"""Enhanced Beancount data structures for Conservancy

The classes in this module are interface-compatible with Beancount's core data
structures, and provide additional business logic that we want to use
throughout Conservancy tools.
"""
# Copyright © 2020  Brett Smith
# License: AGPLv3-or-later WITH Beancount-Plugin-Additional-Permission-1.0
#
# Full copyright and licensing details can be found at toplevel file
# LICENSE.txt in the repository.

import collections
import datetime
import decimal
import functools
import re

from beancount.core import account as bc_account
from beancount.core import amount as bc_amount
from beancount.core import convert as bc_convert
from beancount.core import data as bc_data
from beancount.core import position as bc_position
from beancount.parser import options as bc_options

from typing import (
    cast,
    overload,
    Callable,
    Hashable,
    Iterable,
    Iterator,
    MutableMapping,
    Optional,
    Pattern,
    Sequence,
    TypeVar,
    Union,
)

from .beancount_types import (
    Close,
    Currency,
    Directive,
    Meta,
    MetaKey,
    MetaValue,
    Open,
    OptionsMap,
    Posting as BasePosting,
    Transaction,
)

DecimalCompat = Union[decimal.Decimal, int]

EQUITY_ACCOUNTS = frozenset([
    'Equity',
    'Expenses',
    'Income',
])
FUND_ACCOUNTS = EQUITY_ACCOUNTS | frozenset([
    'Assets:Prepaid',
    'Assets:Receivable',
    'Liabilities:Payable',
    'Liabilities:UnearnedIncome',
])
LINK_METADATA = frozenset([
    'approval',
    'bank-statement',
    'check',
    'contract',
    'invoice',
    'purchase-order',
    'receipt',
    'rt-id',
    'statement',
    'tax-reporting',
    'tax-statement',
])

class AccountMeta(MutableMapping[MetaKey, MetaValue]):
    """Access account metadata

    This class provides a consistent interface to all the metadata provided by
    Beancount's ``open`` and ``close`` directives: open and close dates,
    used currencies, booking method, and the metadata associated with each.

    For convenience, you can use this class as a Mapping to access the ``open``
    directive's metadata directly.
    """
    __slots__ = ('_opening', '_closing')

    def __init__(self, opening: Open, closing: Optional[Close]=None) -> None:
        self._opening = opening
        self._closing: Optional[Close] = None
        if closing is not None:
            self.add_closing(closing)

    def add_closing(self, closing: Close) -> None:
        if self._closing is not None and self._closing is not closing:
            raise ValueError(f"{self.account} already closed by {self._closing!r}")
        elif closing.account != self.account:
            raise ValueError(f"cannot close {self.account} with {closing.account}")
        elif closing.date < self.open_date:
            raise ValueError(f"close date {closing.date} predates open date {self.open_date}")
        else:
            self._closing = closing

    def __iter__(self) -> Iterator[MetaKey]:
        return iter(self._opening.meta)

    def __len__(self) -> int:
        return len(self._opening.meta)

    def __getitem__(self, key: MetaKey) -> MetaValue:
        return self._opening.meta[key]

    def __setitem__(self, key: MetaKey, value: MetaValue) -> None:
        self._opening.meta[key] = value

    def __delitem__(self, key: MetaKey) -> None:
        del self._opening.meta[key]

    @property
    def account(self) -> 'Account':
        return Account(self._opening.account)

    @property
    def booking(self) -> Optional[bc_data.Booking]:
        return self._opening.booking

    @property
    def close_date(self) -> Optional[datetime.date]:
        return None if self._closing is None else self._closing.date

    @property
    def close_meta(self) -> Optional[Meta]:
        return None if self._closing is None else self._closing.meta

    @property
    def currencies(self) -> Optional[Sequence[Currency]]:
        return self._opening.currencies

    @property
    def open_date(self) -> datetime.date:
        return self._opening.date

    @property
    def open_meta(self) -> Meta:
        return self._opening.meta


class Account(str):
    """Account name string

    This is a string that names an account, like Assets:Bank:Checking
    or Income:Donations. This class provides additional methods for common
    account name parsing and queries.
    """
    __slots__ = ()

    ACCOUNT_RE: Pattern
    SEP = bc_account.sep
    _meta_map: MutableMapping[str, AccountMeta] = {}
    _options_map: OptionsMap

    @classmethod
    def load_options_map(cls, options_map: OptionsMap) -> None:
        cls._options_map = options_map
        roots: Sequence[str] = bc_options.get_account_types(options_map)
        cls.ACCOUNT_RE = re.compile(
            r'^(?:{})(?:{}[A-Z0-9][-A-Za-z0-9]*)+$'.format(
                '|'.join(roots), cls.SEP,
            ))

    @classmethod
    def load_opening(cls, opening: Open) -> None:
        cls._meta_map[opening.account] = AccountMeta(opening)

    @classmethod
    def load_closing(cls, closing: Close) -> None:
        try:
            cls._meta_map[closing.account].add_closing(closing)
        except KeyError:
            raise ValueError(
                f"tried to load {closing.account} close directive before open",
            ) from None

    @classmethod
    def load_openings_and_closings(cls, entries: Iterable[Directive]) -> None:
        for entry in entries:
            # type ignores because Beancount's directives aren't type-checkable.
            if isinstance(entry, bc_data.Open):
                cls.load_opening(entry)  # type:ignore[arg-type]
            elif isinstance(entry, bc_data.Close):
                cls.load_closing(entry)  # type:ignore[arg-type]

    @classmethod
    def load_from_books(cls, entries: Iterable[Directive], options_map: OptionsMap) -> None:
        cls.load_options_map(options_map)
        cls.load_openings_and_closings(entries)

    @classmethod
    def is_account(cls, s: str) -> bool:
        return cls.ACCOUNT_RE.fullmatch(s) is not None

    @classmethod
    def iter_accounts_by_classification(cls, s: str) -> Iterator['Account']:
        class_re = re.compile(f'^{re.escape(s)}(?::|$)')
        for name, meta in cls._meta_map.items():
            try:
                match = class_re.match(meta['classification'])
            except KeyError:
                match = None
            if match:
                yield cls(name)

    @classmethod
    def iter_accounts_by_hierarchy(cls, s: str) -> Iterator['Account']:
        for name in cls._meta_map:
            account = cls(name)
            if account.is_under(s):
                yield account

    @classmethod
    def iter_accounts(cls, s: Optional[str]=None) -> Iterator['Account']:
        """Iterate account objects by name or classification

        With no argument, returns an iterator of all known account names.
        If you pass in a root account name, or a valid account string, returns
        an iterator of all accounts under that account in the hierarchy.
        Otherwise, returns an iterator of all accounts with the given
        ``classification`` metadata.
        """
        if s is None:
            return (cls(acct) for acct in cls._meta_map)
        # We append a stub subaccount to match root accounts.
        elif cls.is_account(f'{s}:RootsOK'):
            return cls.iter_accounts_by_hierarchy(s)
        else:
            return cls.iter_accounts_by_classification(s)

    @property
    def meta(self) -> AccountMeta:
        return self._meta_map[self]

    def is_cash_equivalent(self) -> bool:
        return (
            self.is_under('Assets:') is not None
            and self.is_under('Assets:Prepaid', 'Assets:Receivable') is None
        )

    def is_checking(self) -> bool:
        return self.is_cash_equivalent() and ':Check' in self

    def is_credit_card(self) -> bool:
        return self.is_under('Liabilities:CreditCard') is not None

    def is_open_on_date(self, date: datetime.date) -> Optional[bool]:
        """Return true if this account is open on the given date.

        This method considers the dates on the account's open and close
        directives. If there is no close directive, it just checks the date is
        on or after the opening date. If neither exists, returns None.
        """
        try:
            meta = self.meta
        except KeyError:
            return None
        close_date = meta.close_date
        if close_date is None:
            close_date = date + datetime.timedelta(days=1)
        return meta.open_date <= date < close_date

    def is_opening_equity(self) -> bool:
        return self.is_under('Equity:Funds', 'Equity:OpeningBalance') is not None

    def is_under(self, *acct_seq: str) -> Optional[str]:
        """Return a match if this account is "under" a part of the hierarchy

        Pass in any number of account name strings as arguments. If this
        account is under one of those strings in the account hierarchy, the
        first matching string will be returned. Otherwise, None is returned.

        You can use the return value of this method as a boolean if you don't
        care which account string is matched.

        An account is considered to be under itself:

          Account('Expenses:Tax').is_under('Expenses:Tax') # returns 'Expenses:Tax'

        To do a "strictly under" search, end your search strings with colons:

          Account('Expenses:Tax').is_under('Expenses:Tax:') # returns None
          Account('Expenses:Tax').is_under('Expenses:') # returns 'Expenses:'

        This method does check that all the account boundaries match:

          Account('Expenses:Tax').is_under('Exp') # returns None
        """
        for prefix in acct_seq:
            if self.startswith(prefix) and (
                prefix.endswith(self.SEP)
                or self == prefix
                or self[len(prefix)] == self.SEP
            ):
                return prefix
        return None

    def keeps_balance(self) -> bool:
        return self.is_under(
            self._options_map['name_assets'],
            self._options_map['name_liabilities'],
        ) is not None

    def _find_part_slice(self, index: int) -> slice:
        if index < 0:
            raise ValueError(f"bad part index {index!r}")
        start = 0
        for _ in range(index):
            try:
                start = self.index(self.SEP, start) + 1
            except ValueError:
                raise IndexError("part index {index!r} out of range") from None
        try:
            stop = self.index(self.SEP, start + 1)
        except ValueError:
            stop = len(self)
        return slice(start, stop)

    def count_parts(self) -> int:
        return self.count(self.SEP) + 1

    @overload
    def slice_parts(self, start: None=None, stop: None=None) -> Sequence[str]: ...

    @overload
    def slice_parts(self, start: slice, stop: None=None) -> Sequence[str]: ...

    @overload
    def slice_parts(self, start: int, stop: int) -> Sequence[str]: ...

    @overload
    def slice_parts(self, start: int, stop: None=None) -> str: ...

    def slice_parts(self,
                    start: Optional[Union[int, slice]]=None,
                    stop: Optional[int]=None,
    ) -> Sequence[str]:
        """Slice the account parts like they were a list

        Given a single index, return that part of the account name as a string.
        Otherwise, return a list of part names sliced according to the arguments.
        """
        if start is None:
            part_slice = slice(None)
        elif isinstance(start, slice):
            part_slice = start
        elif stop is None:
            return self[self._find_part_slice(start)]
        else:
            part_slice = slice(start, stop)
        return self.split(self.SEP)[part_slice]

    def root_part(self, count: int=1) -> str:
        """Return the first part(s) of the account name as a string"""
        try:
            stop = self._find_part_slice(count - 1).stop
        except IndexError:
            return self
        else:
            return self[:stop]
Account.load_options_map(bc_options.OPTIONS_DEFAULTS)


class Amount(bc_amount.Amount):
    """Beancount amount after processing

    Beancount's native Amount class declares number to be Optional[Decimal],
    because the number is None when Beancount first parses a posting that does
    not have an amount, because the user wants it to be automatically balanced.

    As part of the loading process, Beancount replaces those None numbers
    with the calculated amount, so it will always be a Decimal. This class
    overrides the type declaration accordingly, so the type checker knows
    that our code doesn't have to consider the possibility that number is
    None.
    """
    number: decimal.Decimal

    # beancount.core._Amount is the plain namedtuple.
    # beancore.core.Amount adds instance methods to it.
    # b.c.Amount.__New__ calls `b.c._Amount.__new__`, which confuses type
    # checking. See <https://github.com/python/mypy/issues/1279>.
    # It works fine if you use super(), which is better practice anyway.
    # So we override __new__ just to call _Amount.__new__ this way.
    def __new__(cls, number: decimal.Decimal, currency: str) -> 'Amount':
        return super(bc_amount.Amount, Amount).__new__(cls, number, currency)


class Metadata(MutableMapping[MetaKey, MetaValue]):
    """Transaction or posting metadata

    This class wraps a Beancount metadata dictionary with additional methods
    for common parsing and query tasks.
    """
    __slots__ = ('meta',)
    _HUMAN_NAMES: MutableMapping[MetaKey, str] = {
        # Initialize this dict with special cases.
        # We use it as a cache for other metadata names as they're queried.
        'check-id': 'Check Number',
        'paypal-id': 'PayPal ID',
        'rt-id': 'Ticket',
    }
    _HUMAN_TRANSLATIONS = str.maketrans('-_', '  ')

    def __init__(self, source: MutableMapping[MetaKey, MetaValue]) -> None:
        self.meta = source

    def __iter__(self) -> Iterator[MetaKey]:
        return iter(self.meta)

    def __len__(self) -> int:
        return len(self.meta)

    def __getitem__(self, key: MetaKey) -> MetaValue:
        return self.meta[key]

    def __setitem__(self, key: MetaKey, value: MetaValue) -> None:
        self.meta[key] = value

    def __delitem__(self, key: MetaKey) -> None:
        del self.meta[key]

    def get_links(self, key: MetaKey) -> Sequence[str]:
        try:
            value = self.meta[key]
        except KeyError:
            return ()
        if isinstance(value, str):
            return value.split()
        else:
            raise TypeError("{} metadata is a {}, not str".format(
                key, type(value).__name__,
            ))

    def report_links(self, key: MetaKey) -> Sequence[str]:
        """Return a sequence of link strings under the named metadata key

        get_links raises a TypeError if the metadata is not a string.
        This method simply returns the empty sequence.
        Validation code (like in the plugin) usually uses get_links()
        while reporting code uses report_links().
        """
        try:
            return self.get_links(key)
        except TypeError:
            return ()

    @overload
    def first_link(self, key: MetaKey, default: None=None) -> Optional[str]: ...

    @overload
    def first_link(self, key: MetaKey, default: str) -> str: ...

    def first_link(self, key: MetaKey, default: Optional[str]=None) -> Optional[str]:
        try:
            return self.get_links(key)[0]
        except (IndexError, TypeError):
            return default

    @classmethod
    def human_name(cls, key: MetaKey) -> str:
        """Return the "human" version of a metadata name

        This is usually the metadata key with punctuation replaced with spaces,
        and then titlecased, with a few special cases. The return value is
        suitable for using in reports.
        """
        try:
            retval = cls._HUMAN_NAMES[key]
        except KeyError:
            retval = key.translate(cls._HUMAN_TRANSLATIONS).title()
            retval = re.sub(r'\bId$', 'ID', retval)
            cls._HUMAN_NAMES[key] = retval
        return retval


class PostingMeta(Metadata):
    """Combined access to posting metadata with its parent transaction metadata

    This lets you access posting metadata through a single dict-like object.
    If you try to look up metadata that doesn't exist on the posting, it will
    look for the value in the parent transaction metadata instead.

    You can set and delete metadata as well. Changes only affect the metadata
    of the posting, never the transaction. Changes are propagated to the
    underlying Beancount data structures.

    Functionally, you can think of this as identical to:

      collections.ChainMap(post.meta, txn.meta)

    Under the hood, this class does a little extra work to avoid creating
    posting metadata if it doesn't have to.
    """
    __slots__ = ('txn', 'index', 'post')

    def __init__(self,
                 txn: Transaction,
                 index: int,
                 post: Optional[BasePosting]=None,
    ) -> None:
        if post is None:
            post = txn.postings[index]
        self.txn = txn
        self.index = index
        self.post = post
        self.meta: collections.ChainMap = collections.ChainMap(txn.meta)
        if post.meta is not None:
            self.meta = self.meta.new_child(post.meta)

    def __getitem__(self, key: MetaKey) -> MetaValue:
        try:
            return super().__getitem__(key)
        except KeyError:
            if key == 'entity' and self.txn.payee is not None:
                return self.txn.payee
            else:
                raise

    def __setitem__(self, key: MetaKey, value: MetaValue) -> None:
        if len(self.meta.maps) == 1:
            self.post = self.post._replace(meta={key: value})
            assert self.post.meta is not None
            self.txn.postings[self.index] = self.post
            self.meta = self.meta.new_child(self.post.meta)
        else:
            super().__setitem__(key, value)

    def __delitem__(self, key: MetaKey) -> None:
        if len(self.meta.maps) == 1:
            raise KeyError(key)
        else:
            super().__delitem__(key)

    # This is arguably cheating a litttle bit, but I'd argue the date of
    # the parent transaction still qualifies as posting metadata, and
    # it's something we want to access so often it's good to have it
    # within easy reach.
    @property
    def date(self) -> datetime.date:
        return self.txn.date

    def detached(self) -> 'PostingMeta':
        """Create a copy of this PostingMeta detached from the original post

        Changes you make to the detached copy will not propagate to the
        underlying data structures. This is mostly useful for reporting code
        that may want to "split" and manipulate the metadata multiple times.
        """
        retval = type(self)(self.txn, self.index, self.post)
        retval.meta = self.meta.new_child()
        return retval


class Posting(BasePosting):
    """Enhanced Posting objects

    This class is a subclass of Beancount's native Posting class where
    specific fields are replaced with enhanced versions:

    * The `account` field is an Account object
    * The `units` field is our Amount object (which simply declares that the
      number is always a Decimal—see that docstring for details)
    * The `meta` field is a PostingMeta object
    """
    __slots__ = ()

    account: Account
    units: Amount
    cost: Optional[bc_position.Cost]
    # mypy correctly complains that our MutableMapping is not compatible
    # with Beancount's meta type declaration of Optional[Dict]. IMO
    # Beancount's type declaration is a smidge too specific: I think its type
    # declaration should also use MutableMapping, because it would be very
    # unusual for code to specifically require a Dict over that.
    # If it did, this declaration would pass without issue.
    meta: PostingMeta  # type:ignore[assignment]

    @classmethod
    def from_beancount(cls,
                       txn: Transaction,
                       index: int,
                       post: Optional[BasePosting]=None,
    ) -> 'Posting':
        if post is None:
            post = txn.postings[index]
        return cls(
            Account(post.account),
            *post[1:5],
            # see rationale above about Posting.meta
            PostingMeta(txn, index, post), # type:ignore[arg-type]
        )

    @classmethod
    def from_txn(cls, txn: Transaction) -> Iterator['Posting']:
        """Yield an enhanced Posting object for every posting in the transaction"""
        for index, post in enumerate(txn.postings):
            yield cls.from_beancount(txn, index, post)

    @classmethod
    def from_entries(cls, entries: Iterable[Directive]) -> Iterator['Posting']:
        """Yield an enhanced Posting object for every posting in these entries"""
        for entry in entries:
            # Because Beancount's own Transaction class isn't type-checkable,
            # we can't statically check this. Might as well rely on duck
            # typing while we're at it: just try to yield postings from
            # everything, and ignore entries that lack a postings attribute.
            try:
                yield from cls.from_txn(entry)  # type:ignore[arg-type]
            except AttributeError:
                pass

    def at_cost(self) -> Amount:
        if self.cost is None:
            return self.units
        else:
            return Amount(self.units.number * self.cost.number, self.cost.currency)


_KT = TypeVar('_KT', bound=Hashable)
_VT = TypeVar('_VT')
class _SizedDict(collections.OrderedDict, MutableMapping[_KT, _VT]):
    def __init__(self, maxsize: int=128) -> None:
        self.maxsize = maxsize
        super().__init__()

    def __setitem__(self, key: _KT, value: _VT) -> None:
        super().__setitem__(key, value)
        for _ in range(self.maxsize, len(self)):
            self.popitem(last=False)


def balance_of(txn: Transaction,
               *preds: Callable[[Account], Optional[bool]],
) -> Amount:
    """Return the balance of specified postings in a transaction.

    Given a transaction and a series of account predicates, balance_of
    returns the balance of the amounts of all postings with accounts that
    match any of the predicates.

    balance_of uses the "weight" of each posting, so the return value will
    use the currency of the postings' cost when available.
    """
    match_posts = [post for post in Posting.from_txn(txn)
                   if any(pred(post.account) for pred in preds)]
    number = decimal.Decimal(0)
    if not match_posts:
        currency = ''
    else:
        weights: Sequence[Amount] = [
            bc_convert.get_weight(post) for post in match_posts
        ]
        number = sum((wt.number for wt in weights), number)
        currency = weights[0].currency
    return Amount(number, currency)

_opening_balance_cache: MutableMapping[str, bool] = _SizedDict()
def is_opening_balance_txn(txn: Transaction) -> bool:
    key = '\0'.join(
        f'{post.account}={post.units}' for post in txn.postings
    )
    try:
        return _opening_balance_cache[key]
    except KeyError:
        pass
    opening_equity = balance_of(txn, Account.is_opening_equity)
    if not opening_equity.currency:
        retval = False
    else:
        rest = balance_of(txn, lambda acct: not acct.is_under(*EQUITY_ACCOUNTS))
        retval = (
            opening_equity.currency == rest.currency
            and abs(opening_equity.number + rest.number) < decimal.Decimal('.01')
        )
    _opening_balance_cache[key] = retval
    return retval