Function load_rules

src/licensedcode/models.py:1218–1306

Return an iterable of rules loaded from rule files in ``rules_data_dir``. Optionally check for consistency if ``with_checks`` is True.

(
    rules_data_dir=rules_data_dir,
    with_checks=True,
    is_builtin=True,
    with_deprecated=False,
)

def load_rules(
    rules_data_dir=rules_data_dir,
    with_checks=True,
    is_builtin=True,
    with_deprecated=False,
):
    """
    Return an iterable of rules loaded from rule files in ``rules_data_dir``.
    Optionally check for consistency if ``with_checks`` is True.
    """
    # TODO: OPTIMIZE: create a graph of rules to account for containment and
    # similarity clusters?
    seen_files = set()
    processed_files = set()
    lower_case_files = set()
    case_problems = set()
    space_problems = []
    model_errors = []

    for rule_file in resource_iter(location=rules_data_dir, with_dirs=False):
        if rule_file.endswith('.RULE'):
            base_name = file_base_name(rule_file)

            if with_checks and ' ' in base_name:
                space_problems.append(rule_file)

            try:
                rule = Rule.from_file(rule_file=rule_file)
                if not with_deprecated and rule.is_deprecated:
                    continue
                else:
                    yield rule

            except Exception as re:
                if with_checks:
                    model_errors.append(str(re))

            if with_checks:
                # accumulate sets to ensures we do not have illegal names or extra
                # orphaned files
                rule_file_lower = rule_file.lower()
                if rule_file_lower in lower_case_files:
                    case_problems.add(rule_file_lower)
                else:
                    lower_case_files.add(rule_file_lower)

                processed_files.add(rule_file)

        if with_checks and not rule_file.endswith('~'):
            seen_files.add(rule_file)

    if with_checks:
        unknown_files = seen_files - processed_files
        if unknown_files or case_problems or model_errors or space_problems:
            msg = ''

            if model_errors:
                errors = '\n'.join(model_errors)