Parsing bad(?) sgf

[trevoke]

Parsing bad(?) sgf

[Trevoke]

Hi, I'm glad you're using it and you're happy with it!

Hmm, interesting. When I built this gem I assumed that AB/AW was to be a unique property that took a list of stones. If I'm right, then goproblems is using an invalid format by having many of these.. But that doesn't solve your problem. Thanks for bringing this to my attention. The SGF Standard is actually silent on this point.

I will ask the SGF people about whether goproblems.com is doing something correct or not (for my own edification), and then fix that problem anyway.

[jphager2]

Yeah. Im just wondering if it has something to do with how goproblems lets you add problems (through a flash application) which might generate this strange syntax. I assume most sgf editors are designed to output correct sgf

[Trevoke]

Do all the problems on goproblems have this issue? Do you have, by chance,
an example of one that doesn't have it?

On Sat, Mar 15, 2014 at 1:57 PM, John Hager [email protected]:

Yeah. Im just wondering if it has something to do with how goproblems lets
you add problems (through a flash application) which might generate this
strange syntax. I assume most sgf editors are designed to output correct sgf

—
Reply to this email directly or view it on GitHubhttps://github.com//issues/23#issuecomment-37732902
.

[jphager2]

Yeah this is problem 1000:

GM[1]FF[4]VW[]AP[Many Faces of Go:10.0]
SZ[13]
HA[0]
ST[1]
DT[2001-08-13]
KM[0.0]
RU[GOE]

AB[lb][lc][ld][kc][jb][ka][ck][dg][ei][fb][fc][fd][ee][db][cd][ej][fk][fl][fh][eg][fg][lj][kk][jj][ij][md][lf][gm][fm][gc]

AW[ib][jc][kd][jd][le][jf][hc][fi][gb][gd][fe][ff][ia][fj][gl][gk][gh][gg][jk][kg][lh][kl][ll][ik][me][ke][hm][hl]

C[After the previous problem, Black has played at 1 rather than
protecting his corner. How can white punish him?
]
PL[W]LB[gc:A]

(;W[mb](;B[ma]
%28;W[la];B[ge];W[ma];B[mc];W[ja];B[ma];W[kb]
C[White has given up too much when black plays A
]LB[ge:A]
%29
%28;W[ja];B[kb];W[la]
C[RIGHT
]%29)
(;B[ja];W[la]))
(;W[ja];B[mb]
C[Black is unconditionally alive.
])
(;W[ge];B[ja]
C[Black has had another chance to repair his defect.
]))

[Trevoke]

I got a response from the SGF folks. I'm quoting it here.

in a correct SGF file every property appears at most once in a single node. Nowadays bad SGF files are rare (goproblems is apparently an exception). But if you want to write a resilient parser, then maybe should take duplicate properties into account. Like Postel's law: be conservative in what you do, be liberal in what you accept from others.

Reading malformed SGF files is ok, although it could get us into trouble in the long term. But the internal representation (if accessible through an API) or the output should be correct SGF. Check files with sgfc, it is my reference implementation of the SGF standard.

I've been busy as well and haven't had time to work on this, but clearly some of these goproblems are malformed. I will have to do some work to handle that, which probably means adding in some kind of notifications for the user, like an array of errors or something.

[Trevoke]

I just pushed a fix to github ( d99b575 ), it's not deployed to rubygems yet. Can you try it from here and see how it works for you?

[Trevoke]

At least this use case is handled. Closing.