Merge branch 'main' into issue-452

docs/extras/vcf_annotator.md

@@ -54,7 +54,7 @@ vrs-annotate vcf input.vcf.gz --vcf_out output.vcf.gz --vrs_pickle_out vrs_objec
 ```
 
 ### Other Options
-`--vrs_attribute`
+`--vrs_attributes`
 >Will include VRS_Start, VRS_End, VRS_State fields in the INFO field.
 
 `--assembly` [TEXT]

src/ga4gh/vrs/extras/vcf_annotation.py

@@ -3,6 +3,7 @@
     $ vrs-annotate vcf input.vcf.gz --vcf_out output.vcf.gz --vrs_pickle_out vrs_objects.pkl
 
 """
+from collections.abc import Callable
 import pathlib
 import logging
 import pickle
@@ -36,10 +37,55 @@ class SeqRepoProxyType(str, Enum):
 @click.group()
 def _cli() -> None:
     """Annotate input files with VRS variation objects."""
+    logging.basicConfig(
+        filename="vrs-annotate.log",
+        level=logging.INFO,
+        format="%(asctime)s - %(name)s - %(levelname)s - %(message)s"
+    )
+
+
+class _LogLevel(str, Enum):
+    """Define legal values for `--log_level` option."""
+    DEBUG = "debug"
+    INFO = "info"
+    WARNING = "warning"
+    ERROR = "error"
+    CRITICAL = "critical"
+
 
+def _log_level_option(func: Callable) -> Callable:
+    """Provide reusable log level CLI option decorator.
+
+    Adds a `--log_level` CLI option to any decorated command. Doesn't pass on any
+    values, just sets the logging level for this module.
+
+    :param func: incoming click command
+    :return: same command, wrapped with log level option
+    """
+    def _set_log_level(ctx: dict, param: str, value: _LogLevel) -> None:  # pylint: disable=unused-argument
+        level_map = {
+            _LogLevel.DEBUG: logging.DEBUG,
+            _LogLevel.INFO: logging.INFO,
+            _LogLevel.WARNING: logging.WARNING,
+            _LogLevel.ERROR: logging.ERROR,
+            _LogLevel.CRITICAL: logging.CRITICAL,
+        }
+        logging.getLogger(__name__).setLevel(level_map[value])
+
+    wrapped_func = click.option(
+        "--log_level",
+        type=click.Choice([v.value for v in _LogLevel.__members__.values()]),
+        default="info",
+        help="Set the logging level.",
+        callback=_set_log_level,
+        expose_value=False,
+        is_eager=True
+    )(func)
+    return wrapped_func
 
 
 @_cli.command(name="vcf")
+@_log_level_option
 @click.argument(
     "vcf_in",
     nargs=1,

src/ga4gh/vrs/models.py

@@ -27,7 +27,7 @@
 from ga4gh.core.pydantic import get_pydantic_root
 
 from canonicaljson import encode_canonical_json
-from pydantic import BaseModel, Field, RootModel, StringConstraints, ConfigDict
+from pydantic import BaseModel, Field, RootModel, StringConstraints, ConfigDict, ValidationInfo, field_validator
 
 from ga4gh.core.pydantic import (
     getattr_in
@@ -366,6 +366,25 @@ class Range(RootModel):
         min_length=2,
     )
 
+    @field_validator("root", mode="after")
+    def validate_range(cls, v: List[Optional[int]]) -> List[Optional[int]]:
+        """Validate range values
+
+        :param v: Root value
+        :raises ValueError: If ``root`` does not include at least one integer or if
+            the first element in ``root`` is greater than the second element in ``root``
+        :return: Inclusive range
+        """
+        if v.count(None) == 2:
+            err_msg = "Must provide at least one integer."
+            raise ValueError(err_msg)
+
+        if v[0] is not None and v[1] is not None:
+            if v[0] > v[1]:
+                err_msg = "The first integer must be less than or equal to the second integer."
+                raise ValueError(err_msg)
+
+        return v
 
 class Residue(RootModel):
     """A character representing a specific residue (i.e., molecular species) or
@@ -498,6 +517,26 @@ class SequenceLocation(_Ga4ghIdentifiableObject):
     )
     sequence: Optional[SequenceString] = Field(None, description="The literal sequence encoded by the `sequenceReference` at these coordinates.")
 
+    @field_validator("start", "end", mode="after")
+    def validate_start_end(cls, v: Optional[Union[Range, int]], info: ValidationInfo) -> Optional[Union[Range, int]]:
+        """Validate ``start`` and ``end`` fields
+
+        :param v: ``start`` or ``end`` value
+        :param info: Validation info
+        :raises ValueError: If ``start`` or ``end`` has a value less than 0
+        :return: Sequence Location
+        """
+        if v is not None:
+            if isinstance(v, int):
+                int_values = [v]
+            else:
+                int_values = [val for val in v.root if val is not None]
+
+            if any(int_val < 0 for int_val in int_values):
+                err_msg = f"The minimum value of `{info.field_name}` is 0."
+                raise ValueError(err_msg)
+        return v
+
     def ga4gh_serialize_as_version(self, as_version: PrevVrsVersion):
         """This method will return a serialized string following the conventions for
         SequenceLocation serialization as defined in the VRS version specified by

tests/test_vrs.py

@@ -96,6 +96,22 @@
 cpb_431012 = models.CisPhasedBlock(**cpb_431012_dict)
 
 
+@pytest.mark.parametrize(
+    "vrs_model, expected_err_msg",
+    [
+        (lambda: models.Range(root=[None, None]), "Must provide at least one integer."),
+        (lambda: models.Range(root=[2, 1]), "The first integer must be less than or equal to the second integer."),
+        (lambda: models.SequenceLocation(sequenceReference=allele_280320.location.sequenceReference, start=-1), "The minimum value of `start` is 0."),
+        (lambda: models.SequenceLocation(sequenceReference=allele_280320.location.sequenceReference, end=[-1, 0]), "The minimum value of `end` is 0."),
+    ]
+)
+def test_model_validation_errors(vrs_model, expected_err_msg):
+    """Test that invalid VRS models raise errors"""
+    with pytest.raises(ValueError) as e:
+        vrs_model()
+    assert str(e.value.errors()[0]["ctx"]["error"]) == expected_err_msg
+
+
 def test_vr():
     assert a.model_dump(exclude_none=True) == allele_dict
     assert is_pydantic_instance(a)