Upgrade pydantic to v2+

pydatalab/pydatalab/apps/xrd/models.py

@@ -1,6 +1,6 @@
 from typing import List, Optional
 
-from pydantic import BaseModel
+from pydantic import BaseModel, ConfigDict
 
 __all__ = ("XRDPattern", "XRDMeasurement")
 
@@ -30,15 +30,13 @@ class XRDProcessing(BaseModel):
     peak_widths: List[float]
 
     baselines: List[List[float]]
-
-    class Config:
-        extra = "allow"
+    model_config = ConfigDict(extra="allow")
 
 
 class XRDMetadata(BaseModel): ...
 
 
 class XRDMeasurement(BaseModel):
-    data: Optional[XRDPattern]
-    processing: Optional[XRDProcessing]
-    metadata: Optional[XRDMetadata]
+    data: Optional[XRDPattern] = None
+    processing: Optional[XRDProcessing] = None
+    metadata: Optional[XRDMetadata] = None

pydatalab/pydatalab/config.py

@@ -9,12 +9,14 @@
 from pydantic import (
     AnyUrl,
     BaseModel,
-    BaseSettings,
+    ConfigDict,
     Field,
     ValidationError,
+    field_validator,
     root_validator,
     validator,
 )
+from pydantic_settings import BaseSettings
 
 from pydatalab.models import Person
 from pydatalab.models.utils import RandomAlphabeticalRefcodeFactory, RefCodeFactory
@@ -49,20 +51,20 @@ def config_file_settings(settings: BaseSettings) -> Dict[str, Any]:
 class DeploymentMetadata(BaseModel):
     """A model for specifying metadata about a datalab deployment."""
 
-    maintainer: Optional[Person]
+    maintainer: Optional[Person] = None
     issue_tracker: Optional[AnyUrl] = Field("https://github.com/datalab-org/datalab/issues")
-    homepage: Optional[AnyUrl]
+    homepage: Optional[AnyUrl] = None
     source_repository: Optional[AnyUrl] = Field("https://github.com/datalab-org/datalab")
 
-    @validator("maintainer")
+    @field_validator("maintainer")
+    @classmethod
     def strip_fields_from_person(cls, v):
         if not v.contact_email:
             raise ValueError("Must provide contact email for maintainer.")
 
         return Person(contact_email=v.contact_email, display_name=v.display_name)
 
-    class Config:
-        extra = "allow"
+    model_config = ConfigDict(extra="allow")
 
 
 class BackupStrategy(BaseModel):
@@ -73,7 +75,8 @@ class BackupStrategy(BaseModel):
         description="Whether this backup strategy is active; i.e., whether it is actually used. All strategies will be disabled in testing scenarios.",
     )
     hostname: str | None = Field(
-        description="The hostname of the SSH-accessible server on which to store the backup (`None` indicates local backups)."
+        None,
+        description="The hostname of the SSH-accessible server on which to store the backup (`None` indicates local backups).",
     )
     location: Path = Field(
         description="The location under which to store the backups on the host. Each backup will be date-stamped and stored in a subdirectory of this location."
@@ -271,6 +274,8 @@ def validate_cache_ages(cls, values):
             )
         return values
 
+    # TODO[pydantic]: We couldn't refactor the `validator`, please replace it by `field_validator` manually.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-validators for more information.
     @validator("IDENTIFIER_PREFIX", pre=True, always=True)
     def validate_identifier_prefix(cls, v, values):
         """Make sure that the identifier prefix is set and is valid, raising clear error messages if not.
@@ -307,7 +312,8 @@ def deactivate_backup_strategies_during_testing(cls, values):
 
         return values
 
-    @validator("LOG_FILE")
+    @field_validator("LOG_FILE")
+    @classmethod
     def make_missing_log_directory(cls, v):
         """Make sure that the log directory exists and is writable."""
         if v is None:
@@ -320,6 +326,8 @@ def make_missing_log_directory(cls, v):
             raise RuntimeError(f"Unable to create log file at {v}") from exc
         return v
 
+    # TODO[pydantic]: We couldn't refactor this class, please create the `model_config` manually.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-config for more information.
     class Config:
         env_prefix = "pydatalab_"
         extra = "allow"

pydatalab/pydatalab/models/cells.py

@@ -7,8 +7,6 @@
 from pydatalab.models.items import Item
 from pydatalab.models.utils import Constituent
 
-# from pydatalab.logger import LOGGER
-
 
 class CellComponent(Constituent): ...
 
@@ -31,25 +29,30 @@ class Cell(Item):
     type: str = Field("cells", const="cells", pattern="^cells$")
 
     cell_format: Optional[CellFormat] = Field(
+        None,
         description="The form factor of the cell, e.g., coin, pouch, in situ or otherwise.",
     )
 
     cell_format_description: Optional[str] = Field(
-        description="Additional human-readable description of the cell form factor, e.g., 18650, AMPIX, CAMPIX"
+        None,
+        description="Additional human-readable description of the cell form factor, e.g., 18650, AMPIX, CAMPIX",
     )
 
-    cell_preparation_description: Optional[str] = Field()
+    cell_preparation_description: Optional[str] = Field(None)
 
     characteristic_mass: Optional[float] = Field(
-        description="The characteristic mass of the cell in milligrams. Can be used to normalize capacities."
+        None,
+        description="The characteristic mass of the cell in milligrams. Can be used to normalize capacities.",
     )
 
     characteristic_chemical_formula: Optional[str] = Field(
-        description="The chemical formula of the active material. Can be used to calculated molar mass in g/mol for normalizing capacities."
+        None,
+        description="The chemical formula of the active material. Can be used to calculated molar mass in g/mol for normalizing capacities.",
     )
 
     characteristic_molar_mass: Optional[float] = Field(
-        description="The molar mass of the active material, in g/mol. Will be inferred from the chemical formula, or can be supplied if it cannot be supplied"
+        None,
+        description="The molar mass of the active material, in g/mol. Will be inferred from the chemical formula, or can be supplied if it cannot be supplied",
     )
 
     positive_electrode: List[CellComponent] = Field([])
@@ -60,6 +63,8 @@ class Cell(Item):
 
     active_ion_charge: float = Field(1)
 
+    # TODO[pydantic]: We couldn't refactor the `validator`, please replace it by `field_validator` manually.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-validators for more information.
     @validator("characteristic_molar_mass", always=True, pre=True)
     def set_molar_mass(cls, v, values):
         from periodictable import formula

pydatalab/pydatalab/models/collections.py

@@ -13,10 +13,10 @@ class Collection(Entry, HasOwner, HasBlocks):
     collection_id: HumanReadableIdentifier = Field(None)
     """A short human-readable/usable name for the collection."""
 
-    title: Optional[str]
+    title: Optional[str] = None
     """A descriptive title for the collection."""
 
-    description: Optional[str]
+    description: Optional[str] = None
     """A description of the collection, either in plain-text or a markup language."""
 
     num_items: Optional[int] = Field(None)

pydatalab/pydatalab/models/entries.py

@@ -1,7 +1,7 @@
 import abc
 from typing import List, Optional
 
-from pydantic import BaseModel, Field, root_validator
+from pydantic import BaseModel, ConfigDict, Field, model_validator
 
 from pydatalab.models.relationships import TypedRelationship
 from pydatalab.models.utils import (
@@ -34,7 +34,8 @@ class Entry(BaseModel, abc.ABC):
     relationships: Optional[List[TypedRelationship]] = None
     """A list of related entries and their types."""
 
-    @root_validator(pre=True)
+    @model_validator(mode="before")
+    @classmethod
     def check_id_names(cls, values):
         """Slightly upsetting hack: this case *should* be covered by the pydantic setting for
         populating fields by alias names.
@@ -63,7 +64,6 @@ def to_reference(self, additional_fields: Optional[List[str]] = None) -> "EntryR
 
         return EntryReference(**data)
 
-    class Config:
-        allow_population_by_field_name = True
-        json_encoders = JSON_ENCODERS
-        extra = "ignore"
+    # TODO[pydantic]: The following keys were removed: `json_encoders`.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-config for more information.
+    model_config = ConfigDict(populate_by_name=True, json_encoders=JSON_ENCODERS, extra="ignore")

pydatalab/pydatalab/models/equipment.py

@@ -10,14 +10,14 @@ class Equipment(Item):
 
     type: str = Field("equipment", const="equipment", pattern="^equipment$")
 
-    serial_numbers: Optional[str]
+    serial_numbers: Optional[str] = None
     """A string describing one or more serial numbers for the instrument."""
 
-    manufacturer: Optional[str]
+    manufacturer: Optional[str] = None
     """The manufacturer of this piece of equipment"""
 
-    location: Optional[str]
+    location: Optional[str] = None
     """Place where the equipment is located"""
 
-    contact: Optional[str]
+    contact: Optional[str] = None
     """Contact information for equipment (e.g., email address or phone number)."""

pydatalab/pydatalab/models/files.py

@@ -13,10 +13,10 @@ class File(Entry, HasOwner, HasRevisionControl):
 
     type: str = Field("files", const="files", pattern="^files$")
 
-    size: Optional[int] = Field(description="The size of the file on disk in bytes.")
+    size: Optional[int] = Field(None, description="The size of the file on disk in bytes.")
 
     last_modified_remote: Optional[IsoformatDateTime] = Field(
-        description="The last date/time at which the remote file was modified."
+        None, description="The last date/time at which the remote file was modified."
     )
 
     item_ids: List[str] = Field(description="A list of item IDs associated with this file.")
@@ -27,27 +27,29 @@ class File(Entry, HasOwner, HasRevisionControl):
 
     extension: str = Field(description="The file extension that the file was uploaded with.")
 
-    original_name: Optional[str] = Field(description="The raw filename as uploaded.")
+    original_name: Optional[str] = Field(None, description="The raw filename as uploaded.")
 
-    location: Optional[str] = Field(description="The location of the file on disk.")
+    location: Optional[str] = Field(None, description="The location of the file on disk.")
 
-    url_path: Optional[str] = Field(description="The path to a remote file.")
+    url_path: Optional[str] = Field(None, description="The path to a remote file.")
 
     source: Optional[str] = Field(
-        description="The source of the file, e.g. 'remote' or 'uploaded'."
+        None, description="The source of the file, e.g. 'remote' or 'uploaded'."
     )
 
     time_added: datetime.datetime = Field(description="The timestamp for the original file upload.")
 
-    metadata: Optional[Dict[Any, Any]] = Field(description="Any additional metadata.")
+    metadata: Optional[Dict[Any, Any]] = Field(None, description="Any additional metadata.")
 
-    representation: Optional[Any] = Field()
+    representation: Optional[Any] = Field(None)
 
     source_server_name: Optional[str] = Field(
-        description="The server name at which the file is stored."
+        None, description="The server name at which the file is stored."
     )
 
-    source_path: Optional[str] = Field(description="The path to the file on the remote resource.")
+    source_path: Optional[str] = Field(
+        None, description="The path to the file on the remote resource."
+    )
 
     is_live: bool = Field(
         description="Whether or not the file should be watched for future updates."

pydatalab/pydatalab/models/items.py

@@ -30,21 +30,23 @@ class Item(Entry, HasOwner, HasRevisionControl, IsCollectable, HasBlocks, abc.AB
     item_id: HumanReadableIdentifier
     """A locally unique, human-readable identifier for the entry. This ID is mutable."""
 
-    description: Optional[str]
+    description: Optional[str] = None
     """A description of the item, either in plain-text or a markup language."""
 
-    date: Optional[IsoformatDateTime]
+    date: Optional[IsoformatDateTime] = None
     """A relevant 'creation' timestamp for the entry (e.g., purchase date, synthesis date)."""
 
-    name: Optional[str]
+    name: Optional[str] = None
     """An optional human-readable/usable name for the entry."""
 
-    files: Optional[List[File]]
+    files: Optional[List[File]] = None
     """Any files attached to this sample."""
 
     file_ObjectIds: List[PyObjectId] = Field([])
     """Links to object IDs of files stored within the database."""
 
+    # TODO[pydantic]: We couldn't refactor the `validator`, please replace it by `field_validator` manually.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-validators for more information.
     @validator("refcode", pre=True, always=True)
     def refcode_validator(cls, v):
         """Generate a refcode if not provided."""

pydatalab/pydatalab/models/people.py

@@ -1,9 +1,9 @@
 from enum import Enum
-from typing import List, Optional
+from typing import List, Literal, Optional
 
 import bson
 import bson.errors
-from pydantic import BaseModel, ConstrainedStr, Field, parse_obj_as, validator
+from pydantic import BaseModel, ConstrainedStr, Field, field_validator, parse_obj_as, validator
 from pydantic import EmailStr as PydanticEmailStr
 
 from pydatalab.models.entries import Entry
@@ -36,9 +36,11 @@ class Identity(BaseModel):
     verified: bool = Field(False)
     """Whether the identity has been verified (by some means, e.g., OAuth2 or email)"""
 
-    display_name: Optional[str]
+    display_name: Optional[str] = None
     """The user's display name associated with the identity, also to be exposed in free text searches."""
 
+    # TODO[pydantic]: We couldn't refactor the `validator`, please replace it by `field_validator` manually.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-validators for more information.
     @validator("name", pre=True, always=True)
     def add_missing_name(cls, v, values):
         """If the identity is created without a free-text 'name', then
@@ -55,6 +57,8 @@ def add_missing_name(cls, v, values):
 
         return v
 
+    # TODO[pydantic]: We couldn't refactor the `validator`, please replace it by `field_validator` manually.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-validators for more information.
     @validator("verified", pre=True, always=True)
     def add_missing_verification(cls, v):
         """Fills in missing value for `verified` if not given."""
@@ -101,32 +105,35 @@ class AccountStatus(str, Enum):
 class Person(Entry):
     """A model that describes an individual and their digital identities."""
 
-    type: str = Field("people", const=True)
+    type: Literal["people"] = "people"
     """The entry type as a string."""
 
     identities: List[Identity] = Field(default_factory=list)
     """A list of identities attached to this person, e.g., email addresses, OAuth accounts."""
 
-    display_name: Optional[DisplayName]
+    display_name: Optional[DisplayName] = None
     """The user-chosen display name."""
 
-    contact_email: Optional[EmailStr]
+    contact_email: Optional[EmailStr] = None
     """In the case of multiple *verified* email identities, this email will be used as the primary contact."""
 
-    managers: Optional[List[PyObjectId]]
+    managers: Optional[List[PyObjectId]] = None
     """A list of user IDs that can manage this person's items."""
 
     account_status: AccountStatus = Field(AccountStatus.UNVERIFIED)
     """The status of the user's account."""
 
+    # TODO[pydantic]: We couldn't refactor the `validator`, please replace it by `field_validator` manually.
+    # Check https://docs.pydantic.dev/dev-v2/migration/#changes-to-validators for more information.
     @validator("type", pre=True, always=True)
     def add_missing_type(cls, v):
         """Fill in missing `type` field if not provided."""
         if v is None:
             v = "people"
         return v
 
-    @validator("type", pre=True)
+    @field_validator("type", mode="before")
+    @classmethod
     def set_default_type(cls, _):
         return "people"