Require Pydantic 2 (#1391)

Closes #1199 with a minor version bump to 0.12

docs/source/conf.py

@@ -26,7 +26,7 @@
 author = "Charles Tapley Hoyt"
 
 # The full version, including alpha/beta/rc tags.
-release = "0.11.36-dev"
+release = "0.12.0-dev"
 
 # The short X.Y version.
 parsed_version = re.match(

pyproject.toml

@@ -21,7 +21,7 @@ markers = [
 
 [project]
 name = "bioregistry"
-version = "0.11.36-dev"
+version = "0.12.0-dev"
 description = "Integrated registry of biological databases and nomenclatures"
 readme = "README.md"
 authors = [
@@ -70,7 +70,7 @@ dependencies = [
     "pystow>=0.1.13",
     "click",
     "more_click>=0.1.2",
-    "pydantic",
+    "pydantic>=2.0",
     "curies>=0.7.0"
 ]
 
@@ -268,7 +268,7 @@ known-first-party = [
 docstring-code-format = true
 
 [tool.bumpversion]
-current_version = "0.11.36-dev"
+current_version = "0.12.0-dev"
 parse = "(?P<major>\\d+)\\.(?P<minor>\\d+)\\.(?P<patch>\\d+)(?:-(?P<release>[0-9A-Za-z-]+(?:\\.[0-9A-Za-z-]+)*))?(?:\\+(?P<build>[0-9A-Za-z-]+(?:\\.[0-9A-Za-z-]+)*))?"
 serialize = [
     "{major}.{minor}.{patch}-{release}+{build}",

src/bioregistry/app/api.py

@@ -29,7 +29,6 @@
 )
 
 from .utils import FORMAT_MAP, _autocomplete, _search
-from ..utils import pydantic_dict
 
 __all__ = [
     "api_router",
@@ -57,8 +56,7 @@ class YAMLResponse(Response):
     def render(self, content: Any) -> bytes:
         """Render content as YAML."""
         if isinstance(content, BaseModel):
-            content = pydantic_dict(
-                content,
+            content = content.model_dump(
                 exclude_none=True,
                 exclude_unset=True,
             )

src/bioregistry/app/impl.py

@@ -18,7 +18,6 @@
 from .api import api_router
 from .constants import BIOSCHEMAS
 from .ui import ui_blueprint
-from ..constants import PYDANTIC_1
 
 if TYPE_CHECKING:
     import bioregistry
@@ -194,7 +193,6 @@ def get_app(
         curie_to_str=curie_to_str,
         fastapi_url_for=fast_api.url_path_for,
         markdown=markdown,
-        is_pydantic_1=PYDANTIC_1,
     )
 
     if return_flask:

src/bioregistry/app/templates/meta/related.html

@@ -11,15 +11,9 @@
 {% endmacro %}
 
 {% macro registry_field_dl(cls, key) -%}
-    {% if is_pydantic_1 %}
-        {% set field = cls.__fields__[key].field_info %}
-        <dt>{{ field.title or key.replace("_", " ").title() }}</dt>
-        <dd>{{ field.description }}</dd>
-    {% else %}
-        {% set field = cls.model_fields[key] %}
-        <dt>{{ field.title or key.replace("_", " ").title() }}</dt>
-        <dd>{{ field.description }}</dd>
-    {% endif %}
+    {% set field = cls.model_fields[key] %}
+    <dt>{{ field.title or key.replace("_", " ").title() }}</dt>
+    <dd>{{ field.description }}</dd>
 {% endmacro %}
 
 {% block styles %}

src/bioregistry/app/ui.py

@@ -58,7 +58,7 @@
     read_prefix_reviews,
     read_registry_contributions,
 )
-from ..utils import curie_to_str, pydantic_schema
+from ..utils import curie_to_str
 
 __all__ = [
     "ui_blueprint",
@@ -275,7 +275,7 @@ def contexts() -> str:
         "contexts.html",
         rows=manager.contexts.items(),
         formats=FORMATS,
-        schema=pydantic_schema(Context),
+        schema=Context.model_json_schema(),
     )
 
 
@@ -289,7 +289,7 @@ def context(identifier: str) -> str:
         "context.html",
         identifier=identifier,
         entry=entry,
-        schema=pydantic_schema(Context)["properties"],
+        schema=Context.model_json_schema()["properties"],
         formats=FORMATS,
     )
 

src/bioregistry/app/utils.py

@@ -21,7 +21,7 @@
 from bioregistry.resource_manager import Manager
 
 from .proxies import manager
-from ..utils import _norm, pydantic_dict
+from ..utils import _norm
 
 
 def _get_resource_providers(
@@ -178,15 +178,13 @@ def serialize(
 
     if accept == "application/json":
         return current_app.response_class(
-            json.dumps(
-                pydantic_dict(data, exclude_unset=True, exclude_none=True), ensure_ascii=False
-            ),
+            json.dumps(data.model_dump(exclude_unset=True, exclude_none=True), ensure_ascii=False),
             mimetype="application/json",
         )
     elif accept in "application/yaml":
         return current_app.response_class(
             yaml.safe_dump(
-                pydantic_dict(data, exclude_unset=True, exclude_none=True), allow_unicode=True
+                data.model_dump(exclude_unset=True, exclude_none=True), allow_unicode=True
             ),
             mimetype="text/plain",
         )

src/bioregistry/compare.py

@@ -42,7 +42,6 @@
 from bioregistry.constants import DOCS_IMG, EXPORT_REGISTRY
 from bioregistry.license_standardizer import standardize_license
 from bioregistry.schema import Resource
-from bioregistry.utils import pydantic_dict
 
 if TYPE_CHECKING:
     import matplotlib.axes
@@ -481,7 +480,7 @@ def _get_license_and_conflicts() -> tuple[list[str], set[str], set[str], set[str
 def _remap(*, key: str, prefixes: Collection[str]) -> Set[str]:
     br_external_to = {}
     for br_id, resource in read_registry().items():
-        _k = (pydantic_dict(resource).get(key) or {}).get("prefix")
+        _k = (resource.model_dump().get(key) or {}).get("prefix")
         if _k:
             br_external_to[_k] = br_id
 

src/bioregistry/constants.py

@@ -2,7 +2,6 @@
 
 """Constants and utilities for registries."""
 
-import importlib.metadata
 import os
 import pathlib
 import re
@@ -21,8 +20,7 @@
     "RAW_DIRECTORY",
 ]
 
-PYDANTIC_1 = importlib.metadata.version("pydantic").startswith("1.")
-PATTERN_KEY = "regex" if PYDANTIC_1 else "pattern"
+PATTERN_KEY = "pattern"
 ORCID_PATTERN = r"^\d{4}-\d{4}-\d{4}-\d{3}(\d|X)$"
 
 HERE = pathlib.Path(os.path.abspath(os.path.dirname(__file__)))

src/bioregistry/export/warnings_export.py

@@ -17,7 +17,6 @@
 import bioregistry
 from bioregistry.constants import DOCS_DATA, EXTERNAL
 from bioregistry.resolve import get_external
-from bioregistry.utils import pydantic_dict
 
 __all__ = [
     "export_warnings",
@@ -26,7 +25,7 @@
 CURATIONS_PATH = DOCS_DATA.joinpath("curation.yml")
 
 ENTRIES = sorted(
-    (prefix, pydantic_dict(resource, exclude_none=True))
+    (prefix, resource.model_dump(exclude_none=True))
     for prefix, resource in bioregistry.read_registry().items()
 )
 

src/bioregistry/health/check_providers.py

@@ -16,7 +16,7 @@
 
 import bioregistry
 from bioregistry.constants import DOCS_DATA
-from bioregistry.utils import pydantic_dict, secho
+from bioregistry.utils import secho
 
 __all__ = [
     "main",
@@ -159,7 +159,7 @@ def main() -> None:
     database.runs.append(current_run)
     database.runs = sorted(database.runs, key=attrgetter("time"), reverse=True)
 
-    HEALTH_YAML_PATH.write_text(yaml.safe_dump(pydantic_dict(database, exclude_none=True)))
+    HEALTH_YAML_PATH.write_text(yaml.safe_dump(database.model_dump(exclude_none=True)))
     click.echo(f"Wrote to {HEALTH_YAML_PATH}")
 
 

src/bioregistry/schema/struct.py

@@ -32,6 +32,7 @@
 )
 
 from pydantic import BaseModel, Field, PrivateAttr
+from pydantic.json_schema import models_json_schema
 
 from bioregistry import constants as brc
 from bioregistry.constants import (
@@ -40,18 +41,10 @@
     EMAIL_RE,
     ORCID_PATTERN,
     PATTERN_KEY,
-    PYDANTIC_1,
     URI_FORMAT_KEY,
 )
 from bioregistry.license_standardizer import standardize_license
-from bioregistry.utils import (
-    curie_to_str,
-    deduplicate,
-    pydantic_dict,
-    pydantic_parse,
-    removeprefix,
-    removesuffix,
-)
+from bioregistry.utils import curie_to_str, deduplicate, removeprefix, removesuffix
 
 try:
     from typing import Literal
@@ -670,7 +663,7 @@ class Resource(BaseModel):
 
     def get_external(self, metaprefix: str) -> Mapping[str, Any]:
         """Get an external registry."""
-        return pydantic_dict(self).get(metaprefix) or dict()
+        return self.model_dump().get(metaprefix) or dict()
 
     def get_mapped_prefix(self, metaprefix: str) -> Optional[str]:
         """Get the prefix for the given external.
@@ -695,7 +688,7 @@ def get_mapped_prefix(self, metaprefix: str) -> Optional[str]:
 
     def get_prefix_key(self, key: str, metaprefixes: Union[str, Sequence[str]]) -> Any:
         """Get a key enriched by the given external resources' data."""
-        rv = pydantic_dict(self).get(key)
+        rv = self.model_dump().get(key)
         if rv is not None:
             return rv
         if isinstance(metaprefixes, str):
@@ -1339,7 +1332,7 @@ def get_publications(self) -> List[Publication]:
                 )
         if self.uniprot:
             for publication in self.uniprot.get("publications", []):
-                publications.append(pydantic_parse(Publication, publication))
+                publications.append(Publication.model_validate(publication))
         for provider in self.providers or []:
             publications.extend(provider.publications or [])
         return deduplicate_publications(publications)
@@ -2834,9 +2827,6 @@ def _allowed_uri_format(rv: str) -> bool:
 @lru_cache(maxsize=1)
 def get_json_schema() -> dict[str, Any]:
     """Get the JSON schema for the bioregistry."""
-    if PYDANTIC_1:
-        raise NotImplementedError
-
     rv = {
         "$schema": "http://json-schema.org/draft-07/schema#",
         "$id": "https://bioregistry.io/schema.json",
@@ -2860,8 +2850,6 @@ def get_json_schema() -> dict[str, Any]:
     )
 
     # see https://docs.pydantic.dev/latest/usage/json_schema/#general-notes-on-json-schema-generation
-    from pydantic.json_schema import models_json_schema
-
     _, schema_dict = models_json_schema(
         [(model, "validation") for model in models],  # type:ignore
         title=title,
@@ -2895,7 +2883,7 @@ def _get(resource: Resource, key: str) -> Any:
 
 def deduplicate_publications(publications: Iterable[Publication]) -> List[Publication]:
     """Deduplicate publications."""
-    records = [pydantic_dict(publication, exclude_none=True) for publication in publications]
+    records = [publication.model_dump(exclude_none=True) for publication in publications]
     records_deduplicated = deduplicate(records, keys=DEDP_PUB_KEYS)
     return [Publication(**record) for record in records_deduplicated]
 

src/bioregistry/schema/utils.py

@@ -8,8 +8,6 @@
 
 from pydantic import BaseModel
 
-from bioregistry.utils import pydantic_dict
-
 __all__ = [
     "sanitize_dict",
     "sanitize_model",
@@ -31,7 +29,7 @@ def sanitize_dict(d: dict[str, Any]) -> dict[str, Any]:
 
 def sanitize_model(base_model: BaseModel, **kwargs: Any) -> Mapping[str, Any]:
     """Sanitize a single Pydantic model."""
-    return sanitize_dict(pydantic_dict(base_model, **kwargs))
+    return sanitize_dict(base_model.model_dump(**kwargs))
 
 
 def sanitize_mapping(mapping: Mapping[str, BaseModel]) -> Mapping[str, Mapping[str, Any]]:

src/bioregistry/schema_utils.py

@@ -18,7 +18,6 @@
     MISMATCH_PATH,
 )
 from .schema import Collection, Context, Registry, Resource
-from .utils import pydantic_dict, pydantic_parse
 
 logger = logging.getLogger(__name__)
 
@@ -59,7 +58,7 @@ def _registry_from_path(path: Union[str, Path]) -> Mapping[str, Resource]:
         data = json.load(file)
     for prefix, value in data.items():
         value.setdefault("prefix", prefix)
-    return {prefix: pydantic_parse(Resource, value) for prefix, value in data.items()}
+    return {prefix: Resource.model_validate(value) for prefix, value in data.items()}
 
 
 def add_resource(resource: Resource) -> None:
@@ -125,7 +124,7 @@ def write_collections(collections: Mapping[str, Collection]) -> None:
         collection.resources = sorted(set(collection.resources))
     with open(COLLECTIONS_PATH, encoding="utf-8", mode="w") as file:
         json.dump(
-            {"collections": [pydantic_dict(c, exclude_none=True) for c in values]},
+            {"collections": [c.model_dump(exclude_none=True) for c in values]},
             file,
             indent=2,
             sort_keys=True,
@@ -140,7 +139,7 @@ def write_registry(registry: Mapping[str, Resource], *, path: Optional[Path] = N
     with path.open(mode="w", encoding="utf-8") as file:
         json.dump(
             {
-                key: pydantic_dict(resource, exclude_none=True, exclude={"prefix"})
+                key: resource.model_dump(exclude_none=True, exclude={"prefix"})
                 for key, resource in registry.items()
             },
             file,
@@ -155,7 +154,7 @@ def write_metaregistry(metaregistry: Mapping[str, Registry]) -> None:
     values = [v for _, v in sorted(metaregistry.items())]
     with open(METAREGISTRY_PATH, mode="w", encoding="utf-8") as file:
         json.dump(
-            {"metaregistry": [pydantic_dict(m, exclude_none=True) for m in values]},
+            {"metaregistry": [m.model_dump(exclude_none=True) for m in values]},
             fp=file,
             indent=2,
             sort_keys=True,
@@ -167,7 +166,7 @@ def write_contexts(contexts: Mapping[str, Context]) -> None:
     """Write to contexts."""
     with open(CONTEXTS_PATH, mode="w", encoding="utf-8") as file:
         json.dump(
-            {key: pydantic_dict(context, exclude_none=True) for key, context in contexts.items()},
+            {key: context.model_dump(exclude_none=True) for key, context in contexts.items()},
             fp=file,
             indent=2,
             sort_keys=True,

src/bioregistry/summary.py

@@ -12,7 +12,6 @@
 
 from bioregistry import manager
 from bioregistry.constants import TABLES_SUMMARY_LATEX_PATH
-from bioregistry.utils import pydantic_dict
 from bioregistry.version import get_version
 
 
@@ -124,7 +123,7 @@ def make(cls, force_download: bool = False):
         prefixes_curated = sum(
             any(
                 x not in metaprefixes
-                for x, v in pydantic_dict(entry).items()
+                for x, v in entry.model_dump().items()
                 if v is not None and x not in {"prefix", "mappings"}
             )
             for prefix, entry in registry.items()