arize-phoenix 10.0.4__py3-none-any.whl → 12.28.1__py3-none-any.whl

phoenix/__generated__/classification_evaluator_configs/_models.py

@@ -0,0 +1,18 @@
+# This file is generated. Do not edit by hand.
+
+from typing import Literal
+
+from pydantic import BaseModel
+
+
+class PromptMessage(BaseModel):
+    role: Literal["user"]
+    content: str
+
+
+class ClassificationEvaluatorConfig(BaseModel):
+    name: str
+    description: str
+    optimization_direction: Literal["minimize", "maximize"]
+    messages: list[PromptMessage]
+    choices: dict[str, float]

phoenix/__generated__/classification_evaluator_configs/_tool_selection_classification_evaluator_config.py

@@ -0,0 +1,17 @@
+# This file is generated. Do not edit by hand.
+# ruff: noqa: E501
+
+from ._models import ClassificationEvaluatorConfig, PromptMessage
+
+TOOL_SELECTION_CLASSIFICATION_EVALUATOR_CONFIG = ClassificationEvaluatorConfig(
+    name="tool_selection",
+    description="For determining if the correct tool was selected for a given context. Requires conversation context, a list of available tools, and the LLM's tool selections.",
+    optimization_direction="maximize",
+    messages=[
+        PromptMessage(
+            role="user",
+            content="You are an impartial judge evaluating an LLM's tool-calling behavior, specifically whether the LLM selected the most appropriate tool or tools for the task.\nYour task: Determine whether the LLM's tool selection was correct or incorrect based on: - The conversation context - The available tools - The LLM's tool invocation(s)\nCriteria Return \"correct\" only when ALL of the following are true: - The LLM chose the best available tool for the user query OR correctly avoided tools if none were needed. - The tool name exists in the available tools list. - The tool is allowed and safe to call. - The LLM selected the correct number of tools for the task.\nReturn \"incorrect\" if ANY of the following are true: - The LLM used a hallucinated or nonexistent tool. - The LLM selected a tool when none was needed. - The LLM did not use a tool when one was required. - The LLM chose a suboptimal or irrelevant tool. - The LLM selected an unsafe or not-permitted tool. - The tool name does not appear in the available tools list.\nBefore providing your final judgment, explain your reasoning and consider: - What does the input context require? - Can this be answered without tools, or is a tool necessary? - If a tool was selected, does it exist in the available tools? - Does the selected tool's description match the user's needs? - Is the selection safe and appropriate? - Is there a better tool available that should have been chosen instead?\n<data> <context> {{input}} </context>\n<available_tools> {{available_tools}} </available_tools>\n<tool_selection> {{tool_selection}} </tool_selection> </data>\nGiven the above data, is the tool selection correct or incorrect?",
+        )
+    ],
+    choices={"correct": 1.0, "incorrect": 0.0},
+)

phoenix/__init__.py

@@ -84,11 +84,12 @@ class PhoenixTraceOpenAILoader(Loader):
             "The legacy `phoenix.trace.openai` instrumentor module has been removed.\n"
             "Please use OpenInference to instrument the OpenAI SDK. Additionally, the "
             "`phoenix.otel` module can be used to quickly configure OpenTelemetry:\n\n"
-            "https://docs.arize.com/phoenix/tracing/integrations-tracing/openai"
+            "https://arize.com/docs/phoenix/tracing/integrations-tracing/openai"
             "\n\n"
             "Example usage:\n\n"
+            "pip install openinference-instrumentation-openai\n\n"
             "```python\n"
-            "from phoenix.otel register\n"
+            "from phoenix.otel import register\n"
             "from openinference.instrumentation.openai import OpenAIInstrumentor\n\n"
             "tracer_provider = register()\n"
             "OpenAIInstrumentor().instrument(tracer_provider=tracer_provider)\n"
@@ -108,7 +109,7 @@ class PhoenixTraceLangchainLoader(Loader):
             "The legacy `phoenix.trace.langchain` instrumentor module has been removed.\n"
             "Please use OpenInference to instrument the LangChain SDK. Additionally, the "
             "`phoenix.otel` module can be used to quickly configure OpenTelemetry:\n\n"
-            "https://docs.arize.com/phoenix/tracing/integrations-tracing/langchain"
+            "https://arize.com/docs/phoenix/tracing/integrations-tracing/langchain"
             "\n\n"
             "Example usage:\n\n"
             "```python\n"
@@ -129,7 +130,7 @@ class PhoenixTraceLlamaIndexLoader(Loader):
             "The legacy `phoenix.trace.llama_index` instrumentor module has been removed.\n"
             "Please use OpenInference to instrument the LlamaIndex SDK. Additionally, the "
             "`phoenix.otel` module can be used to quickly configure OpenTelemetry:\n\n"
-            "https://docs.arize.com/phoenix/tracing/integrations-tracing/llamaindex"
+            "https://arize.com/docs/phoenix/tracing/integrations-tracing/llamaindex"
             "\n\n"
             "Example usage:\n\n"
             "```python\n"

phoenix/auth.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 
 import re
 from dataclasses import dataclass
-from datetime import datetime, timedelta
+from datetime import datetime, timedelta, timezone
 from enum import Enum, auto
 from hashlib import pbkdf2_hmac
 from typing import Any, Literal, Optional, Protocol
@@ -47,6 +47,18 @@ def is_valid_password(*, password: Secret, salt: bytes, password_hash: bytes) ->
     return password_hash == compute_password_hash(password=password, salt=salt)
 
 
+def sanitize_email(email: str) -> str:
+    """
+    Sanitizes an email address by trimming whitespace and converting to lowercase.
+
+    Args:
+        email (str): the email address to sanitize
+    Returns:
+        str: the sanitized email address
+    """
+    return email.strip().lower()
+
+
 def validate_email_format(email: str) -> None:
     """
     Checks that the email has a valid format.
@@ -117,6 +129,18 @@ def set_oauth2_nonce_cookie(
     )
 
 
+def set_oauth2_code_verifier_cookie(
+    *, response: ResponseType, code_verifier: str, max_age: timedelta
+) -> ResponseType:
+    return _set_cookie(
+        response=response,
+        cookie_name=PHOENIX_OAUTH2_CODE_VERIFIER_COOKIE_NAME,
+        cookie_max_age=max_age,
+        samesite="lax",
+        value=code_verifier,
+    )
+
+
 def _set_cookie(
     *,
     response: ResponseType,
@@ -157,6 +181,11 @@ def delete_oauth2_nonce_cookie(response: ResponseType) -> ResponseType:
     return response
 
 
+def delete_oauth2_code_verifier_cookie(response: ResponseType) -> ResponseType:
+    response.delete_cookie(key=PHOENIX_OAUTH2_CODE_VERIFIER_COOKIE_NAME)
+    return response
+
+
 @dataclass(frozen=True)
 class _PasswordRequirements:
     """
@@ -258,6 +287,8 @@ PHOENIX_OAUTH2_STATE_COOKIE_NAME = "phoenix-oauth2-state"
 """The name of the cookie that stores the state used for the OAuth2 authorization code flow."""
 PHOENIX_OAUTH2_NONCE_COOKIE_NAME = "phoenix-oauth2-nonce"
 """The name of the cookie that stores the nonce used for the OAuth2 authorization code flow."""
+PHOENIX_OAUTH2_CODE_VERIFIER_COOKIE_NAME = "phoenix-oauth2-code-verifier"
+"""The name of the cookie that stores the PKCE code verifier for OAuth2."""
 DEFAULT_OAUTH2_LOGIN_EXPIRY_MINUTES = 15
 """
 The default amount of time in minutes that can elapse between the initial
@@ -300,7 +331,13 @@ class ClaimSet:
 
     @property
     def status(self) -> ClaimSetStatus:
-        if self.expiration_time and self.expiration_time.timestamp() < datetime.now().timestamp():
+        # Per JWT RFC 7519 Section 4.1.4, the expiration time identifies the time
+        # "on or after which" the JWT must not be accepted. Use <= for inclusive check.
+        # https://datatracker.ietf.org/doc/html/rfc7519#section-4.1.4
+        if (
+            self.expiration_time
+            and self.expiration_time.timestamp() <= datetime.now(timezone.utc).timestamp()
+        ):
             return ClaimSetStatus.EXPIRED
         if self.token_id is not None and self.subject is not None:
             return ClaimSetStatus.VALID