Throw Exception at Multiple Stages in Captioning if Transcription is Invalid

deepgram_captions/converters.py

@@ -2,13 +2,32 @@
 from .helpers import chunk_array, replace_text_with_word
 
 
+class ConverterException(Exception):
+    pass
+
+
 class DeepgramConverter:
-    def __init__(self, dg_response):
+    def __init__(self, dg_response, use_exception: bool = True):
         if not isinstance(dg_response, dict):
             self.response = json.loads(dg_response.to_json())
         else:
             self.response = dg_response
 
+        if use_exception:
+            one_valid_transcription = False
+            for channel in self.response["results"]["channels"]:
+                if channel["alternatives"][0]["transcript"] != "":
+                    one_valid_transcription = True
+                    break
+            if "utterances" in self.response["results"]:
+                for utterance in self.response["results"]["utterances"]:
+                    if utterance["transcript"] != "":
+                        one_valid_transcription = True
+                        break
+
+            if not one_valid_transcription:
+                raise ConverterException("No valid transcriptions found in response")
+
     def get_lines(self, line_length):
         results = self.response["results"]
         content = []
@@ -19,7 +38,6 @@ def get_lines(self, line_length):
                     content.extend(chunk_array(utterance["words"], line_length))
                 else:
                     content.append(utterance["words"])
-
         else:
             words = results["channels"][0]["alternatives"][0]["words"]
             diarize = (

deepgram_captions/helpers.py

@@ -1,6 +1,10 @@
 from datetime import datetime
 
 
+class EmptyTranscriptException(Exception):
+    pass
+
+
 def seconds_to_timestamp(seconds, format="%H:%M:%S.%f"):
     seconds = round(seconds, 3)
     dt = datetime.utcfromtimestamp(seconds)

deepgram_captions/srt.py

@@ -1,4 +1,4 @@
-from .helpers import seconds_to_timestamp
+from .helpers import seconds_to_timestamp, EmptyTranscriptException
 
 
 def srt(converter, line_length=None):
@@ -12,6 +12,9 @@ def srt(converter, line_length=None):
 
     current_speaker = None
 
+    if not lines[0]:
+        raise EmptyTranscriptException("No transcript data found")
+
     for words in lines:
         output.append(str(entry))
         entry += 1

deepgram_captions/webvtt.py

@@ -1,4 +1,4 @@
-from .helpers import seconds_to_timestamp
+from .helpers import seconds_to_timestamp, EmptyTranscriptException
 
 
 def webvtt(converter, line_length=None):
@@ -22,6 +22,9 @@ def webvtt(converter, line_length=None):
     if hasattr(converter, "get_lines") and callable(getattr(converter, "get_lines")):
         lines = converter.get_lines(line_length)
 
+        if not lines[0]:
+            raise EmptyTranscriptException("No transcript data found")
+
         speaker_labels = "speaker" in lines[0][0]
 
         for words in lines: