# This file is part of beets.
# Copyright 2016, Adrian Sampson.
#
# Permission is hereby granted, free of charge, to any person obtaining
# a copy of this software and associated documentation files (the
# "Software"), to deal in the Software without restriction, including
# without limitation the rights to use, copy, modify, merge, publish,
# distribute, sublicense, and/or sell copies of the Software, and to
# permit persons to whom the Software is furnished to do so, subject to
# the following conditions:
#
# The above copyright notice and this permission notice shall be
# included in all copies or substantial portions of the Software.

"""Parsing of strings into DBCore queries.
"""

import re
import itertools
from . import query

PARSE_QUERY_PART_REGEX = re.compile(
    # Non-capturing optional segment for the keyword.
    r'(-|\^)?'   # Negation prefixes.

    r'(?:'
    r'(\S+?)'    # The field key.
    r'(?<!\\):'  # Unescaped :
    r')?'

    r'(.*)',         # The term itself.

    re.I  # Case-insensitive.
)


def parse_query_part(part, query_classes={}, prefixes={},
                     default_class=query.SubstringQuery):
    """Parse a single *query part*, which is a chunk of a complete query
    string representing a single criterion.

    A query part is a string consisting of:
    - A *pattern*: the value to look for.
    - Optionally, a *field name* preceding the pattern, separated by a
      colon. So in `foo:bar`, `foo` is the field name and `bar` is the
      pattern.
    - Optionally, a *query prefix* just before the pattern (and after the
      optional colon) indicating the type of query that should be used. For
      example, in `~foo`, `~` might be a prefix. (The set of prefixes to
      look for is given in the `prefixes` parameter.)
    - Optionally, a negation indicator, `-` or `^`, at the very beginning.

    Both prefixes and the separating `:` character may be escaped with a
    backslash to avoid their normal meaning.

    The function returns a tuple consisting of:
    - The field name: a string or None if it's not present.
    - The pattern, a string.
    - The query class to use, which inherits from the base
      :class:`Query` type.
    - A negation flag, a bool.

    The three optional parameters determine which query class is used (i.e.,
    the third return value). They are:
    - `query_classes`, which maps field names to query classes. These
      are used when no explicit prefix is present.
    - `prefixes`, which maps prefix strings to query classes.
    - `default_class`, the fallback when neither the field nor a prefix
      indicates a query class.

    So the precedence for determining which query class to return is:
    prefix, followed by field, and finally the default.

    For example, assuming the `:` prefix is used for `RegexpQuery`:
    - `'stapler'` -> `(None, 'stapler', SubstringQuery, False)`
    - `'color:red'` -> `('color', 'red', SubstringQuery, False)`
    - `':^Quiet'` -> `(None, '^Quiet', RegexpQuery, False)`, because
      the `^` follows the `:`
    - `'color::b..e'` -> `('color', 'b..e', RegexpQuery, False)`
    - `'-color:red'` -> `('color', 'red', SubstringQuery, True)`
    """
    # Apply the regular expression and extract the components.
    part = part.strip()
    match = PARSE_QUERY_PART_REGEX.match(part)

    assert match  # Regex should always match
    negate = bool(match.group(1))
    key = match.group(2)
    term = match.group(3).replace('\\:', ':')

    # Check whether there's a prefix in the query and use the
    # corresponding query type.
    for pre, query_class in prefixes.items():
        if term.startswith(pre):
            return key, term[len(pre):], query_class, negate

    # No matching prefix, so use either the query class determined by
    # the field or the default as a fallback.
    query_class = query_classes.get(key, default_class)
    return key, term, query_class, negate


def construct_query_part(model_cls, prefixes, query_part):
    """Parse a *query part* string and return a :class:`Query` object.

    :param model_cls: The :class:`Model` class that this is a query for.
      This is used to determine the appropriate query types for the
      model's fields.
    :param prefixes: A map from prefix strings to :class:`Query` types.
    :param query_part: The string to parse.

    See the documentation for `parse_query_part` for more information on
    query part syntax.
    """
    # A shortcut for empty query parts.
    if not query_part:
        return query.TrueQuery()

    # Use `model_cls` to build up a map from field (or query) names to
    # `Query` classes.
    query_classes = {}
    for k, t in itertools.chain(model_cls._fields.items(),
                                model_cls._types.items()):
        query_classes[k] = t.query
    query_classes.update(model_cls._queries)  # Non-field queries.

    # Parse the string.
    key, pattern, query_class, negate = \
        parse_query_part(query_part, query_classes, prefixes)

    # If there's no key (field name) specified, this is a "match
    # anything" query.
    if key is None:
        if issubclass(query_class, query.FieldQuery):
            # The query type matches a specific field, but none was
            # specified. So we use a version of the query that matches
            # any field.
            out_query = query.AnyFieldQuery(pattern, model_cls._search_fields,
                                            query_class)
        else:
            # Non-field query type.
            out_query = query_class(pattern)

    # Field queries get constructed according to the name of the field
    # they are querying.
    elif issubclass(query_class, query.FieldQuery):
        key = key.lower()
        out_query = query_class(key.lower(), pattern, key in model_cls._fields)

    # Non-field (named) query.
    else:
        out_query = query_class(pattern)

    # Apply negation.
    if negate:
        return query.NotQuery(out_query)
    else:
        return out_query


def query_from_strings(query_cls, model_cls, prefixes, query_parts):
    """Creates a collection query of type `query_cls` from a list of
    strings in the format used by parse_query_part. `model_cls`
    determines how queries are constructed from strings.
    """
    subqueries = []
    for part in query_parts:
        subqueries.append(construct_query_part(model_cls, prefixes, part))
    if not subqueries:  # No terms in query.
        subqueries = [query.TrueQuery()]
    return query_cls(subqueries)


def construct_sort_part(model_cls, part, case_insensitive=True):
    """Create a `Sort` from a single string criterion.

    `model_cls` is the `Model` being queried. `part` is a single string
    ending in ``+`` or ``-`` indicating the sort. `case_insensitive`
    indicates whether or not the sort should be performed in a case
    sensitive manner.
    """
    assert part, "part must be a field name and + or -"
    field = part[:-1]
    assert field, "field is missing"
    direction = part[-1]
    assert direction in ('+', '-'), "part must end with + or -"
    is_ascending = direction == '+'

    if field in model_cls._sorts:
        sort = model_cls._sorts[field](model_cls, is_ascending,
                                       case_insensitive)
    elif field in model_cls._fields:
        sort = query.FixedFieldSort(field, is_ascending, case_insensitive)
    else:
        # Flexible or computed.
        sort = query.SlowFieldSort(field, is_ascending, case_insensitive)
    return sort


def sort_from_strings(model_cls, sort_parts, case_insensitive=True):
    """Create a `Sort` from a list of sort criteria (strings).
    """
    if not sort_parts:
        sort = query.NullSort()
    elif len(sort_parts) == 1:
        sort = construct_sort_part(model_cls, sort_parts[0], case_insensitive)
    else:
        sort = query.MultipleSort()
        for part in sort_parts:
            sort.add_sort(construct_sort_part(model_cls, part,
                                              case_insensitive))
    return sort


def parse_sorted_query(model_cls, parts, prefixes={},
                       case_insensitive=True):
    """Given a list of strings, create the `Query` and `Sort` that they
    represent.
    """
    # Separate query token and sort token.
    query_parts = []
    sort_parts = []

    # Split up query in to comma-separated subqueries, each representing
    # an AndQuery, which need to be joined together in one OrQuery
    subquery_parts = []
    for part in parts + [',']:
        if part.endswith(','):
            # Ensure we can catch "foo, bar" as well as "foo , bar"
            last_subquery_part = part[:-1]
            if last_subquery_part:
                subquery_parts.append(last_subquery_part)
            # Parse the subquery in to a single AndQuery
            # TODO: Avoid needlessly wrapping AndQueries containing 1 subquery?
            query_parts.append(query_from_strings(
                query.AndQuery, model_cls, prefixes, subquery_parts
            ))
            del subquery_parts[:]
        else:
            # Sort parts (1) end in + or -, (2) don't have a field, and
            # (3) consist of more than just the + or -.
            if part.endswith(('+', '-')) \
                    and ':' not in part \
                    and len(part) > 1:
                sort_parts.append(part)
            else:
                subquery_parts.append(part)

    # Avoid needlessly wrapping single statements in an OR
    q = query.OrQuery(query_parts) if len(query_parts) > 1 else query_parts[0]
    s = sort_from_strings(model_cls, sort_parts, case_insensitive)
    return q, s