@kubb/adapter-oas 5.0.0-beta.3 → 5.0.0-beta.31

package/src/refs.ts

@@ -1,6 +1,8 @@
 import { isReference } from './guards.ts'
 import type { Document } from './types.ts'
 
+const _refCache = new WeakMap<Document, Map<string, unknown>>()
+
 /**
  * Resolves a local JSON pointer reference from a document.
  *
@@ -18,11 +20,19 @@ export function resolveRef<T = unknown>(document: Document, $ref: string): T | n
   if ($ref === '') {
     return null
   }
-  if ($ref.startsWith('#')) {
-    $ref = globalThis.decodeURIComponent($ref.substring(1))
-  } else {
-    return null
+  if (!$ref.startsWith('#')) return null
+  $ref = globalThis.decodeURIComponent($ref.substring(1))
+
+  let docCache = _refCache.get(document)
+  if (!docCache) {
+    docCache = new Map()
+    _refCache.set(document, docCache)
   }
+
+  if (docCache.has($ref)) {
+    return docCache.get($ref) as T
+  }
+
   const current = $ref
     .split('/')
     .filter(Boolean)
@@ -31,6 +41,8 @@ export function resolveRef<T = unknown>(document: Document, $ref: string): T | n
   if (!current) {
     throw new Error(`Could not find a definition for ${origRef}.`)
   }
+
+  docCache.set($ref, current)
   return current as T
 }
 

package/src/resolvers.ts

@@ -86,7 +86,7 @@ export type OperationsOptions = {
  * ```
  */
 export function getParameters(document: Document, operation: Operation): Array<ParameterObject> {
-  const resolveParams = (params: unknown[]): Array<ParameterObject> =>
+  const resolveParams = (params: Array<unknown>): Array<ParameterObject> =>
     params.map((p) => dereferenceWithRef(document, p)).filter((p): p is ParameterObject => !!p && typeof p === 'object' && 'in' in p && 'name' in p)
 
   const operationParams = resolveParams(operation.schema?.parameters || [])
@@ -108,7 +108,7 @@ export function getParameters(document: Document, operation: Operation): Array<P
   return Array.from(paramMap.values())
 }
 
-function getResponseBody(responseBody: boolean | ResponseObject, contentType?: string): MediaTypeObject | false | [string, MediaTypeObject, ...string[]] {
+function getResponseBody(responseBody: boolean | ResponseObject, contentType?: string): MediaTypeObject | false | [string, MediaTypeObject, ...Array<string>] {
   if (!responseBody) return false
   if (isReference(responseBody)) return false
 
@@ -260,7 +260,7 @@ function hasStructuralKeywords(fragment: SchemaObject): boolean {
 export function flattenSchema(schema: SchemaObject | null): SchemaObject | null {
   if (!schema?.allOf || schema.allOf.length === 0) return schema ?? null
 
-  const allOfFragments = schema.allOf as SchemaObject[]
+  const allOfFragments = schema.allOf as Array<SchemaObject>
   if (allOfFragments.some((item) => isRef(item))) return schema
   if (allOfFragments.some(hasStructuralKeywords)) return schema
 
@@ -305,27 +305,25 @@ export function extractSchemaFromContent(content: Record<string, unknown> | unde
 /**
  * Walks a schema tree and collects the names of all `#/components/schemas/<name>` `$ref`s.
  */
-function collectRefs(schema: unknown, refs = new Set<string>()): Set<string> {
+function* collectRefs(schema: unknown): Generator<string, void, undefined> {
   if (Array.isArray(schema)) {
-    for (const item of schema) collectRefs(item, refs)
-    return refs
+    for (const item of schema) yield* collectRefs(item)
+    return
   }
 
   if (schema && typeof schema === 'object') {
     for (const key in schema) {
       const value = (schema as Record<string, unknown>)[key]
-      if (key === '$ref' && typeof value === 'string') {
-        if (value.startsWith(SCHEMA_REF_PREFIX)) {
-          const name = value.slice(SCHEMA_REF_PREFIX.length)
-          if (name) refs.add(name)
-        }
-      } else {
-        collectRefs(value, refs)
+      if (!(key === '$ref' && typeof value === 'string')) {
+        yield* collectRefs(value)
+        continue
+      }
+      if (value.startsWith(SCHEMA_REF_PREFIX)) {
+        const name = value.slice(SCHEMA_REF_PREFIX.length)
+        if (name) yield name
       }
     }
   }
-
-  return refs
 }
 
 /**
@@ -341,13 +339,13 @@ function collectRefs(schema: unknown, refs = new Set<string>()): Set<string> {
  * ```
  */
 export function sortSchemas(schemas: Record<string, SchemaObject>): Record<string, SchemaObject> {
-  const deps = new Map<string, string[]>()
+  const deps = new Map<string, Array<string>>()
 
   for (const [name, schema] of Object.entries(schemas)) {
-    deps.set(name, Array.from(collectRefs(schema)))
+    deps.set(name, [...new Set(collectRefs(schema))])
   }
 
-  const sorted: string[] = []
+  const sorted: Array<string> = []
   const visited = new Set<string>()
 
   function visit(name: string, stack: Set<string>) {
@@ -404,7 +402,7 @@ function resolveSchemaRef(document: Document, schema: SchemaObject): SchemaObjec
 export function getSchemas(document: Document, { contentType }: GetSchemasOptions): GetSchemasResult {
   const components = document.components
 
-  const candidates: SchemaWithMetadata[] = [
+  const candidates: Array<SchemaWithMetadata> = [
     ...Object.entries((components?.schemas as Record<string, SchemaObject>) ?? {}).map(([name, schema]) => ({
       schema: resolveSchemaRef(document, schema),
       source: 'schemas' as const,
@@ -426,7 +424,7 @@ export function getSchemas(document: Document, { contentType }: GetSchemasOption
     ),
   ]
 
-  const normalizedNames = new Map<string, SchemaWithMetadata[]>()
+  const normalizedNames = new Map<string, Array<SchemaWithMetadata>>()
   for (const item of candidates) {
     const key = pascalCase(item.originalName)
     const bucket = normalizedNames.get(key) ?? []
@@ -514,6 +512,7 @@ export function buildSchemaNode(schema: SchemaObject, name: string | null | unde
     writeOnly: schema.writeOnly,
     default: defaultValue,
     example: schema.example,
+    format: schema.format,
   } as const
 }
 
@@ -530,7 +529,7 @@ export function buildSchemaNode(schema: SchemaObject, name: string | null | unde
  * // ['application/json', 'multipart/form-data']
  * ```
  */
-export function getRequestBodyContentTypes(document: Document, operation: Operation): string[] {
+export function getRequestBodyContentTypes(document: Document, operation: Operation): Array<string> {
   if (operation.schema.requestBody) {
     operation.schema.requestBody = dereferenceWithRef(document, operation.schema.requestBody)
   }
@@ -542,3 +541,33 @@ export function getRequestBodyContentTypes(document: Document, operation: Operat
   // Do not bail out on isReference — the content is already present on the merged object.
   return body.content ? Object.keys(body.content) : []
 }
+
+/**
+ * Returns all response content type keys for an operation at a given status code.
+ *
+ * Response `$ref`s are resolved in-place first — the same mutation `getResponseSchema` performs —
+ * so the returned list reflects the available content types even for referenced responses.
+ *
+ * @example
+ * ```ts
+ * getResponseBodyContentTypes(document, operation, 200)
+ * // ['application/json', 'application/xml']
+ * ```
+ */
+export function getResponseBodyContentTypes(document: Document, operation: Operation, statusCode: string | number): Array<string> {
+  if (operation.schema.responses) {
+    const responses = operation.schema.responses
+    for (const key in responses) {
+      const schema = responses[key]
+      if (schema && isReference(schema)) {
+        responses[key] = resolveRef<any>(document, schema.$ref)
+      }
+    }
+  }
+
+  const responseObj = operation.getResponseByStatusCode(statusCode)
+  if (!responseObj || typeof responseObj !== 'object' || isReference(responseObj)) return []
+
+  const body = responseObj as { content?: Record<string, unknown> }
+  return body.content ? Object.keys(body.content) : []
+}

package/src/stream.ts

@@ -0,0 +1,276 @@
+import { ast } from '@kubb/core'
+import type BaseOas from 'oas'
+import { SCHEMA_REF_PREFIX } from './constants.ts'
+import { buildDiscriminatorChildMap, patchDiscriminatorNode } from './discriminator.ts'
+import type { SchemaParser } from './parser.ts'
+import { resolveServerUrl } from './resolvers.ts'
+import type { DiscriminatorTarget } from './discriminator.ts'
+import type { AdapterOas, Document, SchemaObject } from './types.ts'
+
+export type PreScanResult = {
+  refAliasMap: Map<string, ast.SchemaNode>
+  enumNames: Array<string>
+  circularNames: Array<string>
+  discriminatorChildMap: Map<string, DiscriminatorTarget> | null
+  dedupePlan: ast.DedupePlan | null
+}
+
+/**
+ * Builds the deduplication plan from the already-parsed top-level schema nodes plus a
+ * single extra parse pass over operations (so duplicates in request/response bodies are seen).
+ *
+ * Only enums and objects are candidates, and object shapes that reference a circular schema are
+ * rejected to avoid hoisting recursive structures. Inline shapes reuse their context-derived
+ * name (collision-resolved against existing component names); shapes without a name stay inline.
+ */
+function createDedupePlan({
+  schemaNodes,
+  operationNodes,
+  schemaNames,
+  circularNames,
+}: {
+  schemaNodes: Array<ast.SchemaNode>
+  operationNodes: Array<ast.OperationNode>
+  schemaNames: Array<string>
+  circularNames: Array<string>
+}): ast.DedupePlan {
+  const circularSchemas = new Set(circularNames)
+  const usedNames = new Set(schemaNames)
+
+  return ast.buildDedupePlan([...schemaNodes, ...operationNodes], {
+    isCandidate: (node) => {
+      if (node.type === 'enum') return true
+      if (node.type !== 'object') return false
+      // Skip object shapes that are part of a circular chain — hoisting them would break the cycle.
+      if (node.name && circularSchemas.has(node.name)) return false
+      return !ast.containsCircularRef(node, { circularSchemas })
+    },
+    nameFor: (node) => {
+      const base = node.name
+      if (!base) return null
+
+      let name = base
+      let counter = 2
+      while (usedNames.has(name)) {
+        name = `${base}${counter++}`
+      }
+      usedNames.add(name)
+      return name
+    },
+    refFor: (name) => `${SCHEMA_REF_PREFIX}${name}`,
+  })
+}
+
+/**
+ * Reads the server URL from the document's `servers` array at `serverIndex`,
+ * interpolating any `serverVariables` into the URL template.
+ *
+ * Returns `null` when `serverIndex` is omitted or out of range.
+ *
+ * @example Resolve the first server
+ * `resolveBaseUrl({ document, serverIndex: 0 })`
+ *
+ * @example Override a path variable
+ * `resolveBaseUrl({ document, serverIndex: 0, serverVariables: { version: 'v2' } })`
+ */
+export function resolveBaseUrl({
+  document,
+  serverIndex,
+  serverVariables,
+}: {
+  document: Document
+  serverIndex?: number
+  serverVariables?: Record<string, string>
+}): string | null {
+  const server = serverIndex !== undefined ? document.servers?.at(serverIndex) : undefined
+  return server?.url ? resolveServerUrl(server, serverVariables) : null
+}
+
+/**
+ * Parses every schema once to build the lookup structures that streaming needs upfront.
+ *
+ * Three things happen in this single pass:
+ * - `refAliasMap` records schemas that are pure `$ref` aliases so the streaming pass can inline them.
+ * - `enumNames` collects the names of every enum schema so plugins skip re-scanning the stream.
+ * - `circularNames` runs cycle detection, which requires all nodes in memory simultaneously.
+ *   The `allNodes` array is local and drops out of scope as soon as this function returns.
+ *
+ * After this call, only `refAliasMap` and `discriminatorChildMap` stay alive in the adapter closure.
+ * Both are proportional to the number of aliases or discriminator parents, not total schema count.
+ *
+ * Each schema is parsed again during the streaming pass — this is intentional.
+ * Holding the parsed nodes in memory here would defeat the streaming memory benefit.
+ *
+ * @example
+ * ```ts
+ * const { refAliasMap, enumNames, circularNames } = preScan({
+ *   schemas,
+ *   parseSchema,
+ *   parserOptions,
+ *   discriminator: 'strict',
+ * })
+ * ```
+ */
+export function preScan({
+  schemas,
+  parseSchema,
+  parseOperation,
+  baseOas,
+  parserOptions,
+  discriminator,
+  dedupe,
+}: {
+  schemas: Record<string, SchemaObject>
+  parseSchema: (entry: { schema: SchemaObject; name: string }, options: ast.ParserOptions) => ast.SchemaNode
+  parseOperation: SchemaParser['parseOperation']
+  baseOas: BaseOas
+  parserOptions: ast.ParserOptions
+  discriminator: AdapterOas['options']['discriminator']
+  dedupe: boolean
+}): PreScanResult {
+  const allNodes: Array<ast.SchemaNode> = []
+  const refAliasMap = new Map<string, ast.SchemaNode>()
+  const enumNames: Array<string> = []
+  const discriminatorParentNodes: Array<ast.SchemaNode> = []
+
+  for (const [name, schema] of Object.entries(schemas)) {
+    const node = parseSchema({ schema, name }, parserOptions)
+    allNodes.push(node)
+    if (node.type === 'ref' && node.name && node.name !== name) {
+      refAliasMap.set(name, node)
+    }
+    if (ast.narrowSchema(node, ast.schemaTypes.enum) && node.name) {
+      enumNames.push(node.name)
+    }
+    if (discriminator === 'inherit' && (schema.oneOf ?? schema.anyOf) && schema.discriminator?.propertyName) {
+      discriminatorParentNodes.push(node)
+    }
+  }
+
+  const circularNames = [...ast.findCircularSchemas(allNodes)]
+  const discriminatorChildMap = discriminatorParentNodes.length > 0 ? buildDiscriminatorChildMap(discriminatorParentNodes) : null
+
+  let dedupePlan: ast.DedupePlan | null = null
+  if (dedupe) {
+    // One extra parse pass over operations so duplicates in request/response bodies are seen.
+    // Reuses the already-parsed `allNodes` for schemas — no second schema parse.
+    const operationNodes: Array<ast.OperationNode> = []
+    for (const methods of Object.values(baseOas.getPaths())) {
+      for (const operation of Object.values(methods)) {
+        if (!operation) continue
+        const operationNode = parseOperation(parserOptions, operation)
+        if (operationNode) operationNodes.push(operationNode)
+      }
+    }
+
+    dedupePlan = createDedupePlan({ schemaNodes: allNodes, operationNodes, schemaNames: Object.keys(schemas), circularNames })
+
+    for (const definition of dedupePlan.hoisted) {
+      if (definition.type === 'enum' && definition.name) enumNames.push(definition.name)
+    }
+  }
+
+  return { refAliasMap, enumNames, circularNames, discriminatorChildMap, dedupePlan }
+}
+
+/**
+ * Creates a lazy `InputStreamNode` from already-resolved adapter state.
+ *
+ * The schema and operation iterables each start a fresh parse pass on every
+ * `[Symbol.asyncIterator]()` call. This lets multiple plugins consume the same
+ * stream object independently without sharing a cursor or holding all nodes in memory.
+ *
+ * Ref aliases in `refAliasMap` are inlined during iteration: an alias entry is replaced
+ * with its target's parsed node (but keeps the alias name) so plugins never receive bare `ref` nodes.
+ *
+ * @example
+ * ```ts
+ * const streamNode = createInputStream({ schemas, parseSchema, parseOperation, baseOas, parserOptions, refAliasMap, discriminatorChildMap, meta })
+ * for await (const schema of streamNode.schemas) {
+ *   // each call to for-await restarts from the first schema
+ * }
+ * ```
+ */
+export function createInputStream({
+  schemas,
+  parseSchema,
+  parseOperation,
+  baseOas,
+  parserOptions,
+  refAliasMap,
+  discriminatorChildMap,
+  dedupePlan,
+  meta,
+}: {
+  schemas: Record<string, SchemaObject>
+  parseSchema: SchemaParser['parseSchema']
+  parseOperation: SchemaParser['parseOperation']
+  baseOas: BaseOas
+  parserOptions: ast.ParserOptions
+  refAliasMap: Map<string, ast.SchemaNode>
+  discriminatorChildMap: Map<string, DiscriminatorTarget> | null
+  dedupePlan: ast.DedupePlan | null
+  meta: ast.InputMeta
+}): ast.InputStreamNode {
+  // Rewrites a top-level schema against the dedupe plan: a structurally identical sibling
+  // becomes a `ref` alias to the canonical one (keeping its own name); otherwise nested
+  // duplicates are collapsed while the schema's own root is preserved.
+  const rewriteTopLevelSchema = (node: ast.SchemaNode): ast.SchemaNode => {
+    if (!dedupePlan) return node
+
+    const canonical = dedupePlan.canonicalBySignature.get(ast.schemaSignature(node))
+    if (canonical && canonical.name !== node.name) {
+      return ast.createSchema({
+        type: 'ref',
+        name: node.name ?? null,
+        ref: canonical.ref,
+        description: node.description,
+        deprecated: node.deprecated,
+      })
+    }
+
+    return ast.applyDedupe(node, dedupePlan.canonicalBySignature, true)
+  }
+
+  const schemasIterable: AsyncIterable<ast.SchemaNode> = {
+    [Symbol.asyncIterator]() {
+      return (async function* () {
+        // Hoisted canonical definitions are emitted first so the schema list owns the shared shapes.
+        if (dedupePlan) {
+          for (const definition of dedupePlan.hoisted) yield definition
+        }
+
+        for (const [name, schema] of Object.entries(schemas)) {
+          // Inline ref aliases: replace the alias entry with its target's parsed node
+          // (keeping the alias name). Skip the first parse entirely for alias entries
+          // since that result is never used.
+          const alias = refAliasMap.get(name)
+          if (alias?.name && schemas[alias.name]) {
+            yield rewriteTopLevelSchema({ ...parseSchema({ schema: schemas[alias.name]!, name: alias.name }, parserOptions), name })
+            continue
+          }
+
+          const parsed = parseSchema({ schema, name }, parserOptions)
+          const node = discriminatorChildMap?.get(name) ? patchDiscriminatorNode(parsed, discriminatorChildMap.get(name)!) : parsed
+          yield rewriteTopLevelSchema(node)
+        }
+      })()
+    },
+  }
+
+  const operationsIterable: AsyncIterable<ast.OperationNode> = {
+    [Symbol.asyncIterator]() {
+      return (async function* () {
+        for (const methods of Object.values(baseOas.getPaths())) {
+          for (const operation of Object.values(methods)) {
+            if (!operation) continue
+            const node = parseOperation(parserOptions, operation)
+            if (node) yield dedupePlan ? ast.applyDedupe(node, dedupePlan.canonicalBySignature) : node
+          }
+        }
+      })()
+    },
+  }
+
+  return ast.createStreamInput(schemasIterable, operationsIterable, meta)
+}

package/src/types.ts

@@ -123,8 +123,9 @@ export type ResponseObject = OASResponseObject
 export type MediaTypeObject = OASMediaTypeObject
 
 /**
- * Configuration options for the OpenAPI adapter.
- * Controls spec validation, content-type selection, and server URL resolution.
+ * Configuration options for the OpenAPI adapter. Controls spec validation,
+ * content-type selection, server URL resolution, and how types are derived
+ * from the spec.
  *
  * @example
  * ```ts
@@ -138,23 +139,28 @@ export type MediaTypeObject = OASMediaTypeObject
  */
 export type AdapterOasOptions = {
   /**
-   * Validate the OpenAPI spec before parsing.
+   * Validate the OpenAPI spec with `@readme/openapi-parser` before parsing.
+   * Set to `false` only when you have a known-invalid spec you still want to
+   * generate from.
+   *
    * @default true
    */
   validate?: boolean
   /**
-   * Preferred content-type used when extracting request/response schemas.
-   * Defaults to the first valid JSON media type found in the spec.
+   * Preferred media type when an operation defines several. Defaults to the
+   * first JSON-compatible media type found in the spec.
    */
   contentType?: ContentType
   /**
-   * Index into `oas.api.servers` for computing `baseURL`.
-   * `0` → first server, `1` → second server. Omit to leave `baseURL` undefined.
+   * Index into the `servers` array from your OpenAPI spec. Used to compute the
+   * base URL for plugins that need it. Most projects pick `0` for the primary
+   * server. Omit to leave `baseURL` undefined.
    */
   serverIndex?: number
   /**
    * Override values for `{variable}` placeholders in the selected server URL.
-   * Only used when `serverIndex` is set.
+   * Only used when `serverIndex` is set. Variables you do not provide use
+   * their `default` value from the spec.
    *
    * @example
    * ```ts
@@ -165,12 +171,28 @@ export type AdapterOasOptions = {
    */
   serverVariables?: Record<string, string>
   /**
-   * How the discriminator field is interpreted.
-   * - `'strict'`  — uses `oneOf` schemas as written in the spec.
-   * - `'inherit'` — propagates discriminator values into child schemas from `discriminator.mapping`.
+   * How the `discriminator` field on `oneOf`/`anyOf` schemas is interpreted.
+   * - `'strict'` — child schemas stay exactly as written; the discriminator
+   *   narrows types at the call site but child shapes are not modified.
+   * - `'inherit'` — Kubb propagates the discriminator property as a literal
+   *   value into each child schema, so each branch's discriminator field is
+   *   precisely typed.
+   *
    * @default 'strict'
    */
   discriminator?: 'strict' | 'inherit'
+  /**
+   * Collapse structurally identical schemas and enums into a single shared definition.
+   *
+   * Duplicated inline shapes (especially enums repeated across many properties) are hoisted
+   * into one named schema; every other occurrence — and any structurally identical top-level
+   * component — becomes a `ref` to it. Equality is shape-only: documentation such as
+   * `description` and `example` is ignored. Enabled by default; set to `false` to keep every
+   * occurrence inline and produce byte-for-byte identical output to earlier versions.
+   *
+   * @default true
+   */
+  dedupe?: boolean
 } & Partial<ast.ParserOptions>
 
 /**
@@ -182,6 +204,7 @@ export type AdapterOasResolvedOptions = {
   serverIndex: AdapterOasOptions['serverIndex']
   serverVariables: AdapterOasOptions['serverVariables']
   discriminator: NonNullable<AdapterOasOptions['discriminator']>
+  dedupe: NonNullable<AdapterOasOptions['dedupe']>
   dateType: NonNullable<AdapterOasOptions['dateType']>
   integerType: NonNullable<AdapterOasOptions['integerType']>
   unknownType: NonNullable<AdapterOasOptions['unknownType']>