import os
import scipy.io as sio
import numpy as np
import h5py


def remending(q):
    return q.replace(".mat","").replace(".npz","")

def listfiles():
    for zw in os.listdir("adata"):
        fn="adata/"+zw
        yield remending(zw),fn

def loadfile73(fn):
    f = h5py.File(fn,'r')
    x = f.get('X')
    x = np.array(x) # For converting to a NumPy array
    y = f.get('y')
    y = np.array(y) # For converting to a NumPy array
    x=np.transpose(x)
    y=np.transpose(y)
    return x,y
def loadfile(fn="thyroid.mat"):
    if ".npz" in fn:
        f=np.load(fn)
        return f["x"],f["y"]
    try:
        mat=sio.loadmat(fn)
        return mat["X"],mat["y"]
    except:
        return loadfile73(fn)

def loadfiles():
    for f,fn in listfiles():
        yield f,*loadfile(fn)

def filterfiles():
    for f,x,y in loadfiles():
        cou=np.sum(y)
        if cou<10:continue
        yield f,x,y
if __name__ == '__main__':
    if False:
        for f,x,y in filterfiles():
            #print(f,x.shape[1],y.shape[0],np.mean(y))
            les=[len(set(x[:,i]))/len(x) for i in range(x.shape[1])]
            lestr=f"{np.mean(les)}+-{np.std(les)}"
            print(f'"{f}":{int(x.shape[1]/2)+1},{lestr}')

    for f,x,y in filterfiles():
        if "wbc" in f:break