Fix a couple incorrect link detections ('e.g.' and 'foo.jpg') (close #13)

__tests__/string-utils.ts

@@ -31,6 +31,11 @@ describe('extractEntities', () => {
     'start middle end.com/foo/bar?baz=bux#hash',
     'newline1.com\nnewline2.com',
     'not.. a..url ..here',
+    'e.g.',
+    'something-cool.jpg',
+    'website.com.jpg',
+    'e.g./foo',
+    'website.com.jpg/foo',
   ]
   interface Output {
     type: string
@@ -80,6 +85,11 @@ describe('extractEntities', () => {
       {type: 'link', value: 'newline2.com', noScheme: true},
     ],
     [],
+    [],
+    [],
+    [],
+    [],
+    [],
   ]
   it('correctly handles a set of text inputs', () => {
     for (let i = 0; i < inputs.length; i++) {
@@ -145,6 +155,12 @@ describe('detectLinkables', () => {
     'start middle end.com/foo/bar?baz=bux#hash',
     'newline1.com\nnewline2.com',
     'not.. a..url ..here',
+    'e.g.',
+    'e.g. real.com fake.notreal',
+    'something-cool.jpg',
+    'website.com.jpg',
+    'e.g./foo',
+    'website.com.jpg/foo',
   ]
   const outputs = [
     ['no linkable'],
@@ -171,6 +187,12 @@ describe('detectLinkables', () => {
     ['start middle ', {link: 'end.com/foo/bar?baz=bux#hash'}],
     [{link: 'newline1.com'}, '\n', {link: 'newline2.com'}],
     ['not.. a..url ..here'],
+    ['e.g.'],
+    ['e.g. ', {link: 'real.com'}, ' fake.notreal'],
+    ['something-cool.jpg'],
+    ['website.com.jpg'],
+    ['e.g./foo'],
+    ['website.com.jpg/foo'],
   ]
   it('correctly handles a set of text inputs', () => {
     for (let i = 0; i < inputs.length; i++) {

package.json

@@ -46,7 +46,8 @@
     "react-native-svg": "^12.4.0",
     "react-native-tab-view": "^3.3.0",
     "react-native-url-polyfill": "^1.3.0",
-    "react-native-web": "^0.17.7"
+    "react-native-web": "^0.17.7",
+    "tlds": "^1.234.0"
   },
   "devDependencies": {
     "@babel/core": "^7.12.9",
@@ -74,7 +75,9 @@
   },
   "jest": {
     "preset": "react-native",
-    "setupFiles": ["./jest.js"],
+    "setupFiles": [
+      "./jest.js"
+    ],
     "moduleFileExtensions": [
       "ts",
       "tsx",

src/lib/strings.ts

@@ -1,6 +1,7 @@
 import {AtUri} from '../third-party/uri'
 import {Entity} from '../third-party/api/src/client/types/app/bsky/feed/post'
 import {PROD_SERVICE} from '../state'
+import TLDs from 'tlds'
 
 export const MAX_DISPLAY_NAME = 64
 export const MAX_DESCRIPTION = 256
@@ -57,6 +58,14 @@ export function ago(date: number | string | Date): string {
   }
 }
 
+export function isValidDomain(str: string): boolean {
+  return !!TLDs.find(tld => {
+    let i = str.lastIndexOf(tld)
+    if (i === -1) return false
+    return str.charAt(i - 1) === '.' && i === str.length - tld.length
+  })
+}
+
 export function extractEntities(
   text: string,
   knownHandles?: Set<string>,
@@ -85,10 +94,14 @@ export function extractEntities(
   {
     // links
     const re =
-      /(^|\s)((https?:\/\/[\S]+)|([a-z][a-z0-9]*(\.[a-z0-9]+)+[\S]*))(\b)/dg
+      /(^|\s)((https?:\/\/[\S]+)|((?<domain>[a-z][a-z0-9]*(\.[a-z0-9]+)+)[\S]*))(\b)/dg
     while ((match = re.exec(text))) {
       let value = match[2]
       if (!value.startsWith('http')) {
+        const domain = match.groups?.domain
+        if (!domain || !isValidDomain(domain)) {
+          continue
+        }
         value = `https://${value}`
       }
       ents.push({
@@ -110,7 +123,7 @@ interface DetectedLink {
 type DetectedLinkable = string | DetectedLink
 export function detectLinkables(text: string): DetectedLinkable[] {
   const re =
-    /((^|\s)@[a-z0-9\.-]*)|((^|\s)https?:\/\/[\S]+)|((^|\s)[a-z][a-z0-9]*(\.[a-z0-9]+)+[\S]*)/gi
+    /((^|\s)@[a-z0-9\.-]*)|((^|\s)https?:\/\/[\S]+)|((^|\s)(?<domain>[a-z][a-z0-9]*(\.[a-z0-9]+)+)[\S]*)/gi
   const segments = []
   let match
   let start = 0
@@ -118,6 +131,10 @@ export function detectLinkables(text: string): DetectedLinkable[] {
     let matchIndex = match.index
     let matchValue = match[0]
 
+    if (match.groups?.domain && !isValidDomain(match.groups?.domain)) {
+      continue
+    }
+
     if (/\s/.test(matchValue)) {
       // HACK
       // skip the starting space

yarn.lock

@@ -11708,6 +11708,11 @@ thunky@^1.0.2:
   resolved "https://registry.yarnpkg.com/thunky/-/thunky-1.1.0.tgz#5abaf714a9405db0504732bbccd2cedd9ef9537d"
   integrity sha512-eHY7nBftgThBqOyHGVN+l8gF0BucP09fMo0oO/Lb0w1OF80dJv+lDVpXG60WMQvkcxAkNybKsrEIE3ZtKGmPrA==
 
+tlds@^1.234.0:
+  version "1.234.0"
+  resolved "https://registry.yarnpkg.com/tlds/-/tlds-1.234.0.tgz#f61fe73f6e85c51f8503181f47dcfbd18c6910db"
+  integrity sha512-TNDfeyDIC+oroH44bMbWC+Jn/2qNrfRvDK2EXt1icOXYG5NMqoRyUosADrukfb4D8lJ3S1waaBWSvQro0erdng==
+
 tmpl@1.0.5:
   version "1.0.5"
   resolved "https://registry.yarnpkg.com/tmpl/-/tmpl-1.0.5.tgz#8683e0b902bb9c20c4f726e3c0b69f36518c07cc"