格式化大字符串时出错 - 如何检测错误？

Question

我对Python很新，我目前正在使用我需要的大型字符串格式。

发生此问题，因为我不明白大字符串格式中发生错误的位置。更确切地说，我得到一个形式的错误

Traceback (most recent call last): 
    File "/usr/lib/python2.7/dist-packages/IPython/core/interactiveshell.py", line 2731, in run_code 
    exec code_obj in self.user_global_ns, self.user_ns 
    File "<ipython-input-3-f6a3bb7fe2f9>", line 13, in <module> 
    trainCV = trainCV % (train_params) 
ValueError: unsupported format character ',' (0x2c) at index 2726 

有没有一种方法来精确检测线路得到错误发生的地方？

我的完整代码如下所示：

trainCV = open('Conv_Constructor.yaml','r').read() 
train_params = {'batch_size': 100, 
        'output_channels_h2': 64, 
        'conv_kernel_size': 8, 
        'pool_size': 2, 
        'stride_size': 1, 
        'output_channels_h3': 64, 
        'num_classes': 6, 
        'valid_stop': 4200, 
        'test_start': 4200, 
        'test_stop': 4400, 
        'max_epochs': 5} 

trainCV = trainCV % (train_params) 
print trainCV 

而且Conv_Constructor.yaml文件我想作为一个字符串格式如下

# ---------- INPUTS --------- 
# 
#  batch_size 
#  output_channels_h2 
#  conv_kernel_size 
#  pool_size 
#  stride_size 
#  output_channels_h3 
#  num_classes 
#  valid_stop 
#  test_start 
#  test_stop 
#  max_epochs 

################################################################## 

!obj:pylearn2.train.Train { 
    dataset: !obj:pylearn2.official_train_data.load_data { 
     start: 0, 
     stop: 4000 
     # one_hot: 1, 
    }, 

    model: !obj:pylearn2.models.mlp.MLP { 
     batch_size: %(batch_size)i, 
     input_space: !obj:pylearn2.space.Conv2DSpace { 
      shape: [32, 32], 
      num_channels: 1, 
      axes = ('b',0,1,'c') 
     }, 

     layers: [ !obj:pylearn2.models.mlp.ConvRectifiedLinear { 
        layer_name: 'h2', 
        output_channels: %(output_channels_h2)i, 
        #params : !pkl: 'dae_layer_1_weights.plk', 
        irange: .05, 

        kernel_shape: [%(conv_kernel_size)i, %(conv_kernel_size)i], 

        pool_shape: [%(pool_size)i, %(pool_size)i], 

        pool_stride: [%(stride_size)i, %(stride_size)i], 

        max_kernel_norm: 1.9365 
       }, !obj:pylearn2.models.mlp.ConvRectifiedLinear { 
        layer_name: 'h3', 
        output_channels: %(output_channels_h3)i, 
        #params : !pkl: 'dae_layer_1_weights.plk', 
        irange: .05, 
        kernel_shape: %(conv_kernel_size)i, %(conv_kernel_size)i], 

        pool_shape:[%(pool_size)i, %(pool_size)i], 

        pool_stride: [%(stride_size)i, %(stride_size)i], 
        max_kernel_norm: 1.9365 


       }, !obj:pylearn2.models.mlp.Softmax { 
        max_col_norm: 1.9365, 
        layer_name: 'y', 
        n_classes: %(num_classes)i, 
        istdev: .05 
       } 
       ], 
    }, 

    algorithm: !obj:pylearn2.training_algorithms.sgd.SGD { 
     batch_size: %(batch_size)i, 
     learning_rate: .01, 
     init_momentum: .5, 
     monitoring_dataset: 
      { 
       'valid' : !obj:pylearn2.official_train_data.load_data { 
         start: 4000, 
         stop: %(valid_stop)i 
         #one_hot: 1, 
       }, 
       'test' : !obj:pylearn2.official_train_data.load_data { 
         start: %(test_start), 
         stop: %(test_stop) 
         #one_hot: 1, 
       } 
      }, 

     cost: !obj:pylearn2.costs.cost.SumOfCosts { costs: [ 
      !obj:pylearn2.costs.cost.MethodCost { 
       method: 'cost_from_X' 
      }, !obj:pylearn2.costs.mlp.WeightDecay { 
       coeffs: [ .00005, .00005, .00005 ] 
      } 
      ] 
     }, 
     termination_criterion: !obj:pylearn2.termination_criteria.And { 
      criteria: [ 
       !obj:pylearn2.termination_criteria.MonitorBased { 
        channel_name: "valid_y_misclass", 
        prop_decrease: 0.50, 
        N: 50 
       }, 
       !obj:pylearn2.termination_criteria.EpochCounter { 
        max_epochs: %(max_epochs)i 
       }, 
      ] 
     }, 
    }, 

    extensions: 
     [ !obj:pylearn2.train_extensions.best_params.MonitorBasedSaveBest { 
      channel_name: 'valid_y_misclass', 
      save_path: "%(save_path)s/convolutional_network_best.pkl" 
     }, !obj:pylearn2.training_algorithms.sgd.MomentumAdjustor { 
      start: 1, 
      saturate: 10, 
      final_momentum: .99 
     } 
    ] 
} 

Answer 1

你可以找到更多的错误很容易通过单独处理每条线而不是整个字符串。 更换

trainCV = trainCV % (train_params) 

与

trainCV = trainCV.split('\n') 
t1=[] 
try: 
    for i, t in enumerate(trainCV): 
     t1.append(t % (train_params)) 
except : 
    print 'Error in line {}:'.format(i) 
    print t[i] 
    raise 

，你会得到下面的输出：

78 
         start: %(test_start), 

这意味着你的字符串格式化也不太工作（在这种情况下，我认为是支架丢失后的i）。以这种方式调试你的大字符串，你应该有一个工作代码。

之后做到这一点，你可以通过加入列表打印：

print '\n'.join(t1)