fix: walking dag-cbor paths

packages/verified-fetch/src/utils/get-resolved-accept-header.ts

@@ -0,0 +1,42 @@
+import { isExplicitAcceptHeader, isExplicitFormatQuery, isExplicitIpldAcceptRequest } from './is-accept-explicit.js'
+import { queryFormatToAcceptHeader } from './select-output-type.js'
+import type { ParsedUrlStringResults } from './parse-url-string.js'
+import type { ComponentLogger } from '@libp2p/interface'
+
+export interface ResolvedAcceptHeaderOptions {
+  query?: ParsedUrlStringResults['query']
+  headers?: RequestInit['headers']
+  logger?: ComponentLogger
+}
+
+export function getResolvedAcceptHeader ({ query, headers, logger }: ResolvedAcceptHeaderOptions): string | undefined {
+  const log = logger?.forComponent('helia:verified-fetch:get-resolved-accept-header')
+  const requestHeaders = new Headers(headers)
+  const incomingAcceptHeader = requestHeaders.get('accept') ?? undefined
+
+  if (incomingAcceptHeader != null) {
+    log?.('incoming accept header "%s"', incomingAcceptHeader)
+  }
+
+  if (!isExplicitIpldAcceptRequest({ query, headers: requestHeaders })) {
+    log?.('no explicit IPLD content-type requested, returning incoming accept header %s', incomingAcceptHeader)
+    return incomingAcceptHeader
+  }
+
+  const queryFormatMapping = queryFormatToAcceptHeader(query?.format)
+
+  if (query?.format != null) {
+    log?.('incoming query format "%s", mapped to %s', query.format, queryFormatMapping)
+  }
+
+  let acceptHeader = incomingAcceptHeader
+  // if the incomingAcceptHeader is autogenerated by the requesting client (browser/curl/fetch/etc) then we may need to override it if query.format is specified
+  if (!isExplicitAcceptHeader(requestHeaders) && isExplicitFormatQuery(query)) {
+    log?.('accept header not recognized, but query format provided, setting accept header to %s', queryFormatMapping)
+    acceptHeader = queryFormatMapping
+  }
+
+  log?.('resolved accept header to "%s"', acceptHeader)
+
+  return acceptHeader
+}

packages/verified-fetch/src/utils/is-accept-explicit.ts

@@ -0,0 +1,32 @@
+import { FORMAT_TO_MIME_TYPE } from './select-output-type.js'
+import type { ParsedUrlStringResults } from './parse-url-string.js'
+
+export interface IsAcceptExplicitOptions {
+
+  query?: ParsedUrlStringResults['query']
+  headers: Headers
+}
+
+export function isExplicitAcceptHeader (headers: Headers): boolean {
+  const incomingAcceptHeader = headers.get('accept')
+  if (incomingAcceptHeader != null && Object.values(FORMAT_TO_MIME_TYPE).includes(incomingAcceptHeader)) {
+    return true
+  }
+  return false
+}
+
+export function isExplicitFormatQuery (query?: ParsedUrlStringResults['query']): boolean {
+  const formatQuery = query?.format
+  if (formatQuery != null && Object.keys(FORMAT_TO_MIME_TYPE).includes(formatQuery)) {
+    return true
+  }
+  return false
+}
+
+/**
+ * The user can provide an explicit `accept` header in the request headers or a `format` query parameter in the URL.
+ * If either of these are provided, this function returns true.
+ */
+export function isExplicitIpldAcceptRequest ({ query, headers }: IsAcceptExplicitOptions): boolean {
+  return isExplicitAcceptHeader(headers) || isExplicitFormatQuery(query)
+}

packages/verified-fetch/src/utils/responses.ts

@@ -85,6 +85,19 @@ export function notAcceptableResponse (url: string, body?: SupportedBodyTypes, i
   return response
 }
 
+export function notFoundResponse (url: string, body?: SupportedBodyTypes, init?: ResponseInit): Response {
+  const response = new Response(body, {
+    ...(init ?? {}),
+    status: 404,
+    statusText: 'Not Found'
+  })
+
+  setType(response, 'basic')
+  setUrl(response, url)
+
+  return response
+}
+
 /**
  * if body is an Error, it will be converted to a string containing the error message.
  */

packages/verified-fetch/src/utils/select-output-type.ts

@@ -55,6 +55,7 @@ const CID_TYPE_MAP: Record<number, string[]> = {
     'application/octet-stream',
     'application/vnd.ipld.raw',
     'application/vnd.ipfs.ipns-record',
+    'application/vnd.ipld.dag-json',
     'application/vnd.ipld.car',
     'application/x-tar'
   ]
@@ -145,7 +146,7 @@ function parseQFactor (str?: string): number {
   return factor
 }
 
-const FORMAT_TO_MIME_TYPE: Record<RequestFormatShorthand, string> = {
+export const FORMAT_TO_MIME_TYPE: Record<RequestFormatShorthand, string> = {
   raw: 'application/vnd.ipld.raw',
   car: 'application/vnd.ipld.car',
   'dag-json': 'application/vnd.ipld.dag-json',

packages/verified-fetch/src/utils/walk-path.ts

@@ -1,4 +1,5 @@
-import { walkPath as exporterWalk, type ExporterOptions, type ReadableStorage, type UnixFSEntry } from 'ipfs-unixfs-exporter'
+import { CodeError } from '@libp2p/interface'
+import { walkPath as exporterWalk, type ExporterOptions, type ReadableStorage, type ObjectNode, type UnixFSEntry } from 'ipfs-unixfs-exporter'
 import type { CID } from 'multiformats/cid'
 
 export interface PathWalkerOptions extends ExporterOptions {
@@ -24,11 +25,15 @@
   }
 
   if (terminalElement == null) {
-    throw new Error('No terminal element found')
+    throw new CodeError('No terminal element found', 'NO_TERMINAL_ELEMENT')
   }
 
   return {
     ipfsRoots,
     terminalElement
   }
 }
+
+export function objectNodeGuard (node: UnixFSEntry): node is ObjectNode {
+  return node.type === 'object'
+}

packages/verified-fetch/src/verified-fetch.ts

@@ -20,19 +20,20 @@
 import { dagCborToSafeJSON } from './utils/dag-cbor-to-safe-json.js'
 import { getContentDispositionFilename } from './utils/get-content-disposition-filename.js'
 import { getETag } from './utils/get-e-tag.js'
+import { getResolvedAcceptHeader } from './utils/get-resolved-accept-header.js'
 import { getStreamFromAsyncIterable } from './utils/get-stream-from-async-iterable.js'
 import { tarStream } from './utils/get-tar-stream.js'
 import { parseResource } from './utils/parse-resource.js'
 import { setCacheControlHeader } from './utils/response-headers.js'
-import { badRequestResponse, movedPermanentlyResponse, notAcceptableResponse, notSupportedResponse, okResponse, badRangeResponse, okRangeResponse, badGatewayResponse } from './utils/responses.js'
-import { selectOutputType, queryFormatToAcceptHeader } from './utils/select-output-type.js'
-import { walkPath } from './utils/walk-path.js'
+import { badRequestResponse, movedPermanentlyResponse, notAcceptableResponse, notSupportedResponse, okResponse, badRangeResponse, okRangeResponse, badGatewayResponse, notFoundResponse } from './utils/responses.js'
+import { selectOutputType } from './utils/select-output-type.js'
+import { objectNodeGuard, walkPath } from './utils/walk-path.js'
 import type { CIDDetail, ContentTypeParser, Resource, VerifiedFetchInit as VerifiedFetchOptions } from './index.js'
 import type { RequestFormatShorthand } from './types.js'
 import type { ParsedUrlStringResults } from './utils/parse-url-string'
 import type { Helia } from '@helia/interface'
 import type { DNSResolver } from '@multiformats/dns/resolvers'
-import type { UnixFSEntry } from 'ipfs-unixfs-exporter'
+import type { ObjectNode, UnixFSEntry } from 'ipfs-unixfs-exporter'
 import type { CID } from 'multiformats/cid'
 
 interface VerifiedFetchComponents {
@@ -93,6 +94,7 @@
  * skipped and set to these values.
  */
 const RAW_HEADERS = [
+  'application/vnd.ipld.dag-json',
   'application/vnd.ipld.raw',
   'application/octet-stream'
 ]
@@ -103,8 +105,9 @@
  * type. This avoids the user from receiving something different when they
  * signal that they want to `Accept` a specific mime type.
  */
-function getOverridenRawContentType (headers?: HeadersInit): string | undefined {
-  const acceptHeader = new Headers(headers).get('accept') ?? ''
+function getOverridenRawContentType ({ headers, accept }: { headers?: HeadersInit, accept?: string }): string | undefined {
+  // accept has already been resolved by getResolvedAcceptHeader, if we have it, use it.
+  const acceptHeader = accept ?? new Headers(headers).get('accept') ?? ''
 
   // e.g. "Accept: text/html, application/xhtml+xml, application/xml;q=0.9, image/webp, */*;q=0.8"
   const acceptHeaders = acceptHeader.split(',')
@@ -233,8 +236,33 @@
 
   private async handleDagCbor ({ resource, cid, path, accept, options }: FetchHandlerFunctionArg): Promise<Response> {
     this.log.trace('fetching %c/%s', cid, path)
+    let terminalElement: ObjectNode | undefined
+    let ipfsRoots: CID[] | undefined
+
+    // need to walk path, if it exists, to get the terminal element
+    try {
+      const pathDetails = await walkPath(this.helia.blockstore, `${cid.toString()}/${path}`, options)
+      ipfsRoots = pathDetails.ipfsRoots
+      const potentialTerminalElement = pathDetails.terminalElement
+      if (potentialTerminalElement == null) {
+        return notFoundResponse(resource.toString())
+      }
+      if (objectNodeGuard(potentialTerminalElement)) {
+        terminalElement = potentialTerminalElement
+      }
+    } catch (err: any) {
+      if (options?.signal?.aborted === true) {
+        throw new AbortError('signal aborted by user')
+      }
+      if (['ERR_NO_PROP', 'NO_TERMINAL_ELEMENT'].includes(err.code)) {
+        return notFoundResponse(resource.toString())
+      }
+
+      this.log.error('error walking path %s', path, err)
+      return badGatewayResponse(resource.toString(), 'Error walking path')
+    }
+    const block = terminalElement?.node ?? await this.helia.blockstore.get(cid, options)
 
-    const block = await this.helia.blockstore.get(cid, options)
     let body: string | Uint8Array
 
     if (accept === 'application/octet-stream' || accept === 'application/vnd.ipld.dag-cbor' || accept === 'application/cbor') {
@@ -274,9 +302,14 @@
 
     response.headers.set('content-type', accept)
 
+    if (ipfsRoots != null) {
+      response.headers.set('X-Ipfs-Roots', ipfsRoots.map(cid => cid.toV1().toString()).join(',')) // https://specs.ipfs.tech/http-gateways/path-gateway/#x-ipfs-roots-response-header
+    }
+
     return response
   }
 
+  // eslint-disable-next-line complexity
   private async handleDagPb ({ cid, path, resource, options }: FetchHandlerFunctionArg): Promise<Response> {
     let terminalElement: UnixFSEntry | undefined
     let ipfsRoots: CID[] | undefined
@@ -291,6 +324,9 @@
       if (options?.signal?.aborted === true) {
         throw new AbortError('signal aborted by user')
       }
+      if (['ERR_NO_PROP', 'NO_TERMINAL_ELEMENT'].includes(err.code)) {
+        return notFoundResponse(resource.toString())
+      }
       this.log.error('error walking path %s', path, err)
 
       return badGatewayResponse(resource.toString(), 'Error walking path')
@@ -385,7 +421,7 @@
     }
   }
 
-  private async handleRaw ({ resource, cid, path, options }: FetchHandlerFunctionArg): Promise<Response> {
+  private async handleRaw ({ resource, cid, path, options, accept }: FetchHandlerFunctionArg): Promise<Response> {
     const byteRangeContext = new ByteRangeContext(this.helia.logger, options?.headers)
     const result = await this.helia.blockstore.get(cid, options)
     byteRangeContext.setBody(result)
@@ -396,7 +432,7 @@
     // if the user has specified an `Accept` header that corresponds to a raw
     // type, honour that header, so for example they don't request
     // `application/vnd.ipld.raw` but get `application/octet-stream`
-    const overriddenContentType = getOverridenRawContentType(options?.headers)
+    const overriddenContentType = getOverridenRawContentType({ headers: options?.headers, accept })
     if (overriddenContentType != null) {
       response.headers.set('content-type', overriddenContentType)
     } else {
@@ -499,20 +535,8 @@
 
     options?.onProgress?.(new CustomProgressEvent<CIDDetail>('verified-fetch:request:resolve', { cid, path }))
 
-    const requestHeaders = new Headers(options?.headers)
-    const incomingAcceptHeader = requestHeaders.get('accept')
-
-    if (incomingAcceptHeader != null) {
-      this.log('incoming accept header "%s"', incomingAcceptHeader)
-    }
-
-    const queryFormatMapping = queryFormatToAcceptHeader(query.format)
-
-    if (query.format != null) {
-      this.log('incoming query format "%s", mapped to %s', query.format, queryFormatMapping)
-    }
+    const acceptHeader = getResolvedAcceptHeader({ query, headers: options?.headers, logger: this.helia.logger })
 
-    const acceptHeader = incomingAcceptHeader ?? queryFormatMapping
     const accept = selectOutputType(cid, acceptHeader)
     this.log('output type %s', accept)
 
@@ -523,7 +547,7 @@
     let response: Response
     let reqFormat: RequestFormatShorthand | undefined
 
-    const handlerArgs = { resource: resource.toString(), cid, path, accept, options }
+    const handlerArgs: FetchHandlerFunctionArg = { resource: resource.toString(), cid, path, accept, options }
 
     if (accept === 'application/vnd.ipfs.ipns-record') {
       // the user requested a raw IPNS record

packages/verified-fetch/test/verified-fetch.spec.ts

@@ -710,4 +710,94 @@ describe('@helia/verifed-fetch', () => {
       expect(output).to.deep.equal(obj)
     })
   })
+
+  describe('?format', () => {
+    let helia: Helia
+    let verifiedFetch: VerifiedFetch
+    let contentTypeParser: Sinon.SinonStub
+
+    beforeEach(async () => {
+      contentTypeParser = Sinon.stub()
+      helia = await createHelia()
+      verifiedFetch = new VerifiedFetch({
+        helia
+      }, {
+        contentTypeParser
+      })
+    })
+
+    afterEach(async () => {
+      await stop(helia, verifiedFetch)
+    })
+
+    it('cbor?format=dag-json should be able to override curl/browser default accept header when query parameter is provided', async () => {
+      const obj = {
+        hello: 'world'
+      }
+      const c = dagCbor(helia)
+      const cid = await c.add(obj)
+
+      const resp = await verifiedFetch.fetch(`http://example.com/ipfs/${cid}?format=dag-json`, {
+        headers: {
+          // see https://github.com/ipfs/helia-verified-fetch/issues/35
+          // accept: '*/*'
+        }
+      })
+      expect(resp.headers.get('content-type')).to.equal('application/vnd.ipld.dag-json')
+      const data = ipldDagJson.decode(await resp.arrayBuffer())
+      expect(data).to.deep.equal(obj)
+    })
+
+    it('raw?format=dag-json should be able to override curl/browser default accept header when query parameter is provided', async () => {
+      const finalRootFileContent = uint8ArrayFromString(JSON.stringify({
+        hello: 'world'
+      }))
+      const cid = CID.createV1(raw.code, await sha256.digest(finalRootFileContent))
+      await helia.blockstore.put(cid, finalRootFileContent)
+
+      const resp = await verifiedFetch.fetch(`http://example.com/ipfs/${cid}?format=dag-json`, {
+        headers: {
+          // see https://github.com/ipfs/helia-verified-fetch/issues/35
+          accept: '*/*'
+        }
+      })
+      expect(resp).to.be.ok()
+      expect(resp.status).to.equal(200)
+      expect(resp.statusText).to.equal('OK')
+      const data = await resp.arrayBuffer()
+      expect(resp.headers.get('content-type')).to.equal('application/vnd.ipld.dag-json')
+      expect(new Uint8Array(data)).to.equalBytes(finalRootFileContent)
+    })
+  })
+
+  describe('404 paths', () => {
+    let helia: Helia
+    let verifiedFetch: VerifiedFetch
+    let contentTypeParser: Sinon.SinonStub
+
+    beforeEach(async () => {
+      contentTypeParser = Sinon.stub()
+      helia = await createHelia()
+      verifiedFetch = new VerifiedFetch({
+        helia
+      }, {
+        contentTypeParser
+      })
+    })
+
+    afterEach(async () => {
+      await stop(helia, verifiedFetch)
+    })
+
+    it('returns a 404 when walking dag-cbor for non-existent path', async () => {
+      const obj = {
+        hello: 'world'
+      }
+      const c = dagCbor(helia)
+      const cid = await c.add(obj)
+
+      const resp = await verifiedFetch.fetch(`http://example.com/ipfs/${cid}/foo/i-do-not-exist`)
+      expect(resp.status).to.equal(404)
+    })
+  })
 })