payi 0.1.0a80__tar.gz → 0.1.0a82__tar.gz

payi-0.1.0a82/.release-please-manifest.json

@@ -0,0 +1,3 @@
+{
+  ".": "0.1.0-alpha.82"
+}

{payi-0.1.0a80 → payi-0.1.0a82}/CHANGELOG.md

@@ -1,5 +1,21 @@
 # Changelog
 
+## 0.1.0-alpha.82 (2025-06-06)
+
+Full Changelog: [v0.1.0-alpha.81...v0.1.0-alpha.82](https://github.com/Pay-i/pay-i-python/compare/v0.1.0-alpha.81...v0.1.0-alpha.82)
+
+### Bug Fixes
+
+* vertex, google.genai report tokens when details missing ([#314](https://github.com/Pay-i/pay-i-python/issues/314)) ([54dd25b](https://github.com/Pay-i/pay-i-python/commit/54dd25b1abd93e6fdec22dd7bb8dae39236ee532))
+
+## 0.1.0-alpha.81 (2025-06-06)
+
+Full Changelog: [v0.1.0-alpha.80...v0.1.0-alpha.81](https://github.com/Pay-i/pay-i-python/compare/v0.1.0-alpha.80...v0.1.0-alpha.81)
+
+### Features
+
+* logger support ([#312](https://github.com/Pay-i/pay-i-python/issues/312)) ([f353bb2](https://github.com/Pay-i/pay-i-python/commit/f353bb2cbc24d28114c3bd770a8243f8b115d1fc))
+
 ## 0.1.0-alpha.80 (2025-06-04)
 
 Full Changelog: [v0.1.0-alpha.79...v0.1.0-alpha.80](https://github.com/Pay-i/pay-i-python/compare/v0.1.0-alpha.79...v0.1.0-alpha.80)

{payi-0.1.0a80 → payi-0.1.0a82}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: payi
-Version: 0.1.0a80
+Version: 0.1.0a82
 Summary: The official Python library for the payi API
 Project-URL: Homepage, https://github.com/Pay-i/pay-i-python
 Project-URL: Repository, https://github.com/Pay-i/pay-i-python

{payi-0.1.0a80 → payi-0.1.0a82}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "payi"
-version = "0.1.0-alpha.80"
+version = "0.1.0-alpha.82"
 description = "The official Python library for the payi API"
 dynamic = ["readme"]
 license = "Apache-2.0"

{payi-0.1.0a80 → payi-0.1.0a82}/src/payi/_version.py

@@ -1,4 +1,4 @@
 # File generated from our OpenAPI spec by Stainless. See CONTRIBUTING.md for details.
 
 __title__ = "payi"
-__version__ = "0.1.0-alpha.80"  # x-release-please-version
+__version__ = "0.1.0-alpha.82"  # x-release-please-version

{payi-0.1.0a80 → payi-0.1.0a82}/src/payi/lib/AnthropicInstrumentor.py

@@ -1,4 +1,3 @@
-import logging
 from typing import Any, Union, Optional, Sequence
 from typing_extensions import override
 
@@ -48,7 +47,7 @@ class AnthropicInstrumentor:
             )
 
         except Exception as e:
-            logging.debug(f"Error instrumenting anthropic: {e}")
+            instrumentor._logger.debug(f"Error instrumenting anthropic: {e}")
             return
 
 
@@ -60,6 +59,7 @@ def messages_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("Anthropic messages wrapper")
     return instrumentor.invoke_wrapper(
         _AnthropicProviderRequest(instrumentor=instrumentor, streaming_type=_StreamingType.iterator, instance=instance),
         _IsStreaming.kwargs,
@@ -77,6 +77,7 @@ def stream_messages_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("Anthropic stream wrapper")
     return instrumentor.invoke_wrapper(
         _AnthropicProviderRequest(instrumentor=instrumentor, streaming_type=_StreamingType.stream_manager, instance=instance),
         _IsStreaming.true,
@@ -94,6 +95,7 @@ async def amessages_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("aync Anthropic messages wrapper")
     return await instrumentor.async_invoke_wrapper(
         _AnthropicProviderRequest(instrumentor=instrumentor, streaming_type=_StreamingType.iterator, instance=instance),
         _IsStreaming.kwargs,
@@ -111,6 +113,7 @@ async def astream_messages_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("aync Anthropic stream wrapper")
     return await instrumentor.async_invoke_wrapper(
         _AnthropicProviderRequest(instrumentor=instrumentor, streaming_type=_StreamingType.stream_manager, instance=instance),
         _IsStreaming.true,
@@ -202,7 +205,7 @@ class _AnthropicProviderRequest(_ProviderRequest):
                     self._estimated_prompt_tokens = estimated_token_count
      
             except Exception:
-                logging.warning("Error getting encoding for cl100k_base")
+                self._instrumentor._logger.warning("Error getting encoding for cl100k_base")
 
         return True
 
@@ -233,7 +236,7 @@ class _AnthropicProviderRequest(_ProviderRequest):
                         self._ingest["provider_response_json"] = text
 
         except Exception as e:
-            logging.debug(f"Error processing exception: {e}")
+            self._instrumentor._logger.debug(f"Error processing exception: {e}")
             return False
 
         return True

{payi-0.1.0a80 → payi-0.1.0a82}/src/payi/lib/BedrockInstrumentor.py

@@ -1,6 +1,5 @@
 import os
 import json
-import logging
 from typing import Any, Sequence
 from functools import wraps
 from typing_extensions import override
@@ -38,12 +37,13 @@ class BedrockInstrumentor:
             )
 
         except Exception as e:
-            logging.debug(f"Error instrumenting bedrock: {e}")
+            instrumentor._logger.debug(f"Error instrumenting bedrock: {e}")
             return
 
 @_PayiInstrumentor.payi_wrapper
 def create_client_wrapper(instrumentor: _PayiInstrumentor, wrapped: Any, instance: Any, *args: Any, **kwargs: Any) -> Any: #  noqa: ARG001
     if kwargs.get("service_name") != "bedrock-runtime":
+        instrumentor._logger.debug(f"skipping client wrapper creation for {kwargs.get('service_name', '')} service")
         return wrapped(*args, **kwargs)
 
     try:
@@ -53,13 +53,16 @@ def create_client_wrapper(instrumentor: _PayiInstrumentor, wrapped: Any, instanc
         client.converse = wrap_converse(instrumentor, client.converse)
         client.converse_stream = wrap_converse_stream(instrumentor, client.converse_stream)
 
+        instrumentor._logger.debug(f"Instrumented bedrock client")
+
         if BedrockInstrumentor._instrumentor._proxy_default:
             # Register client callbacks to handle the Pay-i extra_headers parameter in the inference calls and redirect the request to the Pay-i endpoint
             _register_bedrock_client_callbacks(client)
+            instrumentor._logger.debug(f"Registered bedrock client callbaks for proxy")
 
         return client
     except Exception as e:
-        logging.debug(f"Error instrumenting bedrock client: {e}")
+        instrumentor._logger.debug(f"Error instrumenting bedrock client: {e}")
     
     return wrapped(*args, **kwargs)
 
@@ -148,6 +151,7 @@ def wrap_invoke(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
         modelId:str = kwargs.get("modelId", "") # type: ignore
 
         if _is_supported_model(modelId):
+            instrumentor._logger.debug(f"bedrock invoke wrapper, modelId: {modelId}")
             return instrumentor.invoke_wrapper(
                 _BedrockInvokeSynchronousProviderRequest(instrumentor=instrumentor),
                 _IsStreaming.false,
@@ -156,6 +160,8 @@ def wrap_invoke(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
                 args,
                 kwargs,
             )   
+
+        instrumentor._logger.debug(f"bedrock invoke wrapper, unsupported modelId: {modelId}")
         return wrapped(*args, **kwargs)
     
     return invoke_wrapper
@@ -166,6 +172,7 @@ def wrap_invoke_stream(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
         modelId: str = kwargs.get("modelId", "") # type: ignore
 
         if _is_supported_model(modelId):
+            instrumentor._logger.debug(f"bedrock invoke stream wrapper, modelId: {modelId}")
             return instrumentor.invoke_wrapper(
                 _BedrockInvokeStreamingProviderRequest(instrumentor=instrumentor, model_id=modelId),
                 _IsStreaming.true,
@@ -174,6 +181,7 @@ def wrap_invoke_stream(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
                 args,
                 kwargs,
             )
+        instrumentor._logger.debug(f"bedrock invoke stream wrapper, unsupported modelId: {modelId}")
         return wrapped(*args, **kwargs)
 
     return invoke_wrapper
@@ -184,6 +192,7 @@ def wrap_converse(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
         modelId:str = kwargs.get("modelId", "") # type: ignore
 
         if _is_supported_model(modelId):
+            instrumentor._logger.debug(f"bedrock converse wrapper, modelId: {modelId}")
             return instrumentor.invoke_wrapper(
                 _BedrockConverseSynchronousProviderRequest(instrumentor=instrumentor),
                 _IsStreaming.false,
@@ -192,6 +201,7 @@ def wrap_converse(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
                 args,
                 kwargs,
         )
+        instrumentor._logger.debug(f"bedrock converse wrapper, unsupported modelId: {modelId}")
         return wrapped(*args, **kwargs)
     
     return invoke_wrapper
@@ -202,6 +212,7 @@ def wrap_converse_stream(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
         modelId: str = kwargs.get("modelId", "") # type: ignore
 
         if _is_supported_model(modelId):
+            instrumentor._logger.debug(f"bedrock converse stream wrapper, modelId: {modelId}")
             return instrumentor.invoke_wrapper(
                 _BedrockConverseStreamingProviderRequest(instrumentor=instrumentor),
                 _IsStreaming.true,
@@ -210,6 +221,7 @@ def wrap_converse_stream(instrumentor: _PayiInstrumentor, wrapped: Any) -> Any:
                 args,
                 kwargs,
             )
+        instrumentor._logger.debug(f"bedrock converse stream wrapper, unsupported modelId: {modelId}")
         return wrapped(*args, **kwargs)
 
     return invoke_wrapper
@@ -251,7 +263,7 @@ class _BedrockProviderRequest(_ProviderRequest):
             return True
 
         except Exception as e:
-            logging.debug(f"Error processing exception: {e}")
+            self._instrumentor._logger.debug(f"Error processing exception: {e}")
             return False
 
 class _BedrockInvokeStreamingProviderRequest(_BedrockProviderRequest):

{payi-0.1.0a80 → payi-0.1.0a82}/src/payi/lib/GoogleGenAiInstrumentor.py

@@ -1,6 +1,5 @@
 import json
 import math
-import logging
 from typing import Any, List, Union, Optional, Sequence
 from typing_extensions import override
 
@@ -41,7 +40,7 @@ class GoogleGenAiInstrumentor:
             )
 
         except Exception as e:
-            logging.debug(f"Error instrumenting vertex: {e}")
+            instrumentor._logger.debug(f"Error instrumenting vertex: {e}")
             return
 
 @_PayiInstrumentor.payi_wrapper
@@ -52,6 +51,7 @@ def generate_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("genai generate_content wrapper")
     return instrumentor.invoke_wrapper(
         _GoogleGenAiRequest(instrumentor),
         _IsStreaming.false,
@@ -69,6 +69,7 @@ def generate_stream_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("genai generate_content_stream wrapper")
     return instrumentor.invoke_wrapper(
         _GoogleGenAiRequest(instrumentor),
         _IsStreaming.true,
@@ -86,6 +87,7 @@ async def agenerate_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("async genai generate_content wrapper")
     return await instrumentor.async_invoke_wrapper(
         _GoogleGenAiRequest(instrumentor),
         _IsStreaming.false,
@@ -103,6 +105,7 @@ async def agenerate_stream_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("async genai generate_content_stream wrapper")
     return await instrumentor.async_invoke_wrapper(
         _GoogleGenAiRequest(instrumentor),
         _IsStreaming.true,
@@ -306,15 +309,15 @@ class _GoogleGenAiRequest(_ProviderRequest):
         usage = response_dict.get("usage_metadata", {})
         input = usage.get("prompt_token_count", 0)
 
-        prompt_tokens_details: list[dict[str, Any]] = usage.get("prompt_tokens_details")
-        candidates_tokens_details: list[dict[str, Any]] = usage.get("candidates_tokens_details")
+        prompt_tokens_details: list[dict[str, Any]] = usage.get("prompt_tokens_details", [])
+        candidates_tokens_details: list[dict[str, Any]] = usage.get("candidates_tokens_details", [])
 
         model: str = response_dict.get("model_version", "")
+        
+        # for character billing only
+        large_context = "" if input < 128000 else "_large_context"
 
         if self._is_character_billing_model(model):
-            # gemini 1.0 and 1.5 units are reported in characters, per second, per image, etc...
-            large_context = "" if input < 128000 else "_large_context"
-        
             for details in prompt_tokens_details:
                 modality = details.get("modality", "")
                 if not modality:
@@ -354,7 +357,7 @@ class _GoogleGenAiRequest(_ProviderRequest):
                     audio_seconds = math.ceil(modality_token_count / 25)
                     self.add_units("audio"+large_context, input=audio_seconds)
 
-        elif model.startswith("gemini-2.0"):
+        else:
             for details in prompt_tokens_details:
                 modality = details.get("modality", "")
                 if not modality:
@@ -373,3 +376,19 @@ class _GoogleGenAiRequest(_ProviderRequest):
                 modality_token_count = details.get("token_count", 0)
                 if modality in ("VIDEO", "AUDIO", "TEXT", "IMAGE"):
                     self.add_units(modality.lower(), output=modality_token_count)
+
+        if not self._ingest["units"]:
+            input = usage.get("prompt_token_count", 0)
+            output = usage.get("candidates_token_count", 0) * 4
+            
+            if self._is_character_billing_model(model):
+                if self._prompt_character_count > 0:
+                    input = self._prompt_character_count
+                else:
+                    input *= 4
+
+                # if no units were added, add a default unit and assume 4 characters per token
+                self._ingest["units"]["text"+large_context] = Units(input=input, output=output)
+            else:
+                # if no units were added, add a default unit
+                self._ingest["units"]["text"] = Units(input=input, output=output)

{payi-0.1.0a80 → payi-0.1.0a82}/src/payi/lib/OpenAIInstrumentor.py

@@ -1,5 +1,4 @@
 import json
-import logging
 from typing import Any, Union, Optional, Sequence
 from typing_extensions import override
 from importlib.metadata import version
@@ -62,7 +61,7 @@ class OpenAiInstrumentor:
             )
 
         except Exception as e:
-            logging.debug(f"Error instrumenting openai: {e}")
+            instrumentor._logger.debug(f"Error instrumenting openai: {e}")
             return
 
 
@@ -74,6 +73,7 @@ def embeddings_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("OpenAI Embeddings wrapper")
     return instrumentor.invoke_wrapper(
         _OpenAiEmbeddingsProviderRequest(instrumentor),
         _IsStreaming.false,
@@ -91,6 +91,7 @@ async def aembeddings_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("async OpenAI Embeddings wrapper")
     return await instrumentor.async_invoke_wrapper(
         _OpenAiEmbeddingsProviderRequest(instrumentor),
         _IsStreaming.false,
@@ -108,6 +109,7 @@ def chat_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("OpenAI completions wrapper")
     return instrumentor.invoke_wrapper(
         _OpenAiChatProviderRequest(instrumentor),
         _IsStreaming.kwargs,
@@ -125,6 +127,7 @@ async def achat_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("async OpenAI completions wrapper")
     return await instrumentor.async_invoke_wrapper(
         _OpenAiChatProviderRequest(instrumentor),
         _IsStreaming.kwargs,
@@ -142,6 +145,7 @@ def responses_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("OpenAI responses wrapper")
     return instrumentor.invoke_wrapper(
         _OpenAiResponsesProviderRequest(instrumentor),
         _IsStreaming.kwargs,
@@ -159,6 +163,7 @@ async def aresponses_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("async OpenAI responses wrapper")
     return await instrumentor.async_invoke_wrapper(
         _OpenAiResponsesProviderRequest(instrumentor),
         _IsStreaming.kwargs,
@@ -207,12 +212,12 @@ class _OpenAiProviderRequest(_ProviderRequest):
             del extra_headers[PayiHeaderNames.resource_scope]
             
         if not price_as_resource and not price_as_category:
-            logging.error("Azure OpenAI requires price as resource and/or category to be specified, not ingesting")
+            self._instrumentor._logger.error("Azure OpenAI requires price as resource and/or category to be specified, not ingesting")
             return False
 
         if resource_scope:
             if not(resource_scope in ["global", "datazone"] or resource_scope.startswith("region")):
-                logging.error("Azure OpenAI invalid resource scope, not ingesting")
+                self._instrumentor._logger.error("Azure OpenAI invalid resource scope, not ingesting")
                 return False
 
             self._ingest["resource_scope"] = resource_scope
@@ -256,7 +261,7 @@ class _OpenAiProviderRequest(_ProviderRequest):
                         self._ingest["provider_response_json"] = text
 
         except Exception as e:
-            logging.debug(f"Error processing exception: {e}")
+            self._instrumentor._logger.debug(f"Error processing exception: {e}")
             return False
 
         return True
@@ -372,7 +377,7 @@ class _OpenAiChatProviderRequest(_OpenAiProviderRequest):
                 try:
                     enc = tiktoken.get_encoding("o200k_base") # type: ignore
                 except Exception:
-                    logging.warning("Error getting encoding for fallback o200k_base")
+                    self._instrumentor._logger.warning("Error getting encoding for fallback o200k_base")
                     enc = None
             
             if enc:
@@ -450,7 +455,7 @@ class _OpenAiResponsesProviderRequest(_OpenAiProviderRequest):
             try:
                 enc = tiktoken.get_encoding("o200k_base") # type: ignore
             except Exception:
-                logging.warning("Error getting encoding for fallback o200k_base")
+                self._instrumentor._logger.warning("Error getting encoding for fallback o200k_base")
                 enc = None
         
         # find each content..type="input_text" and count tokens

{payi-0.1.0a80 → payi-0.1.0a82}/src/payi/lib/VertexInstrumentor.py

@@ -1,6 +1,5 @@
 import json
 import math
-import logging
 from typing import Any, List, Union, Optional, Sequence
 from typing_extensions import override
 
@@ -43,7 +42,7 @@ class VertexInstrumentor:
             )
 
         except Exception as e:
-            logging.debug(f"Error instrumenting vertex: {e}")
+            instrumentor._logger.debug(f"Error instrumenting vertex: {e}")
             return
 
 @_PayiInstrumentor.payi_wrapper
@@ -54,6 +53,7 @@ def generate_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("vertexai generate_content wrapper")
     return instrumentor.invoke_wrapper(
         _GoogleVertexRequest(instrumentor),
         _IsStreaming.kwargs,
@@ -71,6 +71,7 @@ async def agenerate_wrapper(
     *args: Any,
     **kwargs: Any,
 ) -> Any:
+    instrumentor._logger.debug("async vertexai generate_content wrapper")
     return await instrumentor.async_invoke_wrapper(
         _GoogleVertexRequest(instrumentor),
         _IsStreaming.kwargs,
@@ -252,15 +253,16 @@ class _GoogleVertexRequest(_ProviderRequest):
         usage = response_dict.get("usage_metadata", {})
         input = usage.get("prompt_token_count", 0)
 
-        prompt_tokens_details: list[dict[str, Any]] = usage.get("prompt_tokens_details")
-        candidates_tokens_details: list[dict[str, Any]] = usage.get("candidates_tokens_details")
+        prompt_tokens_details: list[dict[str, Any]] = usage.get("prompt_tokens_details", [])
+        candidates_tokens_details: list[dict[str, Any]] = usage.get("candidates_tokens_details", [])
 
         model: str = response_dict.get("model_version", "")
+        
+        # for character billing only
+        large_context = "" if input < 128000 else "_large_context"
 
         if self._is_character_billing_model(model):
             # gemini 1.0 and 1.5 units are reported in characters, per second, per image, etc...
-            large_context = "" if input < 128000 else "_large_context"
-        
             for details in prompt_tokens_details:
                 modality = details.get("modality", "")
                 if not modality:
@@ -300,7 +302,7 @@ class _GoogleVertexRequest(_ProviderRequest):
                     audio_seconds = math.ceil(modality_token_count / 25)
                     self.add_units("audio"+large_context, input=audio_seconds)
 
-        elif model.startswith("gemini-2.0"):
+        else:
             for details in prompt_tokens_details:
                 modality = details.get("modality", "")
                 if not modality:
@@ -319,3 +321,19 @@ class _GoogleVertexRequest(_ProviderRequest):
                 modality_token_count = details.get("token_count", 0)
                 if modality in ("VIDEO", "AUDIO", "TEXT", "IMAGE"):
                     self.add_units(modality.lower(), output=modality_token_count)
+
+        if not self._ingest["units"]:
+            input = usage.get("prompt_token_count", 0)
+            output = usage.get("candidates_token_count", 0) * 4
+            
+            if self._is_character_billing_model(model):
+                if self._prompt_character_count > 0:
+                    input = self._prompt_character_count
+                else:
+                    input *= 4
+
+                # if no units were added, add a default unit and assume 4 characters per token
+                self._ingest["units"]["text"+large_context] = Units(input=input, output=output)
+            else:
+                # if no units were added, add a default unit
+                self._ingest["units"]["text"] = Units(input=input, output=output)