Improve author connector

Handling twitter names

Author: Kikobeats

__snapshots__/index.js.snap-shot

@@ -648,7 +648,7 @@ exports['the-verge 1'] = {
 }
 
 exports['twitter 1'] = {
-  "author": "K I K O B E a T S ⚡️",
+  "author": "Kikobeats",
   "date": "2017-06-28T21:01:00.000Z",
   "description": "“Experimenting with Clearbit API + Apple TV 3D Parallax https://t.co/Qsm163k4mJ https://t.co/5bcuqoEyAa”",
   "image": "https://pbs.twimg.com/profile_images/876748424690925568/14xPNAvw_400x400.jpg",
@@ -658,6 +658,17 @@ exports['twitter 1'] = {
   "url": "https://twitter.com/Kikobeats/status/880139124791029763"
 }
 
+exports['twitter 2'] = {
+  "author": "K4rliky",
+  "date": "2017-11-25T19:04:00.000Z",
+  "description": "“Lo mejor de @codemotion_es #codemotionMadrid es estar con la gente que quieres 😍@ladyCircus”",
+  "image": "https://pbs.twimg.com/media/DPfzNMGXUAESoL3.jpg:large",
+  "logo": "https://abs.twimg.com/icons/apple-touch-icon-192x192.png",
+  "publisher": "Twitter",
+  "title": "Asume la realidad: no hay cuchara. on Twitter",
+  "url": "https://twitter.com/k4rliky/status/934482867480121345"
+}
+
 exports['usa-today 1'] = {
   "author": "Editors, Usa Today",
   "date": "2016-05-24T16:29:19.230Z",

src/get-data/rules/author.js

@@ -5,7 +5,7 @@ const {isString, flow} = require('lodash')
 const toTitle = require('to-title-case')
 const {isUrl} = require('../util')
 
-const REGEX_BY = /^[\s\n]*by[\s\n]*/im
+const REGEX_BY = /^[\s\n]*by|@[\s\n]*/im
 const REGEX_STRICT = /^\S+\s+\S+/
 
 const removeBy = value => value.replace(REGEX_BY, '')
@@ -43,11 +43,16 @@ const wrap = rule => $ => {
 
 const strict = rule => $ => {
   const value = rule($)
-
   if (!REGEX_STRICT.test(value)) return
   return value
 }
 
+const getFirst = ($, collection) =>
+  collection
+    .filter((i, el) => $(el).text().trim())
+    .first()
+    .text()
+
 /**
  * Rules.
  */
@@ -63,21 +68,10 @@ module.exports = [
   wrap($ => $('meta[property="book:author"]').attr('content')),
   strict(wrap($ => $('a[class*="author"]').first().text())),
   strict(wrap($ => $('[class*="author"] a').first().text())),
-
-  strict(wrap($ => $('a[href*="/author/"]')
-    .filter((i, el) => $(el).text().trim())
-    .first()
-    .text()
-  )),
-
+  strict(wrap($ => getFirst($, $('a[href*="/author/"]')))),
+  wrap($ => $('a[class*="screenname"]').first().text()),
   strict(wrap($ => $('[class*="author"]').first().text())),
   strict(wrap($ => $('[class*="byline"]').first().text())),
-
-  wrap($ => $('.fullname')
-    .filter((i, el) => $(el).text().trim())
-    .first()
-    .text()
-  ),
-
+  wrap($ => getFirst($, $('.fullname'))),
   wrap($ => $('[class*="user-info"]').text())
 ]

test/web/twitter/index.js renamed to test/web/twitter-gif/index.js

@@ -0,0 +1,18 @@
+'use strict'
+
+const snapshot = require('snap-shot')
+const {promisify} = require('util')
+const {resolve} = require('path')
+
+const fs = require('fs')
+
+const getMetaData = require('../../..')
+const readFile = promisify(fs.readFile)
+
+const url = 'https://twitter.com/k4rliky/status/934482867480121345'
+
+it('twitter', async () => {
+  const html = await readFile(resolve(__dirname, 'input.html'))
+  const metadata = await getMetaData({html, url})
+  snapshot(metadata)
+})