OverScrap

This piece of JS aims to scrap a playoverwatch.com profile page and to produce a JSON digest of the stats found in it for a specific account. When getting a profile's stats, one can choose between regions & game mode, just like on the page itself. The JSON produced is basically a "distraction free" version of what is seen on the page.

Right now, the tool focuses on the Career stats section and can be used for the quickplay & competitive game modes for any region you wish

With OverScrap, you get 2 versions of the same tool :

1. The main lib that basically requests a profile page, parses it, and spits out the almost raw data it was sent (just converts ints & floats on the fly so the resulting JSON is a bit more easy to use, but does not touch anything else).
2. A "server" mode that pops a server and allows for 2 behaviours w/ 2 distinct endpoints :
   • /api : Just the "I don't wanna configure a server myself" version of 1.
   • /graphql : Injects some cool features like the ability to filter the data you want. It also regroups "Hero Specific" stats under the same section rather than having it scattered in different sections.

Originally, I wrote this tool to allow me to track very specific stats (like my K/D ratio) without all the clutter existing applications provide. This tool 1st objective to reflect what is on a player's playoverwatch profile page with as much fidelity as possible, then, to add value by combining the data gathered to come up with composite stats that'll try to add value to the overall service.

Notes :

• I'm figuring this as I go so there are probably issues I haven't yet seen, feel free to get in touch w/ me if you'd like sth added / reworked and I'll see what I can do =)
• As Blizzard, understandably, does not seem to allow cross-origin for playoverwatch.com, this probably means that the requesting will have to be done server-side (which is what the "server" tool is doing, in fact) before handing the parsed data over to a user-facing app.

What does the scrapped data look like?

The general structure of the file is as follows :

{
    "profile":{
        "currentSR":"<current profile's SR>"
    }
    "heroesStats":{
        "<stats broken down by hero + 'All heroes'>"
    }
}

Here's an example for one character in the heroesStats section

{
...
  "Pharah": 
   { "hero Specific": 
      { "Rocket Direct Hits": 12,
        "Rocket Direct Hits - Most in Game": 12,
        "Rocket Direct Hits - Average": 90.38 },
     "Combat": 
      { "Shots Fired": 27,
        "Shots Hit": 21,
        "Damage Done": 2287,
        "Weapon Accuracy": "77%" },
     "Best": 
      { "Damage Done - Most in Life": 1733,
        "Weapon Accuracy - Best in Game": "77%",
        "Damage Done - Most in Game": 2287,
        "Objective Time - Most in Game": "00:14" },
     "Average": 
      { "Deaths - Average": 7.53,
        "Objective Time - Average": "01:47",
        "Damage Done - Average": 17224.04 },
     "Deaths": { "Death": "1", "Environmental Death": 1 },
     "Match Awards": { "Medals - Silver": 0, "Medals - Gold": 0, "Medals": 1 },
     "Game": 
      { "Time Played": "1 minute",
        "Games Played": 0,
        "Objective Time": "00:14" },
     "Miscellaneous": { "Games Lost": 0 } },
...
}

Note that numerical values are parsed now, which removes the hassle of making your own conversions after the fact.

The stats collected also include the All heroes section (as seen on the original page).

How do I use it ?

As a lib in a nodejs app

import OverScrap from 'over-scrap'
 
const overscrap = new OverScrap();
 
overscrap.loadDataFromProfile('Hoshin#2365', 'eu', 'competitive')
.then(playerStats => {
    // from then on, the `playerStats` object contains all heroes data parsed & organized as shown in the previous section
})
 

Using the cli tool

./bin/overscrap-cli.js <Battle tag> [ region [ game mode ] ]
#^this'll output the raw JSON parsed from the playoverwatch profile page 
 

Using the GraphQL server

Right now, OverScrap ships with a small GQL server so that it is possible to fine-tune what you want to request even more.

You can either use the server components on their own, or run it through the bin/overscrap.server.js script.

If ran through the provided binary, the server will start immediately and take any random port available, unless the OVERSCRAP_PORT env variable is set.

Let's say that the server booted up on localhost and took port #4242, you can access the graphiql interface at : localhost:4242/graphql

From then on, you should be able to use graphiql for a simple statsByHeroName query. GraphiQL will provide autocompletion on most items but the "hero specific" ones (which are just raw JSON).

For example, if you're a D.Va player and want to track how good of a "deleter/bully" you are, you could build a query like this :

query {
  statsByHeroName(heroName:"D.Va", mode:"competitive", battleTag:"Hoshin#2365", region:"eu"){
    name
    combat {
        eliminations
        finalBlows
        soloKills
        objectiveKills
        multikills
        environmentalKills
        eliminationsPerLife
    }
    best {
        eliminationsMostinLife
        allDamageDoneMostinLife
        killStreakBest
        eliminationsMostinGame
        finalBlowsMostinGame
        objectiveKillsMostinGame
        soloKillsMostinGame
    }
    average {
        allDamageDoneAvgper10Min
    }
    deaths
  }
}

And then get this kind of result :

{
  "data": {
    "statsByHeroName": {
      "name": "D.Va",
      "combat": {
        "eliminations": 1209,
        "finalBlows": 474,
        "soloKills": 85,
        "objectiveKills": 653,
        "multikills": 17,
        "environmentalKills": 13,
        "eliminationsPerLife": null
      },
      "best": {
        "eliminationsMostInLife": 23,
        "allDamageDoneMostInLife": 11893,
        "killStreakBest": 23,
        "eliminationsMostInGame": 61,
        "finalBlowsMostInGame": 29,
        "objectiveKillsMostInGame": 46,
        "soloKillsMostInGame": 6
      },
      "average": {
        "allDamageDoneAvgPer10Min": 16.89
      },
      "deaths": 280
    }
  }
}

To this effect, all hero specific data has been moved into the "hero specific" category (so for example Reinhardt charge kills don't end up un the "best" category but rather in a specific section where you can get all things purely Reinhardt, D.Va, Pharah ...

For example, getting raw stats for D.Va would get you :

query {
    statsByHeroName(heroName:"D.Va", mode:"competitive", battleTag:"Hoshin#2365", region:"eu"){
        name
        heroSpecific {
            raw # Here it's not possible to get autocompletion, just raw data as JSON
        }
    }
}

Produces:

{
  "data": {
    "statsByHeroName": {
      "name": "D.Va",
      "heroSpecific": {
        "raw": {
          "mechsCalled": 253,
          "mechsCalledMostInGame": 11,
          "damageBlockedMostInGame": 28847,
          "damageBlocked": 569955,
          "mechDeaths": 365,
          "meleeFinalBlowMostInGame": 1
        }
      }
    }
  }
}
 

Whereas doing the same for Reinhardt :

query {
    statsByHeroName(heroName:"Reinhardt", mode:"competitive", battleTag:"Hoshin#2365", region:"eu"){
        name
        heroSpecific {
            raw 
        }
    }
}
 

Produces this :

{
  "data": {
    "statsByHeroName": {
      "name": "Reinhardt",
      "heroSpecific": {
        "raw": {
          "damageBlocked": 188695,
          "damageBlockedMostInGame": 25436,
          "chargeKills": 44,
          "chargeKillsMostInGame": 10,
          "fireStrikeKills": 85,
          "fireStrikeKillsMostInGame": 18,
          "earthshatterKills": 50,
          "earthshatterKillsMostInGame": 11
        }
      }
    }
  }
}

Note about the server : it is configured to use a "sample file" while not explicitly set to work for a production environment. To enable "production" mode, you just have to export/set the OVERSCRAP_ENV env variable and set it to production. The "premade" CLI and server tools you find in ./bin already have that variable overriden so that you hit the live data and not mocks.