Attempt to fix metaschema caching (#576)

* Fix load_schema, which was incorrectly modifying the object from get_metaschema 

We're using custom cache to pre-load documents in schema_loader, but there are different documents with the same id, so for that to work, we can't have a shared index.

* Add test for load_schema with cache

Author: tetron

schema_salad/schema.py

@@ -249,7 +249,14 @@ def load_schema(
 
     metaschema_names, _metaschema_doc, metaschema_loader = get_metaschema()
     if cache is not None:
-        metaschema_loader.cache.update(cache)
+        # we want to replace some items in the cache, so we need to
+        # make a new Loader with an empty index.
+        for k, v in metaschema_loader.cache.items():
+            if k not in cache:
+                cache[k] = v
+        metaschema_loader = Loader(
+            ctx=metaschema_loader.ctx, cache=cache, session=metaschema_loader.session
+        )
     schema_doc, schema_metadata = metaschema_loader.resolve_ref(schema_ref, "")
 
     if not isinstance(schema_doc, MutableSequence):

schema_salad/tests/test_examples.py

@@ -527,7 +527,7 @@ class TestExp(Exception):
     except TestExp as e:
         assert str(e).endswith("frag.yml:3:3: Whoops"), e
     except Exception as exc:
-        assert False, exc
+        raise AssertionError("Unexpected exception" + str(exc))
 
 
 def test_cmap() -> None:

schema_salad/tests/test_misc.py

@@ -1,5 +1,7 @@
 from schema_salad.avro.schema import Names
 from schema_salad.schema import load_schema
+from typing import Optional, Dict, Union
+from rdflib.graph import Graph
 
 from .util import get_data
 
@@ -9,3 +11,32 @@ def test_misc() -> None:
     assert path
     document_loader, avsc_names, schema_metadata, metaschema_loader = load_schema(path)
     assert isinstance(avsc_names, Names)
+
+
+def test_load_schema_cache() -> None:
+
+    schemaid = "http://commonwl.org/schema_salad/test/schema.yml"
+
+    path1 = get_data("tests/test_schema/misc_schema_v1.yml")
+    assert path1
+
+    with open(path1, "rt") as f:
+        cache1: Optional[Dict[str, Union[str, Graph, bool]]] = {schemaid: f.read()}
+
+    document_loader, avsc_names, schema_metadata, metaschema_loader = load_schema(
+        schemaid, cache=cache1
+    )
+    assert isinstance(avsc_names, Names)
+    assert "EmptyType" not in document_loader.vocab
+
+    path2 = get_data("tests/test_schema/misc_schema_v2.yml")
+    assert path2
+
+    with open(path2, "rt") as f:
+        cache2: Optional[Dict[str, Union[str, Graph, bool]]] = {schemaid: f.read()}
+
+    document_loader, avsc_names, schema_metadata, metaschema_loader = load_schema(
+        schemaid, cache=cache2
+    )
+    assert isinstance(avsc_names, Names)
+    assert "EmptyType" in document_loader.vocab

schema_salad/tests/test_schema/misc_schema_v1.yml

@@ -0,0 +1,25 @@
+$base: "https://w3id.org/cwl/cwltest#"
+$graph:
+  - name: TestCase
+    type: record
+    documentRoot: true
+    fields:
+      id:
+        type: [int, string]
+        jsonldPredicate:
+          _type: "@id"
+          identity: true
+      label: string?
+      doc: string?
+      tags: string[]?
+      tool:
+        type: string
+        jsonldPredicate:
+          _type: "@id"
+      job:
+        type: string?
+        jsonldPredicate:
+          _type: "@id"
+      should_fail: boolean?
+      output:
+        type: Any?

schema_salad/tests/test_schema/misc_schema_v2.yml

@@ -0,0 +1,29 @@
+$base: "https://w3id.org/cwl/cwltest#"
+$graph:
+  - name: TestCase
+    type: record
+    documentRoot: true
+    fields:
+      id:
+        type: [int, string]
+        jsonldPredicate:
+          _type: "@id"
+          identity: true
+      label: string?
+      doc: string?
+      tags: string[]?
+      tool:
+        type: string
+        jsonldPredicate:
+          _type: "@id"
+      job:
+        type: string?
+        jsonldPredicate:
+          _type: "@id"
+      should_fail: boolean?
+      output:
+        type: Any?
+
+  - name: EmptyType
+    documentRoot: true
+    type: record