MAINT: Convert leaf model types to Pydanticv2 (microsoft#1769)

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: rlundeen2

pyrit/memory/memory_models.py

@@ -824,7 +824,7 @@ def __init__(self, *, entry: AttackResult) -> None:
 
         # Retry events
         self.retry_events_json = (
-            json.dumps([evt.to_dict() for evt in entry.retry_events]) if entry.retry_events else None
+            json.dumps([evt.model_dump(mode="json") for evt in entry.retry_events]) if entry.retry_events else None
         )
         self.total_retries = entry.total_retries
 
@@ -923,7 +923,7 @@ def get_attack_result(self) -> AttackResult:
         if self.retry_events_json:
             from pyrit.models.retry_event import RetryEvent
 
-            retry_events = [RetryEvent.from_dict(evt_dict) for evt_dict in json.loads(self.retry_events_json)]
+            retry_events = [RetryEvent.model_validate(evt_dict) for evt_dict in json.loads(self.retry_events_json)]
 
         return AttackResult(
             conversation_id=self.conversation_id,

pyrit/models/attack_result.py

@@ -250,15 +250,15 @@ def to_dict(self) -> dict[str, Any]:
             "outcome_reason": self.outcome_reason,
             "timestamp": self.timestamp.isoformat(),
             "related_conversations": sorted(
-                [ref.to_dict() for ref in self.related_conversations],
+                [ref.model_dump(mode="json") for ref in self.related_conversations],
                 key=lambda r: r["conversation_id"],
             ),
             "metadata": self.metadata,
             "labels": self.labels,
             "error_message": self.error_message,
             "error_type": self.error_type,
             "error_traceback": self.error_traceback,
-            "retry_events": [e.to_dict() for e in self.retry_events],
+            "retry_events": [e.model_dump(mode="json") for e in self.retry_events],
             "total_retries": self.total_retries,
         }
 
@@ -291,13 +291,15 @@ def from_dict(cls, data: dict[str, Any]) -> AttackResult:
             timestamp=(
                 datetime.fromisoformat(data["timestamp"]) if data.get("timestamp") else datetime.now(timezone.utc)
             ),
-            related_conversations={ConversationReference.from_dict(r) for r in data.get("related_conversations", [])},
+            related_conversations={
+                ConversationReference.model_validate(r) for r in data.get("related_conversations", [])
+            },
             metadata=data.get("metadata", {}),
             labels=data.get("labels", {}),
             error_message=data.get("error_message"),
             error_type=data.get("error_type"),
             error_traceback=data.get("error_traceback"),
-            retry_events=[RetryEvent.from_dict(e) for e in data.get("retry_events", [])],
+            retry_events=[RetryEvent.model_validate(e) for e in data.get("retry_events", [])],
             total_retries=data.get("total_retries", 0),
         )
 

pyrit/models/conversation_reference.py

@@ -3,10 +3,13 @@
 
 from __future__ import annotations
 
-from dataclasses import dataclass
 from enum import Enum
 from typing import Optional
 
+from pydantic import BaseModel, ConfigDict
+
+from pyrit.common.deprecation import print_deprecation_message
+
 
 class ConversationType(Enum):
     """Types of conversations that can be associated with an attack."""
@@ -17,15 +20,15 @@ class ConversationType(Enum):
     CONVERTER = "converter"
 
 
-@dataclass(frozen=True)
-class ConversationReference:
+class ConversationReference(BaseModel):
     """Immutable reference to a conversation that played a role in the attack."""
 
+    model_config = ConfigDict(frozen=True)
+
     conversation_id: str
     conversation_type: ConversationType
     description: Optional[str] = None
 
-    # Allow use in set / dict
     def __hash__(self) -> int:
         """
         Return a hash derived from conversation ID.
@@ -36,45 +39,55 @@ def __hash__(self) -> int:
         """
         return hash(self.conversation_id)
 
+    def __eq__(self, other: object) -> bool:
+        """
+        Compare two references by conversation ID.
+
+        Args:
+            other (object): Other object to compare.
+
+        Returns:
+            bool: True when the other object is a matching ConversationReference.
+
+        """
+        return isinstance(other, ConversationReference) and self.conversation_id == other.conversation_id
+
     def to_dict(self) -> dict[str, str | None]:
         """
         Serialize to a JSON-compatible dictionary.
 
+        .. deprecated::
+            Use :meth:`model_dump` with ``mode="json"`` instead. This method
+            will be removed in version 0.16.0.
+
         Returns:
             dict[str, str | None]: Dictionary with conversation_id, conversation_type, and description.
         """
-        return {
-            "conversation_id": self.conversation_id,
-            "conversation_type": self.conversation_type.value,
-            "description": self.description,
-        }
+        print_deprecation_message(
+            old_item=ConversationReference.to_dict,
+            new_item='ConversationReference.model_dump(mode="json")',
+            removed_in="0.16.0",
+        )
+        return self.model_dump(mode="json")
 
     @classmethod
     def from_dict(cls, data: dict[str, str | None]) -> ConversationReference:
         """
         Reconstruct a ConversationReference from a dictionary.
 
+        .. deprecated::
+            Use :meth:`model_validate` instead. This method will be removed
+            in version 0.16.0.
+
         Args:
-            data (dict[str, str | None]): Dictionary as produced by to_dict().
+            data (dict[str, str | None]): Dictionary as produced by ``model_dump(mode="json")``.
 
         Returns:
             ConversationReference: Reconstructed instance.
         """
-        return cls(
-            conversation_id=str(data["conversation_id"]),
-            conversation_type=ConversationType(data["conversation_type"]),
-            description=data.get("description"),
+        print_deprecation_message(
+            old_item=ConversationReference.from_dict,
+            new_item="ConversationReference.model_validate",
+            removed_in="0.16.0",
         )
-
-    def __eq__(self, other: object) -> bool:
-        """
-        Compare two references by conversation ID.
-
-        Args:
-            other (object): Other object to compare.
-
-        Returns:
-            bool: True when the other object is a matching ConversationReference.
-
-        """
-        return isinstance(other, ConversationReference) and self.conversation_id == other.conversation_id
+        return cls.model_validate(data)

pyrit/models/conversation_stats.py

@@ -1,26 +1,24 @@
 # Copyright (c) Microsoft Corporation.
 # Licensed under the MIT license.
 
-from __future__ import annotations
+from datetime import datetime
+from typing import ClassVar, Optional
 
-from dataclasses import dataclass, field
-from typing import TYPE_CHECKING, ClassVar, Optional
+from pydantic import BaseModel, ConfigDict, Field
 
-if TYPE_CHECKING:
-    from datetime import datetime
 
-
-@dataclass(frozen=True)
-class ConversationStats:
+class ConversationStats(BaseModel):
     """
     Lightweight aggregate statistics for a conversation.
 
     Used to build attack summaries without loading full message pieces.
     """
 
+    model_config = ConfigDict(frozen=True)
+
     PREVIEW_MAX_LEN: ClassVar[int] = 100
 
     message_count: int = 0
     last_message_preview: Optional[str] = None
-    labels: dict[str, str] = field(default_factory=dict)
+    labels: dict[str, str] = Field(default_factory=dict)
     created_at: Optional[datetime] = None

pyrit/models/harm_definition.py

@@ -9,19 +9,18 @@
 
 import logging
 import re
-from dataclasses import dataclass, field
 from pathlib import Path
 from typing import Optional, Union
 
 import yaml
+from pydantic import BaseModel, Field
 
 from pyrit.common.path import HARM_DEFINITION_PATH
 
 logger = logging.getLogger(__name__)
 
 
-@dataclass
-class ScaleDescription:
+class ScaleDescription(BaseModel):
     """
     A single scale description entry from a harm definition.
 
@@ -35,8 +34,7 @@ class ScaleDescription:
     description: str
 
 
-@dataclass
-class HarmDefinition:
+class HarmDefinition(BaseModel):
     """
     A harm definition loaded from a YAML file.
 
@@ -54,8 +52,8 @@ class HarmDefinition:
 
     version: str
     category: str
-    scale_descriptions: list[ScaleDescription] = field(default_factory=list)
-    source_path: Optional[str] = field(default=None, kw_only=True)
+    scale_descriptions: list[ScaleDescription] = Field(default_factory=list)
+    source_path: Optional[str] = None
 
     def get_scale_description(self, score_value: str) -> Optional[str]:
         """
@@ -92,7 +90,6 @@ def validate_category(category: str, *, check_exists: bool = False) -> bool:
             False otherwise.
 
         """
-        # Check if category matches pattern: only lowercase letters and underscores
         if not re.match(r"^[a-z_]+$", category):
             return False
 
@@ -127,7 +124,6 @@ def from_yaml(cls, harm_definition_path: Union[str, Path]) -> "HarmDefinition":
         """
         path = Path(harm_definition_path)
 
-        # If it's just a filename (no directory separators), look in the standard directory
         resolved_path = HARM_DEFINITION_PATH / path if path.parent == Path(".") else path
 
         if not resolved_path.exists():
@@ -145,15 +141,13 @@ def from_yaml(cls, harm_definition_path: Union[str, Path]) -> "HarmDefinition":
         if not isinstance(data, dict):
             raise ValueError(f"Harm definition file {resolved_path} must contain a YAML mapping/dictionary.")
 
-        # Validate required fields
         if "version" not in data:
             raise ValueError(f"Harm definition file {resolved_path} is missing required 'version' field.")
         if "category" not in data:
             raise ValueError(f"Harm definition file {resolved_path} is missing required 'category' field.")
         if "scale_descriptions" not in data:
             raise ValueError(f"Harm definition file {resolved_path} is missing required 'scale_descriptions' field.")
 
-        # Parse scale descriptions
         scale_descriptions = []
         for item in data["scale_descriptions"]:
             if not isinstance(item, dict) or "score_value" not in item or "description" not in item:

pyrit/models/json_response_config.py

@@ -4,9 +4,10 @@
 from __future__ import annotations
 
 import json
-from dataclasses import dataclass
 from typing import Any, Optional
 
+from pydantic import BaseModel, ConfigDict
+
 # Would prefer StrEnum, but.... Python 3.10
 _METADATAKEYS = {
     "RESPONSE_FORMAT": "response_format",
@@ -16,8 +17,7 @@
 }
 
 
-@dataclass
-class _JsonResponseConfig:
+class _JsonResponseConfig(BaseModel):
     """
     Configuration for JSON responses (with OpenAI).
 
@@ -27,8 +27,10 @@ class _JsonResponseConfig:
     https://platform.openai.com/docs/api-reference/responses/create#responses_create-text
     """
 
+    model_config = ConfigDict(extra="forbid")
+
     enabled: bool = False
-    schema: Optional[dict[str, Any]] = None
+    json_schema: Optional[dict[str, Any]] = None
     schema_name: str = "CustomSchema"
     strict: bool = True
 
@@ -53,7 +55,7 @@ def from_metadata(cls, *, metadata: Optional[dict[str, Any]]) -> _JsonResponseCo
 
             return cls(
                 enabled=True,
-                schema=schema,
+                json_schema=schema,
                 schema_name=metadata.get(_METADATAKEYS["JSON_SCHEMA_NAME"], "CustomSchema"),
                 strict=metadata.get(_METADATAKEYS["JSON_SCHEMA_STRICT"], True),
             )