wip

src/ga4gh/core/__init__.py

@@ -8,10 +8,13 @@
 from ._internal.digests import sha512t24u
 from ._internal.enderef import ga4gh_enref, ga4gh_deref
 from ._internal.exceptions import GA4GHError
-from ._internal.identifiers import (ga4gh_digest, ga4gh_identify, ga4gh_serialize, is_ga4gh_identifier,
-                                    parse_ga4gh_identifier)
-from ._internal.jsonschema import (build_models, build_class_referable_attribute_map, is_pjs_class, is_pjs_instance,
-                                   is_curie, is_identifiable, is_literal, pjs_copy)
+from ._internal.identifiers import (
+    ga4gh_digest, ga4gh_identify, ga4gh_serialize, is_ga4gh_identifier,
+    parse_ga4gh_identifier
+)
+from ._internal.pydantic import (
+    is_pydantic_instance, is_curie, is_identifiable, is_literal, pjs_copy
+)
 
 try:
     __version__ = get_distribution(__name__).version

src/ga4gh/core/_internal/enderef.py

@@ -9,12 +9,7 @@
 
 """
 
-import logging
-
 from .identifiers import ga4gh_identify, is_ga4gh_identifier
-from .jsonschema import is_array, is_curie, is_identifiable, is_pjs_instance, pjs_copy
-
-_logger = logging.getLogger(__name__)
 
 
 def ga4gh_enref(o, cra_map, object_store=None):
@@ -40,7 +35,7 @@ def _enref(o):
         ref_att_names = cra_map.get(o.type, [])
         for ran in ref_att_names:
             v = o[ran]
-            if is_array(v):
+            if is_list(v):
                 o[ran] = [_enref(o2) for o2 in v]
             elif isinstance(v, str):
                 pass
@@ -53,7 +48,7 @@ def _enref(o):
 
         return _id_and_store(o)
 
-    if not is_pjs_instance(o):
+    if not is_pydantic_instance(o):
         raise ValueError("Called ga4gh_enref() with non-python_jsonschema_object instance")
     if not is_identifiable(o):
         raise ValueError("Called ga4gh_enref() with non-identifiable object")
@@ -82,7 +77,7 @@ def _deref(o):
         ref_att_names = cra_map[o.type]
         for ran in ref_att_names:
             v = o[ran]
-            if is_array(v):
+            if is_list(v):
                 o[ran] = [_deref(object_store[str(curie)]) for curie in v]
             elif is_ga4gh_identifier(v):
                 o[ran] = _deref(object_store[str(v)])
@@ -91,7 +86,7 @@ def _deref(o):
 
         return o
 
-    if not is_pjs_instance(o):
+    if not is_pydantic_instance(o):
         raise ValueError("Called ga4gh_deref() with non-python_jsonschema_object instance")
     if not is_identifiable(o):
         raise ValueError("Called ga4gh_deref() with non-identifiable object")

src/ga4gh/core/_internal/identifiers.py

@@ -14,32 +14,15 @@
 For that reason, they are implemented here in one file.
 
 """
-
-import logging
-import os
 import re
 
 from canonicaljson import encode_canonical_json
-import pkg_resources
-import yaml
 
 from .digests import sha512t24u
-from .jsonschema import is_array, is_pjs_instance, is_curie_type, is_identifiable, is_literal
+from .pydantic import is_list, is_pydantic_instance, is_curie_type, is_identifiable, is_literal
 
 __all__ = "ga4gh_digest ga4gh_identify ga4gh_serialize is_ga4gh_identifier parse_ga4gh_identifier".split()
 
-_logger = logging.getLogger(__name__)
-
-# Assume that ga4gh.yaml and vrs.yaml files are in the same directory for now
-schema_dir = os.environ.get("VRSATILE_SCHEMA_DIR", pkg_resources.resource_filename(__name__, "data/schemas/vrsatile"))
-cfg = yaml.safe_load(open(schema_dir + "/merged.yaml"))
-defs = cfg["definitions"]
-
-type_prefix_map_default = dict()
-for k,v in defs.items():
-    if "ga4gh_prefix" in v:
-        type_prefix_map_default[k] = v["ga4gh_prefix"]
-
 namespace = "ga4gh"
 curie_sep = ":"
 ref_sep = "."
@@ -96,15 +79,8 @@ def ga4gh_identify(vro, type_prefix_map=None):
     'ga4gh:VSL.u5fspwVbQ79QkX6GHLF8tXPCAXFJqRPx'
 
     """
-
-    if type_prefix_map is None:
-        type_prefix_map = type_prefix_map_default
-    try:
-        pfx = type_prefix_map[vro.type]
-    except KeyError:
-        _logger.debug("No identifier prefix is defined for %s; check ga4gh.yaml", vro.type)
-        return None
     digest = ga4gh_digest(vro)
+    pfx = vro.prefix
     ir = f"{namespace}{curie_sep}{pfx}{ref_sep}{digest}"
     return ir
 
@@ -173,17 +149,16 @@ def dictify(vro, enref=True):
                 v = v.split(ref_sep, 1)[1]
             return v
 
-        if is_pjs_instance(vro):
+        if is_pydantic_instance(vro):
             if is_identifiable(vro) and enref:
                 return ga4gh_digest(vro)
+
             d = {k: dictify(vro[k], enref=True)
                  for k in vro
-                 if not (k.startswith("_") or
-                         (k == "id" and vro.type in type_prefix_map_default) or
-                         vro[k] is None)}
+                 if k in vro.ga4gh_digest_keys}
             return d
 
-        if is_array(vro):
+        if is_list(vro):
             if is_curie_type(vro[0]):
                 return sorted(dictify(o) for o in vro.data)
             return sorted([dictify(o) for o in vro.typed_elems])

src/ga4gh/core/_internal/pydantic.py

@@ -0,0 +1,25 @@
+def is_identifiable(o: any) -> bool:
+    """Determine if object is identifiable. An object is considered identifiable if
+    contains a `ga4gh_digest_keys` attribute
+
+    :param o: Object
+    :return: `True` if `o` has `ga4gh_digest_keys` attribute. `False` otherwise.
+    """
+    return hasattr(o, "ga4gh_digest_keys")
+
+
+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:
+    # return isinstance(o, CURIE)
+    pass
+
+def is_pydantic_instance(o: any) -> bool:
+    # return isinstance(o, BaseModel)
+    pass

src/ga4gh/vrs/normalize.py

@@ -7,7 +7,7 @@
 import logging
 
 from bioutils.normalize import normalize as _normalize, NormalizationMode
-from ga4gh.core import is_pjs_instance, pjs_copy, ga4gh_digest
+from ga4gh.core import is_pydantic_instance, pjs_copy, ga4gh_digest
 
 from ._internal import models
 from .dataproxy import SequenceProxy
@@ -71,7 +71,7 @@ def _normalize_variationset(o, data_proxy=None):
 def normalize(vo, data_proxy=None):
     """normalize given vrs object, regardless of type"""
 
-    assert is_pjs_instance(vo)
+    assert is_pydantic_instance(vo)
     vo_type = vo.type._value
 
     if vo_type in handlers:

tests/test_vrs.py

@@ -1,4 +1,4 @@
-from ga4gh.core import sha512t24u, ga4gh_digest, ga4gh_serialize, ga4gh_identify, is_pjs_instance
+from ga4gh.core import sha512t24u, ga4gh_digest, ga4gh_serialize, ga4gh_identify, is_pydantic_instance
 from ga4gh.vrs import models, vrs_deref, vrs_enref
 
 allele_dict = {
@@ -22,7 +22,7 @@ def test_vr():
 
     assert a.as_dict() == allele_dict
 
-    assert is_pjs_instance(a.location)
+    assert is_pydantic_instance(a.location)
 
     assert ga4gh_serialize(
         a.location