chore: formatting.

Author: Khurdhula-Harshavardhan

jigsawstack/async_request.py

@@ -184,11 +184,11 @@ def __get_headers(self) -> Dict[str, str]:
             h["x-jigsaw-no-request-log"] = "true"
 
         _headers = h.copy()
-        
-        #don't override Content-Type if using multipart
+
+        # don't override Content-Type if using multipart
         if self.files and "Content-Type" in self.headers:
             self.headers.pop("Content-Type")
-            
+
         _headers.update(self.headers)
 
         return _headers
@@ -251,38 +251,29 @@ async def make_request(
         _form_data = None
 
         if verb.lower() in ["get", "delete"]:
-            #convert params for URL encoding if needed
+            # convert params for URL encoding if needed
             _params = self.__convert_params(params)
         elif files:
             # for multipart requests - matches request.py behavior
             _form_data = aiohttp.FormData()
-            
+
             # add file(s) to form data
             for field_name, file_data in files.items():
                 if isinstance(file_data, bytes):
                     # just pass the blob without filename
                     _form_data.add_field(
-                        field_name,
-                        BytesIO(file_data),
-                        content_type="application/octet-stream"
+                        field_name, BytesIO(file_data), content_type="application/octet-stream"
                     )
                 elif isinstance(file_data, tuple):
                     # if tuple format (filename, data, content_type)
                     filename, content, content_type = file_data
                     _form_data.add_field(
-                        field_name,
-                        content,
-                        filename=filename,
-                        content_type=content_type
+                        field_name, content, filename=filename, content_type=content_type
                     )
-            
+
             # add params as 'body' field in multipart form (JSON stringified)
             if params and isinstance(params, dict):
-                _form_data.add_field(
-                    "body",
-                    json.dumps(params),
-                    content_type="application/json"
-                )
+                _form_data.add_field("body", json.dumps(params), content_type="application/json")
         elif data:
             # for binary data without multipart
             _data = data

jigsawstack/audio.py

@@ -80,8 +80,8 @@ def speech_to_text(
     ) -> Union[SpeechToTextResponse, SpeechToTextWebhookResponse]:
         options = options or {}
         path = "/ai/transcribe"
-        params= options or {}
-        if isinstance(blob, dict): 
+        params = options or {}
+        if isinstance(blob, dict):
             # URL or file_store_key based request
             resp = Request(
                 config=self.config,
@@ -143,7 +143,7 @@ async def speech_to_text(
                 verb="post",
             ).perform_with_content()
             return resp
-        
+
         files = {"file": blob}
         resp = await AsyncRequest(
             config=self.config,

jigsawstack/embedding.py

@@ -66,7 +66,6 @@ def execute(
             ).perform_with_content()
             return resp
 
-        
         files = {"file": blob}
         resp = Request(
             config=self.config,

jigsawstack/request.py

@@ -158,7 +158,7 @@ def __get_headers(self) -> Dict[Any, Any]:
             "Accept": "application/json",
             "x-api-key": f"{self.api_key}",
         }
-        
+
         # Only add Content-Type if not using multipart (files)
         if not self.files and not self.data:
             h["Content-Type"] = "application/json"
@@ -167,11 +167,11 @@ def __get_headers(self) -> Dict[Any, Any]:
             h["x-jigsaw-no-request-log"] = "true"
 
         _headers = h.copy()
-        
+
         # Don't override Content-Type if using multipart
         if self.files and "Content-Type" in self.headers:
             self.headers.pop("Content-Type")
-            
+
         _headers.update(self.headers)
 
         return _headers

jigsawstack/translate.py

@@ -108,7 +108,6 @@ def image(
             ).perform_with_content()
             return resp
 
-        
         files = {"file": blob}
         resp = Request(
             config=self.config,

jigsawstack/validate.py

@@ -111,7 +111,7 @@ def nsfw(
                 verb="post",
             ).perform_with_content()
             return resp
-    
+
         files = {"file": blob}
         resp = Request(
             config=self.config,
@@ -198,7 +198,7 @@ async def nsfw(
                 verb="post",
             ).perform_with_content()
             return resp
-        
+
         files = {"file": blob}
         resp = await AsyncRequest(
             config=self.config,

jigsawstack/vision.py

@@ -160,7 +160,7 @@ class OCRResponse(BaseResponse):
     tags: List[str]
     has_text: bool
     sections: List[object]
-    total_pages: Optional[int] 
+    total_pages: Optional[int]
     page_range: Optional[
         List[int]
     ]  # Only available if page_range is set in the request parameters.
@@ -205,7 +205,7 @@ def vocr(
             ).perform_with_content()
             return resp
 
-        files ={"file": blob}
+        files = {"file": blob}
         resp = Request(
             config=self.config,
             path=path,

tests/test_vocr.py

@@ -13,8 +13,12 @@
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-jigsaw = jigsawstack.JigsawStack(api_url="http://localhost:3000/api/", api_key=os.getenv("JIGSAWSTACK_API_KEY"))
-async_jigsaw = jigsawstack.AsyncJigsawStack(api_url="http://localhost:3000/api/", api_key=os.getenv("JIGSAWSTACK_API_KEY"))
+jigsaw = jigsawstack.JigsawStack(
+    api_url="http://localhost:3000/api/", api_key=os.getenv("JIGSAWSTACK_API_KEY")
+)
+async_jigsaw = jigsawstack.AsyncJigsawStack(
+    api_url="http://localhost:3000/api/", api_key=os.getenv("JIGSAWSTACK_API_KEY")
+)
 
 IMAGE_URL = "https://jigsawstack.com/preview/vocr-example.jpg"
 
@@ -46,7 +50,7 @@
             "prompt": [
                 "What is the main heading?",
                 "Extract any dates mentioned",
-                "What are the key points?"
+                "What are the key points?",
             ]
         },
     },
@@ -57,25 +61,19 @@
             "prompt": {
                 "title": "Extract the main title",
                 "content": "What is the main content?",
-                "metadata": "Extract any metadata or additional information"
+                "metadata": "Extract any metadata or additional information",
             }
         },
     },
     {
         "name": "url_with_string_prompt",
-        "params": {
-            "url": IMAGE_URL,
-            "prompt": "Summarize the text content"
-        },
+        "params": {"url": IMAGE_URL, "prompt": "Summarize the text content"},
         "blob": None,
         "options": None,
     },
     {
         "name": "url_with_list_prompt",
-        "params": {
-            "url": IMAGE_URL,
-            "prompt": ["Extract headers", "Extract body text"]
-        },
+        "params": {"url": IMAGE_URL, "prompt": ["Extract headers", "Extract body text"]},
         "blob": None,
         "options": None,
     },
@@ -85,31 +83,20 @@
 PDF_TEST_CASES = [
     {
         "name": "pdf_with_page_range",
-        "params": {
-            "url": PDF_URL,
-            "page_range": [1, 3],
-            "prompt": "Extract text from these pages"
-        },
+        "params": {"url": PDF_URL, "page_range": [1, 3], "prompt": "Extract text from these pages"},
         "blob": None,
         "options": None,
     },
     {
         "name": "pdf_single_page",
-        "params": {
-            "url": PDF_URL,
-            "page_range": [1, 1],
-            "prompt": "What is on the first page?"
-        },
+        "params": {"url": PDF_URL, "page_range": [1, 1], "prompt": "What is on the first page?"},
         "blob": None,
         "options": None,
     },
     {
         "name": "pdf_blob_with_page_range",
         "blob": PDF_URL,
-        "options": {
-            "page_range": [1, 3],
-            "prompt": "what is this about?"
-        },
+        "options": {"page_range": [1, 3], "prompt": "what is this about?"},
     },
 ]
 
@@ -135,7 +122,7 @@ def test_vocr(self, test_case):
                 result = jigsaw.vision.vocr(test_case["params"])
 
             print(f"Test {test_case['name']}: Success={result.get('success')}")
-            
+
             # Verify response structure
             assert result["success"] is True
             if "prompt" in (test_case.get("params") or {}):
@@ -147,13 +134,11 @@ def test_vocr(self, test_case):
             assert isinstance(result["tags"], list)
             assert "sections" in result
             assert isinstance(result["sections"], list)
-            
+
         except JigsawStackError as e:
             pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
 
-    @pytest.mark.parametrize(
-        "test_case", pdf_test_cases, ids=[tc["name"] for tc in pdf_test_cases]
-    )
+    @pytest.mark.parametrize("test_case", pdf_test_cases, ids=[tc["name"] for tc in pdf_test_cases])
     def test_vocr_pdf(self, test_case):
         """Test synchronous VOCR with PDF inputs"""
         try:
@@ -164,19 +149,21 @@ def test_vocr_pdf(self, test_case):
             else:
                 # Use params directly
                 result = jigsaw.vision.vocr(test_case["params"])
-            
+
             # Verify response structure
             assert result["success"] is True
             if "prompt" in (test_case.get("params") or {}):
                 assert "context" in result
             assert "total_pages" in result
-            
-            if test_case.get("params", {}).get("page_range") or test_case.get("options", {}).get("page_range"):
+
+            if test_case.get("params", {}).get("page_range") or test_case.get("options", {}).get(
+                "page_range"
+            ):
                 assert "page_range" in result
                 assert isinstance(result["page_range"], list)
 
             logger.info(f"Test {test_case['name']}: total_pages={result.get('total_pages')}")
-            
+
         except JigsawStackError as e:
             pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
 
@@ -197,15 +184,13 @@ async def test_vocr_async(self, test_case):
             if test_case.get("blob"):
                 # Download blob content
                 blob_content = requests.get(test_case["blob"]).content
-                result = await async_jigsaw.vision.vocr(
-                    blob_content, test_case.get("options", {})
-                )
+                result = await async_jigsaw.vision.vocr(blob_content, test_case.get("options", {}))
             else:
                 # Use params directly
                 result = await async_jigsaw.vision.vocr(test_case["params"])
 
             print(f"Test {test_case['name']}: Success={result.get('success')}")
-            
+
             # Verify response structure
             assert result["success"] is True
             if "prompt" in (test_case.get("params") or {}):
@@ -217,44 +202,44 @@ async def test_vocr_async(self, test_case):
             assert isinstance(result["tags"], list)
             assert "sections" in result
             assert isinstance(result["sections"], list)
-            
+
             # Log some details
-            logger.info(f"Test {test_case['name']}: has_text={result['has_text']}, tags={result['tags'][:3] if result['tags'] else []}")
-            
+            logger.info(
+                f"Test {test_case['name']}: has_text={result['has_text']}, tags={result['tags'][:3] if result['tags'] else []}"
+            )
+
         except JigsawStackError as e:
             pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
 
-    @pytest.mark.parametrize(
-        "test_case", pdf_test_cases, ids=[tc["name"] for tc in pdf_test_cases]
-    )
+    @pytest.mark.parametrize("test_case", pdf_test_cases, ids=[tc["name"] for tc in pdf_test_cases])
     @pytest.mark.asyncio
     async def test_vocr_pdf_async(self, test_case):
         """Test asynchronous VOCR with PDF inputs"""
         try:
             if test_case.get("blob"):
                 # Download blob content
                 blob_content = requests.get(test_case["blob"]).content
-                result = await async_jigsaw.vision.vocr(
-                    blob_content, test_case.get("options", {})
-                )
+                result = await async_jigsaw.vision.vocr(blob_content, test_case.get("options", {}))
             else:
                 # Use params directly
                 result = await async_jigsaw.vision.vocr(test_case["params"])
 
             print(f"Test {test_case['name']}: Success={result.get('success')}")
-            
+
             # Verify response structure
             assert result["success"] is True
             if "prompt" in (test_case.get("params") or {}):
                 assert "context" in result
             assert "total_pages" in result  # PDF specific
-            
+
             # Check if page_range is in response when requested
-            if test_case.get("params", {}).get("page_range") or test_case.get("options", {}).get("page_range"):
+            if test_case.get("params", {}).get("page_range") or test_case.get("options", {}).get(
+                "page_range"
+            ):
                 assert "page_range" in result
                 assert isinstance(result["page_range"], list)
 
             logger.info(f"Test {test_case['name']}: total_pages={result.get('total_pages')}")
-            
+
         except JigsawStackError as e:
-            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")
+            pytest.fail(f"Unexpected JigsawStackError in {test_case['name']}: {e}")