🏗️(project) migrate to pydantic v2 and switch tests to polyfactory

Migrating to `pydantic` v2 should speed up processing and allow
interoperability with projects such as `warren`. This migration makes the
hypothesis package used in tests obsolete, which is why we introduce
`polyfactory`.

Author: Leobouloc

CHANGELOG.md

@@ -8,6 +8,9 @@ and this project adheres to
 
 ## [Unreleased]
 
+- Upgrade `pydantic` to `2.5.3`
+- Migrate model tests from hypothesis strategies to polyfactory
+
 ## [4.0.0] - 2024-01-23
 
 ### Added

pyproject.toml

@@ -31,7 +31,8 @@ dependencies = [
     # By default, we only consider core dependencies required to use Ralph as a
     # library (mostly models).
     "langcodes>=3.2.0",
-    "pydantic[dotenv,email]>=1.10.0, <2.0",
+    "pydantic[email]>=2.5.3,<3.0",
+    "pydantic_settings>=2.1.0,<3.0",
     "rfc3987>=1.3.0",
 ]
 dynamic = ["version"]
@@ -91,7 +92,6 @@ dev = [
     "black==23.12.1",
     "cryptography==41.0.7",
     "factory-boy==3.3.0",
-    "hypothesis<6.92.0", # pin as hypothesis 6.92.0 observability feature seems broken
     "logging-gelf==0.0.31",
     "mike==2.0.0",
     "mkdocs==1.5.3",
@@ -103,6 +103,7 @@ dev = [
     "neoteroi-mkdocs==1.0.4",
     "pyfakefs==5.3.2",
     "pymdown-extensions==10.7",
+    "polyfactory==2.14.1",
     "pytest==7.4.4",
     "pytest-asyncio==0.23.3",
     "pytest-cov==4.1.0",

src/ralph/api/__init__.py

@@ -50,4 +50,7 @@ async def whoami(
     user: AuthenticatedUser = Depends(get_authenticated_user),
 ) -> Dict[str, Any]:
     """Return the current user's username along with their scopes."""
-    return {"agent": user.agent, "scopes": user.scopes}
+    return {
+        "agent": user.agent.model_dump(mode="json", exclude_none=True),
+        "scopes": user.scopes,
+    }

src/ralph/api/auth/basic.py

@@ -1,6 +1,7 @@
 """Basic authentication & authorization related tools for the Ralph API."""
 
 import logging
+import os
 from functools import lru_cache
 from pathlib import Path
 from threading import Lock
@@ -10,7 +11,7 @@
 from cachetools import TTLCache, cached
 from fastapi import Depends, HTTPException, status
 from fastapi.security import HTTPBasic, HTTPBasicCredentials
-from pydantic import BaseModel, root_validator
+from pydantic import RootModel, model_validator
 from starlette.authentication import AuthenticationError
 
 from ralph.api.auth.user import AuthenticatedUser
@@ -40,45 +41,42 @@ class UserCredentials(AuthenticatedUser):
     username: str
 
 
-class ServerUsersCredentials(BaseModel):
+class ServerUsersCredentials(RootModel[List[UserCredentials]]):
     """Custom root pydantic model.
 
     Describe expected list of all server users credentials as stored in
     the credentials file.
 
     Attributes:
-        __root__ (List): Custom root consisting of the
+        root (List): Custom root consisting of the
                         list of all server users credentials.
     """
 
-    __root__: List[UserCredentials]
-
     def __add__(self, other) -> Any:  # noqa: D105
-        return ServerUsersCredentials.parse_obj(self.__root__ + other.__root__)
+        return ServerUsersCredentials.model_validate(self.root + other.root)
 
     def __getitem__(self, item: int) -> UserCredentials:  # noqa: D105
-        return self.__root__[item]
+        return self.root[item]
 
     def __len__(self) -> int:  # noqa: D105
-        return len(self.__root__)
+        return len(self.root)
 
     def __iter__(self) -> Iterator[UserCredentials]:  # noqa: D105
-        return iter(self.__root__)
+        return iter(self.root)
 
-    @root_validator
-    @classmethod
-    def ensure_unique_username(cls, values: Any) -> Any:
+    @model_validator(mode="after")
+    def ensure_unique_username(self) -> Any:
         """Every username should be unique among registered users."""
-        usernames = [entry.username for entry in values.get("__root__")]
+        usernames = [entry.username for entry in self.root]
         if len(usernames) != len(set(usernames)):
             raise ValueError(
                 "You cannot create multiple credentials with the same username"
             )
-        return values
+        return self
 
 
 @lru_cache()
-def get_stored_credentials(auth_file: Path) -> ServerUsersCredentials:
+def get_stored_credentials(auth_file: os.PathLike) -> ServerUsersCredentials:
     """Helper to read the credentials/scopes file.
 
     Read credentials from JSON file and stored them to avoid reloading them with every
@@ -96,7 +94,9 @@ def get_stored_credentials(auth_file: Path) -> ServerUsersCredentials:
         msg = "Credentials file <%s> not found."
         logger.warning(msg, auth_file)
         raise AuthenticationError(msg.format(auth_file))
-    return ServerUsersCredentials.parse_file(auth_file)
+
+    with open(auth_file, encoding=settings.LOCALE_ENCODING) as f:
+        return ServerUsersCredentials.model_validate_json(f.read())
 
 
 @cached(

src/ralph/api/auth/oidc.py

@@ -9,7 +9,7 @@
 from fastapi.security import HTTPBearer, OpenIdConnect
 from jose import ExpiredSignatureError, JWTError, jwt
 from jose.exceptions import JWTClaimsError
-from pydantic import AnyUrl, BaseModel, Extra
+from pydantic import AnyUrl, BaseModel, ConfigDict
 from typing_extensions import Annotated
 
 from ralph.api.auth.user import AuthenticatedUser, UserScopes
@@ -44,13 +44,11 @@ class IDToken(BaseModel):
 
     iss: str
     sub: str
-    aud: Optional[str]
+    aud: Optional[str] = None
     exp: int
     iat: int
-    scope: Optional[str]
-
-    class Config:  # noqa: D106
-        extra = Extra.ignore
+    scope: Optional[str] = None
+    model_config = ConfigDict(extra="ignore")
 
 
 @lru_cache()
@@ -142,7 +140,7 @@ def get_oidc_user(
             headers={"WWW-Authenticate": "Bearer"},
         ) from exc
 
-    id_token = IDToken.parse_obj(decoded_token)
+    id_token = IDToken.model_validate(decoded_token)
 
     user = AuthenticatedUser(
         agent={"openid": f"{id_token.iss}/{id_token.sub}"},

src/ralph/api/auth/user.py

@@ -1,8 +1,10 @@
 """Authenticated user for the Ralph API."""
 
-from typing import Dict, FrozenSet, Literal
+from typing import FrozenSet, Literal
 
-from pydantic import BaseModel
+from pydantic import BaseModel, RootModel
+
+from ralph.models.xapi.base.agents import BaseXapiAgent
 
 Scope = Literal[
     "statements/write",
@@ -18,7 +20,7 @@
 ]
 
 
-class UserScopes(FrozenSet[Scope]):
+class UserScopes(RootModel[FrozenSet[Scope]]):
     """Scopes available to users."""
 
     def is_authorized(self, requested_scope: Scope):
@@ -47,19 +49,11 @@ def is_authorized(self, requested_scope: Scope):
         }
 
         expanded_user_scopes = set()
-        for scope in self:
+        for scope in self.root:
             expanded_user_scopes.update(expanded_scopes.get(scope, {scope}))
 
         return requested_scope in expanded_user_scopes
 
-    @classmethod
-    def __get_validators__(cls):  # noqa: D105
-        def validate(value: FrozenSet[Scope]):
-            """Transform value to an instance of UserScopes."""
-            return cls(value)
-
-        yield validate
-
 
 class AuthenticatedUser(BaseModel):
     """Pydantic model for user authentication.
@@ -69,5 +63,5 @@ class AuthenticatedUser(BaseModel):
         scopes (list): The scopes the user has access to.
     """
 
-    agent: Dict
+    agent: BaseXapiAgent
     scopes: UserScopes

src/ralph/api/forwarding.py

@@ -42,16 +42,16 @@ async def forward_xapi_statements(
             try:
                 # NB: post or put
                 req = await getattr(client, method)(
-                    forwarding.url,
+                    str(forwarding.url),
                     json=statements,
                     auth=(forwarding.basic_username, forwarding.basic_password),
                     timeout=forwarding.timeout,
                 )
                 req.raise_for_status()
                 msg = "Forwarded %s statements to %s with success."
                 if isinstance(statements, list):
-                    logger.debug(msg, len(statements), forwarding.url)
+                    logger.debug(msg, len(statements), str(forwarding.url))
                 else:
-                    logger.debug(msg, 1, forwarding.url)
+                    logger.debug(msg, 1, str(forwarding.url))
             except (RequestError, HTTPStatusError) as error:
                 logger.error("Failed to forward xAPI statements. %s", error)

src/ralph/api/models.py

@@ -6,7 +6,7 @@
 from typing import Optional, Union
 from uuid import UUID
 
-from pydantic import AnyUrl, BaseModel, Extra
+from pydantic import AnyUrl, BaseModel, ConfigDict
 
 from ..models.xapi.base.agents import BaseXapiAgent
 from ..models.xapi.base.groups import BaseXapiGroup
@@ -29,13 +29,7 @@ class BaseModelWithLaxConfig(BaseModel):
     we receive statements through the API.
     """
 
-    class Config:
-        """Enable extra properties.
-
-        Useful for not having to perform comprehensive validation.
-        """
-
-        extra = Extra.allow
+    model_config = ConfigDict(extra="allow", coerce_numbers_to_str=True)
 
 
 class LaxObjectField(BaseModelWithLaxConfig):
@@ -64,6 +58,6 @@ class LaxStatement(BaseModelWithLaxConfig):
     """
 
     actor: Union[BaseXapiAgent, BaseXapiGroup]
-    id: Optional[UUID]
+    id: Optional[UUID] = None
     object: LaxObjectField
     verb: LaxVerbField

src/ralph/api/routers/health.py

@@ -47,7 +47,7 @@ async def heartbeat(response: Response) -> Heartbeat:
 
     Return a 200 if all checks are successful.
     """
-    statuses = Heartbeat.construct(
+    statuses = Heartbeat.model_construct(
         database=await await_if_coroutine(BACKEND_CLIENT.status())
     )
     if not statuses.is_alive:

src/ralph/api/routers/statements.py

@@ -10,6 +10,7 @@
 from fastapi import (
     APIRouter,
     BackgroundTasks,
+    Body,
     Depends,
     HTTPException,
     Query,
@@ -19,7 +20,7 @@
     status,
 )
 from fastapi.dependencies.models import Dependant
-from pydantic import parse_obj_as
+from pydantic import TypeAdapter
 from pydantic.types import Json
 from typing_extensions import Annotated
 
@@ -98,14 +99,17 @@ def _enrich_statement_with_authority(
 ) -> None:
     # authority: Information about whom or what has asserted the statement is true.
     # https://github.com/adlnet/xAPI-Spec/blob/master/xAPI-Data.md#249-authority
-    statement["authority"] = current_user.agent
+    statement["authority"] = current_user.agent.model_dump(
+        exclude_none=True, mode="json"
+    )
 
 
 def _parse_agent_parameters(agent_obj: dict) -> AgentParameters:
     """Parse a dict and return an AgentParameters object to use in queries."""
     # Transform agent to `dict` as FastAPI cannot parse JSON (seen as string)
 
-    agent = parse_obj_as(BaseXapiAgent, agent_obj)
+    adapter = TypeAdapter(BaseXapiAgent)
+    agent = adapter.validate_python(agent_obj)
 
     agent_query_params = {}
     if isinstance(agent, BaseXapiAgentWithMbox):
@@ -119,7 +123,7 @@ def _parse_agent_parameters(agent_obj: dict) -> AgentParameters:
         agent_query_params["account__home_page"] = agent.account.homePage
 
     # Overwrite `agent` field
-    return AgentParameters.construct(**agent_query_params)
+    return AgentParameters.model_construct(**agent_query_params)
 
 
 def strict_query_params(request: Request) -> None:
@@ -141,7 +145,7 @@ def strict_query_params(request: Request) -> None:
 
 @router.get("")
 @router.get("/")
-async def get(  # noqa: PLR0913
+async def get(  # noqa: PLR0912,PLR0913
     request: Request,
     current_user: Annotated[
         AuthenticatedUser,
@@ -169,7 +173,7 @@ async def get(  # noqa: PLR0913
         None,
         description="Filter, only return Statements matching the specified Verb id",
     ),
-    activity: Optional[IRI] = Query(
+    activity: Optional[Annotated[IRI, Body()]] = Query(
         None,
         description=(
             "Filter, only return Statements for which the Object "
@@ -334,7 +338,14 @@ async def get(  # noqa: PLR0913
         # Overwrite `agent` field
         query_params["agent"] = _parse_agent_parameters(
             json.loads(query_params["agent"])
-        )
+        ).model_dump(mode="json", exclude_none=True)
+
+    # Coerce `verb` and `activity` as IRI
+    if query_params.get("verb"):
+        query_params["verb"] = IRI(query_params["verb"])
+
+    if query_params.get("activity"):
+        query_params["activity"] = IRI(query_params["activity"])
 
     # mine: If using scopes, only restrict users with limited scopes
     if settings.LRS_RESTRICT_BY_SCOPES:
@@ -346,7 +357,9 @@ async def get(  # noqa: PLR0913
 
     # Filter by authority if using `mine`
     if mine:
-        query_params["authority"] = _parse_agent_parameters(current_user.agent)
+        query_params["authority"] = _parse_agent_parameters(
+            current_user.agent.model_dump(mode="json")
+        ).model_dump(mode="json", exclude_none=True)
 
     if "mine" in query_params:
         query_params.pop("mine")
@@ -355,7 +368,7 @@ async def get(  # noqa: PLR0913
     try:
         query_result = await await_if_coroutine(
             BACKEND_CLIENT.query_statements(
-                RalphStatementsQuery.construct(**{**query_params, "limit": limit})
+                RalphStatementsQuery.model_construct(**{**query_params, "limit": limit})
             )
         )
     except BackendException as error:
@@ -415,7 +428,7 @@ async def put(
     LRS Specification:
     https://github.com/adlnet/xAPI-Spec/blob/1.0.3/xAPI-Communication.md#211-put-statements
     """
-    statement_as_dict = statement.dict(exclude_unset=True)
+    statement_as_dict = statement.model_dump(exclude_unset=True, mode="json")
     statement_id = str(statement_id)
 
     statement_as_dict.update(id=str(statement_as_dict.get("id", statement_id)))
@@ -504,7 +517,9 @@ async def post(  # noqa: PLR0912
 
     # Enrich statements before forwarding
     statements_dict = {}
-    for statement in (x.dict(exclude_unset=True) for x in statements):
+    for statement in (
+        x.model_dump(exclude_unset=True, mode="json") for x in statements
+    ):
         _enrich_statement_with_id(statement)
         # Requests with duplicate statement IDs are considered invalid
         if statement["id"] in statements_dict: