Load released test schemas from the bundle

spec_tests/test_hed_cache.py

@@ -15,7 +15,6 @@ def setUpClass(cls):
         cls.saved_cache_folder = hed_cache.HED_CACHE_DIRECTORY
         schema.set_cache_directory(cls.hed_cache_dir)
 
-        cls.default_xml_base_filename = "HED8.0.0t.xml"
         cls.hed_test_version = "7.1.1"
         cls.hed_invalid_version = "4.6.7"
 

tests/data/schema_tests/README.md

@@ -0,0 +1,23 @@
+# Schema test data
+
+Tests load released standard schemas from the copies bundled with hedtools in `hed/schema/schema_data/`, with
+`load_schema_version("8.2.0")`, never from a copy here. A test that needs the file itself uses
+`bundled_standard_path` in `tests/schema/schema_test_helpers.py`. A test that names a released standard inside
+an `xml_folder` (a version string is looked up only in that folder) builds a temporary folder with
+`make_xml_folder` in the same module. `load_schema_version` returns one shared cached object per version, so a
+test that changes its schema works on a `copy.deepcopy`.
+
+## Full standard schemas kept as files
+
+Each is released 8.0.0 with a deliberate edit, so the bundle cannot replace it:
+
+- `HED8.0.0_2_value_classes.xml` - a tag with two value classes (edge-case format roundtrip test).
+- `unknown_attribute.xml` - attributes the schema does not define (edge-case format roundtrip test).
+- `HED8.0.0_2_value_classes.mediawiki`, `unknown_attribute.mediawiki` - the MediaWiki sources of the two above;
+  no test reads them.
+- `../validator_tests/HED8.0.0_added_tests.mediawiki` - test additions for the validator tests (for example
+  `Item/Currency-test`).
+
+Everything else here is a small edited fixture or fragment (`added_prop*`, `duplicate_unit*`,
+`prologue_tests/`, `wiki_tests/`, `prerelease/` stubs) or a test library (`merge_tests/`,
+`merge_group_tests/`, `HED_test*`).

tests/data/schema_tests/merge_group_tests/README.md

@@ -3,14 +3,11 @@
 Test-only library schemas for the schema merge-group tests
 (`tests/schema/test_schema_merge.py`), copied from hed-tests
 `json_test_data/test_schemas/<library>/hedxml_unmerged/` at hed-tests commit
-ccfb308, plus the two vendored standard schemas (`HED8.4.0.xml`,
-`HED8.5.0.xml`) from that repository's `json_test_data/test_schemas/hedxml/`
-folder so that loading is hermetic. `HED8.5.0.xml` is a prerelease snapshot: refreshed 2026-09-13 from
-hed-schemas `standard_schema/prerelease/HED8.5.0.xml` at hed-schemas commit
-c049843 (anyUnits and Quantity; via hed-tests `convert_test_schemas.py --refresh`,
-hed-tests PR #56, merged as 0f2afd4); re-copy it when
-8.5.0 is released and keep the vendored copy after that. The libraries are in UNMERGED form and use the cache-convention
-file names, so this folder can be passed as `xml_folder` to
-`load_schema_version`. See the hed-tests README in that folder for what each
+ccfb308. The libraries are in UNMERGED form and use the cache-convention file
+names. Their partner standards (8.4.0 and 8.5.0) are not copied here: the tests
+build a temporary `xml_folder` from this folder plus the copies bundled with
+hedtools in `hed/schema/schema_data/` (`make_xml_folder` in
+`tests/schema/schema_test_helpers.py`), so loading stays hermetic. When
+re-copying from hed-tests, do not copy its `HED8.*.xml` standards back in. See the hed-tests README in that folder for what each
 library version probes; the `.mediawiki` sources there are the source of
 truth and are regenerated with `src/scripts/convert_test_schemas.py`.

tests/models/test_base_input.py

@@ -38,16 +38,13 @@ def setUpClass(cls):
         bids_root_path = os.path.realpath(
             os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/bids_tests/eeg_ds003645s_hed")
         )
-        schema_path = os.path.realpath(
-            os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/schema_tests/HED8.2.0.xml")
-        )
         cls.bids_root_path = bids_root_path
         json_path = os.path.realpath(os.path.join(bids_root_path, "task-FacePerception_events.json"))
         events_path = os.path.realpath(
             os.path.join(bids_root_path, "sub-002/eeg/sub-002_task-FacePerception_run-1_events.tsv")
         )
 
-        cls.hed_schema = schema.load_schema(schema_path)
+        cls.hed_schema = schema.load_schema_version("8.2.0")
         sidecar1 = Sidecar(json_path, name="face_sub1_json")
         mapper1 = ColumnMapper(sidecar=sidecar1, optional_tag_columns=["HED"], warn_on_missing_column=False)
         cls.input_data1 = BaseInput(

tests/models/test_column_mapper.py

@@ -1,7 +1,7 @@
 import os
 import unittest
 
-from hed import load_schema
+from hed import load_schema_version
 from hed.errors import ValidationErrors
 from hed.models import ColumnType, HedString
 from hed.models.column_mapper import ColumnMapper
@@ -11,10 +11,7 @@
 class Test(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
-        base_data_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/")
-        schema_file = "schema_tests/HED8.0.0t.xml"
-
-        cls.hed_schema = load_schema(os.path.join(base_data_dir, schema_file))
+        cls.hed_schema = load_schema_version("8.0.0")
         cls.integer_key_dictionary = {0: "one", 1: "two", 2: "three"}
         cls.zero_based_row_column_count = 3
         cls.column_prefix_dictionary = {2: "Event/Description/", 3: "Event/Label/", 4: "Event/Category/"}

tests/models/test_hed_group.py

@@ -10,8 +10,7 @@ class Test(unittest.TestCase):
     @classmethod
     def setUpClass(cls):
         cls.base_data_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/")
-        hed_xml_file = os.path.join(cls.base_data_dir, "schema_tests/HED8.0.0t.xml")
-        cls.hed_schema = schema.load_schema(hed_xml_file)
+        cls.hed_schema = schema.load_schema_version("8.0.0")
 
     def test_remove_groups(self):
         from hed.models.definition_dict import DefTagNames

tests/models/test_query_handler.py

@@ -23,8 +23,7 @@ class TestParser(unittest.TestCase):
     def setUpClass(cls):
         base_data_dir = os.path.realpath(os.path.join(os.path.dirname(__file__), "../data/"))
         cls.base_data_dir = base_data_dir
-        hed_xml_file = os.path.join(base_data_dir, "schema_tests/HED8.0.0t.xml")
-        cls.hed_schema = schema.load_schema(hed_xml_file)
+        cls.hed_schema = schema.load_schema_version("8.0.0")
 
     def base_test(self, parse_expr, search_strings):
         expression = QueryHandler(parse_expr)

tests/models/test_query_service.py

@@ -1,6 +1,5 @@
 """Tests for query_service module: get_query_handlers and search_hed_objs."""
 
-import os
 import unittest
 
 from hed import schema
@@ -79,9 +78,7 @@ class TestSearchHedObjs(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        base_data_dir = os.path.realpath(os.path.join(os.path.dirname(__file__), "../data/"))
-        hed_xml_file = os.path.join(base_data_dir, "schema_tests/HED8.0.0t.xml")
-        cls.hed_schema = schema.load_schema(hed_xml_file)
+        cls.hed_schema = schema.load_schema_version("8.0.0")
 
     def test_basic_search(self):
         hed_objs = [

tests/models/test_query_util.py

@@ -10,7 +10,6 @@
 - children = [Red] (only the matched tag)
 """
 
-import os
 import unittest
 from unittest.mock import Mock
 
@@ -300,9 +299,7 @@ class TestSearchResultGroupIdentitySemantics(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        base_data_dir = os.path.realpath(os.path.join(os.path.dirname(__file__), "../data/"))
-        hed_xml_file = os.path.join(base_data_dir, "schema_tests/HED8.0.0t.xml")
-        cls.hed_schema = schema.load_schema(hed_xml_file)
+        cls.hed_schema = schema.load_schema_version("8.0.0")
 
     def setUp(self):
         # Two HedStrings with identical content parsed independently.

tests/models/test_schema_lookup.py

@@ -19,12 +19,10 @@ class TestSchemaLookupBase(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        base_data_dir = os.path.realpath(os.path.join(os.path.dirname(__file__), "../data/"))
-        hed_xml_file = os.path.join(base_data_dir, "schema_tests/HED8.0.0t.xml")
         try:
             from hed import schema as hed_schema
 
-            cls.hed_schema = hed_schema.load_schema(hed_xml_file)
+            cls.hed_schema = hed_schema.load_schema_version("8.0.0")
             cls.lookup = generate_schema_lookup(cls.hed_schema)
         except Exception as exc:
             cls.hed_schema = None
@@ -194,13 +192,11 @@ class TestGenerateSchemaLookupGroup(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        base_data_dir = os.path.realpath(os.path.join(os.path.dirname(__file__), "../data/"))
-        hed_xml_file = os.path.join(base_data_dir, "schema_tests/HED8.0.0t.xml")
         try:
             from hed import schema as hed_schema
             from hed.schema.hed_schema_group import HedSchemaGroup
 
-            single = hed_schema.load_schema(hed_xml_file)
+            single = hed_schema.load_schema_version("8.0.0")
             cls.group = HedSchemaGroup([single])
             cls.single_lookup = generate_schema_lookup(single)
             cls.group_lookup = generate_schema_lookup(cls.group)

tests/models/test_sidecar.py

@@ -14,8 +14,7 @@ class Test(unittest.TestCase):
     def setUpClass(cls):
         base_data_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/")
         cls.base_data_dir = base_data_dir
-        hed_xml_file = os.path.join(base_data_dir, "schema_tests/HED8.0.0t.xml")
-        cls.hed_schema = schema.load_schema(hed_xml_file)
+        cls.hed_schema = schema.load_schema_version("8.0.0")
         json_filename = os.path.join(base_data_dir, "sidecar_tests/both_types_events.json")
         cls.json_filename = json_filename
         json_def_filename = os.path.join(base_data_dir, "sidecar_tests/both_types_events_with_defs.json")

tests/models/test_spreadsheet_input.py

@@ -17,8 +17,7 @@ class Test(unittest.TestCase):
     def setUpClass(cls):
         base = os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/")
         cls.base_data_dir = base
-        hed_xml_file = os.path.join(base, "schema_tests/HED8.0.0t.xml")
-        cls.hed_schema = schema.load_schema(hed_xml_file)
+        cls.hed_schema = schema.load_schema_version("8.0.0")
         default = os.path.join(
             os.path.dirname(os.path.realpath(__file__)), "../data/spreadsheet_validator_tests/ExcelMultipleSheets.xlsx"
         )

tests/models/test_string_search.py

@@ -12,7 +12,6 @@
 schema lookup table from a real schema file.
 """
 
-import os
 import unittest
 
 import pandas as pd
@@ -982,12 +981,10 @@ class TestSchemaLookupAncestorSearch(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        base_data_dir = os.path.realpath(os.path.join(os.path.dirname(__file__), "../data/"))
-        hed_xml_file = os.path.join(base_data_dir, "schema_tests/HED8.0.0t.xml")
         try:
             from hed import schema as hed_schema
 
-            cls.hed_schema = hed_schema.load_schema(hed_xml_file)
+            cls.hed_schema = hed_schema.load_schema_version("8.0.0")
             cls.lookup = generate_schema_lookup(cls.hed_schema)
         except Exception:
             cls.hed_schema = None

tests/models/test_tabular_input.py

@@ -17,9 +17,6 @@ def setUpClass(cls):
         bids_root_path = os.path.realpath(
             os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/bids_tests/eeg_ds003645s_hed")
         )
-        schema_path = os.path.realpath(
-            os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/schema_tests/HED8.2.0.xml")
-        )
         sidecar1_path = os.path.realpath(os.path.join(bids_root_path, "task-FacePerception_events.json"))
         cls.events_path = os.path.realpath(
             os.path.join(bids_root_path, "sub-002/eeg/sub-002_task-FacePerception_run-1_events.tsv")
@@ -29,7 +26,7 @@ def setUpClass(cls):
                 os.path.dirname(os.path.realpath(__file__)), "../data/other_tests/task-FacePerceptionSmall_events.json"
             )
         )
-        cls.hed_schema = schema.load_schema(schema_path)
+        cls.hed_schema = schema.load_schema_version("8.2.0")
         cls.sidecar1 = Sidecar(sidecar1_path, name="face_sub1_json")
         cls.sidecar2 = Sidecar(sidecar2_path, name="face_small_json")
         cls.base_output_folder = base_output_folder

tests/schema/schema_test_helpers.py

@@ -1,13 +1,16 @@
 """
 Shared utility functions for schema tests.
 
-Provides helper decorators and functions for creating temporary files in tests.
+Provides helper decorators and functions for creating temporary files and schema folders in tests.
 """
 
 import functools
 import os
+import shutil
 import tempfile
 
+from hed import schema
+
 
 def get_temp_filename(extension):
     """
@@ -57,3 +60,44 @@ def wrapper(*args, **kwargs):
         return wrapper
 
     return decorator
+
+
+def bundled_standard_path(version):
+    """
+    Return the path of a released standard schema bundled with hedtools.
+
+    Parameters:
+        version (str): A released standard version (e.g., "8.4.0").
+
+    Returns:
+        str: Path to the bundled XML file in hed/schema/schema_data.
+    """
+    return os.path.join(os.path.dirname(schema.__file__), "schema_data", f"HED{version}.xml")
+
+
+def make_xml_folder(versions, copies=()):
+    """
+    Build a temporary folder to pass as xml_folder: fixture files plus bundled standard schemas.
+
+    A version string loaded with xml_folder is looked up only in that folder, so a test that names a
+    released standard there needs the file on disk; it comes from the bundle instead of a copy in tests/data.
+
+    Parameters:
+        versions (list[str]): Released standard versions copied from the bundle into the folder root.
+        copies (list[tuple[str, str]]): (source, target) pairs; a source file or folder is copied into the
+            target subfolder ("" for the root).
+
+    Returns:
+        str: Path to the new folder; the caller removes it with shutil.rmtree.
+    """
+    folder = tempfile.mkdtemp()
+    for source, target in copies:
+        destination = os.path.join(folder, target)
+        if os.path.isdir(source):
+            shutil.copytree(source, destination, dirs_exist_ok=True)
+        else:
+            os.makedirs(destination, exist_ok=True)
+            shutil.copy(source, destination)
+    for version in versions:
+        shutil.copy(bundled_standard_path(version), folder)
+    return folder

tests/schema/test_base_tag.py

@@ -1,18 +1,16 @@
-import os
 import unittest
 
 from hed.errors import ErrorHandler
 from hed.models import HedString
-from hed.schema import load_schema
+from hed.schema import load_schema_version
 
 
 class TestBaseTagBase(unittest.TestCase):
-    schema_file = "../data/schema_tests/HED8.0.0t.xml"
+    schema_version = "8.0.0"
 
     @classmethod
     def setUpClass(cls):
-        hed_xml = os.path.join(os.path.dirname(os.path.realpath(__file__)), cls.schema_file)
-        cls.hed_schema = load_schema(hed_xml)
+        cls.hed_schema = load_schema_version(cls.schema_version)
         cls.error_handler = ErrorHandler()
 
     def tag_form_base(self, test_strings, expected_results, expected_errors):

tests/schema/test_convert_tags.py

@@ -6,7 +6,7 @@
 
 
 class TestTagFormat(TestHedBase):
-    schema_file = "../data/schema_tests/HED8.0.0t.xml"
+    schema_version = "8.0.0"
 
 
 class TestConvertTag(TestTagFormat):

tests/schema/test_hed_schema.py

@@ -1,21 +1,17 @@
 import os
 import unittest
 
+import hed
 from hed.errors import HedFileError
 from hed.models import HedTag
 from hed.schema import HedKey, HedSchemaGroup, get_hed_xml_version, load_schema, load_schema_version
 
 
 class TestHedSchema(unittest.TestCase):
-    schema_file_3g_xml = "../data/schema_tests/HED8.0.0t.xml"
-    schema_file_3g = "../data/schema_tests/HED8.2.0.mediawiki"
-
     @classmethod
     def setUpClass(cls):
-        cls.hed_xml_3g = os.path.join(os.path.dirname(os.path.realpath(__file__)), cls.schema_file_3g_xml)
-        cls.hed_wiki_3g = os.path.join(os.path.dirname(os.path.realpath(__file__)), cls.schema_file_3g)
-        cls.hed_schema_3g_wiki = load_schema(cls.hed_wiki_3g)
-        cls.hed_schema_3g = load_schema(cls.hed_xml_3g)
+        cls.hed_xml_3g = os.path.join(os.path.dirname(hed.schema.__file__), "schema_data", "HED8.0.0.xml")
+        cls.hed_schema_3g = load_schema_version("8.0.0")
 
         schema_file = "../data/validator_tests/HED8.0.0_added_tests.mediawiki"
         hed_xml = os.path.join(os.path.dirname(os.path.realpath(__file__)), schema_file)

tests/schema/test_hed_schema_io.py

@@ -14,7 +14,7 @@
 from hed.errors.error_types import SchemaErrors
 from hed.schema import HedKey, HedSchema, HedSchemaGroup, hed_cache, load_schema, load_schema_version
 from hed.schema.hed_schema_io import _load_schema_version, from_string, parse_version_list
-from tests.schema.schema_test_helpers import get_temp_filename, with_temp_file
+from tests.schema.schema_test_helpers import get_temp_filename, make_xml_folder, with_temp_file
 from tests.schema.util_test_schemas import TEST_SCHEMAS_DIR, load_test_schema, test_schemas_available
 
 
@@ -50,9 +50,7 @@ def test_load_schema_invalid_parameters(self):
             load_schema(bad_filename)
 
     def test_load_schema_name(self):
-        schema_path = os.path.join(
-            os.path.dirname(os.path.realpath(__file__)), "../data/schema_tests/HED8.2.0.mediawiki"
-        )
+        schema_path = os.path.join(os.path.dirname(schema.__file__), "schema_data", "HED8.2.0.xml")
 
         hed_schema = load_schema(schema_path, schema_namespace="testspace", name="Test Name")
         self.assertEqual(hed_schema.schema_namespace, "testspace:")
@@ -1098,8 +1096,19 @@ class TestPrereleaseSchemaLoading(unittest.TestCase):
 
     @classmethod
     def setUpClass(cls):
-        """Set up test fixtures."""
-        cls.schema_dir = os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/schema_tests/")
+        """Set up test fixtures: the bundled 8.2.0 at the root beside the fixture schemas."""
+        fixtures = os.path.join(os.path.dirname(os.path.realpath(__file__)), "../data/schema_tests/")
+        cls.schema_dir = make_xml_folder(
+            ["8.2.0"],
+            [
+                (os.path.join(fixtures, "HED_testlocal_2.1.0.xml"), ""),
+                (os.path.join(fixtures, "prerelease"), "prerelease"),
+            ],
+        )
+
+    @classmethod
+    def tearDownClass(cls):
+        shutil.rmtree(cls.schema_dir, ignore_errors=True)
 
     def test_load_regular_schema(self):
         """Test that a regular schema loads correctly."""