Git ref imports

pyproject.toml

@@ -28,7 +28,16 @@ keywords = [
     "variation"
 ]
 requires-python = ">=3.12"
-dynamic = ["version", "dependencies"]
+dependencies = [
+    "pyyaml"
+]
+dynamic = ["version"]
+
+[project.optional-dependencies]
+dev = [
+    "pytest",
+    "ruff==0.7.2"
+]
 
 [project.urls]
 Homepage = "https://github.com/ga4gh/gks-metaschema"
@@ -37,9 +46,6 @@ Changelog = "https://github.com/ga4gh/gks-metaschema/releases"
 Source = "https://github.com/ga4gh/gks-metaschema"
 "Bug Tracker" = "https://github.com/ga4gh/gks-metaschema/issues"
 
-[tool.setuptools.dynamic]
-dependencies = {file = [".requirements.txt"]}
-
 [tool.setuptools_scm]
 
 [project.scripts]

src/ga4gh/gks/metaschema/tools/schema_import.py

@@ -0,0 +1,24 @@
+import urllib.request
+import tempfile
+
+def construct_url(spec):
+    return f"https://raw.githubusercontent.com/{spec["org"]}/{spec["repo"]}/{spec["sha"]}/{spec["path"]}"
+
+def retrieve_import(spec):
+    url = construct_url(spec)
+    response = urllib.request.urlopen(url)
+    with tempfile.NamedTemporaryFile(delete=False) as t:
+        t.write(response.read())
+    return t.name
+
+## Leaving 'Rich' comments:
+
+# test_spec = {"org":"clingen-data-model",
+#              "repo":"gks-metaschema",
+#              "path":"tests/data/vrs/vrs.yaml",
+#              "sha":"fe2995181d06896f567f81ca4550b3b02830b1e5"}
+
+# path = retrieve_import(test_spec)
+
+# print(path)
+

src/ga4gh/gks/metaschema/tools/source_proc.py

@@ -7,6 +7,7 @@
 from pathlib import Path
 from collections import defaultdict
 from urllib.parse import urlparse
+from ga4gh.gks.metaschema.tools.schema_import import retrieve_import
 
 SCHEMA_DEF_KEYWORD_BY_VERSION = {
     "https://json-schema.org/draft-07/schema": "definitions",
@@ -183,17 +184,31 @@ def load_schema(schema_fp):
             schema = yaml.load(f, Loader=yaml.SafeLoader)
         return schema
 
+    def import_dependency_from_path(self, dependency, fp):
+        if self.imported:
+            root_fp = self.root_schema_fp
+        else:
+            root_fp = self.schema_fp
+        self.imports[dependency] = YamlSchemaProcessor(fp, root_fp=root_fp)    
+
+    def import_embedded_dependency(self, dependency):
+        fp = Path(self.raw_schema['imports'][dependency])
+        if not fp.is_absolute():
+            base_path = self.schema_fp.parent
+            fp = base_path.joinpath(fp)
+        self.import_dependency_from_path(dependency, fp)
+
+    def import_github_reference_dependency(self, dependency):
+        spec = self.raw_schema['imports'][dependency]
+        ref_path = retrieve_import(spec)
+        self.import_dependency_from_path(dependency, ref_path)
+
     def import_dependencies(self):
         for dependency in self.raw_schema.get('imports', list()):
-            fp = Path(self.raw_schema['imports'][dependency])
-            if not fp.is_absolute():
-                base_path = self.schema_fp.parent
-                fp = base_path.joinpath(fp)
-            if self.imported:
-                root_fp = self.root_schema_fp
+            if isinstance(self.raw_schema['imports'][dependency], str):
+                self.import_embedded_dependency(dependency)
             else:
-                root_fp = self.schema_fp
-            self.imports[dependency] = YamlSchemaProcessor(fp, root_fp=root_fp)
+                self.import_github_reference_dependency(dependency)
 
     def process_schema(self):
         if self.defs is None:

tests/data/vrs/vrs-source.yaml

@@ -15,8 +15,13 @@ type: object
 strict: true
 
 imports:
-  gks.core: ../gks-common/core-source.yaml
-
+  # gks.core: ../gks-common/core-source.yaml
+  gks.core:
+    org: clingen-data-model
+    repo: gks-metaschema
+    path: tests/data/gks-common/core-source.yaml
+    sha: fe2995181d06896f567f81ca4550b3b02830b1e5
+
 namespaces:
   gks.common: ../gks-common/$defs/
 

tests/test_basic.py

@@ -4,11 +4,15 @@
 import os
 from pathlib import Path
 
+import ga4gh.gks.metaschema.tools.schema_import
+
 from ga4gh.gks.metaschema.tools.source_proc import YamlSchemaProcessor
+from ga4gh.gks.metaschema.tools.schema_import import construct_url
 from ga4gh.gks.metaschema.scripts.y2t import main as y2t
 from ga4gh.gks.metaschema.scripts.source2splitjs import split_defs_to_js
 from ga4gh.gks.metaschema.scripts.source2classes import main as s2c
 
+
 root = Path(__file__).parent
 
 processor = YamlSchemaProcessor(root / 'data/vrs/vrs-source.yaml')
@@ -63,3 +67,13 @@ def test_docs_create():
     os.makedirs(defs)
     y2t(processor)
     assert True
+
+def test_imports():
+    spec = {"org":"clingen-data-model",
+            "repo":"gks-metaschema",
+            "path":"tests/data/vrs/vrs.yaml",
+            "sha":"fe2995181d06896f567f81ca4550b3b02830b1e5"}
+    target_string = "https://raw.githubusercontent.com/clingen-data-model/gks-metaschema/fe2995181d06896f567f81ca4550b3b02830b1e5/tests/data/vrs/vrs.yaml"
+    assert target_string == construct_url(spec)
+
+