feat: Add support for items as array to get_schema_fields

Author: jpmckinney

docs/changelog.rst

@@ -1,6 +1,14 @@
 Changelog
 =========
 
+1.3.2 (2024-12-17)
+------------------
+
+Added
+~~~~~
+
+-  :func:`ocdskit.schema.get_schema_fields`: Add support for ``items`` as array.
+
 1.3.1 (2024-12-16)
 ------------------
 

docs/conf.py

@@ -23,7 +23,7 @@
 author = "Open Contracting Partnership"
 
 # The short X.Y version
-version = "1.3.1"
+version = "1.3.2"
 # The full version, including alpha/beta/rc tags
 release = version
 

ocdskit/schema.py

@@ -112,15 +112,27 @@ def get_schema_fields(
 
     if items := schema.get('items'):
         # `items` advances the pointer and sets array context (for the next level only).
-        yield from get_schema_fields(
-            items,
-            f'{pointer}/items',
-            path_components,
-            definition,
-            deprecated,
-            whole_list_merge=whole_list_merge,
-            array=True,
-        )
+        if isinstance(items, dict):
+            yield from get_schema_fields(
+                items,
+                f'{pointer}/items',
+                path_components,
+                definition,
+                deprecated,
+                whole_list_merge=whole_list_merge,
+                array=True,
+            )
+        else:
+            for i, subschema in enumerate(items):
+                yield from get_schema_fields(
+                    subschema,
+                    f'{pointer}/items/{i}',
+                    path_components,
+                    definition,
+                    deprecated,
+                    whole_list_merge=whole_list_merge,
+                    array=True,
+                )
 
     for keyword in ('anyOf', 'allOf', 'oneOf'):
         if elements := schema.get(keyword):
@@ -152,7 +164,7 @@ def get_schema_fields(
             prop_pointer = f'{pointer}/properties/{name}'
             prop_path_components = (*path_components, name)
             prop_deprecated = _deprecated(subschema)
-            prop_items = subschema.get('items', {})
+            prop_codelist, prop_open_codelist = _codelist(subschema)
 
             # To date, codelist and openCodelist in OCDS aren't set on `items`.
             yield Field(
@@ -163,8 +175,8 @@ def get_schema_fields(
                 definition=definition,
                 deprecated_self=prop_deprecated,
                 deprecated=deprecated or prop_deprecated,
-                codelist=subschema.get('codelist') or prop_items.get('codelist', ''),
-                open_codelist=subschema.get('openCodelist') or prop_items.get('openCodelist', False),
+                codelist=prop_codelist,
+                open_codelist=prop_open_codelist,
                 multilingual=name in multilingual,
                 required=name in required,
                 merge_by_id=name == 'id' and array and not whole_list_merge,
@@ -186,7 +198,7 @@ def get_schema_fields(
         prop_pointer = f'{pointer}/patternProperties/{name}'
         prop_path_components = (*path_components, name)
         prop_deprecated = _deprecated(subschema)
-        prop_items = subschema.get('items', {})
+        prop_codelist, prop_open_codelist = _codelist(subschema)
 
         yield Field(
             name=name,
@@ -196,8 +208,8 @@ def get_schema_fields(
             definition=definition,
             deprecated_self=prop_deprecated,
             deprecated=deprecated or prop_deprecated,
-            codelist=subschema.get('codelist') or prop_items.get('codelist', ''),
-            open_codelist=subschema.get('openCodelist') or prop_items.get('openCodelist', False),
+            codelist=prop_codelist,
+            open_codelist=prop_open_codelist,
             pattern=True,
             # `patternProperties` can't be multilingual, required, or "id".
         )
@@ -222,6 +234,15 @@ def get_schema_fields(
                     yield from get_schema_fields(subschema, f'/{keyword}/{name}', definition=name)
 
 
+def _codelist(subschema):
+    if codelist := subschema.get('codelist'):
+        return codelist, subschema.get('openCodelist', False)
+    # The behavior hasn't been decided if `items` is an array (e.g. with conflicting codelist-related values).
+    if (items := subschema.get('items')) and isinstance(items, dict):
+        return items.get('codelist', ''), items.get('openCodelist', False)
+    return '', False
+
+
 def _deprecated(value):
     return value.get('deprecated') or (hasattr(value, '__reference__') and value.__reference__.get('deprecated')) or {}
 

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "ocdskit"
-version = "1.3.1"
+version = "1.3.2"
 authors = [{name = "Open Contracting Partnership", email = "data@open-contracting.org"}]
 description = "A suite of command-line tools for working with OCDS data"
 readme = "README.rst"

tests/__init__.py

@@ -11,8 +11,8 @@
 import ocdskit.util
 
 
-def path(filename):
-    return os.path.join('tests', 'fixtures', filename)
+def path(*args):
+    return os.path.join('tests', 'fixtures', *args)
 
 
 def read(filename, mode='rt', **kwargs):

tests/commands/test_mapping_sheet.py

@@ -103,13 +103,13 @@ def test_command_oc4ids(capsys, monkeypatch):
 
 def test_command_bods(capsys, monkeypatch):
     assert_command(capsys, monkeypatch, main,
-                   ['mapping-sheet', '--order-by', 'path', path('bods/person-statement.json')],
+                   ['mapping-sheet', '--order-by', 'path', path('bods', 'person-statement.json')],
                    'mapping-sheet_bods.csv')
 
 
 def test_command_sedl(capsys, monkeypatch):
     assert_command(capsys, monkeypatch, main,
-                   ['mapping-sheet', path('sedl-schema.json')],
+                   ['mapping-sheet', path('sedl', 'schema-alpha.json')],
                    'mapping-sheet_sedl.csv')
 
 

tests/fixtures/schema-items-array.json

@@ -0,0 +1,106 @@
+{
+  "properties": {
+    "json_schema_example_fields": {
+      "type": "object",
+      "properties": {
+        "minimum": {
+          "minimum": 2
+        },
+        "minimum2": {
+          "exclusiveMinimum": true,
+          "minimum": 2
+        },
+        "maximum": {
+          "maximum": 2
+        },
+        "maximum2": {
+          "exclusiveMaximum": true,
+          "maximum": 2
+        },
+        "minItems": {
+          "type": "array",
+          "minItems": 2
+        },
+        "maxItems": {
+          "type": "array",
+          "maxItems": 2
+        },
+        "minLength": {
+          "type": "string",
+          "minLength": 2
+        },
+        "maxLength": {
+          "type": "string",
+          "maxLength": 2
+        },
+        "maxProperties": {
+          "type": "object",
+          "maxProperties": 2
+        },
+        "multipleOf": {
+          "type": "number",
+          "multipleOf": 3
+        },
+        "not": {
+          "not": {
+            "type": "string"
+          }
+        },
+        "anyOf": {
+          "anyOf": [
+            {"type": "array"},
+            {"type": "object"}
+          ]
+        },
+        "allOf": {
+          "anyOf": [
+            {"type": "array"},
+            {"type": "object"}
+          ]
+        },
+        "oneOf": {
+          "oneOf": [
+            {"type": "array"},
+            {"type": "object"}
+          ]
+        },
+        "oneOf2": {
+          "oneOf": [
+            {"type": "number"},
+            {"type": "integer"}
+          ]
+        },
+        "additionalItems": {
+          "type": "array",
+          "items": [{
+            "type": "string"
+          }],
+          "additionalItems": false
+        },
+        "additionalProperties": {
+          "type": "object",
+          "additionalProperties": false
+        },
+        "additionalProperties2": {
+          "type": "object",
+          "patternProperties": {
+            "okay": {
+              "type": "string"
+            }
+          },
+          "additionalProperties": false
+        },
+        "dependencies": {
+          "type": "object",
+          "dependencies": {
+            "b": ["a"]
+          }
+        },
+        "format": {
+          "type": "string",
+          "format": "email"
+        }
+      }
+    }
+  }
+}

tests/fixtures/sedl-schema.json renamed to tests/fixtures/sedl/schema-alpha.json

@@ -5,6 +5,35 @@
 from tests import load
 
 
+def test_items_array():
+    schema = load("schema-items-array.json")
+
+    assert {field.path_components for field in get_schema_fields(schema)} == {
+        ("json_schema_example_fields", "additionalItems"),
+        ("json_schema_example_fields", "additionalProperties"),
+        ("json_schema_example_fields", "additionalProperties2"),
+        ("json_schema_example_fields", "additionalProperties2", "okay"),
+        ("json_schema_example_fields", "allOf"),
+        ("json_schema_example_fields", "anyOf"),
+        ("json_schema_example_fields", "dependencies"),
+        ("json_schema_example_fields", "format"),
+        ("json_schema_example_fields", "maximum"),
+        ("json_schema_example_fields", "maximum2"),
+        ("json_schema_example_fields", "maxItems"),
+        ("json_schema_example_fields", "maxLength"),
+        ("json_schema_example_fields", "maxProperties"),
+        ("json_schema_example_fields", "minimum"),
+        ("json_schema_example_fields", "minimum2"),
+        ("json_schema_example_fields", "minItems"),
+        ("json_schema_example_fields", "minLength"),
+        ("json_schema_example_fields", "multipleOf"),
+        ("json_schema_example_fields", "not"),
+        ("json_schema_example_fields", "oneOf"),
+        ("json_schema_example_fields", "oneOf2"),
+        ("json_schema_example_fields",),
+    }
+
+
 @pytest.mark.parametrize(
     ("path", "expected"),
     [