Merge branch 'vrs/2.1.0-snapshot.2026-02' into issue-602

Author: korikuzma

pyproject.toml

@@ -40,7 +40,6 @@ dependencies = [
     "bioutils",
     "requests",
     "canonicaljson",
-    "setuptools>=78.1.0", # TODO remove this pin caused by HGVS issue
     "typing-extensions",
 ]
 
@@ -53,7 +52,6 @@ extras = [
     "dill~=0.3.7",
     "click",
     "pysam==0.23.0",  # pinned pending https://github.com/ga4gh/vrs-python/issues/560
-    "typing_extensions",
 ]
 dev = [
     # tests
@@ -71,7 +69,6 @@ dev = [
 ]
 notebooks = [
     "jupyter",
-    "tabulate",
     "pyyaml"
 ]
 

src/ga4gh/core/enderef.py

@@ -11,6 +11,8 @@
 
 import logging
 
+from pydantic.main import BaseModel
+
 from .identifiers import ga4gh_identify, is_ga4gh_identifier
 from .pydantic import (
     get_pydantic_root,
@@ -22,7 +24,12 @@
 _logger = logging.getLogger(__name__)
 
 
-def ga4gh_enref(o, cra_map, object_store=None, return_id_obj_tuple=False) -> tuple:  # noqa: ANN001
+def ga4gh_enref(
+    o,  # noqa: ANN001
+    cra_map,  # noqa: ANN001
+    object_store=None,  # noqa: ANN001
+    return_id_obj_tuple: bool = False,
+) -> tuple:
     """Recursively convert "referable attributes" from inlined to
     referenced form.  Returns a new object.
 
@@ -35,13 +42,13 @@ def ga4gh_enref(o, cra_map, object_store=None, return_id_obj_tuple=False) -> tup
     :raise TypeError: if any object IDs are non-GA4GH CURIEs
     """
 
-    def _id_and_store(o):  # noqa: ANN202 ANN001
+    def _id_and_store(o) -> str | None:  # noqa: ANN001
         _id = ga4gh_identify(o)
         if _id and object_store is not None:
             object_store[_id] = o
         return _id
 
-    def _enref(o):  # noqa: ANN202 ANN001
+    def _enref(o: BaseModel) -> str | None:
         """depth-first recursive, in-place enref of object; returns id of object"""
         ref_att_names = cra_map.get(o.type, [])
         for ran in ref_att_names:
@@ -76,7 +83,7 @@ def _enref(o):  # noqa: ANN202 ANN001
     return (_id, o) if return_id_obj_tuple else o
 
 
-def ga4gh_deref(o, cra_map, object_store):  # noqa: ANN201 ANN001
+def ga4gh_deref(o, cra_map, object_store) -> BaseModel:  # noqa: ANN001
     """Convert "referable attributes" in-place from referenced to inlined
     form.
 
@@ -87,7 +94,7 @@ def ga4gh_deref(o, cra_map, object_store):  # noqa: ANN201 ANN001
 
     """
 
-    def _deref(o):  # noqa: ANN202 ANN001
+    def _deref(o: BaseModel):  # noqa: ANN202
         """depth-first recursive, in-place deref of object; returns id of object"""
         if o.type not in cra_map:
             _logger.warning("%s not in cra_map %s", o.type, cra_map)

src/ga4gh/core/identifiers.py

@@ -85,14 +85,14 @@ class use_ga4gh_compute_identifier_when(ContextDecorator):  # noqa: N801
     def my_method():
     """
 
-    def __init__(self, when: VrsObjectIdentifierIs):
+    def __init__(self, when: VrsObjectIdentifierIs) -> None:
         self.when = when
         self.token = None
 
-    def __enter__(self):  # noqa: ANN204
+    def __enter__(self) -> None:
         self.token = ga4gh_compute_identifier_when.set(self.when)
 
-    def __exit__(self, exc_type, exc, exc_tb):  # noqa: ANN204 ANN001
+    def __exit__(self, exc_type, exc, exc_tb) -> None:  # noqa: ANN001
         ga4gh_compute_identifier_when.reset(self.token)
 
 

src/ga4gh/core/models.py

@@ -144,6 +144,16 @@ class Element(BaseModel, ABC):
         description="A list of extensions to the Entity, that allow for capture of information not directly supported by elements defined in the model.",
     )
 
+    def get_extensions_by_name(self, name: str) -> list[Extension]:
+        """Fetch all contained extension exactly matching the provided name
+
+        :param name: name of extension to fetch
+        :return: a list of all matching extensions, empty if no matches found (or if instance contains no extensions)
+        """
+        if not self.extensions:
+            return []
+        return [e for e in self.extensions if e.name == name]
+
 
 #########################################
 # General-purpose data classes

src/ga4gh/vrs/dataproxy.py

@@ -247,10 +247,11 @@ class SeqRepoRESTDataProxy(_SeqRepoDataProxyBase):
 
     rest_version = "1"
 
-    def __init__(self, base_url: str, disable_healthcheck: bool = False):
+    def __init__(self, base_url: str, disable_healthcheck: bool = False) -> None:
         """Initialize REST-based dataproxy instance.
 
         :param base_url: root URL to server
+        :param disable_healthcheck: Whether healthcheck should be disabled
         """
         super().__init__()
         self.base_url = f"{base_url}/{self.rest_version}/"
@@ -288,12 +289,12 @@ class SequenceProxy(Sequence):
 
     """
 
-    def __init__(self, dp: _DataProxy, alias: str):  # noqa: D107
+    def __init__(self, dp: _DataProxy, alias: str) -> None:  # noqa: D107
         self.dp = dp
         self.alias = alias
         self._md = self.dp.get_metadata(self.alias)
 
-    def __str__(self):  # noqa: D105 ANN204
+    def __str__(self) -> str:  # noqa: D105
         return self.dp.get_sequence(self.alias)
 
     def __len__(self):  # noqa: D105 ANN204
@@ -303,7 +304,7 @@ def __reversed__(self):  # noqa: D105 ANN204
         msg = "Reversed iteration of a SequenceProxy is not implemented"
         raise NotImplementedError(msg)
 
-    def __getitem__(self, key):  # noqa: ANN001 ANN204
+    def __getitem__(self, key) -> str:  # noqa: ANN001
         """Return sequence for key (slice), fetching if necessary"""
         if isinstance(key, int):
             key = slice(key, key + 1)
@@ -344,16 +345,18 @@ def _isoformat(o: datetime.datetime) -> str:
 #         self.base_url = base_url
 
 
-def create_dataproxy(uri: str | None = None) -> _DataProxy:
+def create_dataproxy(
+    uri: str | None = None, disable_healthcheck: bool = False
+) -> _DataProxy:
     """Create a dataproxy from uri or GA4GH_VRS_DATAPROXY_URI
 
-    Currently accepted URI schemes:
-
-    * seqrepo+file:///path/to/seqrepo/root
-    * seqrepo+:../relative/path/to/seqrepo/root
-    * seqrepo+http://localhost:5000/seqrepo
-    * seqrepo+https://somewhere:5000/seqrepo
+    :param uri: Dataproxy URI. Currently accepted URI schemes:
 
+        * seqrepo+file:///path/to/seqrepo/root
+        * seqrepo+:../relative/path/to/seqrepo/root
+        * seqrepo+http://localhost:5000/seqrepo
+        * seqrepo+https://somewhere:5000/seqrepo
+    :param disable_healthcheck: Whether healthcheck should be disabled in REST dataproxy
     :raise ValueError: if URI doesn't match recognized schemes, e.g. is missing provider
         prefix (`"seqrepo+"`)
     """
@@ -379,7 +382,9 @@ def create_dataproxy(uri: str | None = None) -> _DataProxy:
             sr = SeqRepo(root_dir=parsed_uri.path)
             dp = SeqRepoDataProxy(sr)
         elif proto in ("http", "https"):
-            dp = SeqRepoRESTDataProxy(uri[len(provider) + 1 :])
+            dp = SeqRepoRESTDataProxy(
+                uri[len(provider) + 1 :], disable_healthcheck=disable_healthcheck
+            )
         else:
             msg = f"SeqRepo URI scheme {parsed_uri.scheme} not implemented"
             raise ValueError(msg)

src/ga4gh/vrs/enderef.py

@@ -1,9 +1,11 @@
+from pydantic.main import BaseModel
+
 from ga4gh.core import ga4gh_deref, ga4gh_enref
 
 from .models import class_refatt_map
 
 
-def vrs_enref(o, object_store=None, return_id_obj_tuple=False):
+def vrs_enref(o, object_store=None, return_id_obj_tuple: bool = False):
     return ga4gh_enref(
         o,
         cra_map=class_refatt_map,
@@ -12,5 +14,5 @@ def vrs_enref(o, object_store=None, return_id_obj_tuple=False):
     )
 
 
-def vrs_deref(o, object_store):
+def vrs_deref(o, object_store) -> BaseModel:
     return ga4gh_deref(o, cra_map=class_refatt_map, object_store=object_store)

src/ga4gh/vrs/extras/object_store.py

@@ -13,7 +13,9 @@ class Sqlite3MutableMapping(MutableMapping):
     If not used as a contextmanager, user must call commit and/or close.
     """
 
-    def __init__(self, sqlite3_db: str | sqlite3.Connection, autocommit: bool = True):
+    def __init__(
+        self, sqlite3_db: str | sqlite3.Connection, autocommit: bool = True
+    ) -> None:
         """Connect to the sqlite3 database specified by an existing sqlite3.Connection
         or a connection string.
 
@@ -96,10 +98,10 @@ def __len__(self):
         finally:
             cur.close()
 
-    def commit(self):
+    def commit(self) -> None:
         self.db.commit()
 
-    def close(self):
+    def close(self) -> None:
         with self._closed_lock:
             if not self._closed:
                 self.commit()
@@ -110,5 +112,5 @@ def __enter__(self):
         self.db.__enter__()
         return self
 
-    def __exit__(self, exc_type, exc_value, traceback):
+    def __exit__(self, exc_type, exc_value, traceback) -> None:
         self.db.__exit__(exc_type, exc_value, traceback)

src/ga4gh/vrs/extras/translator.py

@@ -75,7 +75,7 @@ def __init__(
         default_assembly_name: str = "GRCh38",
         identify: bool = True,
         rle_seq_limit: int | None = 50,
-    ):
+    ) -> None:
         self.default_assembly_name = default_assembly_name
         self.data_proxy = data_proxy
         self.identify = identify
@@ -175,7 +175,7 @@ def __init__(
         data_proxy: _DataProxy,
         default_assembly_name: str = "GRCh38",
         identify: bool = True,
-    ):
+    ) -> None:
         """Initialize AlleleTranslator class"""
         super().__init__(data_proxy, default_assembly_name, identify)
 
@@ -528,7 +528,7 @@ def __init__(
         data_proxy: _DataProxy,
         default_assembly_name: str = "GRCh38",
         identify: bool = True,
-    ):
+    ) -> None:
         """Initialize CnvTranslator class"""
         super().__init__(data_proxy, default_assembly_name, identify)
         self.from_translators = {

src/ga4gh/vrs/models.py

@@ -50,7 +50,7 @@
 def flatten(vals):
     """Flatten vals recursively, lazily using yield"""
 
-    def is_coll(thing):
+    def is_coll(thing) -> bool:
         """Return True if the thing looks like a collection.
 
         This is not exhaustive, do not use in general.
@@ -77,7 +77,7 @@ def flatten_type(t):
     return [t]
 
 
-def overlaps(a: list, b: list):
+def overlaps(a: list, b: list) -> bool:
     """Return true if there are any elements in common between a and b"""
     return len(set(a).intersection(set(b))) > 0
 
@@ -266,7 +266,7 @@ class _ValueObject(Entity, ABC):
     See https://en.wikipedia.org/wiki/Value_object for more on Value Objects.
     """
 
-    def __hash__(self):
+    def __hash__(self) -> int:
         return encode_canonical_json(self.ga4gh_serialize()).decode("utf-8").__hash__()
 
     def ga4gh_serialize(self) -> dict:
@@ -298,14 +298,14 @@ class Ga4ghIdentifiableObject(_ValueObject, ABC):
         description="A sha512t24u digest created using the VRS Computed Identifier algorithm.",
     )
 
-    def __lt__(self, other):
+    def __lt__(self, other) -> bool:
         return self.get_or_create_digest() < other.get_or_create_digest()
 
     @staticmethod
     def is_ga4gh_identifiable() -> bool:
         return True
 
-    def has_valid_ga4gh_id(self):
+    def has_valid_ga4gh_id(self) -> bool | str | None:
         return self.id and GA4GH_IR_REGEXP.match(self.id) is not None
 
     def compute_digest(
@@ -372,7 +372,7 @@ def get_or_create_ga4gh_identifier(
         else:
             return self.compute_ga4gh_identifier(recompute)
 
-    def compute_ga4gh_identifier(self, recompute: bool = False, as_version=None):
+    def compute_ga4gh_identifier(self, recompute: bool = False, as_version=None) -> str:
         """Return a GA4GH Computed Identifier.
 
         If ``as_version`` is provided, other parameters are ignored and a computed
@@ -629,7 +629,7 @@ def validate_start_end(
                 raise ValueError(err_msg)
         return v
 
-    def ga4gh_serialize_as_version(self, as_version: PrevVrsVersion):
+    def ga4gh_serialize_as_version(self, as_version: PrevVrsVersion) -> str:
         """Return a serialized string following the conventions for SequenceLocation
         serialization as defined in the VRS version specified by ``as_version``.
 
@@ -661,7 +661,7 @@ def ga4gh_serialize_as_version(self, as_version: PrevVrsVersion):
         msg = f"Received an unexpected value for `as_version`: {as_version}. MUST be an instance of `PrevVrsVersion`."
         raise TypeError(msg)
 
-    def get_refget_accession(self):
+    def get_refget_accession(self) -> str | None:
         if isinstance(self.sequenceReference, SequenceReference):
             return self.sequenceReference.refgetAccession
         if isinstance(self.sequenceReference, iriReference):
@@ -769,7 +769,7 @@ class Allele(_VariationBase, BaseModelForbidExtra):
         Field(..., description="An expression of the sequence state")
     )
 
-    def ga4gh_serialize_as_version(self, as_version: PrevVrsVersion):
+    def ga4gh_serialize_as_version(self, as_version: PrevVrsVersion) -> str:
         """Return a serialized string following the conventions for
         Allele serialization as defined in the VRS version specified by 'as_version`.
 

src/ga4gh/vrs/normalize.py

@@ -11,6 +11,7 @@
 
 from bioutils.normalize import NormalizationMode
 from bioutils.normalize import normalize as _normalize
+from pydantic.main import BaseModel
 
 from ga4gh.core import ga4gh_digest, is_pydantic_instance, pydantic_copy
 from ga4gh.vrs import models
@@ -83,7 +84,9 @@ def _get_new_allele_location_pos(
     return val
 
 
-def _normalize_allele(input_allele: models.Allele, data_proxy, rle_seq_limit=50):
+def _normalize_allele(
+    input_allele: models.Allele, data_proxy: _DataProxy, rle_seq_limit: int = 50
+):
     """Normalize Allele using "fully-justified" normalization adapted from NCBI's
     VOCA. Fully-justified normalization expands such ambiguous representation over the
     entire region of ambiguity, resulting in an unambiguous representation that may be
@@ -305,7 +308,7 @@ def denormalize_reference_length_expression(
     return alt
 
 
-def _factor_gen(n):
+def _factor_gen(n: int):
     """Yield all factors of an integer `n`, in descending order"""
     lower_factors = []
     i = 1
@@ -319,7 +322,11 @@ def _factor_gen(n):
 
 
 def _define_rle_allele(
-    allele, length, repeat_subunit_length, rle_seq_limit, extended_alt_seq
+    allele: BaseModel,
+    length: int,
+    repeat_subunit_length: int,
+    rle_seq_limit,
+    extended_alt_seq,
 ):
     # Otherwise, create the Allele as an RLE
     allele.state = models.ReferenceLengthExpression(
@@ -332,7 +339,7 @@ def _define_rle_allele(
     return allele
 
 
-def _is_valid_cycle(template_start, template, target):
+def _is_valid_cycle(template_start, template: str, target) -> bool:
     cycle = itertools.cycle(template[template_start:])
     for char in target[len(template) :]:  # noqa: SIM110
         if char != next(cycle):