Merge pull request #133 from SaarthShah/main

added test cases for features & testing for JSON structure

tests/test_transcription.py

@@ -78,10 +78,178 @@ async def test_transcribe_prerecorded_file():
         response = await deepgram.transcription.prerecorded({"buffer": audio, "mimetype": "audio/wav"})
         assert "results" in response
 
+
+def test_prerecorded_json_structure():
+    """
+    Testing the JSON structure of the Deepgram response. This should be consistent across all outputs
+    unless there is a breaking change in the API itself.
+    """
+    response = deepgram.transcription.sync_prerecorded(
+        {
+            "url": AUDIO_URL
+        },
+        {
+            "model": "nova",
+            "smart_format": True,
+        },
+    )
+
+    # Checks that both results and metadata are present in the base response
+    assert set(response.keys()) == set(['results','metadata'])
+
+    # Checks that metadata contains the expected keys
+    assert set(response['metadata'].keys()) == set(['channels','created','duration','model_info','models','request_id','sha256','transaction_key'])
+
+    # Checks that the results key contains the expected keys
+    assert list(response['results'].keys()) == ['channels']
+
+    # Checks that the channels key contains the expected keys
+    assert len(response['results']['channels']) == 1
+
+    # Checks if alternatives is present in the channels[0] object
+    assert list(response['results']['channels'][0].keys()) == ['alternatives']
+
+    # Checks if alternatives is a list of length 1
+    assert len(response['results']['channels'][0]['alternatives']) == 1
+
+    # Checks if the alternatives[0] object contains the expected keys
+    assert set(response['results']['channels'][0]['alternatives'][0].keys()) == set(['transcript', 'confidence', 'words', 'paragraphs'])
+
+    # Checks if the transcript is a string
+    assert type(response['results']['channels'][0]['alternatives'][0]['transcript']) == str
+
+    # Checks if the confidence is a float
+    assert type(response['results']['channels'][0]['alternatives'][0]['confidence']) == float
+
+    # Checks if the words is a list
+    assert type(response['results']['channels'][0]['alternatives'][0]['words']) == list
+
+    # Checks if the paragraphs is a dict
+    assert type(response['results']['channels'][0]['alternatives'][0]['paragraphs']) == dict
+
+    # Checks if the transcript within the paragraphs object is a string
+    assert type(response['results']['channels'][0]['alternatives'][0]['paragraphs']['transcript']) == str
+
+    # Checks if the paragraphs within the paragraphs object is a list
+    assert type(response['results']['channels'][0]['alternatives'][0]['paragraphs']['paragraphs']) == list
+
+    # Checks if the paragraphs[0] object contains the expected keys
+    assert set(response['results']['channels'][0]['alternatives'][0]['paragraphs']['paragraphs'][0]) == set(['sentences', 'num_words', 'start', 'end'])
+
+
+def test_diarization():
+
+    """
+    Testing the diarization output of the Deepgram response.
+    
+    """
+
+    response = deepgram.transcription.sync_prerecorded(
+        {
+            "url": AUDIO_URL
+        },
+        {
+            "model": "nova",
+            "smart_format": True,
+            "diarize": True
+        },
+    )
+
+    # Checks the keys for the alternatives[0] object
+    assert set(response['results']['channels'][0]['alternatives'][0]['paragraphs']['paragraphs'][0].keys()) == set(['sentences','speaker', 'num_words', 'start', 'end'])
+
+    # Checks the keys in the words object
+    assert set(response['results']['channels'][0]['alternatives'][0]['words'][0].keys()) == set(['word', 'start', 'end', 'confidence', 'speaker', 'speaker_confidence', 'punctuated_word'])
+
+
+def test_summarize():
+    """
+    Checking the summarize output of the Deepgram response, especially for summarize v2.
+    """
+    response = deepgram.transcription.sync_prerecorded(
+        {
+            "url": AUDIO_URL
+        },
+        {
+            "model": "nova",
+            "smart_format": True,
+            "summarize": 'v2'
+        },
+    )
+
+    # Checks if the summary object has the expected keys
+    assert set(response['results']['summary'].keys()) == set(['result', 'short'])
+    # Check if the result is a string
+    assert type(response['results']['summary']['result']) == str
+
+    # Check if the request was successful
+    assert response['results']['summary']['result'] == 'success'
+
+    # Check if the short is a string
+    assert type(response['results']['summary']['short']) == str
+
+
+def test_topic_detection():
+    """
+    Checking the topic detection output of the Deepgram response.
+    """
+    response = deepgram.transcription.sync_prerecorded(
+        {
+            "url": AUDIO_URL
+        },
+        {
+            "model": "nova",
+            "smart_format": True,
+            "detect_topics": True
+        },
+    )
+
+    # Checks if the topics key is present in the alternatives[0] object
+    assert type(response['results']['channels'][0]['alternatives'][0]['topics']) == list
+
+    # Checks if the topics[0] object is a dict
+    assert type(response['results']['channels'][0]['alternatives'][0]['topics'][0]) == dict
+
+    # Checks if the topics[0] object contains the expected keys
+    assert set(response['results']['channels'][0]['alternatives'][0]['topics'][0].keys()) == set(['text', 'start_word', 'end_word', 'topics'])
+
+    # Checks if the topics object is a list
+    assert type(response['results']['channels'][0]['alternatives'][0]['topics'][0]['topics']) == list
+
+
+def test_detect_language():
+    response = deepgram.transcription.sync_prerecorded(
+        {
+            "url": AUDIO_URL
+        },
+        {
+            "model": "nova",
+            "detect_language": True
+        },
+    )
+    assert type(response['results']['channels'][0]['detected_language']) == str
+    assert response['results']['channels'][0]['detected_language'] == 'en'
+
+
+def test_alternatives():
+    response = deepgram.transcription.sync_prerecorded(
+        {
+            "url": AUDIO_URL
+        },
+        {
+            "model": "nova",
+            "alternatives": 2
+        },
+    )
+    assert len(response['results']['channels'][0]['alternatives']) == 2
+
+
 def test_missing_api_key():
     with pytest.raises(DeepgramSetupError):
         Deepgram({})
 
+
 def test_400_error():
     with pytest.raises(DeepgramApiError):
         deepgram.transcription.sync_prerecorded({"url": AUDIO_URL}, {"model": "nova", "language": "ta"})
+