Guess the Move - Githubissues

steinitz-zz commented 9 years ago

Hi mliebelt,

I thought you might be interested to know that I've used your fine work as a foundation for a Guess The Move prototype. I integrated PgnViewerJS into Meteor (a story in itself), added some enhancements and deployed it (just for fun) to chesstraining.meteor.com

Note that it doesn't work on the first move (e4), after that it's ok. I'll fix that soon.

Thanks for making public your excellent work. I'd be interested in your comments. You're welcome to the modified code.

All the best,

Steve

mliebelt commented 9 years ago

Nice, my first user! So keep me informed who your work is going. Are your changes available anywhere? You have not forked my repo, so I don't have the chance to see your changes. I will look at it as soon I see your ideas in code.

And hopefully I will finish the work (edit mode is not finished yet) this year, had too much work lately in the company (and too many hours spent on chess.com and lichess.org).

steinitz-zz commented 9 years ago

Ah, good, your first user. I like it.

I am happy to provide the changes directly to you, if you are interested. You may want to use them. However, I've made many, many formatting changes, so I'm not sure if a fork or pull request would be appropriate. Specifically, I can read and understand code better when it's like this:

function ()
{
    if (condition)
    {
        console.log ("1");
    }
    else
    {
        console.log ("2");
    }
}

but many people find that annoying, or worse -- I saw someone propose putting people to death who format like I do :)

Feel free to send me a private email: steve at dtctc dot com

steinitz-zz commented 9 years ago

I was able to debug the Guess the Move issue with the first move (the solution seems slightly drastic :) ). The fixed version is now running at chesstraining.meteor.com . Again, I'd be delighted to provide the entire Meteor project directly to you. Feel free to contact me at steve at dtctc dot com If you are interested. In the meantime, you can download the modified pgnvjs.js here: https://gist.github.com/steinitz/7ddcbbd8fdd9f2f436b3

The project is too much changed (for Meteor) for a pull request. I suppose I could make a special project for the purpose of a pull request, but perhaps it would be better if you were to look over my changes first. My non-standard (but beautiful :) ) formatting, the fact that Meteor has problems with 'use strict' and other issues may make a pull request inappropriate. I welcome your thoughts.

In any case, please accept my sincere appreciation for your fine work.

steinitz-zz commented 9 years ago

Hi,

I have noticed that the pgn of, say Bxb4 is displayed as Bb4. I'd like to have the 'x'. Did you intentionally remove the x or is it an artifact of chess.js or pgn.js?

Thanks for any hints.

Cheers,

Steve

ps. I have a much changed version of pgnvjs.js for you if you are interested.

steinitz-zz commented 9 years ago

Hi,

I made two simple mods to pgn.js to add 'x' for captures. Here are the details.

In eachMoveVariation, save the flags from chess.js:

...
var fen = game.fen();
move.fen = fen;
move.notation.flags = pgn_move.flags; // SJS - save this to know to draw a capture symbol, 'x'.  

$.each (move.variations,
...

Then modify san() as follows:

var san = function (notation)
{
    if (typeof notation.row == 'undefined')
    {
        return notation.notation; // move like O-O and O-O-O
    }
    var fig     = $.t(notation.fig);
    var capture = notation.flags == game.FLAGS.CAPTURE ? 'x' : '';   // SJS - add 'x' for captures 
    var disc    = notation.disc ? notation.disc : "";
    var check   = notation.check ? notation.check : "";
    var prom    = notation.promotion ? notation.promotion : "";
    return fig + disc + capture + notation.col + notation.row + prom + check;  // SJS - capture
};

Cheers

mliebelt commented 9 years ago

See issue #5 which I have added to separate this issue from the missing capture symbol. I have taken your input, and searched for the "best" place to insert the right code then. Will make some tests, and then publish my changes soon.

khoramshahi commented 4 years ago

Hi @mliebelt,

Thanks for sharing such great stuff. Are you still working on this issue? I would like to integrate such behavior into Anki (user guessing the next moves which is noted in pgn and computer replying with the next move in the pgn).

There are some working solutions for Anki, but rather old or buggy.

mliebelt commented 4 years ago

@khoramshahi Do you have a reference to Anki, I never heard of it. I cannot say when I will have time to work on that issue, but the requirements should be easy:

No variations, only one line.
Don't show any moves at the beginning.
Allow the user to make a move, and deny it, if it is the wrong one.
If the move is correct, play automatically the move of the other side, and wait for the next move.
Until last move given is found
(optional) Give hints, if wished by user.
(optional) Allow to solve (partially).

Am I missing anything here?

Bebul commented 3 years ago

There is a great tool to generate Lichess style PUZZLES from PGN file, which works really well. https://github.com/linrock/chess-puzzle-maker The puzzle output from this is the PGN file starting with FEN position and the first move is basically a BLUNDER, which refutation should be found as a puzzle.

So, all your @mliebelt assumptions, like "No variations, only one line" holds, but moreover:

board should be initiated at the postition after first move
- and the last move must be highlighted
the orientation of the board must match the side which is to solve the puzzle

Dunno, what is the state of this issue, the requirements looks really straightforward.

I am absolutely new to PgnViewerJS, so I can easily misinterpret the ideas. If I decide to implement it, am I right the new modus, like pgnPuzzle, is to be introduced?

mliebelt commented 3 years ago

@Bebul Thank you a lot for investing time in that topic. I think you are right. Here is what I would do if I would tackle that issue:

Create a new mode pgnPuzzle.
Decide what is needed: board, parts of editing (only making moves, not recording new ones), mostly viewer mode.
Take your idea from above: pgn has to be one straight line (variants only, if you want to include feedback to wrong moves, see studies in Lichess, interactive excercise), start after first move.
Necessary changes: show only moves made and accepted; have a feedback for success.

The changes will be only in the viewer module, no changes to reader.

I wish you good luck in your challenge!

Bebul commented 3 years ago

After finishing the puzzle I have disabled mooving pieces, so the chessground is frozen, unless user returns by prevMove. Implementation looks good, because using hasMode('puzzle') makes changes impossible to break other functionality which is most important.

But I must still do:

disable showing moves (or moves after current puzzle progress)
create progress info instead with something like "puzzle failed" - "congratulation" - "in progress"
maybe support of more puzzles in PGN with possibility to navigate nextPuzzle

What I dunno is creating documentation and examples. I have puzzle.html as example with one generated by https://github.com/linrock/chess-puzzle-maker

I am looking forward to finishing it and making PULL Request soon.

ppeloton commented 6 months ago

First of all thank you @mliebelt for this wonderful package and all your work. I have tried to contribute by implementing a guess-the-move mode. I am happy to hear any feedback and further help on this issue as well as on other parts of the package.

mliebelt commented 6 months ago

@ppeloton Thank you a lot for providing that functionality. I have to check it (hopefully on weekend), but on first sight it should be doable. If it seems to be ok, I will create a new version that includes your new mode, and check as well the documentation how that could be expanded. If that is all ok, I will publish a new version that includes your changes.

mliebelt commented 6 months ago

I have merged it, and yes, it may be a first step in the right direction. I think we should now define, what "Guess the move" should support. At the moment, the implementation is there, but there is no additional support for this mode. So the following list could be a starting point, and it contains also things I see in the example / implementation then:

Allow one player to do the next move. Question: How do we know who's turn it is?
Give some hint / indication in what the person that wants to solve the puzzle would do. We could include that in the pgn, and we could just show it in the normal notation, but perhaps there are better ideas available. When I look into lichess, and its way to make in a study a puzzle, they have some good ideas.
How about positive / negative feedback on doing a move? At the moment
- positive just means "do the move"
- negative just means "take the move back" (even if was a valid move)
Check if there is a good default layout that supports puzzle "good enough", or if we should find a new layout to support puzzles much better.

So I leave that ticket open, but will include your change in the next release coming soon.

Things to do before doing a new release then:

Have some minimal documentation about the new mode.
Include the new mode (with an example?) in the Configuration Builder (examples/configuration/config.html).

mliebelt commented 6 months ago

I was working on the Configuration Builder, and it works quite well.

ppeloton commented 6 months ago

@mliebelt: Great! I agree, it is a starting point and there are many things one can improve and add to this functionality (happy to help here!).

I continued testing the feature a little bit, there seems to be still a bug as the fen in the that.mypgn object does not seem to get updated properly. If you move a piece two times in a row, the whole puzzle freezes sometimes - e.g. take the puzzle.html I provided in the PR and play first the move f4 and then fxe5, after which the puzzle freezes. The reason seem to bet that the fen in mypgn is not up-to-date and one cannot get the proper piece information in the function onSnapEnd. I have not figured out yet how to fix it. Maybe you can have a look as you know the package better.

ppeloton commented 5 months ago

@mliebelt: Great! I agree, it is a starting point and there are many things one can improve and add to this functionality (happy to help here!).

I continued testing the feature a little bit, there seems to be still a bug as the fen in the that.mypgn object does not seem to get updated properly. If you move a piece two times in a row, the whole puzzle freezes sometimes - e.g. take the puzzle.html I provided in the PR and play first the move f4 and then fxe5, after which the puzzle freezes. The reason seem to bet that the fen in mypgn is not up-to-date and one cannot get the proper piece information in the function onSnapEnd. I have not figured out yet how to fix it. Maybe you can have a look as you know the package better.

I just made a pull request to fix the issue above. I assume in future it is better to open bugs/tickets for specific issues and keep this ticket for general discussions.

ppeloton commented 5 months ago

@mliebelt: I just added a few sentences for a simple documentation. Regarding the future enhancements I would suggest to do this together with the issue #175. E.g. it would make sense to give hints via API in my opinion and to leave users of the package the freedom to define how to handle those hints in the UI (of course there could be a default UI implementation for hints in this package as well)

mliebelt / pgn-viewer

Guess the Move #2