data load for inference

fluid/DeepASR/data_utils/async_data_reader.py

@@ -102,11 +102,14 @@ class SampleInfoBucket(object):
             feature_bin_path = self._feature_bin_paths[block_idx]
             feature_desc_path = self._feature_desc_paths[block_idx]
 
-            label_desc_lines = open(label_desc_path).readlines()
             feature_desc_lines = open(feature_desc_path).readlines()
+            label_desc_lines = None
+            if label_desc_path != "":
+                label_desc_lines = open(label_desc_path).readlines()
 
-            sample_num = int(label_desc_lines[0].split()[1])
-            assert sample_num == int(feature_desc_lines[0].split()[1])
+            sample_num = int(feature_desc_lines[0].split()[1])
+            if label_desc_path != "":
+                assert sample_num == int(label_desc_lines[0].split()[1])
 
             for i in xrange(sample_num):
                 feature_desc_split = feature_desc_lines[i + 1].split()
@@ -115,6 +118,10 @@ class SampleInfoBucket(object):
                 feature_frame_num = int(feature_desc_split[4])
                 feature_dim = int(feature_desc_split[5])
 
+                label_start = -1
+                label_size = -1
+                label_frame_num = feature_frame_num
+                if label_desc_path != "":
                     label_desc_split = label_desc_lines[i + 1].split()
                     label_start = int(label_desc_split[2])
                     label_size = int(label_desc_split[3])
@@ -187,7 +194,7 @@ class AsyncDataReader(object):
 
     def __init__(self,
                  feature_file_list,
-                 label_file_list,
+                 label_file_list="",
                  drop_frame_len=512,
                  proc_num=10,
                  sample_buffer_size=1024,
@@ -221,9 +228,12 @@ class AsyncDataReader(object):
     def generate_bucket_list(self, is_shuffle):
         if self._block_info_list is None:
             block_feature_info_lines = open(self._feature_file_list).readlines()
-            block_label_info_lines = open(self._label_file_list).readlines()
-            assert len(block_feature_info_lines) == len(block_label_info_lines)
             self._block_info_list = []
+            if self._label_file_list != "":
+                block_label_info_lines = open(self._label_file_list).readlines()
+                #block_label_info_lines = open(self._label_file_list).readlines()
+                assert len(block_feature_info_lines) == len(
+                    block_label_info_lines)
                 for i in xrange(0, len(block_feature_info_lines), 2):
                     block_info = (block_feature_info_lines[i],
                                   block_feature_info_lines[i + 1],
@@ -231,6 +241,12 @@ class AsyncDataReader(object):
                                   block_label_info_lines[i + 1])
                     self._block_info_list.append(
                         map(lambda line: line.strip(), block_info))
+            else:
+                for i in xrange(0, len(block_feature_info_lines), 2):
+                    block_info = (block_feature_info_lines[i],
+                                  block_feature_info_lines[i + 1], "", "")
+                    self._block_info_list.append(
+                        map(lambda line: line.strip(), block_info))
 
         if is_shuffle:
             self._rng.shuffle(self._block_info_list)
@@ -318,19 +334,24 @@ class AsyncDataReader(object):
                         sample_info.feature_dim,
                         len(feature_bytes))
 
+                if sample_info.label_bin_path != "":
                     label_bytes = read_bytes(sample_info.label_bin_path,
                                              sample_info.label_start,
                                              sample_info.label_size)
 
-                assert sample_info.label_frame_num * 4 == len(label_bytes), (
-                    sample_info.label_bin_path, sample_info.label_array,
+                    assert sample_info.label_frame_num * 4 == len(
+                        label_bytes), (sample_info.label_bin_path,
+                                       sample_info.label_array,
                                        len(label_bytes))
 
-                label_array = struct.unpack('I' * sample_info.label_frame_num,
-                                            label_bytes)
+                    label_array = struct.unpack(
+                        'I' * sample_info.label_frame_num, label_bytes)
                     label_data = np.array(
                         label_array, dtype='int64').reshape(
                             (sample_info.label_frame_num, 1))
+                else:
+                    label_data = np.zeros(
+                        (sample_info.label_frame_num, 1), dtype='int64')
 
                 feature_frame_num = sample_info.feature_frame_num
                 feature_dim = sample_info.feature_dim
@@ -345,7 +366,6 @@ class AsyncDataReader(object):
                 for transformer in self._transformers:
                     # @TODO(pkuyym) to make transfomer only accept feature_data
                     sample_data = transformer.perform_trans(sample_data)
-
                 while order_id != out_order[0]:
                     time.sleep(0.001)