Merge pull request #85 from JigsawStack/feat/stt-wordtimestamp

feat(stt): add param and cleanup test suites

Author: Khurdhula-Harshavardhan

.github/workflows/ci.yml

@@ -37,12 +37,9 @@ jobs:
           - test_classification.py
           - test_embedding.py
           - test_file_store.py
-          - test_image_generation.py
           - test_object_detection.py
           - test_prediction.py
-          - test_sentiment.py
           - test_sql.py
-          - test_summary.py
           - test_translate.py
           - test_validate.py
           - test_web.py

jigsawstack/audio.py

@@ -49,6 +49,11 @@ class SpeechToTextParams(TypedDict):
     the duration of each chunk in seconds, maximum value is 15, defaults to 3
     """
 
+    word_timestamps: NotRequired[bool]
+    """
+    When set to true, returns each word as its own entry in the chunks array with its own start and end timestamp. Useful for caption alignment and word-accurate search. Cannot be combined with stream=true.
+    """
+
 
 class ChunkResponse(TypedDict):
     text: str

tests/test_embedding.py

@@ -35,61 +35,6 @@
     "https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf"
 )
 
-# Test cases for Embedding V1
-EMBEDDING_V1_TEST_CASES = [
-    {
-        "name": "text_embedding_basic",
-        "params": {
-            "type": "text",
-            "text": SAMPLE_TEXT,
-        },
-    },
-    {
-        "name": "text_embedding_with_truncate",
-        "params": {
-            "type": "text",
-            "text": SAMPLE_TEXT * 100,  # Long text to test truncation
-            "token_overflow_mode": "truncate",
-        },
-    },
-    {
-        "name": "text_embedding_with_error_mode",
-        "params": {
-            "type": "text",
-            "text": SAMPLE_TEXT,
-            "token_overflow_mode": "error",
-        },
-    },
-    {
-        "name": "image_embedding_from_url",
-        "params": {
-            "type": "image",
-            "url": SAMPLE_IMAGE_URL,
-        },
-    },
-    {
-        "name": "audio_embedding_from_url",
-        "params": {
-            "type": "audio",
-            "url": SAMPLE_AUDIO_URL,
-        },
-    },
-    {
-        "name": "pdf_embedding_from_url",
-        "params": {
-            "type": "pdf",
-            "url": SAMPLE_PDF_URL,
-        },
-    },
-    {
-        "name": "text_other_type",
-        "params": {
-            "type": "text-other",
-            "text": "This is a different text type for embedding",
-        },
-    },
-]
-
 # Test cases for Embedding V2
 EMBEDDING_V2_TEST_CASES = [
     {
@@ -173,80 +118,6 @@
 ]
 
 
-class TestEmbeddingV1Sync:
-    """Test synchronous Embedding V1 methods"""
-
-    sync_test_cases = EMBEDDING_V1_TEST_CASES
-
-    @pytest.mark.parametrize(
-        "test_case", sync_test_cases, ids=[tc["name"] for tc in sync_test_cases]
-    )
-    def test_embedding_v1(self, test_case):
-        """Test synchronous embedding v1 with various inputs"""
-        try:
-            result = jigsaw.embedding(test_case["params"])
-            assert result["success"]
-            assert "embeddings" in result
-            assert isinstance(result["embeddings"], list)
-            if "chunks" in result:
-                assert isinstance(result["chunks"], list)
-        except JigsawStackError as e:
-            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
-
-    @pytest.mark.parametrize(
-        "test_case", BLOB_TEST_CASES, ids=[tc["name"] for tc in BLOB_TEST_CASES]
-    )
-    def test_embedding_v1_blob(self, test_case):
-        """Test synchronous embedding v1 with blob inputs"""
-        try:
-            # Download blob content
-            blob_content = requests.get(test_case["blob_url"]).content
-            result = jigsaw.embedding(blob_content, test_case["options"])
-            assert result["success"]
-            assert "embeddings" in result
-            assert isinstance(result["embeddings"], list)
-        except JigsawStackError as e:
-            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
-
-
-class TestEmbeddingV1Async:
-    """Test asynchronous Embedding V1 methods"""
-
-    async_test_cases = EMBEDDING_V1_TEST_CASES
-
-    @pytest.mark.parametrize(
-        "test_case", async_test_cases, ids=[tc["name"] for tc in async_test_cases]
-    )
-    @pytest.mark.asyncio
-    async def test_embedding_v1_async(self, test_case):
-        """Test asynchronous embedding v1 with various inputs"""
-        try:
-            result = await async_jigsaw.embedding(test_case["params"])
-            assert result["success"]
-            assert "embeddings" in result
-            assert isinstance(result["embeddings"], list)
-            if "chunks" in result:
-                assert isinstance(result["chunks"], list)
-        except JigsawStackError as e:
-            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
-
-    @pytest.mark.parametrize(
-        "test_case", BLOB_TEST_CASES, ids=[tc["name"] for tc in BLOB_TEST_CASES]
-    )
-    @pytest.mark.asyncio
-    async def test_embedding_v1_blob_async(self, test_case):
-        """Test asynchronous embedding v1 with blob inputs"""
-        try:
-            # Download blob content
-            blob_content = requests.get(test_case["blob_url"]).content
-            result = await async_jigsaw.embedding(blob_content, test_case["options"])
-            assert result["success"]
-            assert "embeddings" in result
-            assert isinstance(result["embeddings"], list)
-        except JigsawStackError as e:
-            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
-
-
 class TestEmbeddingV2Sync:
     """Test synchronous Embedding V2 methods"""