main.py 文件源码

def zipfiles2dataframe(zip, rexpr):
    """
    Extract files in the zip that match the regular expression rexpr and load them as a data frame
    """
    dfs = []
    for f in zip.filelist:
        m = rexpr.search(f.filename)
        if not m:
            continue

        df = pd.read_fwf(StringIO(zip.read(f).decode('utf-8')), header=None, skiprows=1)
        df.columns = ['event', 'tau']
        df['user'], df['session'] = m.groups()
        dfs.append(df)

    df = pd.concat(dfs).set_index(['user', 'session'])

    return df