openai-agents 0.2.8__py3-none-any.whl → 0.2.10__py3-none-any.whl

agents/memory/sqlite_session.py

@@ -0,0 +1,275 @@
+from __future__ import annotations
+
+import asyncio
+import json
+import sqlite3
+import threading
+from pathlib import Path
+
+from ..items import TResponseInputItem
+from .session import SessionABC
+
+
+class SQLiteSession(SessionABC):
+    """SQLite-based implementation of session storage.
+
+    This implementation stores conversation history in a SQLite database.
+    By default, uses an in-memory database that is lost when the process ends.
+    For persistent storage, provide a file path.
+    """
+
+    def __init__(
+        self,
+        session_id: str,
+        db_path: str | Path = ":memory:",
+        sessions_table: str = "agent_sessions",
+        messages_table: str = "agent_messages",
+    ):
+        """Initialize the SQLite session.
+
+        Args:
+            session_id: Unique identifier for the conversation session
+            db_path: Path to the SQLite database file. Defaults to ':memory:' (in-memory database)
+            sessions_table: Name of the table to store session metadata. Defaults to
+                'agent_sessions'
+            messages_table: Name of the table to store message data. Defaults to 'agent_messages'
+        """
+        self.session_id = session_id
+        self.db_path = db_path
+        self.sessions_table = sessions_table
+        self.messages_table = messages_table
+        self._local = threading.local()
+        self._lock = threading.Lock()
+
+        # For in-memory databases, we need a shared connection to avoid thread isolation
+        # For file databases, we use thread-local connections for better concurrency
+        self._is_memory_db = str(db_path) == ":memory:"
+        if self._is_memory_db:
+            self._shared_connection = sqlite3.connect(":memory:", check_same_thread=False)
+            self._shared_connection.execute("PRAGMA journal_mode=WAL")
+            self._init_db_for_connection(self._shared_connection)
+        else:
+            # For file databases, initialize the schema once since it persists
+            init_conn = sqlite3.connect(str(self.db_path), check_same_thread=False)
+            init_conn.execute("PRAGMA journal_mode=WAL")
+            self._init_db_for_connection(init_conn)
+            init_conn.close()
+
+    def _get_connection(self) -> sqlite3.Connection:
+        """Get a database connection."""
+        if self._is_memory_db:
+            # Use shared connection for in-memory database to avoid thread isolation
+            return self._shared_connection
+        else:
+            # Use thread-local connections for file databases
+            if not hasattr(self._local, "connection"):
+                self._local.connection = sqlite3.connect(
+                    str(self.db_path),
+                    check_same_thread=False,
+                )
+                self._local.connection.execute("PRAGMA journal_mode=WAL")
+            assert isinstance(self._local.connection, sqlite3.Connection), (
+                f"Expected sqlite3.Connection, got {type(self._local.connection)}"
+            )
+            return self._local.connection
+
+    def _init_db_for_connection(self, conn: sqlite3.Connection) -> None:
+        """Initialize the database schema for a specific connection."""
+        conn.execute(
+            f"""
+            CREATE TABLE IF NOT EXISTS {self.sessions_table} (
+                session_id TEXT PRIMARY KEY,
+                created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
+                updated_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
+            )
+        """
+        )
+
+        conn.execute(
+            f"""
+            CREATE TABLE IF NOT EXISTS {self.messages_table} (
+                id INTEGER PRIMARY KEY AUTOINCREMENT,
+                session_id TEXT NOT NULL,
+                message_data TEXT NOT NULL,
+                created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
+                FOREIGN KEY (session_id) REFERENCES {self.sessions_table} (session_id)
+                    ON DELETE CASCADE
+            )
+        """
+        )
+
+        conn.execute(
+            f"""
+            CREATE INDEX IF NOT EXISTS idx_{self.messages_table}_session_id
+            ON {self.messages_table} (session_id, created_at)
+        """
+        )
+
+        conn.commit()
+
+    async def get_items(self, limit: int | None = None) -> list[TResponseInputItem]:
+        """Retrieve the conversation history for this session.
+
+        Args:
+            limit: Maximum number of items to retrieve. If None, retrieves all items.
+                   When specified, returns the latest N items in chronological order.
+
+        Returns:
+            List of input items representing the conversation history
+        """
+
+        def _get_items_sync():
+            conn = self._get_connection()
+            with self._lock if self._is_memory_db else threading.Lock():
+                if limit is None:
+                    # Fetch all items in chronological order
+                    cursor = conn.execute(
+                        f"""
+                        SELECT message_data FROM {self.messages_table}
+                        WHERE session_id = ?
+                        ORDER BY created_at ASC
+                    """,
+                        (self.session_id,),
+                    )
+                else:
+                    # Fetch the latest N items in chronological order
+                    cursor = conn.execute(
+                        f"""
+                        SELECT message_data FROM {self.messages_table}
+                        WHERE session_id = ?
+                        ORDER BY created_at DESC
+                        LIMIT ?
+                        """,
+                        (self.session_id, limit),
+                    )
+
+                rows = cursor.fetchall()
+
+                # Reverse to get chronological order when using DESC
+                if limit is not None:
+                    rows = list(reversed(rows))
+
+                items = []
+                for (message_data,) in rows:
+                    try:
+                        item = json.loads(message_data)
+                        items.append(item)
+                    except json.JSONDecodeError:
+                        # Skip invalid JSON entries
+                        continue
+
+                return items
+
+        return await asyncio.to_thread(_get_items_sync)
+
+    async def add_items(self, items: list[TResponseInputItem]) -> None:
+        """Add new items to the conversation history.
+
+        Args:
+            items: List of input items to add to the history
+        """
+        if not items:
+            return
+
+        def _add_items_sync():
+            conn = self._get_connection()
+
+            with self._lock if self._is_memory_db else threading.Lock():
+                # Ensure session exists
+                conn.execute(
+                    f"""
+                    INSERT OR IGNORE INTO {self.sessions_table} (session_id) VALUES (?)
+                """,
+                    (self.session_id,),
+                )
+
+                # Add items
+                message_data = [(self.session_id, json.dumps(item)) for item in items]
+                conn.executemany(
+                    f"""
+                    INSERT INTO {self.messages_table} (session_id, message_data) VALUES (?, ?)
+                """,
+                    message_data,
+                )
+
+                # Update session timestamp
+                conn.execute(
+                    f"""
+                    UPDATE {self.sessions_table}
+                    SET updated_at = CURRENT_TIMESTAMP
+                    WHERE session_id = ?
+                """,
+                    (self.session_id,),
+                )
+
+                conn.commit()
+
+        await asyncio.to_thread(_add_items_sync)
+
+    async def pop_item(self) -> TResponseInputItem | None:
+        """Remove and return the most recent item from the session.
+
+        Returns:
+            The most recent item if it exists, None if the session is empty
+        """
+
+        def _pop_item_sync():
+            conn = self._get_connection()
+            with self._lock if self._is_memory_db else threading.Lock():
+                # Use DELETE with RETURNING to atomically delete and return the most recent item
+                cursor = conn.execute(
+                    f"""
+                    DELETE FROM {self.messages_table}
+                    WHERE id = (
+                        SELECT id FROM {self.messages_table}
+                        WHERE session_id = ?
+                        ORDER BY created_at DESC
+                        LIMIT 1
+                    )
+                    RETURNING message_data
+                    """,
+                    (self.session_id,),
+                )
+
+                result = cursor.fetchone()
+                conn.commit()
+
+                if result:
+                    message_data = result[0]
+                    try:
+                        item = json.loads(message_data)
+                        return item
+                    except json.JSONDecodeError:
+                        # Return None for corrupted JSON entries (already deleted)
+                        return None
+
+                return None
+
+        return await asyncio.to_thread(_pop_item_sync)
+
+    async def clear_session(self) -> None:
+        """Clear all items for this session."""
+
+        def _clear_session_sync():
+            conn = self._get_connection()
+            with self._lock if self._is_memory_db else threading.Lock():
+                conn.execute(
+                    f"DELETE FROM {self.messages_table} WHERE session_id = ?",
+                    (self.session_id,),
+                )
+                conn.execute(
+                    f"DELETE FROM {self.sessions_table} WHERE session_id = ?",
+                    (self.session_id,),
+                )
+                conn.commit()
+
+        await asyncio.to_thread(_clear_session_sync)
+
+    def close(self) -> None:
+        """Close the database connection."""
+        if self._is_memory_db:
+            if hasattr(self, "_shared_connection"):
+                self._shared_connection.close()
+        else:
+            if hasattr(self._local, "connection"):
+                self._local.connection.close()

agents/model_settings.py

@@ -55,7 +55,6 @@ Headers: TypeAlias = Mapping[str, Union[str, Omit]]
 ToolChoice: TypeAlias = Union[Literal["auto", "required", "none"], str, MCPToolChoice, None]
 
 
-
 @dataclass
 class ModelSettings:
     """Settings to use when calling an LLM.
@@ -92,7 +91,10 @@ class ModelSettings:
     """
 
     truncation: Literal["auto", "disabled"] | None = None
-    """The truncation strategy to use when calling the model."""
+    """The truncation strategy to use when calling the model.
+    See [Responses API documentation](https://platform.openai.com/docs/api-reference/responses/create#responses_create-truncation)
+    for more details.
+    """
 
     max_tokens: int | None = None
     """The maximum number of output tokens to generate."""
@@ -118,7 +120,10 @@ class ModelSettings:
     """Whether to include usage chunk.
     Only available for Chat Completions API."""
 
-    response_include: list[ResponseIncludable] | None = None
+    # TODO: revisit ResponseIncludable | str if ResponseIncludable covers more cases
+    # We've added str to support missing ones like
+    # "web_search_call.action.sources" etc.
+    response_include: list[ResponseIncludable | str] | None = None
     """Additional output data to include in the model response.
     [include parameter](https://platform.openai.com/docs/api-reference/responses/create#responses-create-include)"""
 

agents/models/__init__.py

@@ -0,0 +1,13 @@
+from .default_models import (
+    get_default_model,
+    get_default_model_settings,
+    gpt_5_reasoning_settings_required,
+    is_gpt_5_default,
+)
+
+__all__ = [
+    "get_default_model",
+    "get_default_model_settings",
+    "gpt_5_reasoning_settings_required",
+    "is_gpt_5_default",
+]

agents/models/chatcmpl_converter.py

@@ -271,11 +271,16 @@ class Converter:
                     raise UserError(
                         f"Only file_data is supported for input_file {casted_file_param}"
                     )
+                if "filename" not in casted_file_param or not casted_file_param["filename"]:
+                    raise UserError(
+                        f"filename must be provided for input_file {casted_file_param}"
+                    )
                 out.append(
                     File(
                         type="file",
                         file=FileFile(
                             file_data=casted_file_param["file_data"],
+                            filename=casted_file_param["filename"],
                         ),
                     )
                 )

agents/models/chatcmpl_stream_handler.py

@@ -28,11 +28,17 @@ from openai.types.responses import (
     ResponseTextDeltaEvent,
     ResponseUsage,
 )
-from openai.types.responses.response_reasoning_item import Summary
+from openai.types.responses.response_reasoning_item import Content, Summary
 from openai.types.responses.response_reasoning_summary_part_added_event import (
     Part as AddedEventPart,
 )
 from openai.types.responses.response_reasoning_summary_part_done_event import Part as DoneEventPart
+from openai.types.responses.response_reasoning_text_delta_event import (
+    ResponseReasoningTextDeltaEvent,
+)
+from openai.types.responses.response_reasoning_text_done_event import (
+    ResponseReasoningTextDoneEvent,
+)
 from openai.types.responses.response_usage import InputTokensDetails, OutputTokensDetails
 
 from ..items import TResponseStreamEvent
@@ -95,7 +101,7 @@ class ChatCmplStreamHandler:
 
             delta = chunk.choices[0].delta
 
-            # Handle reasoning content
+            # Handle reasoning content for reasoning summaries
             if hasattr(delta, "reasoning_content"):
                 reasoning_content = delta.reasoning_content
                 if reasoning_content and not state.reasoning_content_index_and_output:
@@ -138,10 +144,55 @@ class ChatCmplStreamHandler:
                     )
 
                     # Create a new summary with updated text
-                    current_summary = state.reasoning_content_index_and_output[1].summary[0]
-                    updated_text = current_summary.text + reasoning_content
-                    new_summary = Summary(text=updated_text, type="summary_text")
-                    state.reasoning_content_index_and_output[1].summary[0] = new_summary
+                    current_content = state.reasoning_content_index_and_output[1].summary[0]
+                    updated_text = current_content.text + reasoning_content
+                    new_content = Summary(text=updated_text, type="summary_text")
+                    state.reasoning_content_index_and_output[1].summary[0] = new_content
+
+            # Handle reasoning content from 3rd party platforms
+            if hasattr(delta, "reasoning"):
+                reasoning_text = delta.reasoning
+                if reasoning_text and not state.reasoning_content_index_and_output:
+                    state.reasoning_content_index_and_output = (
+                        0,
+                        ResponseReasoningItem(
+                            id=FAKE_RESPONSES_ID,
+                            summary=[],
+                            content=[Content(text="", type="reasoning_text")],
+                            type="reasoning",
+                        ),
+                    )
+                    yield ResponseOutputItemAddedEvent(
+                        item=ResponseReasoningItem(
+                            id=FAKE_RESPONSES_ID,
+                            summary=[],
+                            content=[Content(text="", type="reasoning_text")],
+                            type="reasoning",
+                        ),
+                        output_index=0,
+                        type="response.output_item.added",
+                        sequence_number=sequence_number.get_and_increment(),
+                    )
+
+                if reasoning_text and state.reasoning_content_index_and_output:
+                    yield ResponseReasoningTextDeltaEvent(
+                        delta=reasoning_text,
+                        item_id=FAKE_RESPONSES_ID,
+                        output_index=0,
+                        content_index=0,
+                        type="response.reasoning_text.delta",
+                        sequence_number=sequence_number.get_and_increment(),
+                    )
+
+                    # Create a new summary with updated text
+                    if state.reasoning_content_index_and_output[1].content is None:
+                        state.reasoning_content_index_and_output[1].content = [
+                            Content(text="", type="reasoning_text")
+                        ]
+                    current_text = state.reasoning_content_index_and_output[1].content[0]
+                    updated_text = current_text.text + reasoning_text
+                    new_text_content = Content(text=updated_text, type="reasoning_text")
+                    state.reasoning_content_index_and_output[1].content[0] = new_text_content
 
             # Handle regular content
             if delta.content is not None:
@@ -344,17 +395,30 @@ class ChatCmplStreamHandler:
                         )
 
         if state.reasoning_content_index_and_output:
-            yield ResponseReasoningSummaryPartDoneEvent(
-                item_id=FAKE_RESPONSES_ID,
-                output_index=0,
-                summary_index=0,
-                part=DoneEventPart(
-                    text=state.reasoning_content_index_and_output[1].summary[0].text,
-                    type="summary_text",
-                ),
-                type="response.reasoning_summary_part.done",
-                sequence_number=sequence_number.get_and_increment(),
-            )
+            if (
+                state.reasoning_content_index_and_output[1].summary
+                and len(state.reasoning_content_index_and_output[1].summary) > 0
+            ):
+                yield ResponseReasoningSummaryPartDoneEvent(
+                    item_id=FAKE_RESPONSES_ID,
+                    output_index=0,
+                    summary_index=0,
+                    part=DoneEventPart(
+                        text=state.reasoning_content_index_and_output[1].summary[0].text,
+                        type="summary_text",
+                    ),
+                    type="response.reasoning_summary_part.done",
+                    sequence_number=sequence_number.get_and_increment(),
+                )
+            elif state.reasoning_content_index_and_output[1].content is not None:
+                yield ResponseReasoningTextDoneEvent(
+                    item_id=FAKE_RESPONSES_ID,
+                    output_index=0,
+                    content_index=0,
+                    text=state.reasoning_content_index_and_output[1].content[0].text,
+                    type="response.reasoning_text.done",
+                    sequence_number=sequence_number.get_and_increment(),
+                )
             yield ResponseOutputItemDoneEvent(
                 item=state.reasoning_content_index_and_output[1],
                 output_index=0,

agents/models/default_models.py

@@ -0,0 +1,58 @@
+import copy
+import os
+from typing import Optional
+
+from openai.types.shared.reasoning import Reasoning
+
+from agents.model_settings import ModelSettings
+
+OPENAI_DEFAULT_MODEL_ENV_VARIABLE_NAME = "OPENAI_DEFAULT_MODEL"
+
+# discourage directly accessing this constant
+# use the get_default_model and get_default_model_settings() functions instead
+_GPT_5_DEFAULT_MODEL_SETTINGS: ModelSettings = ModelSettings(
+    # We chose "low" instead of "minimal" because some of the built-in tools
+    # (e.g., file search, image generation, etc.) do not support "minimal"
+    # If you want to use "minimal" reasoning effort, you can pass your own model settings
+    reasoning=Reasoning(effort="low"),
+    verbosity="low",
+)
+
+
+def gpt_5_reasoning_settings_required(model_name: str) -> bool:
+    """
+    Returns True if the model name is a GPT-5 model and reasoning settings are required.
+    """
+    if model_name.startswith("gpt-5-chat"):
+        # gpt-5-chat-latest does not require reasoning settings
+        return False
+    # matches any of gpt-5 models
+    return model_name.startswith("gpt-5")
+
+
+def is_gpt_5_default() -> bool:
+    """
+    Returns True if the default model is a GPT-5 model.
+    This is used to determine if the default model settings are compatible with GPT-5 models.
+    If the default model is not a GPT-5 model, the model settings are compatible with other models.
+    """
+    return gpt_5_reasoning_settings_required(get_default_model())
+
+
+def get_default_model() -> str:
+    """
+    Returns the default model name.
+    """
+    return os.getenv(OPENAI_DEFAULT_MODEL_ENV_VARIABLE_NAME, "gpt-4.1").lower()
+
+
+def get_default_model_settings(model: Optional[str] = None) -> ModelSettings:
+    """
+    Returns the default model settings.
+    If the default model is a GPT-5 model, returns the GPT-5 default model settings.
+    Otherwise, returns the legacy default model settings.
+    """
+    _model = model if model is not None else get_default_model()
+    if gpt_5_reasoning_settings_required(_model):
+        return copy.deepcopy(_GPT_5_DEFAULT_MODEL_SETTINGS)
+    return ModelSettings()

agents/models/interface.py

@@ -48,6 +48,7 @@ class Model(abc.ABC):
         tracing: ModelTracing,
         *,
         previous_response_id: str | None,
+        conversation_id: str | None,
         prompt: ResponsePromptParam | None,
     ) -> ModelResponse:
         """Get a response from the model.
@@ -62,6 +63,7 @@ class Model(abc.ABC):
             tracing: Tracing configuration.
             previous_response_id: the ID of the previous response. Generally not used by the model,
                 except for the OpenAI Responses API.
+            conversation_id: The ID of the stored conversation, if any.
             prompt: The prompt config to use for the model.
 
         Returns:
@@ -81,6 +83,7 @@ class Model(abc.ABC):
         tracing: ModelTracing,
         *,
         previous_response_id: str | None,
+        conversation_id: str | None,
         prompt: ResponsePromptParam | None,
     ) -> AsyncIterator[TResponseStreamEvent]:
         """Stream a response from the model.
@@ -95,6 +98,7 @@ class Model(abc.ABC):
             tracing: Tracing configuration.
             previous_response_id: the ID of the previous response. Generally not used by the model,
                 except for the OpenAI Responses API.
+            conversation_id: The ID of the stored conversation, if any.
             prompt: The prompt config to use for the model.
 
         Returns:

agents/models/openai_chatcompletions.py

@@ -55,7 +55,8 @@ class OpenAIChatCompletionsModel(Model):
         output_schema: AgentOutputSchemaBase | None,
         handoffs: list[Handoff],
         tracing: ModelTracing,
-        previous_response_id: str | None,
+        previous_response_id: str | None = None,  # unused
+        conversation_id: str | None = None,  # unused
         prompt: ResponsePromptParam | None = None,
     ) -> ModelResponse:
         with generation_span(
@@ -142,7 +143,8 @@ class OpenAIChatCompletionsModel(Model):
         output_schema: AgentOutputSchemaBase | None,
         handoffs: list[Handoff],
         tracing: ModelTracing,
-        previous_response_id: str | None,
+        previous_response_id: str | None = None,  # unused
+        conversation_id: str | None = None,  # unused
         prompt: ResponsePromptParam | None = None,
     ) -> AsyncIterator[TResponseStreamEvent]:
         """

agents/models/openai_provider.py

@@ -4,10 +4,12 @@ import httpx
 from openai import AsyncOpenAI, DefaultAsyncHttpxClient
 
 from . import _openai_shared
+from .default_models import get_default_model
 from .interface import Model, ModelProvider
 from .openai_chatcompletions import OpenAIChatCompletionsModel
 from .openai_responses import OpenAIResponsesModel
 
+# This is kept for backward compatiblity but using get_default_model() method is recommended.
 DEFAULT_MODEL: str = "gpt-4o"
 
 
@@ -80,7 +82,7 @@ class OpenAIProvider(ModelProvider):
 
     def get_model(self, model_name: str | None) -> Model:
         if model_name is None:
-            model_name = DEFAULT_MODEL
+            model_name = get_default_model()
 
         client = self._get_client()