feat: ✨ implement rule logic

pyproject.toml

@@ -33,6 +33,7 @@ keywords = [
 requires-python = ">=3.12"
 dependencies = [
     "jsonschema>=4.25.1",
+    "python-jsonpath>=2.0.1",
     "types-jsonschema>=4.25.1.20250822",
 ]
 

src/check_datapackage/check.py

@@ -10,6 +10,7 @@
 )
 from check_datapackage.issue import Issue
 from check_datapackage.read_json import read_json
+from check_datapackage.rule import apply_rules
 
 
 def check(
@@ -38,5 +39,7 @@ class for more details, especially about the default values.
         _add_resource_recommendations(schema)
 
     issues = _check_object_against_json_schema(descriptor, schema)
+    issues += apply_rules(config.rules, descriptor)
     issues = exclude(issues, config.exclude)
+
     return sorted(set(issues))

src/check_datapackage/internals.py

@@ -1,6 +1,9 @@
 import re
+from dataclasses import dataclass
+from itertools import chain
 from typing import Any, Callable, Iterable, Iterator, TypeVar
 
+from jsonpath import JSONPathMatch, finditer
 from jsonschema import Draft7Validator, FormatChecker, ValidationError
 
 from check_datapackage.constants import (
@@ -135,6 +138,34 @@ def _get_full_json_path_from_error(error: ValidationError) -> str:
     return error.json_path
 
 
+@dataclass
+class DescriptorField:
+    """A field in the Data Package descriptor.
+
+    Attributes:
+        jsonpath (str): The direct JSON path to the field.
+        value (str): The value contained in the field.
+    """
+
+    jsonpath: str
+    value: Any
+
+
+def _get_fields_at_jsonpath(
+    jsonpath: str, descriptor: dict[str, Any]
+) -> list[DescriptorField]:
+    """Returns all fields that match the JSON path."""
+    matches = finditer(jsonpath, descriptor)
+    return _map(matches, _create_descriptor_field)
+
+
+def _create_descriptor_field(match: JSONPathMatch) -> DescriptorField:
+    return DescriptorField(
+        jsonpath=match.path.replace("['", ".").replace("']", ""),
+        value=match.obj,
+    )
+
+
 In = TypeVar("In")
 Out = TypeVar("Out")
 
@@ -145,3 +176,8 @@ def _map(x: Iterable[In], fn: Callable[[In], Out]) -> list[Out]:
 
 def _filter(x: Iterable[In], fn: Callable[[In], bool]) -> list[In]:
     return list(filter(fn, x))
+
+
+def _flat_map(items: Iterable[In], fn: Callable[[In], Iterable[Out]]) -> list[Out]:
+    """Maps and flattens the items by one level."""
+    return list(chain.from_iterable(map(fn, items)))

src/check_datapackage/rule.py

@@ -1,6 +1,14 @@
 from dataclasses import dataclass
 from typing import Any, Callable
 
+from check_datapackage.internals import (
+    _filter,
+    _flat_map,
+    _get_fields_at_jsonpath,
+    _map,
+)
+from check_datapackage.issue import Issue
+
 
 @dataclass
 class Rule:
@@ -34,3 +42,39 @@ class Rule:
     message: str
     check: Callable[[Any], bool]
     type: str = "custom"
+
+
+def apply_rules(rules: list[Rule], descriptor: dict[str, Any]) -> list[Issue]:
+    """Checks the descriptor for all rules and creates issues for fields that fail.
+
+    Args:
+        rules: The rules to apply to the descriptor.
+        descriptor: The descriptor to check.
+
+    Returns:
+        A list of `Issue`s.
+    """
+    return _flat_map(
+        rules,
+        lambda rule: _apply_rule(rule, descriptor),
+    )
+
+
+def _apply_rule(rule: Rule, descriptor: dict[str, Any]) -> list[Issue]:
+    """Checks the descriptor against the rule and creates issues for fields that fail.
+
+    Args:
+        rule: The rule to apply to the descriptor.
+        descriptor: The descriptor to check.
+
+    Returns:
+        A list of `Issue`s.
+    """
+    matching_fields = _get_fields_at_jsonpath(rule.jsonpath, descriptor)
+    failed_fields = _filter(matching_fields, lambda field: not rule.check(field.value))
+    return _map(
+        failed_fields,
+        lambda field: Issue(
+            jsonpath=field.jsonpath, type=rule.type, message=rule.message
+        ),
+    )

tests/test_check.py

@@ -2,6 +2,9 @@
 
 from check_datapackage.check import check
 from check_datapackage.config import Config
+from check_datapackage.examples import example_package_descriptor
+from check_datapackage.exclude import Exclude
+from tests.test_rule import lowercase_rule
 
 # Without recommendations
 
@@ -168,3 +171,27 @@ def test_fails_descriptor_violating_recommendations():
         "$.sources[0].title",
         "$.resources[0].name",
     }
+
+
+def test_exclude_not_excluding_rule():
+    descriptor = example_package_descriptor()
+    descriptor["name"] = "ALLCAPS"
+    del descriptor["resources"]
+    exclude_required = Exclude(type="required")
+    config = Config(rules=[lowercase_rule], exclude=[exclude_required])
+
+    issues = check(descriptor, config=config)
+
+    assert len(issues) == 1
+    assert issues[0].type == "lowercase"
+
+
+def test_exclude_excluding_rule():
+    descriptor = example_package_descriptor()
+    descriptor["name"] = "ALLCAPS"
+    exclude_lowercase = Exclude(type=lowercase_rule.type)
+    config = Config(rules=[lowercase_rule], exclude=[exclude_lowercase])
+
+    issues = check(descriptor, config=config)
+
+    assert issues == []

tests/test_rule.py

@@ -0,0 +1,99 @@
+from check_datapackage.check import check
+from check_datapackage.config import Config
+from check_datapackage.examples import (
+    example_package_descriptor,
+    example_resource_descriptor,
+)
+from check_datapackage.issue import Issue
+from check_datapackage.rule import Rule
+
+lowercase_rule = Rule(
+    jsonpath="$.name",
+    message="Name must be lowercase.",
+    check=lambda name: name.islower(),
+    type="lowercase",
+)
+resource_name_rule = Rule(
+    jsonpath="$.resources[*].name",
+    message="Resource name must start with 'woolly'.",
+    check=lambda name: name.startswith("woolly"),
+    type="resource-name",
+)
+
+
+def test_direct_jsonpath():
+    descriptor = example_package_descriptor()
+    descriptor["name"] = "ALLCAPS"
+    config = Config(rules=[lowercase_rule])
+    issues = check(descriptor, config=config)
+
+    assert issues == [
+        Issue(
+            jsonpath=lowercase_rule.jsonpath,
+            type=lowercase_rule.type,
+            message=lowercase_rule.message,
+        )
+    ]
+
+
+def test_indirect_jsonpath():
+    descriptor = example_package_descriptor()
+    descriptor["resources"].append(example_resource_descriptor())
+    descriptor["resources"][1]["name"] = "not starting with woolly"
+
+    config = Config(rules=[resource_name_rule])
+    issues = check(descriptor, config=config)
+
+    assert issues == [
+        Issue(
+            jsonpath="$.resources[1].name",
+            type=resource_name_rule.type,
+            message=resource_name_rule.message,
+        ),
+    ]
+
+
+def test_multiple_rules():
+    descriptor = example_package_descriptor()
+    descriptor["name"] = "ALLCAPS"
+    descriptor["resources"][0]["name"] = "not starting with woolly"
+
+    config = Config(rules=[lowercase_rule, resource_name_rule])
+    issues = check(descriptor, config=config)
+
+    assert issues == [
+        Issue(
+            jsonpath=lowercase_rule.jsonpath,
+            type=lowercase_rule.type,
+            message=lowercase_rule.message,
+        ),
+        Issue(
+            jsonpath="$.resources[0].name",
+            type=resource_name_rule.type,
+            message=resource_name_rule.message,
+        ),
+    ]
+
+
+def test_rules_and_default_checks():
+    descriptor = example_package_descriptor()
+    descriptor["name"] = "ALLCAPS"
+    del descriptor["resources"]
+    config = Config(rules=[lowercase_rule])
+    issues = check(descriptor, config=config)
+
+    assert [issue.type for issue in issues] == ["lowercase", "required"]
+
+
+def test_no_matching_jsonpath():
+    descriptor = example_package_descriptor()
+    rule = Rule(
+        jsonpath="$.missing",
+        message="This check always fails.",
+        check=lambda value: False,
+        type="always-fail",
+    )
+    config = Config(rules=[rule])
+    issues = check(descriptor, config=config)
+
+    assert issues == []

tools/vulture-allowlist.py

@@ -1,4 +1,3 @@
 # ruff: noqa
 # mypy: ignore-errors
-rules  # unused variable (src/check_datapackage/config.py:38)
 version  # unused variable (src/check_datapackage/config.py:40)