"""Python APIs for STIX 2 Pattern Semantic Equivalence.

.. autosummary::
   :toctree: pattern

   compare
   transform

|
"""

from ... import pattern_visitor
from ...version import DEFAULT_VERSION
from .compare.observation import observation_expression_cmp
from .transform import ChainTransformer, SettleTransformer
from .transform.observation import (
    AbsorptionTransformer, DNFTransformer, FlattenTransformer,
    NormalizeComparisonExpressionsTransformer, OrderDedupeTransformer,
)

# Lazy-initialize
_pattern_normalizer = None


def _get_pattern_normalizer():
    """
    Get a normalization transformer for STIX patterns.

    Returns:
        The transformer
    """

    # The transformers are either stateless or contain no state which changes
    # with each use.  So we can setup the transformers once and keep reusing
    # them.
    global _pattern_normalizer

    if not _pattern_normalizer:
        normalize_comp_expr = \
            NormalizeComparisonExpressionsTransformer()

        obs_expr_flatten = FlattenTransformer()
        obs_expr_order = OrderDedupeTransformer()
        obs_expr_absorb = AbsorptionTransformer()
        obs_simplify = ChainTransformer(
            obs_expr_flatten, obs_expr_order, obs_expr_absorb,
        )
        obs_settle_simplify = SettleTransformer(obs_simplify)

        obs_dnf = DNFTransformer()

        _pattern_normalizer = ChainTransformer(
            normalize_comp_expr,
            obs_settle_simplify, obs_dnf, obs_settle_simplify,
        )

    return _pattern_normalizer


def equivalent_patterns(pattern1, pattern2, stix_version=DEFAULT_VERSION):
    """
    Determine whether two STIX patterns are semantically equivalent.

    Args:
        pattern1: The first STIX pattern
        pattern2: The second STIX pattern
        stix_version: The STIX version to use for pattern parsing, as a string
            ("2.0", "2.1", etc).  Defaults to library-wide default version.

    Returns:
        True if the patterns are semantically equivalent; False if not
    """
    patt_ast1 = pattern_visitor.create_pattern_object(
        pattern1, version=stix_version,
    )
    patt_ast2 = pattern_visitor.create_pattern_object(
        pattern2, version=stix_version,
    )

    pattern_normalizer = _get_pattern_normalizer()
    norm_patt1, _ = pattern_normalizer.transform(patt_ast1)
    norm_patt2, _ = pattern_normalizer.transform(patt_ast2)

    result = observation_expression_cmp(norm_patt1, norm_patt2)

    return result == 0


def find_equivalent_patterns(
    search_pattern, patterns, stix_version=DEFAULT_VERSION,
):
    """
    Find patterns from a sequence which are equivalent to a given pattern.
    This is more efficient than using equivalent_patterns() in a loop, because
    it doesn't re-normalize the search pattern over and over.  This works
    on an input iterable and is implemented as a generator of matches.  So you
    can "stream" patterns in and matching patterns will be streamed out.

    Args:
        search_pattern: A search pattern as a string
        patterns: An iterable over patterns as strings
        stix_version: The STIX version to use for pattern parsing, as a string
            ("2.0", "2.1", etc).  Defaults to library-wide default version.

    Returns:
        A generator iterator producing the semantically equivalent patterns
    """
    search_pattern_ast = pattern_visitor.create_pattern_object(
        search_pattern, version=stix_version,
    )

    pattern_normalizer = _get_pattern_normalizer()
    norm_search_pattern_ast, _ = pattern_normalizer.transform(
        search_pattern_ast,
    )

    for pattern in patterns:
        pattern_ast = pattern_visitor.create_pattern_object(
            pattern, version=stix_version,
        )
        norm_pattern_ast, _ = pattern_normalizer.transform(pattern_ast)

        result = observation_expression_cmp(
            norm_search_pattern_ast, norm_pattern_ast,
        )

        if result == 0:
            yield pattern