Merge branch 'main' into issue-433

src/ga4gh/core/__init__.py

@@ -8,12 +8,12 @@
 from .enderef import ga4gh_enref, ga4gh_deref
 from .identifiers import (
     ga4gh_digest, ga4gh_identify, ga4gh_serialize, is_ga4gh_identifier,
-    parse_ga4gh_identifier, VrsObjectIdentifierIs, use_ga4gh_compute_identifier_when,
+    VrsObjectIdentifierIs, use_ga4gh_compute_identifier_when,
     CURIE_NAMESPACE, CURIE_SEP, GA4GH_PREFIX_SEP, GA4GH_IR_REGEXP, GA4GH_DIGEST_REGEXP,
     PrevVrsVersion
 )
 from .pydantic import (
-    is_pydantic_instance, is_curie_type, is_ga4gh_identifiable, is_literal, pydantic_copy
+    is_pydantic_instance, is_curie_type, pydantic_copy
 )
 from . import entity_models, domain_models
 
@@ -25,7 +25,6 @@
     "ga4gh_identify",
     "ga4gh_serialize",
     "is_ga4gh_identifier",
-    "parse_ga4gh_identifier",
     "VrsObjectIdentifierIs",
     "use_ga4gh_compute_identifier_when",
     "CURIE_NAMESPACE",
@@ -36,8 +35,6 @@
     "PrevVrsVersion",
     "is_pydantic_instance",
     "is_curie_type",
-    "is_ga4gh_identifiable",
-    "is_literal",
     "pydantic_copy",
     "entity_models",
     "domain_models"

src/ga4gh/core/domain_models.py

@@ -15,7 +15,7 @@
 
 from pydantic import Field, RootModel
 
-from ga4gh.core.entity_models import _DomainEntity
+from ga4gh.core.entity_models import DomainEntity
 
 
 class CommonDomainType(str, Enum):
@@ -30,7 +30,7 @@ class CommonDomainType(str, Enum):
     TR_COMB = "CombinationTherapy"
     GENE = "Gene"
 
-class Phenotype(_DomainEntity):
+class Phenotype(DomainEntity):
     """An observable characteristic or trait of an organism."""
 
     type: Literal[CommonDomainType.PHENOTYPE] = Field(
@@ -39,7 +39,7 @@ class Phenotype(_DomainEntity):
     )
 
 
-class Disease(_DomainEntity):
+class Disease(DomainEntity):
     """A particular abnormal condition that negatively affects the structure or function
     of all or part of an organism and is not immediately due to any external injury.
     """
@@ -50,7 +50,7 @@ class Disease(_DomainEntity):
     )
 
 
-class TraitSet(_DomainEntity):
+class TraitSet(DomainEntity):
     """A set of phenotype and/or disease concepts that together constitute a condition."""
 
     type: Literal[CommonDomainType.TRAIT_SET] = Field(
@@ -73,7 +73,7 @@ class Condition(RootModel):
     )
 
 
-class TherapeuticAction(_DomainEntity):
+class TherapeuticAction(DomainEntity):
     """A therapeutic action taken that is intended to alter or stop a pathologic process."""
 
     type: Literal[CommonDomainType.TR_ACTION] = Field(
@@ -82,7 +82,7 @@ class TherapeuticAction(_DomainEntity):
     )
 
 
-class TherapeuticAgent(_DomainEntity):
+class TherapeuticAgent(DomainEntity):
     """An administered therapeutic agent that is intended to alter or stop a pathologic process."""
 
     type: Literal[CommonDomainType.TR_AGENT] = Field(
@@ -91,7 +91,7 @@ class TherapeuticAgent(_DomainEntity):
     )
 
 
-class TherapeuticSubstituteGroup(_DomainEntity):
+class TherapeuticSubstituteGroup(DomainEntity):
     """A group of therapeutic procedures that may be treated as substitutes for one another."""
 
     type: Literal[CommonDomainType.TR_SUB] = Field(
@@ -105,7 +105,7 @@ class TherapeuticSubstituteGroup(_DomainEntity):
     )
 
 
-class CombinationTherapy(_DomainEntity):
+class CombinationTherapy(DomainEntity):
     """A therapeutic procedure that involves multiple different therapeutic procedures
     performed in combination.
     """
@@ -133,7 +133,7 @@ class TherapeuticProcedure(RootModel):
     )
 
 
-class Gene(_DomainEntity):
+class Gene(DomainEntity):
     """A basic physical and functional unit of heredity."""
 
     type: Literal[CommonDomainType.GENE] = Field(

src/ga4gh/core/enderef.py

@@ -13,9 +13,7 @@
 from .identifiers import ga4gh_identify, is_ga4gh_identifier
 from .pydantic import (
     is_pydantic_instance,
-    is_list,
     is_curie_type,
-    is_ga4gh_identifiable,
     get_pydantic_root,
     pydantic_copy)
 
@@ -45,7 +43,7 @@ def _enref(o):
         ref_att_names = cra_map.get(o.type, [])
         for ran in ref_att_names:
             v = getattr(o, ran)
-            if is_list(v):
+            if isinstance(v, list):
                 setattr(o, ran, [_enref(o2) for o2 in v])
             elif isinstance(v, str):
                 pass
@@ -60,7 +58,7 @@ def _enref(o):
 
     if not is_pydantic_instance(o):
         raise ValueError("Called ga4gh_enref() with non-pydantic instance")
-    if not is_ga4gh_identifiable(o):
+    if not o.is_ga4gh_identifiable():
         raise ValueError("Called ga4gh_enref() with non-identifiable object")
 
     # in-place replacement on object copy
@@ -88,7 +86,7 @@ def _deref(o):
         ref_att_names = cra_map[o.type]
         for ran in ref_att_names:
             v = getattr(o, ran)
-            if is_list(v):
+            if isinstance(v, list):
                 setattr(o, ran, [_deref(object_store[str(curie)]) for curie in v])
             elif is_ga4gh_identifier(v):
                 v = get_pydantic_root(v)
@@ -101,7 +99,7 @@ def _deref(o):
 
     if not is_pydantic_instance(o):
         raise ValueError("Called ga4gh_deref() with non-pydantic instance")
-    if not is_ga4gh_identifiable(o):
+    if not o.is_ga4gh_identifiable():
         raise ValueError("Called ga4gh_deref() with non-identifiable object")
 
     # in-place replacement on object copy

src/ga4gh/core/entity_models.py

@@ -10,6 +10,7 @@
   * `import ga4gh.core`, and refer to models using the fully-qualified
     module name, e.g., `ga4gh.core.entity_models.Coding`
 """
+from abc import ABC
 from typing import Any, Dict, Annotated, Optional, Union, List
 from enum import Enum
 
@@ -153,7 +154,7 @@ class Expression(BaseModel):
 #########################################
 
 
-class _Entity(BaseModel):
+class Entity(ABC, BaseModel):
     """Entity is the root class of the 'gks-common' core information model classes -
     those that have identifiers and other general metadata like labels, xrefs, urls,
     descriptions, etc. All common classes descend from and inherit its attributes.
@@ -187,7 +188,7 @@ def validate_type(cls, v: str | Enum) -> str:
         return v
 
 
-class _DomainEntity(_Entity):
+class DomainEntity(Entity, ABC):
     """An Entity that is specific to a particular biomedical domain such as disease,
     therapeutics, or genes. Domain Entities are considered as 'concept-level' entities,
     as opposed to particular instances. e.g. 'Lung Cancer', not 'patient123's lung

src/ga4gh/core/identifiers.py

@@ -16,18 +16,15 @@
 """
 
 import contextvars
-import logging
 import re
 from contextlib import ContextDecorator
 from enum import Enum, IntEnum
-from typing import Union, Optional
-from pydantic import BaseModel, RootModel
+from typing import Optional
+from pydantic import BaseModel
 
-from .pydantic import get_pydantic_root
+from ga4gh.core.pydantic import get_pydantic_root
 
-__all__ = "ga4gh_digest ga4gh_identify ga4gh_serialize is_ga4gh_identifier parse_ga4gh_identifier".split()
-
-_logger = logging.getLogger(__name__)
+__all__ = "ga4gh_digest ga4gh_identify ga4gh_serialize is_ga4gh_identifier".split()
 
 CURIE_NAMESPACE = "ga4gh"
 CURIE_SEP = ":"
@@ -36,7 +33,7 @@
 GA4GH_IR_REGEXP = re.compile(r"^ga4gh:(?P<type>[^.]+)\.(?P<digest>[0-9A-Za-z_\-]{32})$")
 GA4GH_DIGEST_REGEXP = re.compile(r"^[0-9A-Za-z_\-]{32}$")
 
-ns_w_sep = CURIE_NAMESPACE + CURIE_SEP
+NS_W_SEP = f"{CURIE_NAMESPACE}{CURIE_SEP}"
 
 
 class VrsObjectIdentifierIs(IntEnum):
@@ -47,9 +44,9 @@ class VrsObjectIdentifierIs(IntEnum):
       GA4GH_INVALID - Compute the identifier if it is missing or is present but syntactically invalid
       MISSING - Only compute the identifier if missing
 
-    The default behavior is safe and ensures that the identifiers are correct, 
-    but at a performance cost. Where the source of inputs to `ga4gh_identify` 
-    are well controlled, for example when annotating a VCF file with VRS IDs, 
+    The default behavior is safe and ensures that the identifiers are correct,
+    but at a performance cost. Where the source of inputs to `ga4gh_identify`
+    are well controlled, for example when annotating a VCF file with VRS IDs,
     using `MISSING` can improve performance.
     """
 
@@ -113,27 +110,7 @@ def is_ga4gh_identifier(ir):
     False
 
     """
-    return str(get_pydantic_root(ir)).startswith(ns_w_sep)
-
-
-def parse_ga4gh_identifier(ir):
-    """
-    Parses a GA4GH identifier, returning a dict with type and digest components
-
-    >>> parse_ga4gh_identifier("ga4gh:SQ.0123abcd")
-    {'type': 'SQ', 'digest': '0123abcd'}
-
-    >>> parse_ga4gh_identifier("notga4gh:SQ.0123abcd")
-    Traceback (most recent call last):
-    ...
-    ValueError: notga4gh:SQ.0123abcd
-
-    """
-
-    try:
-        return GA4GH_IR_REGEXP.match(str(ir)).groupdict()
-    except AttributeError as e:
-        raise ValueError(ir) from e
+    return str(get_pydantic_root(ir)).startswith(NS_W_SEP)
 
 
 def ga4gh_identify(vro, in_place: str = 'default', as_version: PrevVrsVersion | None = None) -> str | None:
@@ -171,7 +148,7 @@ def ga4gh_identify(vro, in_place: str = 'default', as_version: PrevVrsVersion |
             obj_id = getattr(vro, "id", None)
             if when_rule == VrsObjectIdentifierIs.MISSING:
                 do_compute = obj_id is None or obj_id == ""
-            else:  # GA4GHComputeIdentifierIs.GA4GH_INVALID
+            else:  # VrsObjectIdentifierIs.GA4GH_INVALID
                 do_compute = not vro.has_valid_ga4gh_id()
 
         if do_compute:
@@ -182,9 +159,11 @@ def ga4gh_identify(vro, in_place: str = 'default', as_version: PrevVrsVersion |
     return None
 
 
-def ga4gh_digest(vro: BaseModel, overwrite: bool = False, as_version: PrevVrsVersion | None = None) -> str:
+def ga4gh_digest(vro: BaseModel, overwrite: bool = False, as_version: PrevVrsVersion | None = None) -> str | None:
     """Return the GA4GH digest for the object.
 
+    Only GA4GH identifiable objects are GA4GH digestible.
+
     If ``as_version`` is provided, other parameters are ignored and a digest is returned
     following the conventions of the VRS version indicated by ``as_version_``.
     Raises ``ValueError`` if ``as_version`` is not a ``PrevVrsVersion``.
@@ -197,7 +176,7 @@ def ga4gh_digest(vro: BaseModel, overwrite: bool = False, as_version: PrevVrsVer
     """
     PrevVrsVersion.validate(as_version)
 
-    if vro.is_ga4gh_identifiable():  # Only GA4GH identifiable objects are GA4GH digestible
+    if vro.is_ga4gh_identifiable():
         if as_version is None:
             return vro.get_or_create_digest(overwrite)
         else:
@@ -206,30 +185,6 @@ def ga4gh_digest(vro: BaseModel, overwrite: bool = False, as_version: PrevVrsVer
         return None
 
 
-def replace_with_digest(val: dict) -> Union[str, dict]:
-    """
-    If val has a digest computed, return it, else return val
-    """
-    if isinstance(val, dict) and val.get("digest", None) is not None:
-        return val["digest"]
-    return val
-
-
-def collapse_identifiable_values(obj: dict) -> dict:
-    """
-    Replaces dict values with their digests if they are defined.
-    Does not collapse the top level object, only objects it contains.
-    """
-    if isinstance(obj, dict):
-        obj = {
-            k: replace_with_digest(collapse_identifiable_values(obj[k]))
-            for k in obj.keys()
-        }
-    elif isinstance(obj, list) or isinstance(obj, set):
-        obj = [replace_with_digest(collapse_identifiable_values(elem)) for elem in obj]
-    return obj
-
-
 def ga4gh_serialize(obj: BaseModel, as_version: PrevVrsVersion | None = None) -> Optional[bytes]:
     """Serializes an object for use in computed digest computation.
 

src/ga4gh/core/pydantic.py

@@ -21,24 +21,6 @@ def getattr_in(obj, names) -> Any:
     return v
 
 
-def is_ga4gh_identifiable(o: Any) -> bool:
-    """
-    Determine if object is a GA4GH identifiable type.
-
-    :param o: Object
-    :return: `True` if `o` is a GA4GH Identifiable Object. `False` otherwise.
-    """
-    return o.is_ga4gh_identifiable()
-
-
-def is_literal(o: Any) -> bool:
-    return isinstance(o, (str, int, float, complex, bool))
-
-
-def is_list(o: Any) -> bool:
-    return isinstance(o, list)
-
-
 def is_curie_type(o: Any) -> bool:
     """
     Returns true if the object is a str-like matching the CURIE pattern.