Any developers who are interested in piggy-backing off of TappedOut data, I've got a page detailing the API endpoints here: http://tappedout.github.io/
Recently, I've added User API Keys (on your settings page) which you can use to enable sharing of your decks with an external app.
Today I've also added the ability to export all of your collections (lists, decks, cubes, binder, etc) via the developer API.
Also, please, never ever ever scrape my site. If you have a request, please ask. =)
You can use my parser however you like. I abandoned the approach because only the card art was in the files, not the formatted card (but that may not matter to you).
.MSE format -- incredibly finicky and tedious, so I can't really guarantee that parser will work every time. I have emailed Twan about this and he agreed the MSE format could have been cleaner, but he doesn't have the time to get back to it. Thusly why I put it on Github. A perfect importer for the MSE format is possible but the spec is super nasty. Its basically an XML file that wraps a bunch of arbitrarily delimited JSON-like code. But its not anywhere near being JSON either. That said I think the project is really marvelous, so if some hero wanted to get their hands dirty and force that sucker to use JSON or something, that would be great. Short of that the templating language it provides isn't awful.
A perfect spec in my opinion would be simplest is best. I would just use a zipped file containing a JSON encoded spec and a zipped image file containing ideally the cart art and rendered versions (or at least an option within the program to export containing rendered images). The JSON spec would contain a header for application data and then a card blob for card attributes.
https://github.com/subsume/magic-set-editor I put the MSE code online in case anyone was interested in patching up MSE to have a more friendly native format. Talked to the main dev and he was interested but just didn't have the time himself.
I look forward to watching the progress of this thread and I certainly would be glad to adopt such a parser and integrate import/export on TappedOut.
All you gotta do is pass in a JSON list of card names and you will get a JSON response of formats its valid in. If you also pass in a format it will tell you what cards aren't in that format, if any.
Yeago here. I run TappedOut. I created this nifty web-service for developers like me who work with decks and find it a pain to detect and validate the format of a particular deck.
The service runs off the same engine I use to detect formats @ TappedOut. There's some bugs, esp. with modern, but I will be working them out =) Enjoy!
I caught up with yeago in the chat room there and I mentioned this too him. So you never know
Also, Yeago,
Would it be possible to have options to change pack size x packs drafted? I know my cube likes doing a 5 packs by 9 card setup in addition to the 3x 15
someone send me a sample export thru contact form =)
yes, i am adding new draft 'shapes' soon (like 5x9). also rarity shapes, too.
I dunno why you boneheads like keeping your cubes in excel sheets or text lists when you could link to TappedOut and get ninja-like charts, etc. =).
I'll have to check tonight. Its gonna be hard of course, because if we the humans pick based on archtypes that certain can be tough to do, because they might pick some awesome cards but end up with a mash of great cards but not a real great deck.
If yer interested in seeing results, the card pages will contain two rows: 'Cube average' and 'Draft average'. Its still munching away, but it will be done... eventually.
Just throw up a thread or reply here when you areready, if I see it I'm in for whenever.
And yeaGo...
On the card individual pages. If you don't see it on the left it hasn't been aggregated yet. Look for more tonight or tomorrow. They should all be parsed then.
These kinda things should be child's play for most modern cpus (assuming at least a dualcore +) . I would think anyway.
heh. well I have a quadcore but a rough estimate isn't looking good.
12k cards in the database. 12 million picks to consider. but for each pick, one must do between 0-7 additional pick-checks to find how many humans were passed.
so uhh, no not child's play. =) Maybe young adult's play.
But I am not sure this approach will work because, while it will shape things up a little bit for drafts with no data, this same approach causes a problem in the context of solo drafts with data it will tend to blur the important difference between the first 8 cards (which is a lot). blah blah blah.
i'm referring to this
bot passes you the card in terms of its ranking? at least for now, until the AI gets better? or is that not feasible?
Ah-ha. Yes that's another algorithm that I could use. Not the number picked but the number of human passes. Oh boy does that add to the CPU cycles, though.
why factor in bot's picks at all? if i get passed sol ring, it's the first time i've seen the card, and i take it, that should count sol ring being picked highly. when three bots pass sol ring, it shouldn't count for anything because they're bots. i'm not a mathematician or a programmer, but there must be some way to factor in human picks and throw out or at least adjust somewhat to the influence of the bots.
I'm not... read my example again. =). I illustrated how using *only* human picks in a draft will skew the picks unless there's already good data in the system.
Any developers who are interested in piggy-backing off of TappedOut data, I've got a page detailing the API endpoints here: http://tappedout.github.io/
Recently, I've added User API Keys (on your settings page) which you can use to enable sharing of your decks with an external app.
Today I've also added the ability to export all of your collections (lists, decks, cubes, binder, etc) via the developer API.
Also, please, never ever ever scrape my site. If you have a request, please ask. =)
Yeago -- HNIC, TappedOut
.MSE format -- incredibly finicky and tedious, so I can't really guarantee that parser will work every time. I have emailed Twan about this and he agreed the MSE format could have been cleaner, but he doesn't have the time to get back to it. Thusly why I put it on Github. A perfect importer for the MSE format is possible but the spec is super nasty. Its basically an XML file that wraps a bunch of arbitrarily delimited JSON-like code. But its not anywhere near being JSON either. That said I think the project is really marvelous, so if some hero wanted to get their hands dirty and force that sucker to use JSON or something, that would be great. Short of that the templating language it provides isn't awful.
A perfect spec in my opinion would be simplest is best. I would just use a zipped file containing a JSON encoded spec and a zipped image file containing ideally the cart art and rendered versions (or at least an option within the program to export containing rendered images). The JSON spec would contain a header for application data and then a card blob for card attributes.
I'm extremely interested in this project. I recently had my own failings with slurping up MSE files for TappedOut's Community Cards section.
I have written some code on this matter (Python too, heh)
https://github.com/subsume/tappedout-mse-export-template TappedOut's export template for MSE files authored by a kind gentleman who goes by http404error.
https://github.com/subsume/mse-parse Python parser for MSE files. Doesn't grab images, sadly, just data.
https://github.com/subsume/magic-set-editor I put the MSE code online in case anyone was interested in patching up MSE to have a more friendly native format. Talked to the main dev and he was interested but just didn't have the time himself.
I look forward to watching the progress of this thread and I certainly would be glad to adopt such a parser and integrate import/export on TappedOut.
The service runs off the same engine I use to detect formats @ TappedOut. There's some bugs, esp. with modern, but I will be working them out =) Enjoy!
http://validator.tappedout.net
someone send me a sample export thru contact form =)
yes, i am adding new draft 'shapes' soon (like 5x9). also rarity shapes, too.
I dunno why you boneheads like keeping your cubes in excel sheets or text lists when you could link to TappedOut and get ninja-like charts, etc. =).
Gonna shift my focus to making it quicker now. If there's anything urgent that I missed please let me know. Pleasure working with you all!
edit:
like whoa. lightning is here.
poor baby. =)
On the card individual pages. If you don't see it on the left it hasn't been aggregated yet. Look for more tonight or tomorrow. They should all be parsed then.
heh. well I have a quadcore but a rough estimate isn't looking good.
12k cards in the database. 12 million picks to consider. but for each pick, one must do between 0-7 additional pick-checks to find how many humans were passed.
so uhh, no not child's play. =) Maybe young adult's play.
But I am not sure this approach will work because, while it will shape things up a little bit for drafts with no data, this same approach causes a problem in the context of solo drafts with data it will tend to blur the important difference between the first 8 cards (which is a lot). blah blah blah.
Ah-ha. Yes that's another algorithm that I could use. Not the number picked but the number of human passes. Oh boy does that add to the CPU cycles, though.
....will think about it. Good suggestion! =D
I'm not... read my example again. =). I illustrated how using *only* human picks in a draft will skew the picks unless there's already good data in the system.