payi 0.1.0a90__tar.gz → 0.1.0a91__tar.gz

payi-0.1.0a91/.release-please-manifest.json

@@ -0,0 +1,3 @@
+{
+  ".": "0.1.0-alpha.91"
+}

{payi-0.1.0a90 → payi-0.1.0a91}/CHANGELOG.md

@@ -1,5 +1,18 @@
 # Changelog
 
+## 0.1.0-alpha.91 (2025-06-25)
+
+Full Changelog: [v0.1.0-alpha.90...v0.1.0-alpha.91](https://github.com/Pay-i/pay-i-python/compare/v0.1.0-alpha.90...v0.1.0-alpha.91)
+
+### Features
+
+* do not instrument inline base64 image data by default ([#331](https://github.com/Pay-i/pay-i-python/issues/331)) ([8b0455d](https://github.com/Pay-i/pay-i-python/commit/8b0455d24cda3060bc2170b05e5438c72885bf1a))
+
+
+### Chores
+
+* **tests:** skip some failing tests on the latest python versions ([1856b99](https://github.com/Pay-i/pay-i-python/commit/1856b99b5802dc02e4f536818f1a205d9da708c2))
+
 ## 0.1.0-alpha.90 (2025-06-23)
 
 Full Changelog: [v0.1.0-alpha.89...v0.1.0-alpha.90](https://github.com/Pay-i/pay-i-python/compare/v0.1.0-alpha.89...v0.1.0-alpha.90)

{payi-0.1.0a90 → payi-0.1.0a91}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: payi
-Version: 0.1.0a90
+Version: 0.1.0a91
 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-0.1.0a90 → payi-0.1.0a91}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "payi"
-version = "0.1.0-alpha.90"
+version = "0.1.0-alpha.91"
 description = "The official Python library for the payi API"
 dynamic = ["readme"]
 license = "Apache-2.0"

{payi-0.1.0a90 → payi-0.1.0a91}/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.90"  # x-release-please-version
+__version__ = "0.1.0-alpha.91"  # x-release-please-version

{payi-0.1.0a90 → payi-0.1.0a91}/src/payi/lib/AnthropicInstrumentor.py

@@ -173,10 +173,15 @@ class _AnthropicProviderRequest(_ProviderRequest):
 
         messages = kwargs.get("messages")
         if messages:
+
             anthropic_has_image_and_get_texts(self, messages)
 
         return True
 
+    @override
+    def remove_inline_data(self, prompt: 'dict[str, Any]') -> bool:
+        return anthropic_remove_inline_data(prompt)
+
     @override
     def process_exception(self, exception: Exception, kwargs: Any, ) -> bool:
         try:
@@ -352,4 +357,30 @@ def has_image_and_get_texts(encoding: tiktoken.Encoding, content: Union[str, 'li
         token_count = sum(len(encoding.encode(item.get("text", ""))) for item in content if item.get("type") == "text")
         return has_image, token_count
     
-    return False, 0
+    return False, 0
+
+def anthropic_remove_inline_data(prompt: 'dict[str, Any]') -> bool:# noqa: ARG002
+    messages = prompt.get("messages", [])
+    if not messages:
+        return False
+
+    modified = False
+    for message in messages:
+        content = message.get('content', Any)
+        if not content or not isinstance(content, list):
+            continue
+
+        for item in content: # type: ignore
+            if not isinstance(item, dict):
+                continue
+            # item: dict[str, Any]
+            type = item.get("type", "") # type: ignore
+            if type != "image":
+                continue
+
+            source = item.get("source", {}) # type: ignore
+            if source.get("type", "") == "base64": # type: ignore
+                source["data"] = _PayiInstrumentor._not_instrumented
+                modified = True
+
+    return modified

{payi-0.1.0a90 → payi-0.1.0a91}/src/payi/lib/BedrockInstrumentor.py

@@ -356,6 +356,24 @@ class _BedrockInvokeProviderRequest(_BedrockProviderRequest):
 
         return response
 
+    @override
+    def remove_inline_data(self, prompt: 'dict[str, Any]') -> bool:# noqa: ARG002
+        if not self._is_anthropic:
+            return False
+
+        from .AnthropicInstrumentor import anthropic_remove_inline_data
+        body = prompt.get("body", "")
+        if not body:
+            return False
+        
+        body_json = json.loads(body)
+        
+        if anthropic_remove_inline_data(body_json):
+            prompt["body"] = json.dumps(body_json)
+            return True
+
+        return False
+
 class _BedrockConverseProviderRequest(_BedrockProviderRequest):
     @override
     def process_synchronous_response(

{payi-0.1.0a90 → payi-0.1.0a91}/src/payi/lib/OpenAIInstrumentor.py

@@ -301,9 +301,9 @@ class _OpenAiProviderRequest(_ProviderRequest):
         units["text"] = Units(input=input, output=output)
 
     @staticmethod
-    def has_image_and_get_texts(encoding: tiktoken.Encoding, content: Union[str, 'list[Any]'], image_type: str = "image_url", text_type:str = "text") -> 'tuple[bool, int]':
+    def has_image_and_get_texts(encoding: tiktoken.Encoding, content: Union[str, 'list[Any]'], image_type: str, text_type: str) -> 'tuple[bool, int]':
         if isinstance(content, list): # type: ignore
-            has_image = any(item.get("type") == image_type for item in content)
+            has_image = any(item.get("type", "") == image_type for item in content)
             if has_image is False:
                 return has_image, 0
             
@@ -311,6 +311,28 @@ class _OpenAiProviderRequest(_ProviderRequest):
             return has_image, token_count
         return False, 0
 
+    @staticmethod
+    def post_process_request_prompt(content: Union[str, 'list[Any]'], image_type: str, url_subkey: bool) -> bool:
+        modified = False
+        if isinstance(content, list): # type: ignore
+            for item in content:
+                type = item.get("type", "")
+                if type != image_type:
+                    continue
+
+                if url_subkey:
+                    url = item.get("image_url", {}).get("url", "")
+                    if url.startswith("data:"):
+                        item["image_url"]["url"] = _PayiInstrumentor._not_instrumented
+                        modified = True
+                else:
+                    url = item.get("image_url", "")
+                    if url.startswith("data:"):
+                        item["image_url"] = _PayiInstrumentor._not_instrumented
+                        modified = True
+
+        return modified
+
 class _OpenAiEmbeddingsProviderRequest(_OpenAiProviderRequest):
     def __init__(self, instrumentor: _PayiInstrumentor):
         super().__init__(
@@ -397,7 +419,7 @@ class _OpenAiChatProviderRequest(_OpenAiProviderRequest):
             
             if enc:
                 for message in messages:
-                    msg_has_image, msg_prompt_tokens = self.has_image_and_get_texts(enc, message.get('content', ''))
+                    msg_has_image, msg_prompt_tokens = self.has_image_and_get_texts(enc, message.get('content', ''), image_type="image_url", text_type="text")
                     if msg_has_image:
                         has_image = True
                         estimated_token_count += msg_prompt_tokens
@@ -418,6 +440,13 @@ class _OpenAiChatProviderRequest(_OpenAiProviderRequest):
                     self._include_usage_added = True
         return True
 
+    @override
+    def remove_inline_data(self, prompt: 'dict[str, Any]') -> bool:
+        messages = prompt.get("messages", None)
+        if not messages:
+            return False
+        return self.post_process_request_prompt(messages, image_type="image_url", url_subkey=True)
+
     @override
     def process_synchronous_response(
         self,
@@ -531,6 +560,21 @@ class _OpenAiResponsesProviderRequest(_OpenAiProviderRequest):
 
         return True
 
+    @override
+    def remove_inline_data(self, prompt: 'dict[str, Any]') -> bool:
+        modified = False
+        input = prompt.get("input", [])
+        for item in input:
+            if not isinstance(item, dict):
+                continue
+
+            for key, value in item.items(): # type: ignore
+                if key == "content":
+                    if isinstance(value, list):
+                        modified = self.post_process_request_prompt(value, image_type="input_image", url_subkey=False) | modified # type: ignore
+
+        return modified
+
     @override
     def process_synchronous_response(
         self,

{payi-0.1.0a90 → payi-0.1.0a91}/src/payi/lib/VertexRequest.py

@@ -1,6 +1,7 @@
 import json
 import math
 from typing import Any, Optional
+from typing_extensions import override
 
 from payi.lib.helpers import PayiCategories
 from payi.types.ingest_units_params import Units
@@ -62,7 +63,22 @@ class _VertexRequest(_ProviderRequest): # type: ignore
             ingest = True
 
         return _ChunkResult(send_chunk_to_caller=True, ingest=ingest)
-    
+
+    @override
+    def remove_inline_data(self, prompt: 'dict[str, Any]') -> bool:
+        modified = False
+
+        parts: list[dict[str, Any]] = prompt["contents"].get("parts", []) 
+        for part in parts:
+            inline_data = part.get("inline_data", {})
+            if not isinstance(inline_data, dict):
+                continue
+            if "data" in inline_data:
+                inline_data["data"] = _PayiInstrumentor._not_instrumented
+                modified = True
+
+        return modified
+
     def process_response_part_for_function_call(self, part: 'dict[str, Any]') -> None:
         function = part.get("function_call", {})
         if not function:

{payi-0.1.0a90 → payi-0.1.0a91}/src/payi/lib/instrument.py

@@ -67,6 +67,9 @@ class _ProviderRequest:
     def process_initial_stream_response(self, response: Any) -> None:
         pass
 
+    def remove_inline_data(self, prompt: 'dict[str, Any]') -> bool:# noqa: ARG002
+        return False
+
     @property
     def is_aws_client(self) -> bool:
         return self._is_aws_client if self._is_aws_client is not None else False
@@ -133,6 +136,7 @@ class _ProviderRequest:
 class PayiInstrumentConfig(TypedDict, total=False):
     proxy: bool
     global_instrumentation: bool
+    instrument_inline_data: bool
     limit_ids: Optional["list[str]"]
     use_case_name: Optional[str]
     use_case_id: Optional[str]
@@ -183,6 +187,8 @@ class _TrackContext:
             _instrumentor.__exit__(exc_type, exc_val, exc_tb)
 
 class _PayiInstrumentor:
+    _not_instrumented: str = "<not_instrumented>"
+
     def __init__(
         self,
         payi: Optional[Payi],
@@ -219,6 +225,8 @@ class _PayiInstrumentor:
         # default is instrument and ingest metrics
         self._proxy_default: bool = global_config.get("proxy", False)
 
+        self._instrument_inline_data: bool = global_config.get("instrument_inline_data", False)
+
         global_instrumentation = global_config.pop("global_instrumentation", True)
 
         if instruments is None or "*" in instruments:
@@ -345,6 +353,18 @@ class _PayiInstrumentor:
             # convert the function call builder to a list of function calls
             ingest_units["provider_response_function_calls"] = list(request._function_call_builder.values())
 
+        request_json = ingest_units.get('provider_request_json', "")
+        if request_json and self._instrument_inline_data is False:
+            try:
+                prompt_dict = json.loads(request_json)
+                if request.remove_inline_data(prompt_dict):
+                    self._logger.debug(f"Removed inline data from provider_request_json")
+                    # store the modified dict back as JSON string
+                    ingest_units['provider_request_json'] = json.dumps(prompt_dict)
+
+            except Exception as e:
+                self._logger.error(f"Error serializing provider_request_json: {e}")
+
         if int(ingest_units.get("http_status_code") or 0) < 400:
             units = ingest_units.get("units", {})
             if not units or all(unit.get("input", 0) == 0 and unit.get("output", 0) == 0 for unit in units.values()):

{payi-0.1.0a90 → payi-0.1.0a91}/tests/test_client.py

@@ -191,6 +191,7 @@ class TestPayi:
             copy_param = copy_signature.parameters.get(name)
             assert copy_param is not None, f"copy() signature is missing the {name} param"
 
+    @pytest.mark.skipif(sys.version_info >= (3, 10), reason="fails because of a memory leak that started from 3.12")
     def test_copy_build_request(self) -> None:
         options = FinalRequestOptions(method="get", url="/foo")
 
@@ -985,6 +986,7 @@ class TestAsyncPayi:
             copy_param = copy_signature.parameters.get(name)
             assert copy_param is not None, f"copy() signature is missing the {name} param"
 
+    @pytest.mark.skipif(sys.version_info >= (3, 10), reason="fails because of a memory leak that started from 3.12")
     def test_copy_build_request(self) -> None:
         options = FinalRequestOptions(method="get", url="/foo")
 

payi-0.1.0a90/.release-please-manifest.json

@@ -1,3 +0,0 @@
-{
-  ".": "0.1.0-alpha.90"
-}