pydantic-ai 0.3.7__tar.gz → 0.4.1__tar.gz

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pydantic-ai
-Version: 0.3.7
+Version: 0.4.1
 Summary: Agent Framework / shim to use Pydantic with LLMs
 Project-URL: Homepage, https://ai.pydantic.dev
 Project-URL: Source, https://github.com/pydantic/pydantic-ai
@@ -28,9 +28,11 @@ Classifier: Topic :: Internet
 Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Classifier: Topic :: Software Development :: Libraries :: Python Modules
 Requires-Python: >=3.9
-Requires-Dist: pydantic-ai-slim[a2a,anthropic,bedrock,cli,cohere,evals,google,groq,mcp,mistral,openai,vertexai]==0.3.7
+Requires-Dist: pydantic-ai-slim[anthropic,bedrock,cli,cohere,evals,google,groq,mcp,mistral,openai,vertexai]==0.4.1
+Provides-Extra: a2a
+Requires-Dist: fasta2a==0.4.1; extra == 'a2a'
 Provides-Extra: examples
-Requires-Dist: pydantic-ai-examples==0.3.7; extra == 'examples'
+Requires-Dist: pydantic-ai-examples==0.4.1; extra == 'examples'
 Provides-Extra: logfire
 Requires-Dist: logfire>=3.11.0; extra == 'logfire'
 Description-Content-Type: text/markdown

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/pyproject.toml

@@ -46,12 +46,13 @@ requires-python = ">=3.9"
 
 [tool.hatch.metadata.hooks.uv-dynamic-versioning]
 dependencies = [
-    "pydantic-ai-slim[openai,vertexai,google,groq,anthropic,mistral,cohere,bedrock,cli,mcp,evals,a2a]=={{ version }}",
+    "pydantic-ai-slim[openai,vertexai,google,groq,anthropic,mistral,cohere,bedrock,cli,mcp,evals]=={{ version }}",
 ]
 
 [tool.hatch.metadata.hooks.uv-dynamic-versioning.optional-dependencies]
 examples = ["pydantic-ai-examples=={{ version }}"]
 logfire = ["logfire>=3.11.0"]
+a2a = ["fasta2a=={{ version }}"]
 
 [project.urls]
 Homepage = "https://ai.pydantic.dev"
@@ -84,6 +85,7 @@ members = [
 # dev dependencies are defined in `pydantic-ai-slim/pyproject.toml` to allow for minimal testing
 lint = ["mypy>=1.11.2", "pyright>=1.1.390", "ruff>=0.6.9"]
 docs = [
+    "pydantic-ai[a2a]",
     "black>=24.10.0",
     "mkdocs>=1.6.1",
     "mkdocs-glightbox>=0.4.0",

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/evals/test_dataset.py

@@ -1,6 +1,5 @@
 from __future__ import annotations as _annotations
 
-import asyncio
 import json
 import sys
 from dataclasses import dataclass
@@ -8,7 +7,7 @@ from pathlib import Path
 from typing import Any
 
 import pytest
-from dirty_equals import HasRepr
+from dirty_equals import HasRepr, IsNumber
 from inline_snapshot import snapshot
 from pydantic import BaseModel
 
@@ -33,7 +32,7 @@ with try_import() as imports_successful:
         def evaluate(self, ctx: EvaluatorContext[object, object, object]) -> EvaluatorOutput:
             return self.output
 
-    from pydantic_evals.reporting import ReportCase
+    from pydantic_evals.reporting import ReportCase, ReportCaseAdapter
 
 pytestmark = [pytest.mark.skipif(not imports_successful(), reason='pydantic-evals not installed'), pytest.mark.anyio]
 
@@ -178,25 +177,25 @@ async def test_add_evaluator(
     }
 
 
-async def test_evaluate(
+async def test_evaluate_async(
     example_dataset: Dataset[TaskInput, TaskOutput, TaskMetadata],
     simple_evaluator: type[Evaluator[TaskInput, TaskOutput, TaskMetadata]],
 ):
     """Test evaluating a dataset."""
     example_dataset.add_evaluator(simple_evaluator())
 
-    async def mock_task(inputs: TaskInput) -> TaskOutput:
+    async def mock_async_task(inputs: TaskInput) -> TaskOutput:
         if inputs.query == 'What is 2+2?':
             return TaskOutput(answer='4')
         elif inputs.query == 'What is the capital of France?':
             return TaskOutput(answer='Paris')
         return TaskOutput(answer='Unknown')  # pragma: no cover
 
-    report = await example_dataset.evaluate(mock_task)
+    report = await example_dataset.evaluate(mock_async_task)
 
     assert report is not None
     assert len(report.cases) == 2
-    assert report.cases[0].model_dump() == snapshot(
+    assert ReportCaseAdapter.dump_python(report.cases[0]) == snapshot(
         {
             'assertions': {
                 'correct': {
@@ -230,6 +229,58 @@ async def test_evaluate(
     )
 
 
+async def test_evaluate_sync(
+    example_dataset: Dataset[TaskInput, TaskOutput, TaskMetadata],
+    simple_evaluator: type[Evaluator[TaskInput, TaskOutput, TaskMetadata]],
+):
+    """Test evaluating a dataset."""
+    example_dataset.add_evaluator(simple_evaluator())
+
+    def mock_sync_task(inputs: TaskInput) -> TaskOutput:
+        if inputs.query == 'What is 2+2?':
+            return TaskOutput(answer='4')
+        elif inputs.query == 'What is the capital of France?':
+            return TaskOutput(answer='Paris')
+        return TaskOutput(answer='Unknown')  # pragma: no cover
+
+    report = await example_dataset.evaluate(mock_sync_task)
+
+    assert report is not None
+    assert len(report.cases) == 2
+    assert ReportCaseAdapter.dump_python(report.cases[0]) == snapshot(
+        {
+            'assertions': {
+                'correct': {
+                    'name': 'correct',
+                    'reason': None,
+                    'source': {'name': 'SimpleEvaluator', 'arguments': None},
+                    'value': True,
+                }
+            },
+            'attributes': {},
+            'expected_output': {'answer': '4', 'confidence': 1.0},
+            'inputs': {'query': 'What is 2+2?'},
+            'labels': {},
+            'metadata': {'category': 'general', 'difficulty': 'easy'},
+            'metrics': {},
+            'name': 'case1',
+            'output': {'answer': '4', 'confidence': 1.0},
+            'scores': {
+                'confidence': {
+                    'name': 'confidence',
+                    'reason': None,
+                    'source': {'name': 'SimpleEvaluator', 'arguments': None},
+                    'value': 1.0,
+                }
+            },
+            'span_id': '0000000000000003',
+            'task_duration': IsNumber(),  # the runtime behavior is not deterministic due to threading
+            'total_duration': IsNumber(),  # the runtime behavior is not deterministic due to threading
+            'trace_id': '00000000000000000000000000000001',
+        }
+    )
+
+
 async def test_evaluate_with_concurrency(
     example_dataset: Dataset[TaskInput, TaskOutput, TaskMetadata],
     simple_evaluator: type[Evaluator[TaskInput, TaskOutput, TaskMetadata]],
@@ -248,7 +299,7 @@ async def test_evaluate_with_concurrency(
 
     assert report is not None
     assert len(report.cases) == 2
-    assert report.cases[0].model_dump() == snapshot(
+    assert ReportCaseAdapter.dump_python(report.cases[0]) == snapshot(
         {
             'assertions': {
                 'correct': {
@@ -828,8 +879,8 @@ async def test_dataset_evaluate_with_sync_task(example_dataset: Dataset[TaskInpu
     def sync_task(inputs: TaskInput) -> TaskOutput:
         return TaskOutput(answer=inputs.query.upper())
 
-    report = await example_dataset.evaluate(lambda x: asyncio.sleep(0, sync_task(x)))
-    assert report.name == '<lambda>'
+    report = await example_dataset.evaluate(sync_task)
+    assert report.name == 'sync_task'
     assert len(report.cases) == 2
 
 

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/evals/test_otel.py

@@ -65,9 +65,10 @@ async def test_context_subtree_concurrent():
     task1_root = tree1.roots[0]
     assert len(task1_root.children) == 2, 'task1 should have exactly two children'
     task1_child_names = {child.name for child in task1_root.children}
-    assert task1_child_names == {'task1_child1', 'task1_child2'}, (
-        "task1's children should be task1_child1 and task1_child2"
-    )
+    assert task1_child_names == {
+        'task1_child1',
+        'task1_child2',
+    }, "task1's children should be task1_child1 and task1_child2"
 
     # Verify that tree2 only contains spans from task2
     assert len(tree2.roots) == 1, 'tree2 should have exactly one root span'

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/evals/test_reports.py

@@ -12,9 +12,11 @@ with try_import() as imports_successful:
     from pydantic_evals.evaluators import EvaluationResult, Evaluator, EvaluatorContext
     from pydantic_evals.reporting import (
         EvaluationReport,
+        EvaluationReportAdapter,
         RenderNumberConfig,
         RenderValueConfig,
         ReportCase,
+        ReportCaseAdapter,
         ReportCaseAggregate,
     )
 
@@ -157,7 +159,7 @@ async def test_report_case_aggregate():
 async def test_report_serialization(sample_report: EvaluationReport):
     """Test serializing a report to dict."""
     # Serialize the report
-    serialized = sample_report.model_dump()
+    serialized = EvaluationReportAdapter.dump_python(sample_report)
 
     # Check the serialized structure
     assert 'cases' in serialized
@@ -202,7 +204,7 @@ async def test_report_with_error(mock_evaluator: Evaluator[TaskInput, TaskOutput
         name='error_report',
     )
 
-    assert report.cases[0].model_dump() == snapshot(
+    assert ReportCaseAdapter.dump_python(report.cases[0]) == snapshot(
         {
             'assertions': {
                 'error_evaluator': {

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/fasta2a/test_applications.py

@@ -34,12 +34,13 @@ async def test_agent_card():
         assert response.json() == snapshot(
             {
                 'name': 'Agent',
+                'description': 'FastA2A Agent',
                 'url': 'http://localhost:8000',
                 'version': '1.0.0',
+                'protocolVersion': '0.2.5',
                 'skills': [],
                 'defaultInputModes': ['application/json'],
                 'defaultOutputModes': ['application/json'],
                 'capabilities': {'streaming': False, 'pushNotifications': False, 'stateTransitionHistory': False},
-                'authentication': {'schemes': []},
             }
         )

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/models/test_bedrock.py

@@ -1443,9 +1443,9 @@ async def test_bedrock_mistral_tool_result_format(bedrock_provider: BedrockProvi
 
 async def test_bedrock_anthropic_no_tool_choice(bedrock_provider: BedrockProvider):
     my_tool = ToolDefinition(
-        'my_tool',
-        'This is my tool',
-        {'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
+        name='my_tool',
+        description='This is my tool',
+        parameters_json_schema={'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
     )
     mrp = ModelRequestParameters(output_mode='tool', function_tools=[my_tool], allow_text_output=False, output_tools=[])
 

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/models/test_gemini.py

@@ -83,14 +83,18 @@ async def test_model_tools(allow_model_requests: None):
     m = GeminiModel('gemini-1.5-flash', provider=GoogleGLAProvider(api_key='via-arg'))
     tools = [
         ToolDefinition(
-            'foo',
-            'This is foo',
-            {'type': 'object', 'title': 'Foo', 'properties': {'bar': {'type': 'number', 'title': 'Bar'}}},
+            name='foo',
+            description='This is foo',
+            parameters_json_schema={
+                'type': 'object',
+                'title': 'Foo',
+                'properties': {'bar': {'type': 'number', 'title': 'Bar'}},
+            },
         ),
         ToolDefinition(
-            'apple',
-            'This is apple',
-            {
+            name='apple',
+            description='This is apple',
+            parameters_json_schema={
                 'type': 'object',
                 'properties': {
                     'banana': {'type': 'array', 'title': 'Banana', 'items': {'type': 'number', 'title': 'Bar'}}
@@ -99,9 +103,14 @@ async def test_model_tools(allow_model_requests: None):
         ),
     ]
     output_tool = ToolDefinition(
-        'result',
-        'This is the tool for the final Result',
-        {'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}, 'required': ['spam']},
+        name='result',
+        description='This is the tool for the final Result',
+        parameters_json_schema={
+            'type': 'object',
+            'title': 'Result',
+            'properties': {'spam': {'type': 'number'}},
+            'required': ['spam'],
+        },
     )
 
     mrp = ModelRequestParameters(
@@ -148,9 +157,9 @@ async def test_model_tools(allow_model_requests: None):
 async def test_require_response_tool(allow_model_requests: None):
     m = GeminiModel('gemini-1.5-flash', provider=GoogleGLAProvider(api_key='via-arg'))
     output_tool = ToolDefinition(
-        'result',
-        'This is the tool for the final Result',
-        {'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
+        name='result',
+        description='This is the tool for the final Result',
+        parameters_json_schema={'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
     )
     mrp = ModelRequestParameters(
         function_tools=[],
@@ -235,9 +244,9 @@ async def test_json_def_replaced(allow_model_requests: None):
 
     m = GeminiModel('gemini-1.5-flash', provider=GoogleGLAProvider(api_key='via-arg'))
     output_tool = ToolDefinition(
-        'result',
-        'This is the tool for the final Result',
-        json_schema,
+        name='result',
+        description='This is the tool for the final Result',
+        parameters_json_schema=json_schema,
     )
     mrp = ModelRequestParameters(
         function_tools=[],
@@ -320,9 +329,9 @@ async def test_json_def_enum(allow_model_requests: None):
     )
     m = GeminiModel('gemini-1.5-flash', provider=GoogleGLAProvider(api_key='via-arg'))
     output_tool = ToolDefinition(
-        'result',
-        'This is the tool for the final Result',
-        json_schema,
+        name='result',
+        description='This is the tool for the final Result',
+        parameters_json_schema=json_schema,
     )
     mrp = ModelRequestParameters(
         function_tools=[],
@@ -368,9 +377,9 @@ async def test_json_def_replaced_any_of(allow_model_requests: None):
 
     m = GeminiModel('gemini-1.5-flash', provider=GoogleGLAProvider(api_key='via-arg'))
     output_tool = ToolDefinition(
-        'result',
-        'This is the tool for the final Result',
-        json_schema,
+        name='result',
+        description='This is the tool for the final Result',
+        parameters_json_schema=json_schema,
     )
     mrp = ModelRequestParameters(
         function_tools=[],
@@ -437,9 +446,9 @@ async def test_json_def_recursive(allow_model_requests: None):
 
     m = GeminiModel('gemini-1.5-flash', provider=GoogleGLAProvider(api_key='via-arg'))
     output_tool = ToolDefinition(
-        'result',
-        'This is the tool for the final Result',
-        json_schema,
+        name='result',
+        description='This is the tool for the final Result',
+        parameters_json_schema=json_schema,
     )
     with pytest.raises(UserError, match=r'Recursive `\$ref`s in JSON Schema are not supported by Gemini'):
         mrp = ModelRequestParameters(
@@ -476,9 +485,9 @@ async def test_json_def_date(allow_model_requests: None):
 
     m = GeminiModel('gemini-1.5-flash', provider=GoogleGLAProvider(api_key='via-arg'))
     output_tool = ToolDefinition(
-        'result',
-        'This is the tool for the final Result',
-        json_schema,
+        name='result',
+        description='This is the tool for the final Result',
+        parameters_json_schema=json_schema,
     )
     mrp = ModelRequestParameters(
         function_tools=[],

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/models/test_openai.py

@@ -1095,7 +1095,10 @@ def tool_with_tuples(x: tuple[int], y: tuple[str] = ('abc',)) -> str:
                     '$defs': {
                         'MyDefaultRecursiveDc': {
                             'properties': {
-                                'field': {'anyOf': [{'$ref': '#/$defs/MyDefaultRecursiveDc'}, {'type': 'null'}]}
+                                'field': {
+                                    'anyOf': [{'$ref': '#/$defs/MyDefaultRecursiveDc'}, {'type': 'null'}],
+                                    'default': None,
+                                }
                             },
                             'type': 'object',
                         },
@@ -1213,7 +1216,7 @@ def tool_with_tuples(x: tuple[int], y: tuple[str] = ('abc',)) -> str:
                 {
                     '$defs': {
                         'MyDefaultDc': {
-                            'properties': {'x': {'type': 'integer'}},
+                            'properties': {'x': {'default': 1, 'type': 'integer'}},
                             'type': 'object',
                         }
                     },
@@ -1253,7 +1256,7 @@ def tool_with_tuples(x: tuple[int], y: tuple[str] = ('abc',)) -> str:
                 {
                     '$defs': {
                         'MyDefaultDc': {
-                            'properties': {'x': {'type': 'integer'}},
+                            'properties': {'x': {'default': 1, 'type': 'integer'}},
                             'type': 'object',
                         }
                     },
@@ -1293,7 +1296,7 @@ def tool_with_tuples(x: tuple[int], y: tuple[str] = ('abc',)) -> str:
                 {
                     '$defs': {
                         'MyDefaultDc': {
-                            'properties': {'x': {'type': 'integer'}},
+                            'properties': {'x': {'default': 1, 'type': 'integer'}},
                             'type': 'object',
                         }
                     },
@@ -1913,9 +1916,9 @@ def test_openai_model_profile_from_provider():
 
 def test_model_profile_strict_not_supported():
     my_tool = ToolDefinition(
-        'my_tool',
-        'This is my tool',
-        {'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
+        name='my_tool',
+        description='This is my tool',
+        parameters_json_schema={'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
         strict=True,
     )
 

{pydantic_ai-0.3.7 → pydantic_ai-0.4.1}/tests/models/test_openai_responses.py

@@ -473,9 +473,9 @@ async def test_openai_responses_model_instructions(allow_model_requests: None, o
 
 def test_model_profile_strict_not_supported():
     my_tool = ToolDefinition(
-        'my_tool',
-        'This is my tool',
-        {'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
+        name='my_tool',
+        description='This is my tool',
+        parameters_json_schema={'type': 'object', 'title': 'Result', 'properties': {'spam': {'type': 'number'}}},
         strict=True,
     )