Bad data and inconsistencies

[iby]

Thanks for the great lib! I noticed some bad UAs coming up with my filters from time to time, which might caused by fake generic UAs being collected and leaking into releases. Since the core value of this lib is to provide realistic UAs it would be good to filter the bad ones out all together.

1. There's literally the following UA with "{userAgent}" value in v1.0.343:

{appName:"Netscape",connection:{downlink:10,effectiveType:"4g",rtt:50},platform:"MacIntel",pluginsLength:3,vendor:"Google Inc.",userAgent:"{userAgent}",viewportHeight:670,viewportWidth:1190,deviceCategory:"desktop",screenHeight:900,screenWidth:1440,weight:3278510796810622e-20}`

2. There're a few wrongfully identified UAs, here's the platform is MacIntel, but should be Linux:

{appName:"Netscape",connection:{downlink:4.5,effectiveType:"4g",rtt:100},platform:"MacIntel",pluginsLength:3,vendor:"Google Inc.",userAgent:'"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/64.0.3282.39 Safari/537.36"',viewportHeight:800,viewportWidth:1670,deviceCategory:"desktop",screenHeight:1050,screenWidth:1680,weight:6634456214284333e-20}

3. iPad with deviceCategory: 'desktop' and platform: 'MacIntel':

{appName:"Netscape",connection:{downlink:8,effectiveType:"4g",rtt:150},platform:"MacIntel",pluginsLength:0,vendor:"Google Inc.",userAgent:"Mozilla/5.0 (iPad; CPU OS 11_0 like Mac OS X) AppleWebKit/604.1.34 (KHTML, like Gecko) Version/11.0 Mobile/15A5341f Safari/604.1",viewportHeight:810,viewportWidth:1410,deviceCategory:"desktop",screenHeight:1050,screenWidth:1680,weight:8770374263814142e-20}

[theArina]

Same. Sometimes i get wrong platform in ua with platform filter set. #15 also relates.

[kgriffs]

Came across this one just today:

"\t\t\t\t<useragent description=\"Android 7.0 - Chrome 53.0 - Nexus 9\" useragent=\"Mozilla/5.0 (Linux; Android 7.0; Nexus 9 Build/NRD90R) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/53.0.2785.124 Safari/537.36\" appcodename=\"\" appname=\"\" appversion=\"\" platform=\"\" vendor=\"\" vendorsub=\"\"/>"

Full record:

  {
    "appName": "Netscape",
    "connection": {
      "downlink": 10,
      "effectiveType": "4g",
      "rtt": 50
    },
    "platform": "Win32",
    "pluginsLength": 5,
    "vendor": "Google Inc.",
    "userAgent": "\t\t\t\t<useragent description=\"Android 7.0 - Chrome 53.0 - Nexus 9\" useragent=\"Mozilla/5.0 (Linux; Android 7.0; Nexus 9 Build/NRD90R) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/53.0.2785.124 Safari/537.36\" appcodename=\"\" appname=\"\" appversion=\"\" platform=\"\" vendor=\"\" vendorsub=\"\"/>",
    "viewportHeight": 680,
    "viewportWidth": 1010,
    "deviceCategory": "desktop",
    "screenHeight": 820,
    "screenWidth": 1024,
    "weight": 0.00024087090072227242
  },

[sangaline]

I've added filtering for the {userAgent} literal and non-ascii characters. Preventing fake data universally is very difficult. My recommendation is to couple it with third-party user-agent validation libraries and apply filtering. See this section of the README for an example: https://github.com/intoli/user-agents?tab=readme-ov-file#filtering-for-modern-browsers