tykiww commited on
Commit
a7d4d8d
1 Parent(s): aa9a81b

Update utilities/transcripts.py

Browse files
Files changed (1) hide show
  1. utilities/transcripts.py +18 -1
utilities/transcripts.py CHANGED
@@ -114,7 +114,7 @@ class VTTTranscriptLoader:
114
 
115
  return self.manual_document(output, metadata)
116
 
117
- def load(self):
118
  """Processes all VTT files in the directory or the single file and returns a list of results."""
119
  results = []
120
  if os.path.isdir(self.fp):
@@ -129,6 +129,23 @@ class VTTTranscriptLoader:
129
  results.append(transcript)
130
  return results
131
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
132
 
133
  class DocumentEmbedder:
134
  """
 
114
 
115
  return self.manual_document(output, metadata)
116
 
117
+ def directory_load(self):
118
  """Processes all VTT files in the directory or the single file and returns a list of results."""
119
  results = []
120
  if os.path.isdir(self.fp):
 
129
  results.append(transcript)
130
  return results
131
 
132
+ def load(self):
133
+ """Processes all VTT files in the given list of directories or files and returns a list of results."""
134
+ results = []
135
+ for path in self.fp:
136
+ if os.path.isdir(path):
137
+ for root, _, files in os.walk(path):
138
+ for file in files:
139
+ if file.endswith('.vtt'):
140
+ file_path = os.path.join(root, file)
141
+ transcript = self.process_file(file_path)
142
+ results.append(transcript)
143
+ else:
144
+ if path.endswith('.vtt'):
145
+ transcript = self.process_file(path)
146
+ results.append(transcript)
147
+ return results
148
+
149
 
150
  class DocumentEmbedder:
151
  """