collect_rdataframe.py 文件源码

def parse_audit_files():
    audit_paths = glob.glob('audit/workflow_mmlinear_started_20151026_17*')
    audit_data = []
    for path in audit_paths:
        with open(path) as fh:
            cp = ConfigParser()
            cp.readfp(fh)
            for sec in cp.sections():
                dat = dict(cp.items(sec))
                if 'train_lin' in dat['instance_name']:
                    ms = re.match('train_lin_trn([0-9]+|rest)_tst([0-9]+)_c([0-9\.]+)', dat['instance_name'])
                    if ms is None:
                        raise Exception('No match in name: ' + dat['instance_name'])
                    m = ms.groups()
                    dat['training_size'] = m[0]
                    dat['test_size'] = m[1]
                    dat['cost'] = m[2]
                    audit_data.append(dat)
    db = dataset.connect('sqlite:///:memory:')
    tbl = db['audit']
    for d in audit_data:
        tbl.insert(d)
    return tbl