Add Schema definition

pyproject.toml

@@ -27,6 +27,7 @@ boto3 = "^1.24.86"
 requests = "^2.28.1"
 google-cloud-compute = "^1.6.0"
 click = "^8.1.3"
+jsonschema = "^4.16.0"
 
 [tool.poetry.group.dev.dependencies]
 pytest = "^7.1.3"

src/rhelocator/update_images/schema.py

@@ -0,0 +1,107 @@
+from __future__ import annotations
+
+from jsonschema import Draft202012Validator
+from jsonschema import SchemaError
+from jsonschema import ValidationError
+
+
+SCHEMA = {
+    "$schema": "https://json-schema.org/draft/2020-12/schema#",
+    "definitions": {
+        "defaults": {
+            "type": "object",
+            "properties": {
+                "name": {"type": "string", "description": "Human readable image name"},
+                "version": {
+                    "type": "string",
+                    "description": "RHEL image version following MAJOR.MINOR.PATCH",
+                },
+                "imageId": {
+                    "type": "string",
+                    "description": "Image ID used for running the image",
+                },
+                "date": {
+                    "type": "string",
+                    "description": "Date of image release in international date format",
+                },
+                "virt": {"type": "string", "description": "Hypervisor Type"},
+            },
+        },
+        "aws": {
+            "type": "object",
+            "description": "AWS RHEL Image Data",
+            "$ref": "#/definitions/defaults",
+            "properties": {
+                "selflink": {
+                    "description": "URL for direct access from cloud console",
+                    "format": "uri",
+                    "pattern": r"^https://console\.aws\.amazon\.com/ec2/home\?region=[a-z0-9-]*#launchAmi=ami-[a-z0-9]*",  # noqa: E501
+                },
+                "region": {
+                    "type": "string",
+                    "description": "Region of image availability",
+                },
+            },
+            "required": [
+                "name",
+                "version",
+                "imageId",
+                "date",
+                "virt",
+                "selflink",
+                "region",
+            ],
+        },
+        "azure": {
+            "type": "object",
+            "description": "Azure RHEL Image Data",
+            "$ref": "#/definitions/defaults",
+            "required": ["name", "version", "imageId", "date", "virt"],
+        },
+        "google": {
+            "type": "object",
+            "description": "Google RHEL Image Data",
+            "$ref": "#/definitions/defaults",
+            "properties": {
+                "selflink": {
+                    "description": "URL for direct access from cloud console",
+                    "format": "uri",
+                    "pattern": r"^https://console.cloud.google.com/compute/imagesDetail/projects/rhel-cloud/global/images/[a-z0-9-]*",  # noqa: E501
+                }
+            },
+            "required": ["name", "version", "imageId", "date", "virt", "selflink"],
+        },
+    },
+    "additionalProperties": False,
+    "required": ["images"],
+    "description": "Prepared Image Date for cloud image locator",
+    "$ref": "#/definitions/defaults",
+    "properties": {
+        "images": {
+            "type": "object",
+            "additionalProperties": False,
+            "description": "Prepared Image Data",
+            "minProperties": 1,
+            "properties": {
+                "aws": {"type": "array", "items": {"$ref": "#/definitions/aws"}},
+                "azure": {"type": "array", "items": {"$ref": "#/definitions/azure"}},
+                "google": {"type": "array", "items": {"$ref": "#/definitions/google"}},
+            },
+        }
+    },
+}
+
+
+def validate_json(data: str) -> None:
+    try:
+        validator = Draft202012Validator(SCHEMA)
+        errors = sorted(validator.iter_errors(data), key=lambda e: e.path)
+
+        if errors:
+            error_message = "Error validating image data: "
+            for error in errors:
+                error_message += f"[{error.schema_path}]: {error.message}; "
+            raise ValidationError(message=error_message)
+
+    except (ValidationError, SchemaError) as e:
+        raise ValidationError(e.message)

tests/update_images/test_json_schema.py

@@ -0,0 +1,140 @@
+"""Test image updates from remote cloud APIs."""
+from __future__ import annotations
+
+import pytest
+
+from jsonschema import ValidationError
+
+from rhelocator.update_images import schema
+
+
+def test_valid_image_data_validation():
+    """Test image data validation with valid input."""
+    image_data = {
+        "images": {
+            "aws": [
+                {
+                    "name": "some name",
+                    "version": "some version",
+                    "imageId": "some id",
+                    "date": "some date",
+                    "virt": "some virt",
+                    "selflink": "https://console.aws.amazon.com/ec2/home?region=eu-north-1#launchAmi=ami-0e02b69290bcd7b69",  # noqa: E501
+                    "region": "some region",
+                }
+            ],
+            "azure": [
+                {
+                    "name": "some name",
+                    "version": "some version",
+                    "imageId": "some id",
+                    "date": "some date",
+                    "virt": "some virt",
+                }
+            ],
+            "google": [
+                {
+                    "name": "some name",
+                    "version": "some version",
+                    "imageId": "some id",
+                    "date": "some date",
+                    "virt": "some virt",
+                    "selflink": "https://console.cloud.google.com/compute/imagesDetail/projects/rhel-cloud/global/images/rhel-7-v20220719",  # noqa: E501
+                }
+            ],
+        }
+    }
+
+    try:
+        schema.validate_json(image_data)
+    except ValidationError as exc:
+        raise AssertionError(f"Validating proper image data raised an exception! {exc}")
+
+
+def test_incomplete_image_data_validation():
+    """Test image data validation with incomplete input."""
+    image_data = {
+        "images": {
+            "aws": [
+                {
+                    "name": "some name",
+                    "version": "some version",
+                    "imageId": "some id",
+                    "virt": "some virt",
+                    "selflink": "https://console.aws.amazon.com/ec2/home?region=eu-north-1#launchAmi=ami-0e02b69290bcd7b69",  # noqa: E501
+                    "region": "some region",
+                }
+            ]
+        }
+    }
+
+    with pytest.raises(ValidationError, match=r".*'date' is a required property.*"):
+        schema.validate_json(image_data)
+
+
+def test_invalid_image_data_validation():
+    """Test image data validation with invalid input type."""
+    image_data = {
+        "images": {
+            "aws": [
+                {
+                    "name": "some name",
+                    "version": False,
+                    "imageId": "some id",
+                    "virt": "some virt",
+                    "date": "some date",
+                    "selflink": "https://console.aws.amazon.com/ec2/home?region=eu-north-1#launchAmi=ami-0e02b69290bcd7b69",  # noqa: E501
+                    "region": "some region",
+                }
+            ]
+        }
+    }
+
+    with pytest.raises(ValidationError, match=r".*False is not of type 'string'*"):
+        schema.validate_json(image_data)
+
+
+def test_invalid_extra_image_data_validation():
+    """Test image data validation with invalid extrea input data."""
+    image_data = {
+        "images": {
+            "oracle": [
+                {
+                    "name": "some name",
+                    "version": "some version",
+                    "imageId": "some id",
+                    "virt": "some virt",
+                    "date": "some date",
+                    "selflink": "https://console.aws.amazon.com/ec2/home?region=eu-north-1#launchAmi=ami-0e02b69290bcd7b69",  # noqa: E501
+                    "region": "some region",
+                }
+            ]
+        }
+    }
+
+    with pytest.raises(
+        ValidationError, match=r".*Additional properties are not allowed*"
+    ):
+        schema.validate_json(image_data)
+
+
+def test_malicious_url():
+    """Test image data validation with malicious url injection."""
+    image_data = {
+        "images": {
+            "aws": [
+                {
+                    "name": "some name",
+                    "version": "some version",
+                    "imageId": "some id",
+                    "virt": "some virt",
+                    "date": "some date",
+                    "selflink": "http://some-evil-url.com/?https://console.aws.amazon.com/ec2/home?region=eu-north-1#launchAmi=ami-0e02b69290bcd7b69",  # noqa: E501
+                    "region": "some region",
+                }
+            ]
+        }
+    }
+
+    with pytest.raises(ValidationError, match=r".*does not match*"):
+        schema.validate_json(image_data)