @@ function inferComponentLanguages(filePaths: string[], componentPath: string): st
   )
   return unique(
     componentFiles
-      .map((filePath) => LANGUAGE_BY_EXTENSION[path.extname(filePath).toLowerCase()])
+      .map((filePath) => LANGUAGE_BY_EXTENSION[extname(filePath).toLowerCase()])
       .filter((value): value is string => Boolean(value)),
   ).slice(0, 20)
 }
@@ -398,245 +476,256 @@ function buildStructureMarkdown(params: {
   ].join('\n')
 }
 
-export async function extractRepositoryStructure(params: ExtractParams): Promise<ExtractResult> {
-  const filePaths = await collectRelativeFilePaths(params.rootDir)
-  const topLevelEntries = gatherTopLevelEntries(filePaths)
-  const packageJsonPaths = filePaths.filter((filePath) => filePath.endsWith('package.json')).slice(0, 200)
-  const packageJsons = (
-    await Promise.all(
-      packageJsonPaths.map(async (packageJsonPath) => {
-        const data = await readPackageJson(params.rootDir, packageJsonPath)
-        return data ? { path: packageJsonPath, data } : null
-      }),
+function extractRepositoryStructureEffect(
+  params: ExtractParams,
+): Effect.Effect<ExtractResult, Cause.UnknownError, never> {
+  return Effect.gen(function* () {
+    const filePaths = yield* collectRelativeFilePathsEffect(params.rootDir)
+    const topLevelEntries = gatherTopLevelEntries(filePaths)
+    const packageJsonPaths = filePaths.filter((filePath) => filePath.endsWith('package.json')).slice(0, 200)
+    const packageJsons = yield* Effect.forEach(packageJsonPaths, (packageJsonPath) =>
+      readPackageJsonEffect(params.rootDir, packageJsonPath).pipe(
+        Effect.map((data) => (data ? { path: packageJsonPath, data } : null)),
+      ),
+    )
+    const typedPackageJsons = packageJsons.filter(
+      (value): value is { path: string; data: PackageJson } => value !== null,
     )
-  ).filter((value): value is { path: string; data: PackageJson } => value !== null)
-
-  const packageManagers = inferPackageManagerSignals(filePaths)
-  const components: RepositoryStructureComponent[] = packageJsons.map((packageJson) => {
-    const componentPath = packageJson.path === 'package.json' ? '.' : normalizePath(path.dirname(packageJson.path))
-    return {
-      kind: inferComponentKind(componentPath, packageJson.data),
-      name: packageJson.data.name?.trim() || path.basename(componentPath === '.' ? params.rootDir : componentPath),
-      path: componentPath,
-      packageManager: packageManagers[0]?.label,
-      languages: inferComponentLanguages(filePaths, componentPath),
-      frameworks: inferComponentFrameworks(packageJson.data),
-      reason: 'Detected a package manifest in this directory.',
-      evidencePaths: [packageJson.path],
-    }
-  })
 
-  const languages = gatherLanguageSignals(filePaths)
-  const frameworks = inferFrameworkSignals(packageJsons)
-  const routeSignals = buildPathSignals({
-    keyPrefix: 'route',
-    label: 'Routes',
-    reason: 'Detected route-related directories or files.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)(routes|pages|app\/routes)\//.test(filePath)),
-      20,
-    ),
-  })
-  const apiSignals = buildPathSignals({
-    keyPrefix: 'api',
-    label: 'API',
-    reason: 'Detected API or server route surfaces.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)(api|server|openapi|rpc)\//.test(filePath)),
-      20,
-    ),
-  })
-  const databaseSignals = buildPathSignals({
-    keyPrefix: 'database',
-    label: 'Database',
-    reason: 'Detected database schemas, migrations, or data-access directories.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)(db|database|migrations|prisma|drizzle|schema)\b/.test(filePath)),
-      20,
-    ),
-  })
-  const authSignals = buildPathSignals({
-    keyPrefix: 'auth',
-    label: 'Auth',
-    reason: 'Detected authentication-related code or configuration.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)(auth|better-auth|clerk|next-auth)\b/.test(filePath)),
-      20,
-    ),
-  })
-  const jobSignals = buildPathSignals({
-    keyPrefix: 'job',
-    label: 'Jobs',
-    reason: 'Detected jobs, queues, or worker execution surfaces.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)(jobs|queues|workers?)\b/.test(filePath)),
-      20,
-    ),
-  })
-  const envFiles = buildPathSignals({
-    keyPrefix: 'env',
-    label: 'Environment',
-    reason: 'Detected environment variable files.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)\.env(\.|$)/.test(filePath)),
-      20,
-    ),
-  })
-  const dockerFiles = buildPathSignals({
-    keyPrefix: 'docker',
-    label: 'Docker',
-    reason: 'Detected Docker build or compose files.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(Dockerfile|docker-compose|compose\.ya?ml|\.dockerignore)/i.test(filePath)),
-      20,
-    ),
-  })
-  const ciCdFiles = buildPathSignals({
-    keyPrefix: 'cicd',
-    label: 'CI/CD',
-    reason: 'Detected continuous integration or delivery workflow files.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /^\.github\/workflows\/|\.gitlab-ci\.yml$|circleci\//.test(filePath)),
-      20,
-    ),
-  })
-  const deploymentFiles = buildPathSignals({
-    keyPrefix: 'deploy',
-    label: 'Deployment',
-    reason: 'Detected deployment platform or infrastructure descriptors.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) =>
-        /(vercel\.json|fly\.toml|render\.ya?ml|netlify\.toml|railway\.json|k8s|helm)/i.test(filePath),
+    const packageManagers = inferPackageManagerSignals(filePaths)
+    const components: RepositoryStructureComponent[] = typedPackageJsons.map((packageJson) => {
+      const componentPath = packageJson.path === 'package.json' ? '.' : normalizePath(dirname(packageJson.path))
+      return {
+        kind: inferComponentKind(componentPath, packageJson.data),
+        name: packageJson.data.name?.trim() || basename(componentPath === '.' ? params.rootDir : componentPath),
+        path: componentPath,
+        packageManager: packageManagers[0]?.label,
+        languages: inferComponentLanguages(filePaths, componentPath),
+        frameworks: inferComponentFrameworks(packageJson.data),
+        reason: 'Detected a package manifest in this directory.',
+        evidencePaths: [packageJson.path],
+      }
+    })
+
+    const languages = gatherLanguageSignals(filePaths)
+    const frameworks = inferFrameworkSignals(typedPackageJsons)
+    const routeSignals = buildPathSignals({
+      keyPrefix: 'route',
+      label: 'Routes',
+      reason: 'Detected route-related directories or files.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(^|\/)(routes|pages|app\/routes)\//.test(filePath)),
+        20,
       ),
-      20,
-    ),
-  })
-  const testSignals = buildPathSignals({
-    keyPrefix: 'test',
-    label: 'Tests',
-    reason: 'Detected repository test files.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)(__tests__|tests?)\/|(\.test\.|\.(spec)\.)/.test(filePath)),
-      20,
-    ),
-  })
-  const lintSignals = buildPathSignals({
-    keyPrefix: 'lint',
-    label: 'Linting',
-    reason: 'Detected linting configuration files.',
-    filePaths: truncate(
-      findPaths(
-        filePaths,
-        (filePath) => /(eslint|oxlint|biome|lint)/i.test(filePath) && /\.(json|js|ts|mjs|cjs|yaml|yml)$/.test(filePath),
+    })
+    const apiSignals = buildPathSignals({
+      keyPrefix: 'api',
+      label: 'API',
+      reason: 'Detected API or server route surfaces.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(^|\/)(api|server|openapi|rpc)\//.test(filePath)),
+        20,
       ),
-      20,
-    ),
-  })
-  const formatSignals = buildPathSignals({
-    keyPrefix: 'format',
-    label: 'Formatting',
-    reason: 'Detected code-formatting configuration files.',
-    filePaths: truncate(
-      findPaths(
-        filePaths,
-        (filePath) => /(prettier|oxfmt|biome)/i.test(filePath) && /\.(json|js|ts|mjs|cjs|yaml|yml)$/.test(filePath),
+    })
+    const databaseSignals = buildPathSignals({
+      keyPrefix: 'database',
+      label: 'Database',
+      reason: 'Detected database schemas, migrations, or data-access directories.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(^|\/)(db|database|migrations|prisma|drizzle|schema)\b/.test(filePath)),
+        20,
       ),
-      20,
-    ),
-  })
-  const typecheckSignals = buildPathSignals({
-    keyPrefix: 'typecheck',
-    label: 'Typechecking',
-    reason: 'Detected typechecking configuration files.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(tsconfig|mypy|pyright)/i.test(filePath)),
-      20,
-    ),
-  })
-  const entryPoints = buildPathSignals({
-    keyPrefix: 'entry',
-    label: 'Entry point',
-    reason: 'Detected likely application entry points.',
-    filePaths: truncate(
-      findPaths(filePaths, (filePath) => /(^|\/)(main|index|app|server|worker)\.(ts|tsx|js|jsx|py|go)$/.test(filePath)),
-      20,
-    ),
-  })
-
-  const repoShape = inferRepoShape({ topLevelEntries, components, frameworks, filePaths })
-
-  const repoStructure = RepositoryStructureArtifactSchema.parse({
-    schemaVersion: 'repository-structure.v1',
-    extractorVersion: EXTRACTOR_VERSION,
-    generatedAt: new Date().toISOString(),
-    repository: {
-      repositoryKey: params.repositoryKey,
-      repositoryFullName: params.repositoryFullName,
-      defaultBranch: params.defaultBranch,
-      indexBranch: params.indexBranch,
-      commitSha: params.commitSha,
-    },
-    topLevelEntries,
-    repoShape,
-    packageManagers,
-    languages,
-    frameworks,
-    components: truncate(components, 200),
-    entryPoints,
-    routeSignals,
-    apiSignals,
-    databaseSignals,
-    authSignals,
-    jobSignals,
-    envFiles,
-    dockerFiles,
-    ciCdFiles,
-    deploymentFiles,
-    testSignals,
-    lintSignals,
-    formatSignals,
-    typecheckSignals,
-    unknowns: repoShape.kind === 'uncertain' ? ['Repository shape could not be confidently classified.'] : [],
-  })
+    })
+    const authSignals = buildPathSignals({
+      keyPrefix: 'auth',
+      label: 'Auth',
+      reason: 'Detected authentication-related code or configuration.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(^|\/)(auth|better-auth|clerk|next-auth)\b/.test(filePath)),
+        20,
+      ),
+    })
+    const jobSignals = buildPathSignals({
+      keyPrefix: 'job',
+      label: 'Jobs',
+      reason: 'Detected jobs, queues, or worker execution surfaces.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(^|\/)(jobs|queues|workers?)\b/.test(filePath)),
+        20,
+      ),
+    })
+    const envFiles = buildPathSignals({
+      keyPrefix: 'env',
+      label: 'Environment',
+      reason: 'Detected environment variable files.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(^|\/)\.env(\.|$)/.test(filePath)),
+        20,
+      ),
+    })
+    const dockerFiles = buildPathSignals({
+      keyPrefix: 'docker',
+      label: 'Docker',
+      reason: 'Detected Docker build or compose files.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(Dockerfile|docker-compose|compose\.ya?ml|\.dockerignore)/i.test(filePath)),
+        20,
+      ),
+    })
+    const ciCdFiles = buildPathSignals({
+      keyPrefix: 'cicd',
+      label: 'CI/CD',
+      reason: 'Detected continuous integration or delivery workflow files.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /^\.github\/workflows\/|\.gitlab-ci\.yml$|circleci\//.test(filePath)),
+        20,
+      ),
+    })
+    const deploymentFiles = buildPathSignals({
+      keyPrefix: 'deploy',
+      label: 'Deployment',
+      reason: 'Detected deployment platform or infrastructure descriptors.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) =>
+          /(vercel\.json|fly\.toml|render\.ya?ml|netlify\.toml|railway\.json|k8s|helm)/i.test(filePath),
+        ),
+        20,
+      ),
+    })
+    const testSignals = buildPathSignals({
+      keyPrefix: 'test',
+      label: 'Tests',
+      reason: 'Detected repository test files.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(^|\/)(__tests__|tests?)\/|(\.test\.|\.(spec)\.)/.test(filePath)),
+        20,
+      ),
+    })
+    const lintSignals = buildPathSignals({
+      keyPrefix: 'lint',
+      label: 'Linting',
+      reason: 'Detected linting configuration files.',
+      filePaths: truncate(
+        findPaths(
+          filePaths,
+          (filePath) =>
+            /(eslint|oxlint|biome|lint)/i.test(filePath) && /\.(json|js|ts|mjs|cjs|yaml|yml)$/.test(filePath),
+        ),
+        20,
+      ),
+    })
+    const formatSignals = buildPathSignals({
+      keyPrefix: 'format',
+      label: 'Formatting',
+      reason: 'Detected code-formatting configuration files.',
+      filePaths: truncate(
+        findPaths(
+          filePaths,
+          (filePath) => /(prettier|oxfmt|biome)/i.test(filePath) && /\.(json|js|ts|mjs|cjs|yaml|yml)$/.test(filePath),
+        ),
+        20,
+      ),
+    })
+    const typecheckSignals = buildPathSignals({
+      keyPrefix: 'typecheck',
+      label: 'Typechecking',
+      reason: 'Detected typechecking configuration files.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) => /(tsconfig|mypy|pyright)/i.test(filePath)),
+        20,
+      ),
+    })
+    const entryPoints = buildPathSignals({
+      keyPrefix: 'entry',
+      label: 'Entry point',
+      reason: 'Detected likely application entry points.',
+      filePaths: truncate(
+        findPaths(filePaths, (filePath) =>
+          /(^|\/)(main|index|app|server|worker)\.(ts|tsx|js|jsx|py|go)$/.test(filePath),
+        ),
+        20,
+      ),
+    })
+
+    const repoShape = inferRepoShape({ topLevelEntries, components, frameworks, filePaths })
+
+    const repoStructure = RepositoryStructureArtifactSchema.parse({
+      schemaVersion: 'repository-structure.v1',
+      extractorVersion: EXTRACTOR_VERSION,
+      generatedAt: nowIsoDateTimeString(),
+      repository: {
+        repositoryKey: params.repositoryKey,
+        repositoryFullName: params.repositoryFullName,
+        defaultBranch: params.defaultBranch,
+        indexBranch: params.indexBranch,
+        commitSha: params.commitSha,
+      },
+      topLevelEntries,
+      repoShape,
+      packageManagers,
+      languages,
+      frameworks,
+      components: truncate(components, 200),
+      entryPoints,
+      routeSignals,
+      apiSignals,
+      databaseSignals,
+      authSignals,
+      jobSignals,
+      envFiles,
+      dockerFiles,
+      ciCdFiles,
+      deploymentFiles,
+      testSignals,
+      lintSignals,
+      formatSignals,
+      typecheckSignals,
+      unknowns: repoShape.kind === 'uncertain' ? ['Repository shape could not be confidently classified.'] : [],
+    })
+
+    const structureSummary = RepositoryStructureSummarySchema.parse({
+      schemaVersion: 'repository-structure-summary.v1',
+      extractorVersion: EXTRACTOR_VERSION,
+      generatedAt: repoStructure.generatedAt,
+      repository: repoStructure.repository,
+      repoShape: repoStructure.repoShape.kind,
+      topLevelEntries: truncate(repoStructure.topLevelEntries, 20),
+      packageManagers: summarizeSignals(repoStructure.packageManagers),
+      languages: summarizeSignals(repoStructure.languages),
+      frameworks: summarizeSignals(repoStructure.frameworks),
+      components: truncate(
+        repoStructure.components.map((component) => ({
+          kind: component.kind,
+          name: component.name,
+          path: component.path,
+        })),
+        40,
+      ),
+      entryPoints: summarizeSignals(repoStructure.entryPoints),
+      routeSignals: summarizeSignals(repoStructure.routeSignals),
+      apiSignals: summarizeSignals(repoStructure.apiSignals),
+      databaseSignals: summarizeSignals(repoStructure.databaseSignals),
+      authSignals: summarizeSignals(repoStructure.authSignals),
+      jobSignals: summarizeSignals(repoStructure.jobSignals),
+      envFiles: summarizeSignals(repoStructure.envFiles),
+      dockerFiles: summarizeSignals(repoStructure.dockerFiles),
+      ciCdFiles: summarizeSignals(repoStructure.ciCdFiles),
+      testSignals: summarizeSignals(repoStructure.testSignals),
+      lintSignals: summarizeSignals(repoStructure.lintSignals),
+      formatSignals: summarizeSignals(repoStructure.formatSignals),
+      typecheckSignals: summarizeSignals(repoStructure.typecheckSignals),
+      deploymentSignals: summarizeSignals(repoStructure.deploymentFiles),
+      unknowns: repoStructure.unknowns.slice(0, 20),
+    })
 
-  const structureSummary = RepositoryStructureSummarySchema.parse({
-    schemaVersion: 'repository-structure-summary.v1',
-    extractorVersion: EXTRACTOR_VERSION,
-    generatedAt: repoStructure.generatedAt,
-    repository: repoStructure.repository,
-    repoShape: repoStructure.repoShape.kind,
-    topLevelEntries: truncate(repoStructure.topLevelEntries, 20),
-    packageManagers: summarizeSignals(repoStructure.packageManagers),
-    languages: summarizeSignals(repoStructure.languages),
-    frameworks: summarizeSignals(repoStructure.frameworks),
-    components: truncate(
-      repoStructure.components.map((component) => ({
-        kind: component.kind,
-        name: component.name,
-        path: component.path,
-      })),
-      40,
-    ),
-    entryPoints: summarizeSignals(repoStructure.entryPoints),
-    routeSignals: summarizeSignals(repoStructure.routeSignals),
-    apiSignals: summarizeSignals(repoStructure.apiSignals),
-    databaseSignals: summarizeSignals(repoStructure.databaseSignals),
-    authSignals: summarizeSignals(repoStructure.authSignals),
-    jobSignals: summarizeSignals(repoStructure.jobSignals),
-    envFiles: summarizeSignals(repoStructure.envFiles),
-    dockerFiles: summarizeSignals(repoStructure.dockerFiles),
-    ciCdFiles: summarizeSignals(repoStructure.ciCdFiles),
-    testSignals: summarizeSignals(repoStructure.testSignals),
-    lintSignals: summarizeSignals(repoStructure.lintSignals),
-    formatSignals: summarizeSignals(repoStructure.formatSignals),
-    typecheckSignals: summarizeSignals(repoStructure.typecheckSignals),
-    deploymentSignals: summarizeSignals(repoStructure.deploymentFiles),
-    unknowns: repoStructure.unknowns.slice(0, 20),
+    return {
+      repoStructure,
+      structureSummary,
+      structureMarkdown: buildStructureMarkdown({ repoStructure, structureSummary }),
+    }
   })
+}
 
-  return {
-    repoStructure,
-    structureSummary,
-    structureMarkdown: buildStructureMarkdown({ repoStructure, structureSummary }),
-  }
+export function extractRepositoryStructure(params: ExtractParams): Promise<ExtractResult> {
+  return Effect.runPromise(extractRepositoryStructureEffect(params))
 }