improve loading of schema files

3 files changed, 43 insertions, 32 deletions

diff --git a/src/hydrilla/builder/build.py b/src/hydrilla/builder/build.py
index 33838f3..a3c55a7 100644
--- a/src/hydrilla/builder/build.py
+++ b/src/hydrilla/builder/build.py
@@ -51,15 +51,6 @@ here = Path(__file__).resolve().parent
 
 _ = util.translation(here / 'locales').gettext
 
-def index_validator(major_schema_version):
-    """
-    Create an index.json schema validator specific to the requested schema
-    version series.
-    """
-    exact_version = {1: '1.0.1', 2: '2'}[major_schema_version]
-
-    return util.validator_for(f'package_source-{exact_version}.schema.json')
-
 schemas_root = 'https://hydrilla.koszko.org/schemas'
 
 generated_by = {
@@ -365,7 +356,9 @@ class Build:
         files and computed definitions of the source package and items defined
         in it.
         """
-        index_validator(major_schema_version).validate(index_obj)
+        schema_name = f'package_source-{major_schema_version}.schema.json';
+
+        util.validator_for(schema_name).validate(index_obj)
 
         match = re.match(r'.*-((([1-9][0-9]*|0)\.)+)schema\.json$',
                          index_obj['$schema'])
diff --git a/src/hydrilla/util/_util.py b/src/hydrilla/util/_util.py
index 815b7fd..de7435d 100644
--- a/src/hydrilla/util/_util.py
+++ b/src/hydrilla/util/_util.py
@@ -117,14 +117,43 @@ def version_string(ver: list[int], rev: Optional[int]=None) -> str:
     """
     return '.'.join([str(n) for n in ver]) + ('' if rev is None else f'-{rev}')
 
+_schema_name_re = re.compile(r'''
+(?P<name_base>[^/]*)
+-
+(?P<ver>
+  (?P<major>[1-9][0-9]*)
+  (?: # this repeated group matches the remaining version numbers
+    \.
+    (?:[1-9][0-9]*|0)
+  )*
+)
+\.schema\.json
+$
+''', re.VERBOSE)
+
+schema_paths = {}
+for path in (here.parent / 'schemas').rglob('*.schema.json'):
+    match = _schema_name_re.search(path.name)
+    schema_name_base = match.group('name_base')
+    schema_ver_list = match.group('ver').split('.')
+
+    for i in range(len(schema_ver_list)):
+        schema_ver = '.'.join(schema_ver_list[:i+1])
+        schema_paths[f'{schema_name_base}-{schema_ver}.schema.json'] = path
+
+for name, path in [*schema_paths.items()]:
+    schema_paths[f'https://hydrilla.koszko.org/schemas/{name}'] = path
+
 schemas = {}
-for series_dir in (here.parent / 'schemas').glob('*.x'):
-    for path in series_dir.glob("*.schema.json"):
-        schema = json.loads(path.read_text())
-        schemas[schema['$id']] = schema
 
-common_schema_filename = 'common_definitions-1.schema.json'
-common_schema_path = here.parent / "schemas" / common_schema_filename
+def _get_schema(schema_name: str) -> dict:
+    """Return loaded JSON of the requested schema. Cache results."""
+    path = schema_paths[schema_name]
+
+    if path not in schemas:
+        schemas[path] = json.loads(path.read_text())
+
+    return schemas[path]
 
 def validator_for(schema: Union[str, dict]) -> Draft7Validator:
     """
@@ -133,27 +162,16 @@ def validator_for(schema: Union[str, dict]) -> Draft7Validator:
     Other schemas under '../schemas' can be referenced.
     """
     if isinstance(schema, str):
-        schema = schemas[f'https://hydrilla.koszko.org/schemas/{schema}']
+        schema = _get_schema(schema)
 
     resolver = RefResolver(
         base_uri=schema['$id'],
         referrer=schema,
-        handlers={'https': lambda uri: schemas[uri]}
+        handlers={'https': _get_schema}
     )
 
     return Draft7Validator(schema, resolver=resolver)
 
-_major_version_re = re.compile(r'''
--
-(?P<major>[1-9][0-9]*)
-(?: # this repeated group matches the remaining version numbers
-  \.
-  (?:[1-9][0-9]*|0)
-)*
-\.schema\.json
-$
-''', re.VERBOSE)
-
 def load_instance_from_file(path: Path) -> tuple[dict, Optional[int]]:
     """
     Open a file and load its contents as a JSON document (with additional
@@ -167,7 +185,7 @@ def load_instance_from_file(path: Path) -> tuple[dict, Optional[int]]:
     major = None
 
     if type(instance) is dict and type(instance.get('$schema')) is str:
-        match = _major_version_re.search(instance.get('$schema'))
+        match = _schema_name_re.search(instance.get('$schema'))
         major = match and int(match.group('major'))
 
     return instance, major
diff --git a/tests/test_build.py b/tests/test_build.py
index c2d8ced..7fedbd5 100644
--- a/tests/test_build.py
+++ b/tests/test_build.py
@@ -20,7 +20,7 @@ from contextlib import contextmanager
 from jsonschema import ValidationError
 
 from hydrilla import util as hydrilla_util
-from hydrilla.util._util import _major_version_re
+from hydrilla.util._util import _schema_name_re
 from hydrilla.builder import build, _version, local_apt
 from hydrilla.builder.common_errors import *
 
@@ -595,7 +595,7 @@ def try_validate(as_what, instance):
     Select the right JSON schema. Return without errors only if the instance
     validates against it.
     """
-    major = _major_version_re.search(instance['$schema']).group('major')
+    major = _schema_name_re.search(instance['$schema']).group('major')
     exact_schema_version = {'1': '1.0.1', '2': '2'}[major]
     schema_filename = f'{as_what}-{exact_schema_version}.schema.json'
     hydrilla_util.validator_for(schema_filename).validate(instance)