Use local schemas if available

cache hashable calls to yaml loader if possible

Author: sneakers-the-rat

linkml_runtime/__init__.py

@@ -25,6 +25,25 @@
 
 MAIN_SCHEMA_PATH = SCHEMA_DIRECTORY / "meta.yaml"
 
+LINKML_ANNOTATIONS = SCHEMA_DIRECTORY / "annotations.yaml"
+LINKML_ARRAY = SCHEMA_DIRECTORY / "array.yaml"
+LINKML_EXTENSIONS = SCHEMA_DIRECTORY / "extensions.yaml"
+LINKML_MAPPINGS = SCHEMA_DIRECTORY / "mappings.yaml"
+LINKML_TYPES = SCHEMA_DIRECTORY / "types.yaml"
+LINKML_UNITS = SCHEMA_DIRECTORY / "units.yaml"
+LINKML_VALIDATION = SCHEMA_DIRECTORY / "validation.yaml"
+
+
+URI_TO_LOCAL = {
+    'https://w3id.org/linkml/annotations.yaml': str(LINKML_ANNOTATIONS),
+    'https://w3id.org/linkml/array.yaml': str(LINKML_ARRAY),
+    'https://w3id.org/linkml/extensions.yaml': str(LINKML_EXTENSIONS),
+    'https://w3id.org/linkml/mappings.yaml': str(LINKML_MAPPINGS),
+    'https://w3id.org/linkml/meta.yaml': str(MAIN_SCHEMA_PATH),
+    'https://w3id.org/linkml/types.yaml': str(LINKML_TYPES),
+    'https://w3id.org/linkml/units.yaml': str(LINKML_UNITS),
+    'https://w3id.org/linkml/validation.yaml': str(LINKML_VALIDATION),
+}
 
 class MappingError(ValueError):
     """

linkml_runtime/loaders/loader_root.py

@@ -1,11 +1,17 @@
 from abc import ABC, abstractmethod
 from typing import TextIO, Union, Optional, Callable, Dict, Type, Any, List
+from logging import getLogger
+from functools import lru_cache
 
 from pydantic import BaseModel
 from hbreader import FileInfo, hbread
 from jsonasobj2 import as_dict, JsonObj
 
 from linkml_runtime.utils.yamlutils import YAMLRoot
+from linkml_runtime import URI_TO_LOCAL
+
+CACHE_SIZE = 1024
+
 
 
 class Loader(ABC):
@@ -137,20 +143,42 @@ def _construct_target_class(self,
         else:
             return None
 
+
     def _read_source(self,
                      source: Union[str, dict, TextIO], 
                      *, 
                      base_dir: Optional[str] = None, 
                      metadata: Optional[FileInfo] = None, 
                      accept_header: Optional[str] = "text/plain, application/yaml;q=0.9") -> Union[dict, str]:
-        if metadata is None:
-            metadata = FileInfo()
-        if base_dir and not metadata.base_path:
-            metadata.base_path = base_dir
+
+        # avoid instantiating unhashable FileInfo type by getting default base_path, if any
+        if base_dir is None:
+            if metadata is not None:
+                base_dir = str(metadata.base_path) if metadata.base_path is not None else None
+            else:
+                base_dir = str(FileInfo().base_path) if FileInfo().base_path is not None else None
+        elif base_dir is not None:
+            base_dir = str(base_dir)
 
         if not isinstance(source, dict):
-            data = hbread(source, metadata, metadata.base_path, accept_header)
+            # Try to get local version of schema, if one is known to exist
+            try:
+                if str(source) in URI_TO_LOCAL.keys():
+                    source = str(URI_TO_LOCAL[str(source)])
+            except (TypeError, KeyError) as e:
+                # Fine, use original `source` value
+                logger = getLogger('linkml_runtime.loaders.Loader')
+                logger.debug(f"Error converting stringlike source to local linkml file: {source}, got: {e}")
+
+            if not isinstance(metadata, FileInfo):
+                data = self._hashable_read(source, base_dir, accept_header)
+            else:
+                data = hbread(source, metadata, base_dir, accept_header)
         else:
             data = source
 
         return data
+
+    @lru_cache(maxsize=CACHE_SIZE)
+    def _hashable_read(self, source: Union[dict, TextIO], base_dir: Optional[str] = None, accept_header: Optional[str] = None):
+        return hbread(source, None, base_dir, accept_header)

linkml_runtime/loaders/yaml_loader.py

@@ -20,10 +20,6 @@ def load_as_dict(self,
                      *, 
                      base_dir: Optional[str] = None,
                      metadata: Optional[FileInfo] = None) -> Union[dict, List[dict]]:
-        if metadata is None:
-            metadata = FileInfo()
-        if base_dir and not metadata.base_path:
-            metadata.base_path = base_dir
         data = self._read_source(source, base_dir=base_dir, metadata=metadata, accept_header="text/yaml, application/yaml;q=0.9")
         if isinstance(data, str):
             data = StringIO(data)

tests/test_issues/input/issue_1040.yaml

@@ -17,4 +17,4 @@ classes:
   Person:
     in_subset:
        - a
-       -
+       -