ragaai-catalyst 2.1.5b20__tar.gz → 2.1.5b22__tar.gz

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/PKG-INFO

@@ -1,8 +1,8 @@
 Metadata-Version: 2.2
 Name: ragaai_catalyst
-Version: 2.1.5b20
+Version: 2.1.5b22
 Summary: RAGA AI CATALYST
-Author-email: Kiran Scaria <kiran.scaria@raga.ai>, Siddhartha Kosti <siddhartha.kosti@raga.ai>, Vijay Chaurasia <vijay.chaurasia@raga.ai>, Tanaya Pakhale <tanaya.pakhale@raga.ai>, Ritika Goel <ritika.goel@raga.ai>, Kedar Gaikwad <kedar.gaikwad@raga.ai>, Dushyant Mahajan <dushyant.mahajan@raga.ai>
+Author-email: Kiran Scaria <kiran.scaria@raga.ai>, Kedar Gaikwad <kedar.gaikwad@raga.ai>, Dushyant Mahajan <dushyant.mahajan@raga.ai>, Siddhartha Kosti <siddhartha.kosti@raga.ai>, Ritika Goel <ritika.goel@raga.ai>, Vijay Chaurasia <vijay.chaurasia@raga.ai>
 Requires-Python: <3.13,>=3.9
 Description-Content-Type: text/markdown
 License-File: LICENSE

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/examples/travel_agent/agents.py

@@ -20,6 +20,13 @@ class ItineraryAgent:
             cost=0.01,
             latency=0.5,
         )
+        current_span().add_metrics(
+            name="itinerary_planning",
+            score=0.8,
+            reasoning="Planning comprehensive travel itinerary",
+            cost=0.01,
+            latency=0.5,
+        )
         
         # Get weather information
         weather = weather_tool(user_preferences["destination"])

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/pyproject.toml

@@ -9,15 +9,14 @@ readme = "README.md"
 requires-python = ">=3.9,<3.13"
 # license = {file = "LICENSE"}
 
-version = "2.1.5.beta.20"
+version = "2.1.5.beta.22"
 authors = [
     {name = "Kiran Scaria", email = "kiran.scaria@raga.ai"},
-    {name = "Siddhartha Kosti", email = "siddhartha.kosti@raga.ai"},
-    {name = "Vijay Chaurasia", email="vijay.chaurasia@raga.ai"},
-    {name = "Tanaya Pakhale", email = "tanaya.pakhale@raga.ai"},
-    {name = "Ritika Goel", email = "ritika.goel@raga.ai"},
     {name = "Kedar Gaikwad", email = "kedar.gaikwad@raga.ai"},
     {name = "Dushyant Mahajan", email = "dushyant.mahajan@raga.ai"},
+    {name = "Siddhartha Kosti", email = "siddhartha.kosti@raga.ai"},
+    {name = "Ritika Goel", email = "ritika.goel@raga.ai"},
+    {name = "Vijay Chaurasia", email="vijay.chaurasia@raga.ai"}
 ]
 
 dependencies = [

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/ragaai_catalyst/dataset.py

@@ -9,6 +9,10 @@ import pandas as pd
 logger = logging.getLogger(__name__)
 get_token = RagaAICatalyst.get_token
 
+# Job status constants
+JOB_STATUS_FAILED = "failed"
+JOB_STATUS_IN_PROGRESS = "in_progress"
+JOB_STATUS_COMPLETED = "success"
 
 class Dataset:
     BASE_URL = None
@@ -18,6 +22,7 @@ class Dataset:
         self.project_name = project_name
         self.num_projects = 99999
         Dataset.BASE_URL = RagaAICatalyst.BASE_URL
+        self.jobId = None
         headers = {
             "Authorization": f'Bearer {os.getenv("RAGAAI_CATALYST_TOKEN")}',
         }
@@ -219,7 +224,6 @@ class Dataset:
         try:
 
             put_csv_response = put_csv_to_presignedUrl(url)
-            print(put_csv_response)
             if put_csv_response.status_code not in (200, 201):
                 raise ValueError('Unable to put csv to the presignedUrl')
         except Exception as e:
@@ -269,6 +273,7 @@ class Dataset:
                 raise ValueError('Unable to upload csv')
             else:
                 print(upload_csv_response['message'])
+                self.jobId = upload_csv_response['data']['jobId']
         except Exception as e:
             logger.error(f"Error in create_from_csv: {e}")
             raise
@@ -436,6 +441,7 @@ class Dataset:
             response_data = response.json()
             if response_data.get('success', False):
                 print(f"{response_data['message']}")
+                self.jobId = response_data['data']['jobId']
             else:
                 raise ValueError(response_data.get('message', 'Failed to add rows'))
         
@@ -594,6 +600,7 @@ class Dataset:
             
             if response_data.get('success', False):
                 print(f"Column '{column_name}' added successfully to dataset '{dataset_name}'")
+                self.jobId = response_data['data']['jobId']
             else:
                 raise ValueError(response_data.get('message', 'Failed to add column'))
         
@@ -601,3 +608,49 @@ class Dataset:
             print(f"Error adding column: {e}")
             raise
 
+    def get_status(self):
+        headers = {
+            'Content-Type': 'application/json',
+            "Authorization": f"Bearer {os.getenv('RAGAAI_CATALYST_TOKEN')}",
+            'X-Project-Id': str(self.project_id),
+        }
+        try:
+            response = requests.get(
+                f'{Dataset.BASE_URL}/job/status', 
+                headers=headers, 
+                timeout=30)
+            response.raise_for_status()
+            if response.json()["success"]:
+
+                status_json = [item["status"] for item in response.json()["data"]["content"] if item["id"]==self.jobId]
+                status_json = status_json[0]
+                if status_json == "Failed":
+                    print("Job failed. No results to fetch.")
+                    return JOB_STATUS_FAILED
+                elif status_json == "In Progress":
+                    print(f"Job in progress. Please wait while the job completes.\nVisit Job Status: {Dataset.BASE_URL.removesuffix('/api')}/projects/job-status?projectId={self.project_id} to track")
+                    return JOB_STATUS_IN_PROGRESS
+                elif status_json == "Completed":
+                    print(f"Job completed. Fetching results.\nVisit Job Status: {Dataset.BASE_URL.removesuffix('/api')}/projects/job-status?projectId={self.project_id} to check")
+                    return JOB_STATUS_COMPLETED
+                else:
+                    logger.error(f"Unknown status received: {status_json}")
+                    return JOB_STATUS_FAILED
+            else:
+                logger.error("Request was not successful")
+                return JOB_STATUS_FAILED
+        except requests.exceptions.HTTPError as http_err:
+            logger.error(f"HTTP error occurred: {http_err}")
+            return JOB_STATUS_FAILED
+        except requests.exceptions.ConnectionError as conn_err:
+            logger.error(f"Connection error occurred: {conn_err}")
+            return JOB_STATUS_FAILED
+        except requests.exceptions.Timeout as timeout_err:
+            logger.error(f"Timeout error occurred: {timeout_err}")
+            return JOB_STATUS_FAILED
+        except requests.exceptions.RequestException as req_err:
+            logger.error(f"An error occurred: {req_err}")
+            return JOB_STATUS_FAILED
+        except Exception as e:
+            logger.error(f"An unexpected error occurred: {e}")
+            return JOB_STATUS_FAILED

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/ragaai_catalyst/synthetic_data_generation.py

@@ -8,7 +8,9 @@ import markdown
 import pandas as pd
 import json
 from litellm import completion
+import litellm
 from tqdm import tqdm
+import tiktoken
 # import internal_api_completion
 # import proxy_call
 from .internal_api_completion import api_completion as internal_api_completion
@@ -48,13 +50,18 @@ class SyntheticDataGeneration:
         Raises:
             ValueError: If an invalid provider is specified or API key is missing.
         """
+        text_validity = self.validate_input(text)
+        if text_validity:
+            raise ValueError(text_validity)
+
         BATCH_SIZE = 5  # Optimal batch size for maintaining response quality
         provider = model_config.get("provider")
         model = model_config.get("model")
         api_base = model_config.get("api_base")
+        api_version = model_config.get("api_version")
 
         # Initialize the appropriate client based on provider
-        self._initialize_client(provider, api_key, api_base, internal_llm_proxy=kwargs.get("internal_llm_proxy", None))
+        self._initialize_client(provider, api_key, api_base, api_version, internal_llm_proxy=kwargs.get("internal_llm_proxy", None))
 
         # Initialize progress bar
         pbar = tqdm(total=n, desc="Generating QA pairs")
@@ -88,7 +95,7 @@ class SyntheticDataGeneration:
                     pbar.update(len(batch_df))
                     
             except Exception as e:
-                print(f"Batch generation failed.")
+                print(f"Batch generation failed:{str(e)}")
 
                 if any(error in str(e) for error in FAILURE_CASES):
                     raise Exception(f"{e}")
@@ -139,7 +146,7 @@ class SyntheticDataGeneration:
         
         return final_df
 
-    def _initialize_client(self, provider, api_key, api_base=None, internal_llm_proxy=None):
+    def _initialize_client(self, provider, api_key, api_base=None, api_version=None, internal_llm_proxy=None):
         """Initialize the appropriate client based on provider."""
         if not provider:
             raise ValueError("Model configuration must be provided with a valid provider and model.")
@@ -158,7 +165,17 @@ class SyntheticDataGeneration:
             if api_key is None and os.getenv("OPENAI_API_KEY") is None and internal_llm_proxy is None:
                 raise ValueError("API key must be provided for OpenAI.")
             openai.api_key = api_key or os.getenv("OPENAI_API_KEY")
-        
+
+        elif provider == "azure":
+            if api_key is None and os.getenv("AZURE_API_KEY") is None and internal_llm_proxy is None:
+                raise ValueError("API key must be provided for Azure.")
+            litellm.api_key = api_key or os.getenv("AZURE_API_KEY")
+            if api_base is None and os.getenv("AZURE_API_BASE") is None and internal_llm_proxy is None:
+                raise ValueError("API Base must be provided for Azure.")
+            litellm.api_base = api_base or os.getenv("AZURE_API_BASE")
+            if api_version is None and os.getenv("AZURE_API_VERSION") is None and internal_llm_proxy is None:
+                raise ValueError("API version must be provided for Azure.")
+            litellm.api_version = api_version or os.getenv("AZURE_API_VERSION")
         else:
             raise ValueError(f"Provider is not recognized.")
 
@@ -189,7 +206,15 @@ class SyntheticDataGeneration:
             kwargs=kwargs
         )
 
+    def validate_input(self,text):
 
+        if not text.strip():
+            return 'Empty Text provided for qna generation. Please provide valid text'
+        encoding = tiktoken.encoding_for_model("gpt-4")
+        tokens = encoding.encode(text)
+        if len(tokens)<5:
+            return 'Very Small Text provided for qna generation. Please provide longer text'
+        return False
         
           
     def _get_system_message(self, question_type, n):
@@ -274,10 +299,14 @@ class SyntheticDataGeneration:
         # Add optional parameters if they exist in model_config
         if "api_base" in model_config:
             completion_params["api_base"] = model_config["api_base"]
+        if "api_version" in model_config:
+            completion_params["api_version"] = model_config["api_version"]
         if "max_tokens" in model_config:
             completion_params["max_tokens"] = model_config["max_tokens"]
         if "temperature" in model_config:
             completion_params["temperature"] = model_config["temperature"]
+        if 'provider' in model_config:
+            completion_params['model'] = f'{model_config["provider"]}/{model_config["model"]}'
 
         # Make the API call using LiteLLM
         try:
@@ -318,9 +347,13 @@ class SyntheticDataGeneration:
             list_start_index = data.find('[')  # Find the index of the first '['
             substring_data = data[list_start_index:] if list_start_index != -1 else data  # Slice from the list start
             data = substring_data
-
+        elif provider == "azure":
+            data = response.choices[0].message.content.replace('\n', '')
+            list_start_index = data.find('[')  # Find the index of the first '['
+            substring_data = data[list_start_index:] if list_start_index != -1 else data  # Slice from the list start
+            data = substring_data
         else:
-            raise ValueError("Invalid provider. Choose 'groq', 'gemini', or 'openai'.")
+            raise ValueError("Invalid provider. Choose 'groq', 'gemini', 'azure' or 'openai'.")
         try:
             json_data = json.loads(data)
             return pd.DataFrame(json_data)

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/ragaai_catalyst/tracers/agentic_tracing/tracers/agent_tracer.py

@@ -101,7 +101,10 @@ class AgentTracerMixin:
                 original_init = target.__init__
 
                 def wrapped_init(self, *args, **kwargs):
-                    self.gt = kwargs.get("gt", None) if kwargs else None
+                    gt = kwargs.get("gt") if kwargs else None
+                    if gt is not None:
+                        span = self.span(name)
+                        span.add_gt(gt)
                     # Set agent context before initializing
                     component_id = str(uuid.uuid4())
                     hash_id = top_level_hash_id
@@ -159,7 +162,10 @@ class AgentTracerMixin:
                                 @self.file_tracker.trace_decorator
                                 @functools.wraps(method)
                                 def wrapped_method(self, *args, **kwargs):
-                                    self.gt = kwargs.get("gt", None) if kwargs else None
+                                    gt = kwargs.get("gt") if kwargs else None
+                                    if gt is not None:
+                                        span = tracer.span(name)
+                                        span.add_gt(gt)
                                     # Set this agent as current during method execution
                                     token = tracer.current_agent_id.set(
                                         self._agent_component_id
@@ -247,6 +253,7 @@ class AgentTracerMixin:
                             agent_type,
                             version,
                             capabilities,
+                            top_level_hash_id,
                             *args,
                             **kwargs,
                         )
@@ -256,10 +263,9 @@ class AgentTracerMixin:
         return decorator
 
     def _trace_sync_agent_execution(
-        self, func, name, agent_type, version, capabilities, *args, **kwargs
+        self, func, name, agent_type, version, capabilities, top_level_hash_id, *args, **kwargs
     ):
-        # Generate a unique hash_id for this execution context
-        hash_id = str(uuid.uuid4())
+        hash_id = top_level_hash_id
 
         """Synchronous version of agent tracing"""
         if not self.is_active:
@@ -275,7 +281,10 @@ class AgentTracerMixin:
         component_id = str(uuid.uuid4())
 
         # Extract ground truth if present
-        ground_truth = kwargs.pop("gt", None) if kwargs else None
+        ground_truth = kwargs.pop("gt") if kwargs else None
+        if ground_truth is not None:
+            span = self.span(name)
+            span.add_gt(ground_truth)
 
         # Get parent agent ID if exists
         parent_agent_id = self.current_agent_id.get()
@@ -293,7 +302,7 @@ class AgentTracerMixin:
 
         try:
             # Execute the agent
-            result = func(*args, **kwargs)
+            result = self.file_tracker.trace_wrapper(func)(*args, **kwargs)
 
             # Calculate resource usage
             end_memory = psutil.Process().memory_info().rss
@@ -320,9 +329,6 @@ class AgentTracerMixin:
                 children=children,
                 parent_id=parent_agent_id,
             )
-            # Add ground truth to component data if present
-            if ground_truth is not None:
-                agent_component["data"]["gt"] = ground_truth
 
             # Add this component as a child to parent's children list
             parent_children.append(agent_component)
@@ -398,7 +404,10 @@ class AgentTracerMixin:
         component_id = str(uuid.uuid4())
 
         # Extract ground truth if present
-        ground_truth = kwargs.pop("gt", None) if kwargs else None
+        ground_truth = kwargs.pop("gt") if kwargs else None
+        if ground_truth is not None:
+            span = self.span(name)
+            span.add_gt(ground_truth)
 
         # Get parent agent ID if exists
         parent_agent_id = self.current_agent_id.get()
@@ -414,7 +423,7 @@ class AgentTracerMixin:
 
         try:
             # Execute the agent
-            result = await func(*args, **kwargs)
+            result = await self.file_tracker.trace_wrapper(func)(*args, **kwargs)
 
             # Calculate resource usage
             end_memory = psutil.Process().memory_info().rss
@@ -441,10 +450,6 @@ class AgentTracerMixin:
                 parent_id=parent_agent_id,
             )
 
-            # Add ground truth to component data if present
-            if ground_truth is not None:
-                agent_component["data"]["gt"] = ground_truth
-
             # Add this component as a child to parent's children list
             parent_children.append(agent_component)
             self.agent_children.set(parent_children)
@@ -576,8 +581,13 @@ class AgentTracerMixin:
             "interactions": interactions,
         }
 
-        if self.gt:
-            component["data"]["gt"] = self.gt
+        if name in self.span_attributes_dict:
+            span_gt = self.span_attributes_dict[name].gt
+            if span_gt is not None:
+                component["data"]["gt"] = span_gt
+            span_context = self.span_attributes_dict[name].context
+            if span_context:
+                component["data"]["context"] = span_context
 
         # Reset the SpanAttributes context variable
         self.span_attributes_dict[kwargs["name"]] = SpanAttributes(kwargs["name"])

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/ragaai_catalyst/tracers/agentic_tracing/tracers/base.py

@@ -83,6 +83,7 @@ class BaseTracer:
         self.tracking_thread = None
         self.tracking = False
         self.system_monitor = None
+        self.gt = None
 
     def _get_system_info(self) -> SystemInfo:
         return self.system_monitor.get_system_info()
@@ -249,7 +250,8 @@ class BaseTracer:
 
             # Format interactions and add to trace
             interactions = self.format_interactions()
-            trace_data["workflow"] = interactions["workflow"]
+            # trace_data["workflow"] = interactions["workflow"]
+            cleaned_trace_data["workflow"] = interactions["workflow"]
 
             with open(filepath, "w") as f:
                 json.dump(cleaned_trace_data, f, cls=TracerJSONEncoder, indent=2)

{ragaai_catalyst-2.1.5b20 → ragaai_catalyst-2.1.5b22}/ragaai_catalyst/tracers/agentic_tracing/tracers/custom_tracer.py

@@ -45,7 +45,10 @@ class CustomTracerMixin:
             @functools.wraps(func)
             async def async_wrapper(*args, **kwargs):
                 async_wrapper.metadata = metadata
-                self.gt = kwargs.get('gt', None) if kwargs else None
+                gt = kwargs.get('gt') if kwargs else None
+                if gt is not None:
+                    span = self.span(name)
+                    span.add_gt(gt)
                 return await self._trace_custom_execution(
                     func, name or func.__name__, custom_type, version, trace_variables, *args, **kwargs
                 )
@@ -54,7 +57,10 @@ class CustomTracerMixin:
             @functools.wraps(func)
             def sync_wrapper(*args, **kwargs):
                 sync_wrapper.metadata = metadata
-                self.gt = kwargs.get('gt', None) if kwargs else None
+                gt = kwargs.get('gt') if kwargs else None
+                if gt is not None:
+                    span = self.span(name)
+                    span.add_gt(gt)
                 return self._trace_sync_custom_execution(
                     func, name or func.__name__, custom_type, version, trace_variables, *args, **kwargs
                 )
@@ -98,7 +104,7 @@ class CustomTracerMixin:
 
         try:
             # Execute the function
-            result = func(*args, **kwargs)
+            result = self.file_tracker.trace_wrapper(func)(*args, **kwargs)
 
             # Calculate resource usage
             end_time = datetime.now().astimezone().isoformat()
@@ -186,7 +192,7 @@ class CustomTracerMixin:
 
         try:
             # Execute the function
-            result = await func(*args, **kwargs)
+            result = await self.file_tracker.trace_wrapper(func)(*args, **kwargs)
 
             # Calculate resource usage
             end_time = datetime.now().astimezone().isoformat()
@@ -284,9 +290,13 @@ class CustomTracerMixin:
             "interactions": interactions
         }
 
-        if self.gt:
-            component["data"]["gt"] = self.gt
-
+        if kwargs["name"] in self.span_attributes_dict:
+            span_gt = self.span_attributes_dict[kwargs["name"]].gt
+            if span_gt is not None:
+                component["data"]["gt"] = span_gt
+            span_context = self.span_attributes_dict[kwargs["name"]].context
+            if span_context:
+                component["data"]["context"] = span_context
         return component
 
     def start_component(self, component_id):