mirascope 1.21.1__tar.gz → 1.21.3__tar.gz

{mirascope-1.21.1 → mirascope-1.21.3}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mirascope
-Version: 1.21.1
+Version: 1.21.3
 Summary: LLM abstractions that aren't obstructions
 Project-URL: Homepage, https://mirascope.com
 Project-URL: Documentation, https://mirascope.com/WELCOME

{mirascope-1.21.1 → mirascope-1.21.3}/docs/CONTRIBUTING.md

@@ -155,6 +155,15 @@ A few key things to note:
 2. The API reference is generated automatically, but things like new modules still need to be included in the `docs/api` structure for generation to work.
 3. The `docs/tutorials` are written as Jupyter notebooks that get converted into markdown. This conversion will happen on every save when running the server locally, which can make writing docs slow. We recommend setting `strict: false` and commenting out the `mkdocs-jupyter` plugin in `mkdocs.yml` to skip the conversion.
 
+!!! tip "MacOS Users Will Need `cairo`
+
+    If you're developing using MacOS, you'll need to follow these steps:
+
+    1. `brew install cairo`
+    2. `export PKG_CONFIG_PATH="/usr/local/lib/pkgconfig:/opt/homebrew/lib/pkgconfig:$PKG_CONFIG_PATH"`
+    3. `export DYLD_LIBRARY_PATH="/usr/local/lib:/opt/homebrew/lib:$DYLD_LIBRARY_PATH"`
+    4. `source ~/.bash_profile`
+
 ## Testing
 
 All changes to the codebase must be properly unit tested. If a change requires updating an existing unit test, make sure to think through if the change is breaking.

{mirascope-1.21.1 → mirascope-1.21.3}/docs/WELCOME.md

@@ -9,7 +9,7 @@
     <a href="https://github.com/Mirascope/mirascope/blob/dev/LICENSE"><img src="https://img.shields.io/github/license/Mirascope/mirascope.svg" alt="License"/></a>
 </p>
 
-Mirascope is a powerful, flexible, and user-friendly library that simplifies the process of working with LLMs through a unified interface that works across various supported providers, including [OpenAI](https://openai.com/), [Anthropic](https://www.anthropic.com/), [Mistral](https://mistral.ai/), [Gemini](https://gemini.google.com), [Groq](https://groq.com/), [Cohere](https://cohere.com/), [LiteLLM](https://www.litellm.ai/), [Azure AI](https://azure.microsoft.com/en-us/solutions/ai), [Vertex AI](https://cloud.google.com/vertex-ai), and [Bedrock](https://aws.amazon.com/bedrock/).
+Mirascope is a powerful, flexible, and user-friendly library that simplifies the process of working with LLMs through a unified interface that works across various supported providers, including [OpenAI](https://openai.com/), [Anthropic](https://www.anthropic.com/), [Mistral](https://mistral.ai/), [Google (Gemini/Vertex)](https://googleapis.github.io/python-genai/), [Groq](https://groq.com/), [Cohere](https://cohere.com/), [LiteLLM](https://www.litellm.ai/), [Azure AI](https://azure.microsoft.com/en-us/solutions/ai), and [Bedrock](https://aws.amazon.com/bedrock/).
 
 Whether you're generating text, extracting structured information, or developing complex AI-driven agent systems, Mirascope provides the tools you need to streamline your development process and create powerful, robust applications.
 
@@ -89,16 +89,14 @@ Make your first call to an LLM to extract the title and author of a book from th
         ```python hl_lines="19-38 43"
         {% elif provider == "Mistral" %}
         ```python hl_lines="21-46 51"
-        {% elif provider == "Gemini" %}
-        ```python hl_lines="19-57 62"
+        {% elif provider == "Google" %}
+        ```python hl_lines="20-60 65"
         {% elif provider == "Cohere" %}
         ```python hl_lines="19-36 41"
         {% elif provider == "LiteLLM" %}
         ```python hl_lines="16-37 42"
         {% elif provider == "Azure AI" %}
         ```python hl_lines="26-46 51"
-        {% elif provider == "Vertex AI" %}
-        ```python hl_lines="23-62 67"
         {% else %}
         ```python hl_lines="18-39 44"
         {% endif %}

{mirascope-1.21.1 → mirascope-1.21.3}/docs/WHY.md

@@ -95,16 +95,14 @@ Let's compare structured outputs using Mirascope vs. the official SDKs:
         ```python hl_lines="19-38 43"
         {% elif provider == "Mistral" %}
         ```python hl_lines="21-46 51"
-        {% elif provider == "Gemini" %}
-        ```python hl_lines="19-57 62"
+        {% elif provider == "Google" %}
+        ```python hl_lines="20-60 65"
         {% elif provider == "Cohere" %}
         ```python hl_lines="19-36 41"
         {% elif provider == "LiteLLM" %}
         ```python hl_lines="16-37 42"
         {% elif provider == "Azure AI" %}
         ```python hl_lines="26-46 51"
-        {% elif provider == "Vertex AI" %}
-        ```python hl_lines="23-62 67"
         {% elif provider == "Bedrock" %}
         ```python hl_lines="17-48 53"
         {% else %}

{mirascope-1.21.1 → mirascope-1.21.3}/docs/blog/posts/engineers-should-handle-prompting-llms.md

@@ -397,12 +397,11 @@ Beyond OpenAI, Mirascope supports currently these [other LLM providers](https://
 
 - Anthropic
 - Mistral
-- Gemini
+- Google Gemini
 - Groq
 - Cohere
 - LiteLLM
 - Azure AI
-- Vertex AI
 - AWS Bedrock
 
 We've made sure that every example in our documentation shows how to use each provider for easy copy-paste to try it out yourself.

{mirascope-1.21.1 → mirascope-1.21.3}/docs/blog/posts/llm-integration.md

@@ -319,7 +319,7 @@ Mirascope’s call decorator — e.g., `@llm.call()` — simplifies model invoca
 
 When used with `@prompt_template()`, the call decorator injects function arguments into the formatted prompt, removing the need for manually formatting or passing inputs into the LLM call.
 
-It also automates response parsing and its interface lets you hot switch model providers when needed — our growing list includes OpenAI, Anthropic, Mistral, Gemini, Groq, Cohere, LiteLLM, Azure AI, Vertex AI, and Bedrock.
+It also automates response parsing and its interface lets you hot switch model providers when needed — our growing list includes OpenAI, Anthropic, Mistral, Google Gemini, Groq, Cohere, LiteLLM, Azure AI, and Bedrock.
 
 An example of this is shown below. To switch from OpenAI to Anthropic, you can use `llm_override` to override provider settings at runtime to specify a different model.
 

{mirascope-1.21.1 → mirascope-1.21.3}/docs/blog/posts/prompt-engineering-tools.md

@@ -209,12 +209,11 @@ Mirascope also supports interactions with other LLMs besides OpenAI to refine th
 - OpenAI
 - Anthropic
 - Mistral
-- Gemini
+- Google Gemini
 - Groq
 - Cohere
 - LiteLLM
 - Azure AI
-- Vertex AI
 - Bedrock
 
 Mirascope provides wrappers to integrate with a number of popular libraries in the machine learning and LLM space, such as Logfire, Langfuse, Open Telemetry, and Tenacity.

{mirascope-1.21.1 → mirascope-1.21.3}/docs/blog/posts/rag-llm-example.md

@@ -219,7 +219,7 @@ The `ask_alan_turing` function:
 
 We add two Mirascope decorators to turn the function into an LLM API call:
 
-* The `openai.call` decorator makes calls more readable by turning any Python function into an LLM call and is provider agnostic: we could’ve instead written, e.g., `anthropic.call`, `gemini.call`, `vertex.call` or any of a number of others, with minimal code adjustments.
+* The `openai.call` decorator makes calls more readable by turning any Python function into an LLM call and is provider agnostic: we could’ve instead written, e.g., `anthropic.call`, `google.call` or any of a number of others, with minimal code adjustments.
 * Mirascope’s `prompt_template` decorator enables defining the prompt as a template that instructs the model to respond as if it’s Alan Turing and injects a list of the retrieved excerpts.
 
 Both of these decorators, when added to any function, illustrate the best practice of colocation of prompts and calls, which makes the code eminently more readable as the logic for preparing input data, [building prompts](https://mirascope.com/blog/llm-prompt/), and LLM invocation is centralized in a single place.

{mirascope-1.21.1 → mirascope-1.21.3}/docs/learn/async.md

@@ -176,7 +176,7 @@ __Decorator Parameter:__
 
             {% if provider == "LiteLLM" %}
             ```python
-            {% elif provider in ["OpenAI", "Vertex AI"] %}
+            {% elif provider in ["OpenAI"] %}
             ```python hl_lines="2 5"
             {% elif provider == "Azure AI" %}
             ```python hl_lines="1-2 8-10"
@@ -205,7 +205,7 @@ __Dynamic Configuration:__
 
             {% if provider == "LiteLLM" %}
             ```python
-            {% elif provider in ["OpenAI", "Vertex AI"] %}
+            {% elif provider in ["OpenAI"] %}
             ```python hl_lines="2 11"
             {% elif provider == "Azure AI" %}
             ```python hl_lines="1-2 12-14"
@@ -226,7 +226,7 @@ __Dynamic Configuration:__
 
             {% if provider == "LiteLLM" %}
             ```python
-            {% elif provider in ["OpenAI", "Vertex AI"] %}
+            {% elif provider in ["OpenAI"] %}
             ```python hl_lines="2 9"
             {% elif provider == "Azure AI" %}
             ```python hl_lines="1-2 10-12"

{mirascope-1.21.1 → mirascope-1.21.3}/docs/learn/calls.md

@@ -53,8 +53,8 @@ Let's take a look at a basic example using Mirascope vs. official provider SDKs:
 
         {% if provider == "Anthropic" %}
         ```python hl_lines="7-13"
-        {% elif provider in ["Google", "Vertex AI"] %}
-        ```python hl_lines="7-10"
+        {% elif provider in ["Google"] %}
+        ```python hl_lines="7-11"
         {% elif provider == "LiteLLM" %}
         ```python hl_lines="5-9"
         {% elif provider == "Azure AI" %}
@@ -150,17 +150,16 @@ There are also two common methods:
 
 While most LLM providers focus on text outputs, some providers support additional output modalities like audio. The availability of multi-modal outputs varies among providers:
 
-| Provider    | Text | Audio | Image |
-|------------|------|-------|-------|
-| OpenAI     | ✓    | ✓     | -     |
-| Anthropic  | ✓    | -     | -     |
-| Mistral    | ✓    | -     | -     |
-| Gemini     | ✓    | -     | -     |
-| Groq       | ✓    | -     | -     |
-| Cohere     | ✓    | -     | -     |
-| LiteLLM    | ✓    | -     | -     |
-| Azure AI   | ✓    | -     | -     |
-| Vertex AI  | ✓    | -     | -     |
+| Provider      | Text | Audio | Image |
+|---------------|------|-------|-------|
+| OpenAI        | ✓    | ✓     | -     |
+| Anthropic     | ✓    | -     | -     |
+| Mistral       | ✓    | -     | -     |
+| Google Gemini | ✓    | -     | -     |
+| Groq          | ✓    | -     | -     |
+| Cohere        | ✓    | -     | -     |
+| LiteLLM       | ✓    | -     | -     |
+| Azure AI      | ✓    | -     | -     |
 
 Legend: ✓ (Supported), - (Not Supported)
 
@@ -231,14 +230,7 @@ Since `call_params` is just a `TypedDict`, you can always include any additional
     === "{{ method_title }}"
         {% for provider in supported_llm_providers %}
         === "{{ provider }}"
-
-            {% if provider == "Gemini" %}
-            ```python hl_lines="1 7"
-            {% elif provider == "Vertex AI" %}
-            ```python hl_lines="2 7"
-            {% else %}
             ```python hl_lines="4"
-            {% endif %}
             --8<-- "build/snippets/learn/calls/call_params/{{ provider | provider_dir }}/{{ method }}.py::9"
             ```
 

{mirascope-1.21.1 → mirascope-1.21.3}/docs/learn/prompts.md

@@ -138,13 +138,13 @@ Recent advancements in Large Language Model architecture has enabled many model
 
 While Mirascope provides a consistent interface, support varies among providers:
 
-| Type          | Anthropic | Cohere | Gemini | Groq | Mistral | OpenAI |
-|---------------|-----------|--------|--------|------|---------|--------|
-| text          | ✓         | ✓      | ✓      | ✓    | ✓       | ✓      |
-| image         | ✓         | -      | ✓      | ✓    | ✓       | ✓      |
-| audio         | -         | -      | ✓      | -    | -       | ✓      |
-| video         | -         | -      | ✓      | -    | -       | -      |
-| document      | ✓         | -      | -      | -    | -       | -      |
+| Type          | Anthropic | Cohere | Google Gemini | Groq | Mistral | OpenAI |
+|---------------|-----------|--------|---------------|------|---------|--------|
+| text          | ✓         | ✓      | ✓             | ✓    | ✓       | ✓      |
+| image         | ✓         | -      | ✓             | ✓    | ✓       | ✓      |
+| audio         | -         | -      | ✓             | -    | -       | ✓      |
+| video         | -         | -      | ✓             | -    | -       | -      |
+| document      | ✓         | -      | -             | -    | -       | -      |
 
 Legend: ✓ (Supported), - (Not Supported)
 

{mirascope-1.21.1 → mirascope-1.21.3}/docs/learn/response_models.md

@@ -56,8 +56,6 @@ Let's take a look at a basic example using Mirascope vs. official provider SDKs:
         ```python hl_lines="16-37 42"
         {% elif provider == "Azure AI" %}
         ```python hl_lines="26-46 51"
-        {% elif provider == "Vertex AI" %}
-        ```python hl_lines="23-62 67"
         {% elif provider == "Bedrock" %}
         ```python hl_lines="16-48 53"
         {% else %}
@@ -217,10 +215,7 @@ We take advantage of Pydantic's [`Field`](https://docs.pydantic.dev/latest/conce
 
         {% for provider in supported_llm_providers %}
         === "{{ provider }}"
-
-            {% if provider in ["Google", "Vertex AI"] %}
-            ```python
-            {% elif method == "messages" %}
+            {% if method == "messages" %}
             ```python hl_lines="6-7 11-13 27"
             {% elif method == "base_message_param" %}
             ```python hl_lines="6-7 11-13 30"

{mirascope-1.21.1 → mirascope-1.21.3}/docs/learn/streams.md

@@ -190,17 +190,16 @@ While Mirascope provides a consistent interface, you can always access the full,
 
 While most LLM providers focus on text streaming, some providers support streaming additional output modalities like audio. The availability of multi-modal streaming varies among providers:
 
-| Provider    | Text | Audio | Image |
-|------------|------|-------|-------|
-| OpenAI     | ✓    | ✓     | -     |
-| Anthropic  | ✓    | -     | -     |
-| Mistral    | ✓    | -     | -     |
-| Gemini     | ✓    | -     | -     |
-| Groq       | ✓    | -     | -     |
-| Cohere     | ✓    | -     | -     |
-| LiteLLM    | ✓    | -     | -     |
-| Azure AI   | ✓    | -     | -     |
-| Vertex AI  | ✓    | -     | -     |
+| Provider      | Text | Audio | Image |
+|---------------|------|-------|-------|
+| OpenAI        | ✓    | ✓     | -     |
+| Anthropic     | ✓    | -     | -     |
+| Mistral       | ✓    | -     | -     |
+| Google Gemini | ✓    | -     | -     |
+| Groq          | ✓    | -     | -     |
+| Cohere        | ✓    | -     | -     |
+| LiteLLM       | ✓    | -     | -     |
+| Azure AI      | ✓    | -     | -     |
 
 Legend: ✓ (Supported), - (Not Supported)
 
@@ -261,7 +260,7 @@ Error handling in streams is similar to standard non-streaming calls. However, i
         {% for provider in supported_llm_providers %}
         === "{{ provider }}"
 
-            {% if provider in ["Gemini", "Azure AI", "Vertex AI"] %}
+            {% if provider in ["Google", "Azure AI"] %}
             ```python hl_lines="9 12"
             {% elif provider == "Mistral" %}
             ```python hl_lines="2 10 13 16"

{mirascope-1.21.1 → mirascope-1.21.3}/docs/learn/tools.md

@@ -98,8 +98,6 @@ Let's take a look at a basic example of each using Mirascope vs. official provid
         ```python hl_lines="6-12 19-32 34-36"
         {% elif provider == "Azure AI" %}
         ```python hl_lines="16-22 31-43 45-47"
-        {% elif provider == "Vertex AI" %}
-        ```python hl_lines="6-12 18-34 36-43"
         {% elif provider == "Mistral" %}
         ```python hl_lines="10-16 23-36 38-44"
         {% elif provider == "Bedrock" %}
@@ -377,9 +375,7 @@ Just like with [Response Models](./response_models.md#few-shot-examples), you ca
 
             {% for provider in supported_llm_providers %}
             === "{{ provider }}"
-                {% if provider == "Google" %}
-                ```python
-                {% elif tool_method == "function" %}
+                {% if tool_method == "function" %}
                 ```python hl_lines="14 20-21"
                 {% else %}
                 ```python hl_lines="11 15"

{mirascope-1.21.1 → mirascope-1.21.3}/mirascope/core/bedrock/_utils/_setup_call.py

@@ -2,6 +2,7 @@
 
 from __future__ import annotations
 
+import os
 from collections.abc import AsyncGenerator, Awaitable, Callable, Coroutine, Generator
 from functools import wraps
 from typing import Any, ParamSpec, cast, overload
@@ -223,12 +224,23 @@ def setup_call(
 
     call_kwargs |= cast(BedrockCallKwargs, {"modelId": model, "messages": messages})
 
+    env_vars = {}
+    if access_key_id := os.getenv("AWS_ACCESS_KEY_ID"):
+        env_vars["aws_access_key_id"] = access_key_id
+    if secret_access_key := os.getenv("AWS_SECRET_ACCESS_KEY"):
+        env_vars["aws_secret_access_key"] = secret_access_key
+    if session_token := os.getenv("AWS_SESSION_TOKEN"):
+        env_vars["aws_session_token"] = session_token
+    if region_name := os.getenv("AWS_REGION_NAME"):
+        env_vars["region_name"] = region_name
+    if profile_name := os.getenv("AWS_PROFILE"):
+        env_vars["profile_name"] = profile_name
     if client is None:
         if fn_is_async(fn):
-            session = get_session()
+            session = get_session(env_vars=env_vars)
             _client = _AsyncBedrockRuntimeWrappedClient(session, model)
         else:
-            session = Session()
+            session = Session(**env_vars)
             _client = session.client("bedrock-runtime")
     else:
         _client = client

{mirascope-1.21.1 → mirascope-1.21.3}/mirascope/core/google/tool.py

@@ -70,41 +70,50 @@ class GoogleTool(BaseTool):
             fn["parameters"] = model_schema
 
         if "parameters" in fn:
-            # Resolve $defs and $ref
+            # Define a function to handle both ref resolution and examples conversion
+            def resolve_refs_and_fix_examples(
+                schema: dict[str, Any], defs: dict[str, Any] | None = None
+            ) -> dict[str, Any]:
+                """Recursively resolve $ref references and fix examples/example fields."""
+                # If this is a reference, resolve it
+                if "$ref" in schema:
+                    ref = schema["$ref"]
+                    if ref.startswith("#/$defs/") and defs:
+                        ref_key = ref.replace("#/$defs/", "")
+                        if ref_key in defs:
+                            # Merge the definition with the current schema (excluding $ref)
+                            resolved = {
+                                **{k: v for k, v in schema.items() if k != "$ref"},
+                                **resolve_refs_and_fix_examples(defs[ref_key], defs),
+                            }
+                            return resolved
+
+                # Handle examples -> example conversion
+                result = {}
+                for key, value in schema.items():
+                    # Convert "examples" to "example" for Google Schema
+                    if key == "examples":
+                        result["example"] = value
+                    elif isinstance(value, dict):
+                        result[key] = resolve_refs_and_fix_examples(value, defs)
+                    elif isinstance(value, list):
+                        result[key] = [
+                            resolve_refs_and_fix_examples(item, defs)
+                            if isinstance(item, dict)
+                            else item
+                            for item in value
+                        ]
+                    else:
+                        result[key] = value
+                return result
+
+            # Extract $defs if they exist
+            defs = {}
             if "$defs" in fn["parameters"]:
                 defs = fn["parameters"].pop("$defs")
 
-                def resolve_refs(schema: dict[str, Any]) -> dict[str, Any]:
-                    """Recursively resolve $ref references using the $defs dictionary."""
-                    # If this is a reference, resolve it
-                    if "$ref" in schema:
-                        ref = schema["$ref"]
-                        if ref.startswith("#/$defs/"):
-                            ref_key = ref.replace("#/$defs/", "")
-                            if ref_key in defs:
-                                # Merge the definition with the current schema (excluding $ref)
-                                resolved = {
-                                    **{k: v for k, v in schema.items() if k != "$ref"},
-                                    **resolve_refs(defs[ref_key]),
-                                }
-                                return resolved
-
-                    # Process all other keys recursively
-                    result = {}
-                    for key, value in schema.items():
-                        if isinstance(value, dict):
-                            result[key] = resolve_refs(value)
-                        elif isinstance(value, list):
-                            result[key] = [
-                                resolve_refs(item) if isinstance(item, dict) else item
-                                for item in value
-                            ]
-                        else:
-                            result[key] = value
-                    return result
-
-                # Resolve all references in the parameters
-                fn["parameters"] = resolve_refs(fn["parameters"])
+            # Resolve all references in the parameters and fix examples
+            fn["parameters"] = resolve_refs_and_fix_examples(fn["parameters"], defs)
 
             def handle_enum_schema(prop_schema: dict[str, Any]) -> dict[str, Any]:
                 if "enum" in prop_schema:
@@ -112,6 +121,7 @@ class GoogleTool(BaseTool):
                 return prop_schema
 
             # Process properties after resolving references
+            # We're already handling examples -> example conversion recursively above
             fn["parameters"]["properties"] = {
                 prop: {
                     key: value
@@ -120,6 +130,7 @@ class GoogleTool(BaseTool):
                 }
                 for prop, prop_schema in fn["parameters"]["properties"].items()
             }
+
         return Tool(function_declarations=[FunctionDeclaration(**fn)])
 
     @classmethod

{mirascope-1.21.1 → mirascope-1.21.3}/mkdocs.yml

@@ -374,14 +374,6 @@ nav:
               - tool: "api/core/cohere/tool.md"
           - Costs:
               - calculate_cost: "api/core/costs/calculate_cost.md"
-          - Gemini:
-              - call: "api/core/gemini/call.md"
-              - call_params: "api/core/gemini/call_params.md"
-              - call_response: "api/core/gemini/call_response.md"
-              - call_response_chunk: "api/core/gemini/call_response_chunk.md"
-              - dynamic_config: "api/core/gemini/dynamic_config.md"
-              - stream: "api/core/gemini/stream.md"
-              - tool: "api/core/gemini/tool.md"
           - Google:
               - call: "api/core/google/call.md"
               - call_params: "api/core/google/call_params.md"
@@ -422,14 +414,6 @@ nav:
               - dynamic_config: "api/core/openai/dynamic_config.md"
               - stream: "api/core/openai/stream.md"
               - tool: "api/core/openai/tool.md"
-          - Vertex AI:
-              - call: "api/core/vertex/call.md"
-              - call_params: "api/core/vertex/call_params.md"
-              - call_response: "api/core/vertex/call_response.md"
-              - call_response_chunk: "api/core/vertex/call_response_chunk.md"
-              - dynamic_config: "api/core/vertex/dynamic_config.md"
-              - stream: "api/core/vertex/stream.md"
-              - tool: "api/core/vertex/tool.md"
           - xAI:
               - call: "api/core/xai/call.md"
               - call_params: "api/core/xai/call_params.md"

{mirascope-1.21.1 → mirascope-1.21.3}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "mirascope"
-version = "1.21.1"
+version = "1.21.3"
 description = "LLM abstractions that aren't obstructions"
 readme = "README.md"
 license = { file = "LICENSE" }

{mirascope-1.21.1 → mirascope-1.21.3}/tests/core/bedrock/_utils/test_setup_call.py

@@ -493,3 +493,80 @@ async def test_async_bedrock_runtime_wrapped_client_converse_stream():
     mock_client.converse_stream.assert_called_once_with(param2="value2")
 
     mock_session.create_client.assert_called_once_with("bedrock-runtime")
+
+
+@patch("mirascope.core.bedrock._utils._setup_call.Session")
+@patch("mirascope.core.bedrock._utils._setup_call.get_session")
+@patch("mirascope.core.bedrock._utils._setup_call._utils", new_callable=MagicMock)
+def test_setup_call_env_vars(
+    mock_utils: MagicMock,
+    mock_get_session: MagicMock,
+    mock_session_class: MagicMock,
+    mock_base_setup_call: MagicMock,
+    monkeypatch: pytest.MonkeyPatch,
+) -> None:
+    """Test that environment variables are properly passed to Session and get_session."""
+    mock_utils.setup_call = mock_base_setup_call
+    mock_base_setup_call.return_value[1] = [
+        {"role": "user", "content": [{"text": "user test"}]},
+    ]
+    mock_base_setup_call.return_value[3] = {}
+
+    # Set environment variables
+    monkeypatch.setenv("AWS_ACCESS_KEY_ID", "test_access_key")
+    monkeypatch.setenv("AWS_SECRET_ACCESS_KEY", "test_secret_key")
+    monkeypatch.setenv("AWS_SESSION_TOKEN", "test_session_token")
+    monkeypatch.setenv("AWS_REGION_NAME", "us-west-2")
+    monkeypatch.setenv("AWS_PROFILE", "test_profile")
+
+    # Test sync client creation with env vars
+    setup_call(
+        model="anthropic.claude-v2",
+        client=None,
+        fn=MagicMock(),
+        fn_args={},
+        dynamic_config=None,
+        tools=None,
+        json_mode=False,
+        call_params={},
+        response_model=None,
+        stream=False,
+    )
+
+    # Verify Session was called with expected env vars
+    mock_session_class.assert_called_once_with(
+        aws_access_key_id="test_access_key",
+        aws_secret_access_key="test_secret_key",
+        aws_session_token="test_session_token",
+        region_name="us-west-2",
+        profile_name="test_profile",
+    )
+
+    # Test async client creation with env vars
+    async def async_fn(): ...
+
+    mock_session_class.reset_mock()
+
+    setup_call(
+        model="anthropic.claude-v2",
+        client=None,
+        fn=async_fn,
+        fn_args={},
+        dynamic_config=None,
+        tools=None,
+        json_mode=False,
+        call_params={},
+        response_model=None,
+        stream=False,
+    )
+
+    # Verify get_session was called with expected env vars
+    mock_get_session.assert_called_with(
+        env_vars={
+            "aws_access_key_id": "test_access_key",
+            "aws_secret_access_key": "test_secret_key",
+            "aws_session_token": "test_session_token",
+            "region_name": "us-west-2",
+            "profile_name": "test_profile",
+        }
+    )

{mirascope-1.21.1 → mirascope-1.21.3}/tests/core/google/test_tool.py

@@ -10,6 +10,7 @@ from google.genai.types import (  # type: ignore
     Tool,
     Type,
 )
+from pydantic import Field
 
 from mirascope.core.base.tool import BaseTool
 from mirascope.core.google.tool import GoogleTool
@@ -18,10 +19,23 @@ from mirascope.core.google.tool import GoogleTool
 class FormatBook(GoogleTool):
     """Returns the title and author nicely formatted."""
 
-    title: str
+    title: str = Field(..., examples=["The Name of the Wind"])
     author: str
     genre: Literal["fantasy", "scifi"]
 
+    model_config = {
+        "json_schema_extra": {
+            "examples": [
+                {
+                    "title": "The Way of Kings",
+                    "author": "Brandon Sanderson",
+                    "genre": "fantasy",
+                },
+                {"title": "Dune", "author": "Frank Herbert", "genre": "scifi"},
+            ]
+        }
+    }
+
     def call(self) -> str:
         return f"{self.title} by {self.author}"
 
@@ -54,7 +68,9 @@ def test_google_tool() -> None:
                 parameters=Schema(
                     type=Type.OBJECT,
                     properties={
-                        "title": Schema(type=Type.STRING),
+                        "title": Schema(
+                            type=Type.STRING, example=["The Name of the Wind"]
+                        ),
                         "author": Schema(type=Type.STRING),
                         "genre": Schema(
                             type=Type.STRING,
@@ -63,6 +79,14 @@ def test_google_tool() -> None:
                         ),
                     },
                     required=["title", "author", "genre"],
+                    example=[
+                        {
+                            "title": "The Way of Kings",
+                            "author": "Brandon Sanderson",
+                            "genre": "fantasy",
+                        },
+                        {"title": "Dune", "author": "Frank Herbert", "genre": "scifi"},
+                    ],
                 ),
             )
         ],

{mirascope-1.21.1 → mirascope-1.21.3}/uv.lock

@@ -3168,7 +3168,7 @@ wheels = [
 
 [[package]]
 name = "mirascope"
-version = "1.21.0"
+version = "1.21.2"
 source = { editable = "." }
 dependencies = [
     { name = "docstring-parser" },

mirascope-1.21.1/docs/api/core/gemini/call.md

@@ -1,3 +0,0 @@
-# mirascope.core.gemini.call
-
-::: mirascope.core.gemini.call

mirascope-1.21.1/docs/api/core/gemini/call_params.md

@@ -1,3 +0,0 @@
-# mirascope.core.gemini.call_params
-
-::: mirascope.core.gemini.call_params

mirascope-1.21.1/docs/api/core/gemini/call_response.md

@@ -1,3 +0,0 @@
-# mirascope.core.gemini.call_response
-
-::: mirascope.core.gemini.call_response