[ENH][SCHEMA] Use Universal Path Library for Schema Retrieval

tools/schemacode/pyproject.toml

@@ -1,5 +1,9 @@
 [build-system]
-requires = ["pdm-backend", "acres", "pyyaml"]
+requires = [
+  "pdm-backend",
+  "acres",
+  "pyyaml",
+]
 build-backend = "pdm.backend"
 
 [project]
@@ -52,6 +56,7 @@ tests = [
   "coverage[toml]",
   "pytest>6",
   "pytest-cov",
+  "requests",
 ]
 all = [
   "bidsschematools[tests]",

tools/schemacode/src/bidsschematools/schema.py

@@ -9,6 +9,9 @@
 from copy import deepcopy
 from functools import cache, lru_cache
 from pathlib import Path
+import subprocess
+import urllib3
+from tempfile import TemporaryDirectory
 
 from . import _lazytypes as lt
 from . import data, utils
@@ -204,7 +207,9 @@ def flatten_enums(namespace: Namespace, inplace=True) -> Namespace:
 
 
 @lru_cache
-def load_schema(schema_path: lt.Traversable | str | None = None) -> Namespace:
+def load_schema(
+    schema_path: lt.Traversable | str | None = None, bids_version: str = None
+) -> Namespace:
     """Load the schema into a dict-like structure.
 
     This function allows the schema, like BIDS itself, to be specified in
@@ -218,6 +223,9 @@ def load_schema(schema_path: lt.Traversable | str | None = None) -> Namespace:
     schema_path : str, optional
         Directory containing yaml files or yaml file. If ``None``, use the
         default schema packaged with ``bidsschematools``.
+    bids_version: str, optional
+        Version of bids release to load schema from, allows user to have full namespace/dict
+        like access to bids schema given a specific version
 
     Returns
     -------
@@ -228,7 +236,7 @@ def load_schema(schema_path: lt.Traversable | str | None = None) -> Namespace:
     -----
     This function is cached, so it will only be called once per schema path.
     """
-    if schema_path is None:
+    if schema_path is None and bids_version is None:
         # Default to bundled JSON, fall back to bundled YAML directory
         schema_path = data.load.readable("schema.json")
         if not schema_path.is_file():
@@ -239,8 +247,33 @@ def load_schema(schema_path: lt.Traversable | str | None = None) -> Namespace:
                 assert isinstance(schema_path, Path)
                 schema_path = Path.resolve(schema_path.parent / content)
         lgr.info("No schema path specified, defaulting to the bundled schema, `%s`.", schema_path)
-    elif isinstance(schema_path, str):
+    elif isinstance(schema_path, str) and bids_version is None:
         schema_path = Path(schema_path)
+    elif schema_path is None and bids_version:
+        if re.search(r"^\d+\.\d+\.\d+(?:-[0-9A-Za-z.-]+)?$", bids_version):
+            schema_url = "https://bids-specification.readthedocs.io/en/v{}/schema.json".format(bids_version)
+
+        elif bids_version in ("stable", "latest"):
+            schema_url = "https://bids-specification.readthedocs.io/en/{}/schema.json".format(bids_version)
+        else:
+            raise Exception(
+                NameError(
+                    f"BIDS version is limited to `stable`, `latest`, or semantic format `X.X.X`, you gave {bids_version}"
+                )
+            )
+        http = urllib3.PoolManager()
+        response = http.request("GET", schema_url)
+        if response.status != 200:
+            raise urllib3.exceptions.HTTPError(
+                f"Unable to retrieve schema from {schema_url} (status {response.status})"
+            )
+        else:
+            with TemporaryDirectory() as tmp:
+                schema_path = Path(tmp) / "schema.json"
+                with open(schema_path, 'w') as f:
+                    json.dump(response.json(), f)
+
+                return Namespace.from_json(schema_path.read_text())
 
     # JSON file: just load it
     if schema_path.is_file():

tools/schemacode/src/bidsschematools/tests/test_schema.py

@@ -3,8 +3,10 @@
 import json
 import os
 import subprocess
+import requests
 from collections.abc import Mapping
 
+
 import pytest
 from jsonschema.exceptions import ValidationError
 
@@ -47,6 +49,24 @@ def test_load_schema(schema_dir):
     assert "$ref" not in str(schema_obj)
 
 
+@pytest.mark.skipif(
+    requests.get("https://bids-specification.readthedocs.io/").status_code != 200,
+    reason="Unable to reach 'https://bids-specification.readthedocs.io, skipping url retrieval",
+)
+def test_load_schema_from_url():
+    # load using latest and stable keywords
+    assert isinstance(schema.load_schema(bids_version="latest"), Mapping)
+    assert isinstance(schema.load_schema(bids_version="stable"), Mapping)
+
+    # load with known version
+    assert isinstance(schema.load_schema(bids_version="1.11.1"), Mapping)
+
+    # load with bogus url
+    #bad_url = "https://bids-specification.readthedocs.io/en/not-a-real-version/schema.json"
+    #with pytest.raises(Exception):
+    #    schema.load_schema(bad_url)
+
+
 def test_object_definitions(schema_obj):
     """Ensure that object definitions in the schema contain required fields."""
     for obj_type, obj_type_def in schema_obj["objects"].items():