在给定的代码上获取

Question

我得到了转换arff文件的代码。我不得不下载numpy库，现在当我试图用我的文件运行它时，它会给我一些关键错误，比如

“imgInfo[1][clstrDct[clstr]]+=1#增加集群计数 KeyError:“cluster35\r'”

import numpy as np

def xfrm(arFil='KBcls-100-10-20'):
    '''transform a clustered patch arff file to an image training / test file'''
    global imgDct, clstrDct, num, clsts, lne
    imgDct = {}
    clstrDct = {}
    with open(arFil + '.arff', 'r') as ptchFil:
        while True:                         # find Cluster attribute
            lne = ptchFil.readline()
            if lne == '': return 'EOF bfore one'
            if lne.lower().startswith('@attribute cluster'):
                clsts = lne[lne.find('{')+1 : lne.find('}')].split(',')
                num = len(clsts)
                break
        for i in range(len(clsts)):     # map cluster names to integers 0+ w/ inverted mapping also
            clstrDct[clsts[i]] = i
            clstrDct[i] = clsts[i]
        while True:                         # first patch data line
            lne = ptchFil.readline()
            if lne == '': return 'EOF bfore two'
            if lne.startswith('@data'): break
        while True:
            lne = ptchFil.readline()        # read through patch lines
            if lne == '': break             # EOF
            if lne[-1] == '\n': lne=lne[:-1]        # all end with \n except possibly the last line of the file
            attrs = lne.split(',')
            imgId = attrs[0]
            clstr = attrs[-1]
            cls = attrs[-2]
            try: imgInfo = imgDct[imgId]
            except KeyError:
                imgInfo = [cls, np.zeros((num), dtype=int)]     # new cluster counting array
                imgDct[imgId] = imgInfo
            imgInfo[1][clstrDct[clstr]] += 1    # increment the cluster count
        with open(arFil + '-img.arff', 'w') as arFile:
            arFile.write('%    from {0:}.arff: {1:} patch clusters\n%\n'.format(arFil, num))
            arFile.write('@relation Image-Patch-Clusters\n@attribute Image-ID numeric\n')
            for i in range(num):
                arFile.write('@attribute {} numeric\n'.format(clstrDct[i]))         # cluster attributes
            arFile.write('@attribute class {unknown, street, highway}\n@data')
            for imid,iminfo in imgDct.items():
                arFile.write('\n{}, '.format(imid))
                for i in range(num):
                    arFile.write('{}, '.format(iminfo[1][i]))
                arFile.write('{}'.format(iminfo[0]))

if __name__ == "__main__":
    xfrm('Test1Clust')

Answer 1

readline包括与其余内容一起结束的行。这意味着在每个attrs[-1]的末尾有一个额外的\r，\n，或{}。这就是为什么“cluster35\r”中有一个\r。您可以使用strip来消除此问题。在

clstr = attrs[-1].strip()