feat: support for bzz.link cids when downloading files (#350)

* feat: detect and extract bzz.link cids into hash when downloading files

* test: add quite thorough testsuite

src/constants.ts

@@ -1,3 +1,4 @@
 export const META_FILE_NAME = '.swarmgatewaymeta.json'
 export const PREVIEW_FILE_NAME = '.swarmgatewaypreview.jpeg'
 export const PREVIEW_DIMENSIONS = { maxWidth: 250, maxHeight: 175 }
+export const BZZ_LINK_DOMAIN = process.env.REACT_APP_BZZ_LINK_DOMAIN || 'bzz.link'

src/pages/files/Download.tsx

@@ -8,7 +8,7 @@ import { History } from '../../components/History'
 import { Context, defaultUploadOrigin } from '../../providers/File'
 import { Context as SettingsContext } from '../../providers/Settings'
 import { ROUTES } from '../../routes'
-import { extractSwarmHash } from '../../utils'
+import { recognizeSwarmHash } from '../../utils'
 import { determineHistoryName, HISTORY_KEYS, putHistory } from '../../utils/local-storage'
 import { FileNavigation } from './FileNavigation'
 
@@ -71,20 +71,6 @@ export function Download(): ReactElement {
     }
   }
 
-  function recognizeSwarmHash(value: string) {
-    if (value.length < 64) {
-      return value
-    }
-
-    const hash = extractSwarmHash(value)
-
-    if (hash) {
-      return hash
-    }
-
-    return value
-  }
-
   return (
     <>
       <FileNavigation active="DOWNLOAD" />

src/utils/index.test.ts

@@ -0,0 +1,138 @@
+import { extractSwarmHash, extractSwarmCid, recognizeSwarmHash } from './index'
+
+interface TestObject {
+  input: string
+  expectedOutput: string | undefined
+}
+
+const correctHashes: TestObject[] = [
+  // non-encrypted
+  {
+    input: 'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+    expectedOutput: 'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input: 'http://gateway.org/bzz/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+    expectedOutput: 'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input: 'https://gateway.org/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+    expectedOutput: 'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input: 'http://gateway.org/bzz/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f/',
+    expectedOutput: 'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input: 'https://gateway.org/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f/',
+    expectedOutput: 'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  // encrypted
+  {
+    input:
+      'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+    expectedOutput:
+      'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input:
+      'http://gateway.org/bzz/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+    expectedOutput:
+      'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input:
+      'https://gateway.org/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+    expectedOutput:
+      'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input:
+      'http://gateway.org/bzz/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f/',
+    expectedOutput:
+      'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+  {
+    input:
+      'https://gateway.org/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f/',
+    expectedOutput:
+      'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fb7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3f',
+  },
+]
+
+const wrongHashes: string[] = [
+  // one character too long
+  'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fa',
+  'http://gateway.org/bzz/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fa',
+  'https://gateway.org/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fa',
+  'http://gateway.org/bzz/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fa/',
+  'https://gateway.org/b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d3fa/',
+
+  // a bit shorter
+  'b7e53783114e4555384d7fd7154eb8c2e3f7c749c176dcb8f4015b08161b3d',
+]
+
+describe('extractSwarmHash', () => {
+  test('should correctly extract hash', () => {
+    correctHashes.forEach(({ input, expectedOutput }) => {
+      const hash = extractSwarmHash(input)
+      expect(hash).toBe(expectedOutput)
+    })
+  })
+  test('should not extract hash from incorrect inputs', () => {
+    wrongHashes.forEach(input => {
+      const hash = extractSwarmHash(input)
+      expect(hash).toBe(undefined)
+    })
+  })
+})
+
+const correctCids: TestObject[] = [
+  {
+    input: 'https://bah5acgza5afd34lfvo7sowxfjahj4ujeduxgg2ge5u3zo4kcjlzjzi23fhka.bzz.link',
+    expectedOutput: 'e80a3df165abbf275ae5480e9e51241d2e6368c4ed379771424af29ca35b29d4',
+  },
+  {
+    input: 'https://bah5acgza2lzgtqfztvn3xtnzhv7qvbmblljd7bi52l5jiueretcad55vookq.bzz.link',
+    expectedOutput: 'd2f269c0b99d5bbbcdb93d7f0a85815ad23f851dd2fa94509124c401f7b57395',
+  },
+]
+const wrongCids: string[] = [
+  'https://bah5acgza5afd34lfvo7sowxfjahj4ujeduxgg2ge5u3zo4kcjlzjzi23fhka.another.domain',
+  'http://bah5acgza2lzgtqfztvn3xtnzhv7qvbmblljd7bi52l5jiueretcad55vookq.bzz.link',
+  'https://not_cid.bzz.link',
+  'https://bah5acgza5afd34lfvo7sowxfjahj4ujeduxgg2ge5u3zo4kcjlzjzi23fhka.subdomain.bzz.link',
+  'https://bah5acgza5afd34lfvo7sowxfjahj4ujeduxgg2ge5u3zo4kcjlzjzi23fhka.subdomain.bzz.link',
+  'https://bah5acgza2lzgtqfztvn3xtnzhv7qvbmblljd7bi52l5jiueretcad55vook.bzz.link',
+  'https://aah5acgza2lzgtqfztvn3xtnzhv7qvbmblljd7bi52l5jiueretcad55vookq.bzz.link',
+]
+
+describe('extractSwarmCid', () => {
+  test('should correctly extract hash', () => {
+    correctCids.forEach(({ input, expectedOutput }) => {
+      const hash = extractSwarmCid(input)
+      expect(hash).toBe(expectedOutput)
+    })
+  })
+  test('should not extract cid from incorrect urls', () => {
+    wrongCids.forEach(url => {
+      const hash = extractSwarmCid(url)
+      expect(hash).toBe(undefined)
+    })
+  })
+})
+
+describe('recognizeSwarmHash', () => {
+  test('should correctly extract hash', () => {
+    ;[...correctHashes, ...correctCids].forEach(({ input, expectedOutput }) => {
+      const hash = recognizeSwarmHash(input)
+      expect(hash).toBe(expectedOutput)
+    })
+  })
+  test('should not extract hash from incorrect inputs but instead return them', () => {
+    ;[...wrongHashes, ...wrongCids].forEach(url => {
+      const hash = recognizeSwarmHash(url)
+      expect(hash).toBe(url)
+    })
+  })
+})

src/utils/index.ts

@@ -1,5 +1,7 @@
 import { BigNumber } from 'bignumber.js'
 import { Token } from '../models/Token'
+import { decodeCid } from '@ethersphere/swarm-cid'
+import { BZZ_LINK_DOMAIN } from '../constants'
 
 /**
  * Test if value is an integer
@@ -108,10 +110,41 @@ export function makeRetriablePromise<T>(fn: () => Promise<T>, maxRetries = 3, de
   })
 }
 
-export function extractSwarmHash(string: string): string | null {
+// Matches exactly 64 or 128 caracters alphanumeric characters that are surrounded by non-alpha num characters
-  const matches = string.match(/[a-fA-F0-9]{64,128}/)
+const regexpMatchHash = /(?:^|[^a-f0-9]+)([a-f0-9]{64}|[a-f0-9]{128})(?:$|[^a-f0-9]+)/i
 
-  return (matches && matches[0]) || null
+export function extractSwarmHash(string: string): string | undefined {
+  const matches = string.match(regexpMatchHash)
+
+  return (matches && matches[1]) || undefined
+}
+
+// Matches the CID from bzz-link subdomain
+const regexpMatchCID = new RegExp(`https://(bah5acgza[a-z0-9]{52})\\.${BZZ_LINK_DOMAIN}`, 'i')
+
+export function extractSwarmCid(s: string): string | undefined {
+  const matches = s.match(regexpMatchCID)
+
+  if (!matches || !matches[1]) {
+    return
+  }
+
+  const cid = matches[1]
+  try {
+    const decodeResult = decodeCid(cid)
+
+    if (!decodeResult.type) {
+      return
+    }
+
+    return decodeResult.reference
+  } catch (e) {
+    return
+  }
+}
+
+export function recognizeSwarmHash(value: string): string {
+  return extractSwarmHash(value) || extractSwarmCid(value) || value
 }
 
 export function uuidV4(): string {