Ensure video is an absolute video url

Author: Kikobeats

packages/metascraper-video/__snapshots__/index.js.snap-shot

@@ -1,6 +1,6 @@
 exports['video src 1'] = {
   "image": "https://cdn.vox-cdn.com/thumbor/AtQQMyWrexi6-Xyk73jv6nqTO7s=/0x5:1247x658/fit-in/1200x630/cdn.vox-cdn.com/uploads/chorus_asset/file/10079811/Screen_Shot_2018_01_22_at_3.27.50_PM.png",
-  "video": "https://cdn.vox-cdn.com/thumbor/4l0C-7uGFtTfc6lWibo1ITiE2YU=/0x0:1280x720/320x213/filters:focal(538x258:742x462):gifv():no_upscale()/cdn.vox-cdn.com/uploads/chorus_image/image/58416873/2018_01_22_14_19_55.0.gif",
+  "video": null,
   "author": "Rachel Becker",
   "date": "2018-01-22T23:38:17.000Z",
   "description": "The zombies are only released on the weekends, the developers promise",
@@ -26,7 +26,7 @@ exports['source src 1'] = {
 
 exports['og:video 1'] = {
   "image": "https://pbs.twimg.com/tweet_video_thumb/DUQODu8VMAASlkj.jpg",
-  "video": "https://twitter.com/i/videos/955905369242513414?embed_source=facebook",
+  "video": null,
   "author": "_developit",
   "date": "2018-02-18T12:00:00.000Z",
   "description": "“📣 Announcing... Karmatic!\n🦑 Easy automatic (headless) browser testing\n🛠 No config needed, auto-detects webpack settings\n🤹‍♀️ Only one dependency to juggle\n🏎 Powered by Karma, Webpack, Jasmine & Puppeteer\nhttps://t.co/nguAcGaWQP https://t.co/OgxikFBdQ8”",

packages/metascraper-video/index.js

@@ -1,6 +1,10 @@
 'use strict'
 
 const { getUrl, isUrl } = require('@metascraper/helpers')
+const videoExtensions = require('video-extensions')
+const path = require('path')
+
+const isVideoUrl = url => videoExtensions.includes(path.extname(url).substring(1))
 
 /**
  * Wrap a rule with validation and formatting logic.
@@ -9,11 +13,17 @@ const { getUrl, isUrl } = require('@metascraper/helpers')
  * @return {Function} wrapped
  */
 
-const wrap = rule => ({ htmlDom, url }) => {
+const wrapUrl = (url, value) => isUrl(value)
+const wrapVideoUrl = (url, value) => wrapUrl(url, value) && isVideoUrl(value)
+
+const createWrapper = validator => rule => ({ htmlDom, url }) => {
   const value = rule(htmlDom)
-  return isUrl(value) && getUrl(url, value)
+  return validator(url, value) && getUrl(url, value)
 }
 
+const wrap = createWrapper(wrapUrl)
+const wrapVideo = createWrapper(wrapVideoUrl)
+
 /**
  * Rules.
  */
@@ -23,11 +33,11 @@ module.exports = () => ({
     wrap($ => $('video').attr('poster'))
   ],
   video: [
-    wrap($ => $('meta[property="og:video:secure_url"]').attr('content')),
-    wrap($ => $('meta[property="og:video:url"]').attr('content')),
-    wrap($ => $('meta[property="og:video"]').attr('content')),
-    wrap($ => $('meta[property="twitter:player:stream"]').attr('content')),
-    wrap($ => $('video').attr('src')),
-    wrap($ => $('source').attr('src'))
+    wrapVideo($ => $('meta[property="og:video:secure_url"]').attr('content')),
+    wrapVideo($ => $('meta[property="og:video:url"]').attr('content')),
+    wrapVideo($ => $('meta[property="og:video"]').attr('content')),
+    wrapVideo($ => $('meta[property="twitter:player:stream"]').attr('content')),
+    wrapVideo($ => $('video').attr('src')),
+    wrapVideo($ => $('source').attr('src'))
   ]
 })

packages/metascraper-video/package.json

@@ -16,7 +16,8 @@
     "url": "https://github.com/microlinkhq/metascraper/issues"
   },
   "dependencies": {
-    "@metascraper/helpers": "^3.6.0"
+    "@metascraper/helpers": "^3.6.0",
+    "video-extensions": "~1.1.0"
   },
   "devDependencies": {
     "lodash": "latest",