eager resolution of hdmf namespace when loaded from yaml

nwb_linkml/src/nwb_linkml/io/schema.py

@@ -5,6 +5,7 @@ Loading/saving NWB Schema yaml files
 from pathlib import Path
 from pprint import pprint
 from typing import Optional
+import warnings
 
 from linkml_runtime.loaders import yaml_loader
 
@@ -82,6 +83,8 @@ def load_namespace_adapter(
         version (str): Optional: tag or commit to check out namespace is a
             :class:`.NamespaceRepo`. If ``None``, use ``HEAD`` if not already checked out,
             or otherwise use whatever version is already checked out.
+        imported (list[:class:`.NamespacesAdapter`]): Optional: override discovered imports
+            with already-loaded namespaces adapters
 
     Returns:
         :class:`.NamespacesAdapter`
@@ -111,8 +114,15 @@ def load_namespace_adapter(
     for ns in namespaces.namespaces:
         for schema in ns.schema_:
             if schema.source is None:
-                # this is normal, we'll resolve later
+                if imported is None and schema.namespace == "hdmf-common":
+                    # special case - hdmf-common is imported by name without location or version,
+                    # so to get the correct version we have to handle it separately
+                    imported = _resolve_hdmf(namespace, path)
+                    if imported is not None:
+                        imported = [imported]
+                else:
                     continue
+            else:
                 yml_file = (path / schema.source).resolve()
                 sch.append(load_schema_file(yml_file))
 
@@ -124,6 +134,31 @@ def load_namespace_adapter(
     return adapter
 
 
+def _resolve_hdmf(
+    namespace: Path | NamespaceRepo | Namespaces, path: Optional[Path] = None
+) -> Optional[NamespacesAdapter]:
+    if path is None and isinstance(namespace, Namespaces):
+        # cant get any more information from already-loaded namespaces without a path
+        return None
+
+    if isinstance(namespace, NamespaceRepo):
+        # easiest route is if we got a NamespaceRepo
+        if namespace.name == "core":
+            hdmf_path = (path / namespace.imports["hdmf-common"]).resolve()
+            return load_namespace_adapter(namespace=hdmf_path)
+        # otherwise the hdmf-common adapter itself, and it loads common
+        else:
+            return None
+    elif path is not None:
+        # otherwise try and get it from relative paths
+        # pretty much a hack, but hey we are compensating for absence of versioning system here
+        maybe_repo_root = path / NWB_CORE_REPO.imports["hdmf-common"]
+        if maybe_repo_root.exists():
+            return load_namespace_adapter(namespace=maybe_repo_root)
+    warnings.warn(f"Could not locate hdmf-common from namespace {namespace} and path {path}")
+    return None
+
+
 def load_nwb_core(
     core_version: str = "2.7.0", hdmf_version: str = "1.8.0", hdmf_only: bool = False
 ) -> NamespacesAdapter:

nwb_linkml/src/nwb_linkml/providers/git.py

@@ -36,6 +36,14 @@ class NamespaceRepo(BaseModel):
         ),
         default_factory=list,
     )
+    imports: Optional[dict[str, Path]] = Field(
+        None,
+        description=(
+            "Any named imports that are included eg. as submodules within their repository. Dict"
+            " mapping schema name (used in the namespace field) to the namespace file relative to"
+            " the directory containing the **namespace.yaml file** (not the repo root)"
+        ),
+    )
 
     def provide_from_git(self, commit: str | None = None) -> Path:
         """Provide a namespace file from a git repo"""
@@ -61,6 +69,7 @@ NWB_CORE_REPO = NamespaceRepo(
         "2.6.0",
         "2.7.0",
     ],
+    imports={"hdmf-common": Path("../hdmf-common-schema") / "common" / "namespace.yaml"},
 )
 
 HDMF_COMMON_REPO = NamespaceRepo(
@@ -86,7 +95,7 @@ HDMF_COMMON_REPO = NamespaceRepo(
 
 DEFAULT_REPOS = {
     repo.name: repo for repo in [NWB_CORE_REPO, HDMF_COMMON_REPO]
-}  # type: Dict[str, NamespaceRepo]
+}  # type: dict[str, NamespaceRepo]
 
 
 class GitError(OSError):
@@ -112,7 +121,7 @@ class GitRepo:
         self.namespace = namespace
         self._commit = commit
 
-    def _git_call(self, *args: List[str]) -> subprocess.CompletedProcess:
+    def _git_call(self, *args: str) -> subprocess.CompletedProcess:
         res = subprocess.run(["git", "-C", self.temp_directory, *args], capture_output=True)
         if res.returncode != 0:
             raise GitError(
@@ -138,8 +147,11 @@ class GitRepo:
         """
         URL for "origin" remote
         """
+        try:
             res = self._git_call("remote", "get-url", "origin")
             return res.stdout.decode("utf-8").strip()
+        except GitError:
+            return ""
 
     @property
     def active_commit(self) -> str:
@@ -157,6 +169,16 @@ class GitRepo:
         """
         return self.temp_directory / self.namespace.path
 
+    @property
+    def import_namespaces(self) -> dict[str, Path]:
+        """
+        Absolute location of each of the imported namespaces specified in
+        :attr:`.NamespaceRepo.imports`
+        """
+        if self.namespace.imports is None:
+            return {}
+        return {k: (self.namespace_file / v).resolve() for k, v in self.namespace.imports.items()}
+
     @property
     def commit(self) -> Optional[str]:
         """